close

Вход

Забыли?

вход по аккаунту

?

Ferroelectric liquid crystal device based photonic controllers for microwave antenna arrays

код для вставкиСкачать
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
T he quality o f this reproduction is dependent upon the quality of the
copy subm itted.
Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted.
Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FERROELECTRIC LIQUID CRYSTAL DEVICE BASED PHOTONIC CONTROLLERS
FOR MICROWAVE ANTENNA ARRAYS
by
NICHOLAS MADAMOPOULOS
B.S. University of Patra, Greece, 1993
M.S. University of Central Florida, 1996
A dissertation submitted in partial fulfillment of the requirements
for the degree of Doctor of Philosophy
in the School of Optics
at the University of Central Florida
Orlando, Florida
Fall Term
1998
Major Professor: Nabeel A. Riza
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9910800
UMI Microform 9910800
Copyright 1999, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
© 1998 by Nicholas Madamopoulos
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
For the first time, this dissertation proposes, studies, analyzes, and experimentally
demonstrates the use of ferroelectric liquid crystal (FLC) technology for wideband phased
array control applications. Two-dimensional (2-D) FLC devices are used as 2-D
polarization switches in photonic delay lines (PDLs) to control and process optical signals
that drive the elements of a phased array antenna (PAA). This dissertation also studies, for
the first time, fiber optic delay lines based on polarization switching.
The use of photonics for PAA control is, at present, a vital area of applied research.
In particular, the potential impact of compact, low cost, photonically controlled PAAs for
large scale global applications, such as wireless communications, is a major business
growth area. The ultra-short optical wavelength (in microns), as well as the possibility of
using hair thin optical fibers can greatly decrease the size of any PAA control system,
especially with the help of postage-stamp size 2-D liquid crystal (LC) arrays.
This dissertation work concludes with the demonstration of a multichannel 7-bit
PDL system for a wideband PAA such as the Navy’s advanced Aegis radar system. This
multichannel 7-bit PDL system is the first system of its kind. The unique system issues and
problems to be examined and solved in this Ph.D. dissertation include the theoretical
analysis and experimental demonstration of different PDL architectures covering a sub­
nanosecond to several nanoseconds time delay range. This determines which of our
proposed new PDL architectures fits best the Navy system requirements. Note that a
combination of our PDL architectures can be utilized in any time delay system to cover
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
different system requirements. New noise reduction/suppression schemes are proposed,
studied and applied to give record level time delay system performance in terms of signalto-leakage noise ratio (SNR), and switching speeds (e.g., 35 (is) required for fast radar
scan. Issues related with the performance of the system depending on the different optical
modulation technique are investigated. We show that the external modulation FO link gives
more degrees of freedom to the system engineer. We also propose a novel synchronous RF
signal calibration time delay control technique to obtain optimum dynamic range
performance for our PDL. The use of low loss fibers for remoting of the photonic
beamformer, as well as the losses associated with multiple fiber interconnects that limit the
maximum number of array channels in the systems are studied. Different fiber optic
coupling techniques are investigated for enhanced fiber coupling. Multimode fibers are
used, for the first time, at the output plane of the PDL to obtain improved coupling
efficiency. We demonstrate a low -1.7 dB optical insertion loss per bit, which is very close
to the desired insertion loss required for the Navy system. A novel approach for hardware
reduction based on wavelength multiplexing is proposed, where the use of a combination
of wavelength dependent and wavelength independent optical paths provides the required
time delays. Finally, new switching fabric approaches are studied based on polarization
selective holograms (PSH) and their potential use for the implementation of PDLs is
discussed.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To Fotini
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
First, I would like to thank my research advisor Associate Professor Nabeel A. Riza for
trusting me and assigning me such an interesting and challenging project. I also thank him
for his technical insights, academic assistance and consistent encouragement throughout my
studies. I would like to recognize that under him I have developed a constructive scientific
way of thinking. I also thank Dr. Patrick LiKamWa, Dr. M. G. Moharam, and Dr. Robert
Peale for participating in my Ph.D. dissertation committee.
I would like to thank Dr. Christos Christodoulou who helped me when I first came to
Orlando, in August of 1994, and introduced me to Dr. M. G. Moharam, who then
introduced me to the CREOL community. Without their help I would not have reached
where I am today.
I would like to thank Dr. Jinkee Kim and Dr. Shifti Yuan for their help and numerous
discussions on the subject of fiber-optics and fiber-optic coupling. I also thank, my
colleague Sarun Samriddechkajom for sharing with me various tasks in the lab.
I would also like to thank Professor A. T. Georgas and Dr. Nikos A. Vainos, for
encouraging me to take up graduate studies when I was in Greece. Especially, Dr. N. A.
Vainos and his coworkers at the Applied Nonlinear Optics and Optical Processing Systems
Laboratory at the Foundation for Research and Technology-Hellas, helped me obtain
appropriate experience that helped me in my studies and my research.
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Special thanks from the deepest of my heart to Fotini Papageorgiou for being so patient
when I had to spend my days and nights in the laboratory, and for her continuous support
and encouragement throughout my four and a half years of my graduate studies.
I would also like to thank my family for their encouragement in the initial steps of my
studies.
I would also like to acknowledge partial support from the United States Office of Naval
Research (ONR), grant #N000149510988.
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
LIST OF TABLES
xvi
LIST OF FIGURES
xix
GLOSSARY
xli
CHAPTER I: INTRODUCTION
I
CHAPTER 2: PHASED ARRAY ANTENNAS
6
2.1.
Introduction
6
2.2.
The Phased Array Antenna
7
2.3.
Phased Array Antenna Control Techniques
8
2.3.1.
Phase-based Steering
2.3.2.
True Time Delay Steering
8
10
CHAPTER 3: PHOTONIC DELAY LINES
3.1.
14
Introduction
14
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2.
Prior Work in Photonic Delay Lines
3.3.
The Motivation for using Ferroelectric Liquid Crystal Polarization
15
Switching Arrays
18
3.4.
The Proposed Ferroelectric LC-based Photonic Delay Line
19
3.5.
Photonic Delay Line System Issues
21
3.5.1.
Signal-to-Leakage Noise Ratio
21
3.5.2.
Switching Speed
22
3.5.3.
Fiber Optic Links
23
3.5.4.
Phased Array Antenna Remote Control
24
3.5.5.
Fiber Optic Interconnects and Interchannel Isolation
25
3.5.6.
Insertion Loss
32
3.6.
Three Dimensional Photonic Delay Lines Based on Polarization
Switching Arrays
33
3.6.1.
Transmissive PDL Architectures
33
3.6.2.
The Symmetric PDL Architecture
40
3.6.3.
The Reflective Geometry PDL Architecture
43
3.7.
Active Noise Filter Experimental Demonstration
47
3.8.
Conclusion
49
CHAPTER 4: FIBER OPTICS-BASED PHOTONIC DELAY LINE
50
4.1.
Introduction
50
4.2.
The Fiber Optic Photonic Delay Line
52
4.3.
Non-PM Fiber based Photonic Delay Line Theory
53
4.4.
Birefringence Compensation Experimental Demonstration
59
be
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.5.
4.6.
Microwave Band Demonstration of a Reflective Geometry Fiber and
Free Space Photonic Delay Line
63
Conclusion
46
CHAPTER 5: CHARACTERIZATION OF A FERROELECTRIC LIQUID
CRYSTAL BASED TIME DELAY UNIT FOR PHASED ARRAY
ANTENNA APPLICATIONS
74
5.1.
Introduction
74
5.2.
The Ferroelectric Liquid Crystal Principle of Operation
76
5.3.
The Ferroelectric Liquid Crystal-based Photonic Delay Line
80
5.4.
The FLC-based Photonic delay Line Experiment and System Issues
82
5.4.1.
Optical Leakage Noise Reduction
82
5.4.2.
Signal-to-Leakage Noise Ratio and Polarization Extinction Ratio
84
5.4.3.
FLC Switching Speed
86
5.5.
Conclusion
89
CHAPTER 6: PHASED ARRAY ANTENNA MAXIMUM COMPRESSION
REVERSIBLE PHOTONIC BEAMFORMER USING TERNARY
DESIGNS AND MULTIPLE WAVELENGTHS
6.1.
Introduction
6.2.
Maximum Compression Reversible Photonic Beamformer System
91
Architecture
6.2.1.
91
93
Photonic Delay Line Ternary Designs
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
6.2.2.
Compressed Wavelength Multiplexed Reversible Photonic Control
Architecture for 1-D Antenna Beam Steering
6.3.
Compressed Reversible Photonic Control Architecture for 2-D Antenna
Beam Steering
6.4.
106
Photonic Control System Issues Related to the Phased Array
Application
6.5.
96
113
Conclusion
CHAPTER 7:
122
ADAPTABLE-DELAY BALANCED-LOSS BINARY
PHOTONIC DELAY LINE ARCHITECTURES USING
POLARIZATION SWITCHING
124
7.1.
Introduction
124
7.2.
The Adaptable Delay Reflective-Symmetric PDL Architecture
125
7.3.
Adaptable Delay PDL Time Delay Analysis
129
7.4.
Balanced Loss Performance of the Proposed Adaptive Delay PDL Theory
132
7.5.
Experimental Verification of the Balanced Loss PDL Architecture
137
7.6.
Hardware Compressed Versions of the Adaptable Balanced Loss PDL
Architecture
140
7.6.1.
Wavelength Multiplexing-based PDL
140
7.6.2.
Polarization Multiplexing based PDL
141
7.7.
Conclusion
143
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 8:
DIRECTLY MODULATED SEMICONDUCTOR LASER FED
PHOTONIC DELAY LINE USING FERROELECTRIC LIQUID
CRYSTALS
145
8.1.
Introduction
145
8.2.
3-Bit 3-D PDL using FLC Devices, Imaging, and Remoting
146
8.2.1.
Experimental Set-up
8.2.2.
The PDL Free-Space Feedback, Feed-Forward and Symmetric
Delay Architectures
148
The Ferroelectric Liquid Crystal based Polarization Switches
150
Ferroelectric Liquid Crystal PDL Demonstration and System Issues
152
8.2.3.
8.3.
146
8.3.1.
Insertion Loss
152
8.3.2.
Interchannel Crosstalk
154
8.3.3.
PDL RF Signal Measurements
156
8.3.4.
Time Delay Measurements and PDL Switching Speed
160
8.4.
Conclusion
CHAPTER 9:
163
SWITCHED PHOTONIC DELAY LINE PHASED ARRAY
ANTENNA CONTROL USING EXTERNALLY MODULATED
MICROWAVE FIBER-OPTIC LINK
164
9.1.
Introduction
164
9.2.
The External Modulation Fiber-Optic Link Fed Photonic Delay Line
165
9.3.
Photonic Delay Line Optical Loss Budget
168
9.4.
Improved Output Coupling Efficiency using Multi-mode Fibers
170
9.5.
The Leakage Noise Issue
173
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.6.
RF Power Measurements
176
9.7.
S-Band Operation of the Externally Modulated PDL
175
9.8.
Time Delay Measurements
179
9.9.
Interchannel Crosstalk
182
9.10. Conclusion
183
CHAPTER 10: SYNCHRONOUS AMPLITUDE AND TIME CONTROL FOR
AN OPTIMUM DYNAMIC RANGE VARIABLE PHOTONIC
DELAY LINE
10.1. Introduction
184
184
10.2. The Synchronous Signal Calibration and Dynamic Range Loss
Compensation Technique
10.3.
187
Experimental Demonstration of the Signal Calibration and Dynamic
Range Loss Compensation Technique
194
10.4.
Discussion of FO Link-PDL System Performance
204
10.5.
Conclusion
206
CHAPTER 11: THE FIRST DEMONSTRATION OF A 7-BIT 33-CHANNEL
PHOTONIC DELAY LINE FOR PHASED ARRAY RADAR
208
11.1.
Introduction
208
11.2.
Delay Line Requirements
210
11.3.
Multiple Channel Photonic Delay Line Design Issues
211
11.4.
The 7-Bit Photonic Delay Line System Design
223
11.5.
Time Delay Measurements
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.6. 7-bit Photonic Delay Line Insertion Loss Performance
228
11.7
229
Photonic Delay Line Switching Time Response
11.8. Wideband Operation of the 7-bit PDL system
231
11.9.
232
RF Leakage Noise Measurements for the 7-bit PDL
11.10. RF Interchannel Crosstalk for the 7-bit 16-channel PDL System
233
11.11. Conclusion
234
CHAPTER 12: ALL-FIBER CONNECTORIZED FIBER-OPTIC DELAY
MODULES USING 3-D POLARIZATION OPTICS
236
12.1.
Introduction
236
12.2.
Alternative Compact PDL Module Designs
237
12.3. The Bulk Optics Ferroelectric Liquid Crystal based Compact PDL
Module
239
12.4.
Spherical Microlens FO-ColIimator based Switched Compact PDL
240
12.5.
GRIN Lens FO-ColIimators based Switched Compact PDL
244
12.6
Ferroelectric Liquid Crystal Switching Speed
250
12.7
Alternative Compact PDL Module based on Microelectromechanical
12.8.
System Technology
251
Conclusion
252
CHAPTER 13: PHOTONIC DELAY LINES USING POLYMER DISPERSED
LIQUID CRYSTAL TECHNOLOGY
254
13.1.
Introduction
254
13.2.
Basic Characteristics of the Polymer Dispersed Liquid Crystal Devices
255
xiv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.3.
The electrically Switched PDLC-based Photonic Delay Line
258
13.4.
PDLC-based Photonic Delay Line Issues
261
13.4.1. Time Delay Measurements
261
13.4.2. Leakage Noise in the PDLC-based Photonic Delay Line
263
13.5.
Polarization Selective Hologram-based Photonic Delay Lines
266
13.6.
Experimental Demonstration of the PSH-based Photonic Delay Line
269
13.7.
Leakage Noise Measurements and System Improvements
271
13.7.1. Passive Leakage Noise Filter
272
13.7.2. Active Noise Filter
274
13.7.3. Combination of the Active and Passive Noise Filter
275
13.7.4. "Orthogonal Drive" PDLC Device Configuration
276
13.8.
Time Delay Measurements
278
13.9.
Insertion Loss of the PSH-based Photonic Delay Line
280
13.10. A Compact Photonic Delay Line Architecture based on Polarization
Selective Holograms
281
13.10.1. Experimental Set-up
281
13.10.2. Leakage Noise Measurements
282
13.10.3. Time Delay Measurements
283
13.11. Alternative Polarization Selective Hologram Design
284
13.12. Conclusion
288
LIST OF REFERENCES
289
xv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table 3.1: Electrical signal-to-leakage noise ratio and optical extinction ratio
measurements in dB for the delay-path and the straight-path of the optical delay line
with and without the noise reduction scheme.
Table 4.1: SNR and average SNR variation of the output power for the 1-bit fiber
delay line using the birefringence compensation technique.
Table 4.2: Measured transmission/reflection efficiencies of the optical components.
Table 4.3: Expected and measured optical loss for the four settings of the PDL.
Table 4.4: RF SNR for all the different settings of the PDL with and without the
noise reduction scheme.
Table 5.1: Electrical SNR and Polarization ER measurements for the delay path and
straight path of the PDL with and without the novel active noise filter.
Table 7.1: Time delay range for the different PDL architectures.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7.2: Optical and Electrical Signal-to-Noise Ratio for both settings of the
reflective and the new reflective-symmetric adaptable PDL architecture.
Table 8.1. Measured and designed optical losses for each bit and for the overall
PDL.
Table 8.2: RF analyzer-limited C/N ratio measurements for five different time delay
setting of the PDL. These analyzer noise floor limited measurements were obtained
over a 250 kHz analyzer bandwidth using an analyzer resolution bandwidth of 1
kHz.
Table 9.1: Expected and measured optical insertion loss for the binary settings of
each of the four PDL bits.
Table 9.2: Average PDL optical loss.
Table 9.3: Coupling loss for the SM-fiber and the MM-fiber FO collimator system.
Table 9.4: Electrical SNR values of the PDL using (a) GRIN lens pigtailed single­
mode fiber and (b) GRIN lens pigtailed multi-mode fiber coupling system.
Table 11.1: Delay line requirements.
Table 11.2: Switching time requirements for the delay line, after command signal
has been applied to the PDL connectors.
xvii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 11.3: The 7-bit PDL design characteristics.
224
Table 11.4: The desired and experimentally obtained time delays for the 7-bit PDL.
227
Table 11.5: Optical Insertion loss for each bit of the 7-bit PDL system.
228
Table 12.1: Expected and measured optical insertion loss.
243
Table 13.1: Electrical signal-to-leakage noise ratio of the proposed PDLC based
PDL.
264
Table 13.2: Optical SNR measurements for the two PDL settings (without noise
271
reduction).
Table 13.3: Optical SNR for all the different leakage noise filtering approaches. The
SNR without any noise filter is also shown for comparison.
274
Table 13.4: Optical SNR for the PDL with the orthogonal drive configuration and
for all the different leakage noise filter approaches. The SNR without any noise
filter is also shown for comparison.
278
Table 13.5: Optical SNR for the compact reflective PDL with and without the active
283
noise filter.
xviii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES
Figure 2.1: A typical phased array antenna with phase-based control. One
dimension is shown for simplicity.
9
Figure 2.2: A typical Af+1 -element phased array antenna with time delay based
control. One dimension is shown for simplicity.
11
Figure 2.3: A typical N-bit switched time delay line network that requires AM 2x2
switches to implement the 2N different time delay settings. Signal in each bit can
follow either the delay or non-delay path; a microwave phase shifter provides the
fine modulo-2it phase control.
12
Figure 3.1: A M-channel N -bit PDL network for the control of phased array
antennas. (QWP: Quarter wave plate; M: mirror; SLM: spatial light modulator; PBS:
polarizing cube beamsplitter).
20
Figure 3.2: The PDL schematic diagram. (LSB: Least significant bit; MSB: Most
significant bit).
20
xix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3.3: Gaussian beam transformation by a lens. (w,, w2 are the input and
output beam waists, respectively).
26
Figure 3.4: The 4 f imaging system consisting of two lenses of focal lengths / .
The two glass plates simulate additional optical components in the path. (ng : index
of refraction of the glass plate).
27
Figure 3.5: The PBS transmissive feed-forward PDL architecture.
34
Figure 3.6: Ray tracing diagram for total internal reflection from a prism of a pair of
beams focused by a lens.
36
Figure 3.7: PBS based transmissive PDL cascade architecture.
37
Figure 3.8: PBS based transmissive PDL “sandwiched” architecture.
37
Figure 3.9: The single PBS transmissive feed-back PDL architecture.
39
Figure 3.10: The TBS based transmissive PDL architecture.
40
Figure 3.11: The symmetric PDL architecture. The solid line represents the non­
delay path and the dashed line represents the delay path.
Figure 3.12: The proposed compact reflective geometry N-bit switched photonic
delay line for microwave signal processing. This structure uses free space (bit 1),
xx
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
solid optics (bit 2), and non-PM fiber (bit N) delay lines and a novel polarization
noise reduction scheme. The fiber delay design passively compensates for all
externally or internally induced fiber birefringence effects, thus maintaining a high
quality linear polarization at the optical switching and redirection planes of the
switched delay line.
Figure 3.13: The experimental setup for the PDL noise reduction scheme. The NLC
devices are driven by 0-5V, lKHz square wave signal. The “on” or “o ff’ operation
can be selected by setting the Vp drive level.
Figure 4.1: Top view of the proposed fiber birefringence-compensated N-bit Mchannel switched fiber PDL. (QWP: quarter wave plate; P: polarizer, SLM: spatial
light modulator).
Figure 4.2: The geometry of a general retardation plate, such as a non-PM fiber in
the delay path of the photonic delay line. Angle <j>is the angle between the fast axis
“f ’ of the retarder and the x-axis that is parallel to the horizontal or p-polarized light.
The y-axis is parallel to the vertical or ^-polarized light.
Figure 4.3: The birefringence compensation experimental setup. LC3 simulates a
variable birefringence by controling the driving voltage of the NLC device (LC3).
Figure 4.4: Optical Extinction Ratio and Electrical Signal-to-Leakage Noise Ratio
vs. the LC3 applied voltage and the equivalent total induced birefringence noise.
xxi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.5: The fiber birefringence compensation experimental setup. (QWP:
quarter wave plate; PBS: polarization beamsplitter; LC: liquid crystal switch; P:
polarizer; M: mirror).
Figure 4.6: The high optical power AO modulation system being used for (a) single
input-single output microwave transversal filter and (b) single input-multiple output
transmit/broadcast microwave phased array antenna.
Figure 4.7: The experimentally demonstrated 2-bit, 1-channel switched photonic
delay line setup using the compact reflective geometry in the delay paths. The
experiment is performed at a 1 GHz optical modulation frequency using a 633 nm
visible input light beam.
Figure 4.8: The acousto-optic modulator-based light modulation technique for
obtaining the 1 GHz modulated optical input signal for the PDL.
Figure 4.9: (a) Spectrum analyzer trace of the 1 GHz signal modulation measured at
the entrance of the PDL, using the high speed photodetector and (b) spectrum
analyzer trace at the output of the PDL for the first setting (where both bits are set
for the non-delay path setting), indicating the -9.9 dB electrical loss of this setting.
Notice the similar spectrum analyzer floor for both readings, indicating no
extraneous frequency pick-up.
Figure 5.1: Position of the smectic chiral LC molecule in the layer.
xxii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 5.2: The surface-stabilized FLC configuration first demonstrated by Clark
and Lagerwall. n is the molecular director, P the ferroelectric polarization, and z
the layer normal.
77
Figure 5.3: Principle of operation of the FLC polarization switch, (a) “o ff’ state,
(b) “on” state.
79
Figure 5.4: The experimental setup for the photonic time delay unit using
ferroelectric liquid crystal optical switching devices. The dashed line represents the
delay path, while the solid line represents the non-delay path. (P: polarizer; TBS:
Thompson polarization beamsplitters; M: mirror; HWP: half wave plate).
81
Figure 5.5: Oscilloscope traces of the FLC switching time that is controlled by the ±
5 V bipolar waveform. The bottom traces show the bipolar drive signal to FLC 1,
while the top traces shows the photo-detected optical output at port 1; (a) 73 |is time
delay before the FLC device starts responding to the applied voltage, (b) 75.8 (is
(10% to 90% or vice versa) rise or fall time.
87
Figure 5.6: shows oscilloscope traces of the FLC switching time that is controlled
by the specially optimized waveform with a ± 15 V switching transient voltage and
a ± 5 V holding voltage (bottom trace). A 35 (is switching time is shown (top
trace). Note also the -30 (is delay of the FLC responce to the applied waveform.
xxiii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
Figure 6.1: Schematic diagram of the 3-digit ternary optical layout switched PDL,
and a table showing the possible delay settings. MSD: Most Significant Digit, LSD:
Least Significant Digit.
Figure 6.2: (a) A single stage of a PBS based PDL ternary architecture; (b) A single
stage of a TBS based PDL ternary architecture. (TBS: Thompson polarizing
beamsplitter, PBS: Cube polarizing beamsplitter, Ls: Lenses, M: Mirror, SA:
Switching Array, P: Polarizer).
Figure 6.3: (a) Top view of the first stage of a multichannel, fiber-delay, ternary
transmissive PDL; (b) 3-D view of the multichannel, fiber-delay, ternary
transmissive PDL (PBS: Cube polarizing beamsplitter, SA: Switching Array, P:
Polarizer).
Figure 6.4: The geometry of a 2-D phased array antenna that is mechanically steered
in azimuth and electronically steered in height.
Figure 6.5: The novel basic wavelength multiplexed reversible photonic control
system for 1-D steered phased array antennas using a single physical channel Fdigit switched wavelength dependent PDL in cascade with a wavelength
independent G-digit switched PDL with multiple physical channels.
Figure 6.6: (a) TBS based single physical channel, dispersive PM-fiber, ternary
PDL transmissive architecture; (b) PBS based single physical channel, dispersive
non-PM-fiber, ternary PDL reflective architecture.
xxiv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 6.7: PBS based single physical channel fiber grating PM-fiber, ternary PDL
architecture. (SA: switching array: QWP: quarter wave plate; P: polarizer, M
mirror).
101
Figure 6.8: (a) Controller optoelectronic transmit/receive module, and (b) antenna
element optoelectronic transmit/receive module used in our wavelength multiplexing
photonic control systems.
105
Figure 6.9: The novel wavelength multiplexing reversible photonic control system
for 2-D beam steering of phased array antennas using a single physical channel Xdependent ternary PDL and a multichannel A.-independent ternary PDL, for
independent control in the two scan axes. (MUX: multiplexer; DEMUX:
demultiplexer).
108
Figure 6.10: 2-D sub-array partitioning of a 2-D phased array antenna. The antenna
is divided into H sub-arrays, witheach sub-array containing M x N elements.
110
Figure 6.11: The novel CREOL wavelength multiplexed reversible photonic control
system for 2-D beam steering of phased array antennas using 2-D sub-array
partitioning, (a) controller site, and (b) antenna site.
xxv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
111
Figure 6.12: Diagram showing the calculated absolute value of the even time delays
2 t , 4-t , ..., 26-t of the single channel dispersive fiber PDL for 10 wavelengths
with spacing of 1 nm. The dispersion of the fiber is -134 ps/km nm at 1550 nm,
the fiber increment is 30 m. Note that since the fiber has negative dispersion, the
long wavelengths travel faster in the fiber, and thus obtain shorter time delays than
the short wavelengths.
114
Figure 6.13: Calculated optical gain variation across a 40 nm optical bandwidth for
the wavelength multiplexed optical signals that pass through a FLC device
polarization rotator designed for 1310 nm center frequency. Both the 3 dB type plot
and the high resolution/detail plot are shown, indicating essentially no optical signal
variation across the ^.-bandwidth.
120
Figure 6.14: Calculated optical gain variation across a 40 nm optical bandwidth for
the wavelength multiplexed optical signals that pass through a FLC device
polarization rotator designed for 1550 nm center frequency. Both the 3 dB type plot
and the high resolution/detail plot are shown, indicating essentially no optical signal
variation across the ^.-bandwidth.
120
Figure 7.1: Three options of the proposed PDL, using (a) solid optics, (b) free
space, and (c) fiber delay paths for ultra short, moderate, and long time delays,
respectively. (P: Polarizer, QWP: Quarter wave plate, M: Mirror, SA: Switching
Array).
127
XXVI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 7.2: The previously proposed reflective PDL architecture. (P: Polarizer,
QWP: Quarter wave plate, M: Mirror, SA: Switching Array).
Figure 7.3: Sub-picosecond time delay option of our proposed adaptable PDL
architecture using (a) a micrometer resolution translation stage, (b) two glass plates
of different thickness, and (c) a birefringent-mode electrically controlled NLC
device, whose index of refraction can be precisely controlled.
Figure 7.4: The transmission (T), and reflection (R) intensity coefficients of a
typical commercial cube PBS. s: vertical polarization; p: horizontal polarization.
Figure 7.5: SNR analysis of (a) the non-delay mode and (b) the delay mode of a
previously proposed symmetric PDL. Signal is shown with a single arrow, while
noise is shown with a double arrow. (M: Mirror, QWP: Quarter Wave Plate).
Figure 7.6: SNR analysis of the (a) non-delay mode and (b) delay mode of the
proposed PDL. Signal is shown with a single arrow, while noise is shown with a
double arrow. (M: Mirror, QWP: Quarter Wave Plate).
Figure 7.7: (a) The experimental set-up for the proposed adaptable PDL
architecture; (b) The experimental set-up for the previously proposed symmetric
PDL architecture. Delay paths are shown with double arrows, and non-delay paths
with single arrows.
xxvii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 7.8: (a) Wavelength multiplexing technique using dispersive fibers, (b)
Wavelength multiplexing technique using fiber Bragg gratings. QWP: quarter wave
plate; P: polarizer; M: mirror.
141
Figure 7.9: Polarization multiplexing technique using beam-displacing prisms.
(HWP: half wave plate; P: polarizer; TIk: total internal reflection prism; BDP: beam
dispacing prism).
143
Figure 8.1: The experimental 3-bit PDL system using FLC devices, imaging optics,
and fiber-optic remoting. The non-delay paths are represented with solid lines while
the delay paths are represented with dashed lines. (T1R: Total internal reflection
prism; PBS: Polarizing beamsplitter cube; L: Lenses, M: Mirror).
147
Figure 8.2: Our infrared 1310 nm ferroelectric liquid crystal (FLC) polarization
switching device. Each device consists of three FLC cells in cascade.
151
Figure 8.3: Optical interchannel crosstalk relative to the center active channel of the
3-D PDL, with measurements taken along the (a) x and (b) y directions at the PDL
output plane. A maximum optical crosstalk of -27.47 dB (or - 54.94 dB RF) is
measured at the nearest to center channel in the x-direction. These optical power
measurements are directly taken from the PDL output plane, before the GRIN-lensfiber assembly. The 3-D PDL has a channel capacity of 196 channels.
xxviii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
155
Figure 8.4: C/N measurements of the 1 GHz modulation Lasertron QLINK1-051
fiber-optic link (a) without (103 dB/Hz) and (b) with (79.34 dB/Hz) the PDL. The
noise marker is placed at a 100 kHz offset and the analyzer RBW = 1.0 kHz.
157
Figure 8.5: Oscilloscope traces showing the time delayed signals for three different
time delay settings of the PDL. The top trace (trace A) corresponds to the measured
zero delayed signal, the middle trace (trace B) corresponds to a measured 1.66 ns
delayed signal, and the bottom trace (trace C) corresponds to a 5.720 ns measured
delayed signal. The arrows show the points between which the time delay
measurement was done. Scope resolution is 1 ps.
161
Figure 8.6: Oscilloscope traces (a) the 33 |is time delay before the FLC device starts
responding to the applied voltage and (b) the 35 ps (10% to 90% or vice versa) rise
time or fall time. (Top trace: the photodetected optical output showing the FLCdevice time response, Bottom trace: the specially optimized waveform with a ± 15
V switching transient voltage and a ± 5 V holding voltage. Note that the driving
voltage observed on the oscilloscope has been attenuated by
10 dB).
162
Figure 9.1: The experimental set-up of the external modulation fiber-optic link fed
photonic delay line.
165
xxix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 9.2: The experimental 4-bit photonic delay line using FLC devices, imaging
optics, and fiber-optic remoting. The non-delay paths are represented with solid
lines while the delay paths are represented with dashed lines.
Figure 9.3: RF spectrum analyzer traces at 6 GHz showing (a) the 11.98 dBm RF
signal fed to the electo-optic modulator; (b) its 107 dB/Hz C/N measured at 100kHz
offset.
Figure 9.4: RF spectrum analyzer traces showing the 42 dB RF loss of the
externally modulated 6 GHz fiber-optic link, (a) RF power drops at -30.01 dB, (b)
a 104.2 dB/Hz C/N is measured at 100kHz offset.
Figure 9.5: Spectrum analyzer trace showing a 19.16 dB RF loss of the 6 GHz
fiber-optic link with the PDL set for zero delay. This 19.60 dB RF loss is
consistent with the expected RF loss based on the optical losses due to the PDL
setting, the input polarizer, output single-mode fiber and the additional FC/PC
connector.
Figure 9.6: RF spectrum analyzer traces showing the dynamic range loss
compensated 6 GHz (a) RF power at -30.00, and the (b) 104.0 dB/Hz C/N
measured at 100 kHz offset.
Figure 9.7: RF spectrum analyzer traces 6 GHz showing (a) the RF power at -8.33
dBm, and (b) the 103.8 dB/Hz C/N measured at 100 kHz offset.
XXX
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 9.8: RF spectrum analyzer traces of the 3 GHz signal fed into the electro­
optic modulator (a) RF power: 12.17 dBm, and (b) C/N: 108.3 dB/Hz measured at
100 kHz offset.
Figure 9.9: RF spectrum analyzer traces of the 3 GHz photodetected signal after
post amplification (a) RF power: -7.50 dBm, and (b) C/N: 105.0 dB/Hz measured
at 100 kHz offset.
Figure 9.10: Oscilloscope traces at 6 GHz showing a (a) 0.270 ns, (b) 0.105 ns,
and (c) 0.425 ns time delay for the second, third and fourth PDL bits. (Top traces:
the non-delayed signal, bottom traces: the delayed signal).
Figure 9.11: Optical interchannel crosstalk relative to the center active channel when
the PDL is for maximum delay setting, with measurements taken along the (a) x and
(b) y directions at the PDL output plane. A maximum optical interchannel crosstalk
of -42.1 dB (or -84.2 dB RF) is measured at the nearest to center channel in the ydirection.
Figure 10.1: Typical experimental set-up for the externally modulated fiber-optic
link. A fiber-optic attenuator is used to adjust the optical power impinging on the
photodetector.
Figure 10.2: Dynamic range loss compensation method based on high speed
electronic control of the variable optical attenuator in synchronous control with the
PDL settings.
xxxi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 10.3: Basic design of a high speed, variable optical attenuator that operates
in synchronism with the variable PDL. The variable optical attenuator consists of a
cascade of binary attenuator modules
Figure 10.4: The high speed, variable optical attenuator based on fast FLC
polarization switching devices and polarization beamsplitter cubes (T1R: total
internal reflection; P: polarizer, S: polarization switch;
attenuation value
attenuation plate with
A0: attenuation plate with zero attenuation).
Figure 10.5: Phase perturbation-based optical attenuator with 7x7 independently, 0n phase, controlled FLC arrays (SMF: Single mode fiber, GRIN: gradient index
lens).
Figure 10.6: Single stage gray scale optical attenuator based on a holographic
polymer dispersed liquid crystal device with 7x7 independently controlled variable
diffraction efficiency programmable gratings (PGs).
Figure 10.7: The TV-bit electro-optic attenuator based on the photoconductive effect.
Two-dimensional VCSEL array is used to activate each photoconductive bit.
(Si:CPW PCS: coplanar microwave waveguide on a photoconductive silicon
substrate).
xxxii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 10.8: FO link fundamental and two-tone intermoduladon distortion output
versus the link fundamental input at 6 GHz (resolution bandwidth 1 kHz).
Figure 10.9: FO link fundamental and two-tone intermodulation distortion output
versus the link fundamental input at 6 GHz when the PDL is inserted in the optical
path (resolution bandwidth 1 kHz).
Figure 10.10: FO link fundamental and two-tone intermodulation distortion output
versus the link fundamental input at 6 GHz when the dynamic range loss recovery
technique is used (resolution bandwidth 1 kHz).
Figure 10.11: FO link fundamental and two-tone intermodulation distortion output
versus the link fundamental input at 3 GHz (resolution bandwidth 1 kHz).
Figure 10.12: FO link fundamental and two-tone intermodulation distortion output
versus the link fundamental input at 3 GHz when the PDL is inserted in the optical
path (resolution bandwidth 1 kHz).
Figure 10.13: FO link fundamental and two-tone intermodulation distortion output
versus the link fundamental input at 3 GHz when the dynamic range loss recovery
technique is used (resolution bandwidth 1 kHz).
Figure 10.14: Network analyzer plots showing the RF gain of the FO link (a)
without the PDL, (b) with the PDL, and (c) with the PDL and the dynamic range
compensation technique.
xxxiii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 11.1: High density packing hexagonal configuration for a GRIN-lens FOcollimator array, using physical contact of the GRIN-lens ferrules.
212
Figure 11.2: Shift and tilt effects on the output collimated beam of a GRIN-lens
FO-collimator when the SM-fiber has a tilt or shift from the optimum position on
the surface of the GRIN-lens. (a) Optimum SM-fiber position, (b) shift of the SMfiber from the optical axis of the GRIN-lens causes tilt of the output collimated
beam, (c) tilt of the SM-fiber from the optical axis of the GRIN-lens causes shift of
the output collimated beam.
213
Figure 11.3: The OZ-Optics FO-flange used as the building block for our input and
output fiber-optic arrays.
215
Figure 11.4: The fiber array design based on the OZ-Optics FO-flange.
215
Figure 11.5: Telescope design using a set of two plano-convex lenses.
216
Figure 11.6: (a) Position of the optical beams at the input plane of the telescope,
(b) position of the optical beams at the output plane of the telescope.
218
Figure 11.7: The FLC polarization switching array consisting of 33 pixels, (a) FLC
pixel array layout, (b) detail of the pixels.
220
Figure 11.8: The ferroelectric liquid crystal polarization switching arrays, (a) all 33
pixels “off”, (b) 9 pixels “off”, (c) 24 pixels “off”.
xxxiv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
220
Figure 11.9: The 6° tilt in the FLC optical axis in respect to the array axis.
Figure 11.10: (a) The input SM polarization-maintaining fiber array, (b) the output
MM fiber array.
Figure 11.11: The fiber optic remoted photonic controller for phased array antennas.
Figure 11.12: The experimental set-up of our 7-bit 16 active channel channel
photonic delay line.
Figure 11.13: Network analyzer measurements showing the (a) 0.2 ns time delay
and (b) the 0.8 ns time delay.
Figure 11.14: Oscilloscope traces of the switching time of the 33 pixel BNS FLC
devices (a) a 37.6 ps rise time and (b) 100.4 |is fall time. Note also the finite delay
time of 17 |is and 30 ps of the FLC response to the applied waveform.
Figure 11.15: (a) RF spectrum analyzer trace showing the 3 GHz signal of the FOlink without the PDL, (b) RF spectrum analyzer trace showing the 3 GHz signal of
the FO-link with the PDL using gain balancing.
Figure 11.16: (a) RF spectrum analyzer trace showing the 6 GHz signal of the FOlink without the PDL, (b) RF spectrum analyzer trace showing the 6 GHz signal of
the FO-link with the PDL using gain balancing.
xxxv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 11.17: Interchannel crosstalk measurements for the central channel.
233
Figure 12.1: A fiber optically interconnected PDL system using a cascade of single
bit compact PDL modules.
237
Figure 12.2: (a) Electro-mechanical based PDL, and (b) Integrated electro-optic
switch based PDL.
238
Figure 12.3: The reversible bulk optics FLC based compact PDL module.
240
Figure 12.4: The experimental set-up of the compact PDL module based on
spherical microlens fiber-ports.
241
Figure 12.5: Photographs of the compact photonic delay line based on spherical
microlens fiber-ports (a) Top view, (b) perspective view.
242
Figure 12.6: Oscilloscope traces showing (a) the non-delayed and (b) the delayed
signal. Top traces: signal driving the external modulator; Bottom traces:
photodetected signal at the output of the PDL module. The markers have been
positioned at the on-set of the pulse, where the pulse gets to 10% of its maximum
value.
244
Figure 12.7: The experimental set-up of the compact PDL module based on GRIN
lens collimators.
245
xxxvi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 12.8: Measured optical insertion loss data between a pair of GRIN lens FO
collimators as a function of their distance from each other.
Figure 12.9: (a) 0.25 pitch GRIN lens, (b) < 0.25 pitch GRIN lens with an air gap
between the GRIN lens and the fiber.
Figure 12.10: Single physical channel, wavelength dependent, compact PDL for
multichannel operation using FBGs. (P: polarizer, M: mirror, PMF: polarization
maintaining fiber; QWP: quarter wave plate; FBG: fiber Bragg grating; PS:
polarization switch).
Figure 12.11: Delay path option based on alternating long length PMFs without
FBGs, and short length PMFs with FBGs.
Figure 12.12: FLC switching response at 1319 nm (a) a 9.84 ps rise time and (b) a
10 ps fall time.
Figure 12.13: The ultra-compact PDL module based on MEMS technology. Both
ultra-short time delay and extra-long time delay options are shown.
Figure 13.1: The principle of operation of a polymer dispersed liquid crystal device,
(^polymerhost= index of refraction of the polymer host,
= the electricaly controlled
index of refraction of the pore infused with liquid crystal, LC: liquid crystal).
xxxvii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 13.2: The experimental set-up of the proposed PDLC-based photonic delay
line, shown in a single bit configuration. (AOM: acousto-optic modulator; L: lenses;
P: polarizer; HWP: half wave plate; M: mirrors, 0B1: Bragg angle for PDLC1; 0B2:
Bragg angle for PDLC2).
Figure 13.3: Oscilloscope traces showing (a) the non-delayed and (b) the delayed
photodetected RF signal, indicating a 4.7 ns relative time delay. Top traces: signal
driving the AOM; Bottom traces: Photodetected signal at the output of the PDL
module.
Figure 13.4: The proposed PDLC-based PDL, set for no-delay, with active noise
filters to improve the electrical SNR to > 90 dB. Optical signal out = 0.95 x 0.99 x
0.99 x P, optical leakage noise = 0.05 x 0.05 x 0.01 x P (“x ”stands for
multiplication).
Figure 13.5: A single bit of the proposed photonic delay line based on polarization
switching devices and polarization selective holograms. (Dashed lines: delay path;
Solid lines: non-delay path; PS: polarization switch; PSH: polarization selective
hologram; M: mirror, L: lens).
Figure 13.6: The PDLC as a polarization selective hologram (a) horizontally
polarized input “sees” the grating and is deflected into the first order, (b) vertically
polarized input does not “see” the grating and passes through unaffected.
xxxviii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 13.7: The experimental set-up of the proposed PDL using a FLC device as a
polarization switch and PDLC devices as polarization selective holograms. (Dashed
lines: delay path; Solid lines: non-delay path; SMF: Single mode fiber, RF: radiofrequency).
270
Figure 13.8: The PDL experimental set-up showing the passive and active noise
filters. P: polarizer, GRIN: gradient index lens; SMF: single mode fiber, L: lenses;
LD: semiconductor laser.
276
Figure 13.9: Oscilloscope traces showing (a) the non-delayed photodetected signal
and (b) the delayed photodetected signal. (Top traces: reference signal from the
oscilloscope; Bottom traces: the photodetected output signal).
279
Figure 13.10: The compact reflective photonic delay line architecture based on
polarization selective holograms.
282
Figure 13.11: Oscilloscope traces showing (a) the non-delayed photodetected signal
and (b) the delayed photodetected signal for the compact reflective PDLC-based
PDL. (Top traces: reference signal from the oscilloscope; Bottom traces: the
photodetected output signal).
284
Figure 13.12: A PSH via a bireffingent mode nematic liquid crystal based device
used as a polarization dependent diffraction grating.
xxxix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
287
Figure 13.13: Top view of a PSH formed with a thin-film-resistor network based
NLC deflector.
287
xl
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
GLOSSARY
1-D
One-dimensional, or One Dimension
2-D
Two-dimensional, or Two Dimensions
3-D
Three-dimensional, or Three Dimensions
AO
Acousto-optic
AOM
Acousto-Optic Modulator
AR
Anti-Reflection
BDP
Beam-Displacing Prism
C/N
Carrier-to-Noise Ratio
CDR
Compression Dynamic Range
CGH
Computer Generated Holograms
CPW
Coplan ar Waveguide
CW
Continuous Wave
DFB
Distributed Feedback
DR
Dynamic Range
EMI
Electromagnetic Interference
EMP
Electromagnetic Pulse
EO
Electro-optic
ER
Extinction Ratio
FC/PC
Flat Connection-Physical Connection
FL
Focal Length
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FLC
Ferroelectric Liquid Crystal
FO
Fiber Optic
FR
Faraday Rotator
FRM
Faraday Rotator-Mirror
FWHM
Full Width Half Maximum
GRIN
Gradient Index
H-LD
High-Power Laser Diode
IR
Near Infrared
LC
Liquid Crystal
LD
Laser Diode
LSB
Least Significant Bit
LSD
Least Significant Digit
MEMS
Microelectromehanical System
MM
Multimode
MMF
Multimode Fiber
MMIC
Microwave Monolithic Integrated Circuit
MQW
Multiple Quantum Well
MSB
Most Significant Bit
MSD
Most Significant Digit
NA
Numerical Aperture
ND
Neutral Density
NF
Noise Figure
NLC
Nematic Liquid Crystal
OE
Opto-electronic
PAA
Phased Array Antenna
xlii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PBS
Polarizing Beamsplitter
PDL
Photonic Delay Line
PDLC
Polymer Dispersed Liquid Crystal
PG
Programmable Grating
PIPS
Polymerization-Induced-Phase Separation
PM
Polarization Maintaining
PMF
Polarization Maintaining Fiber
PSH
Polarization Selective Holograms
QWP
Quarter Wave Plate
RF
Radio Frequency
RIN
Relative Intensity Noise
SA
Switching Array
SFDR
Spurious Free Dynamic Range
SLM
Spatial Light Modulator
SM
Single Mode
SMF
Single Mode Fiber
SMD
Segmented Mirror Device
SNR
Signal-to-leakage Noise Ratio
SOP
State of Polarization
SSFLC
Surface-Stabilized Ferroelectric Liquid Crystal
T/R
T ransmit/Receive
TBS
Thompson Polarization Beamsplitter
TIR
Total Internal Reflection
TOI
Third Order Intermodulation
VCSEL
Vertical Cavity Surface Emitting Laser
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
VGCA
Variable Gain Control Amplifier
xliv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 1
INTRODUCTION
To detect small objects using a phased array radar, very short microwave frequency
pulses are used, and thus wideband antenna control is required. At present, this is a
difficult task for electronic controllers. Furthermore, there are large size phased array radars
that require long time delays that cannot be implemented using electronics due to the
frequency sensitive, heavy, lossy, and power consuming nature of microwave waveguidetype delay lines. Hence, photonic delay lines are a powerful and important technology for
the implementation of wideband phased array antenna controllers since they solve many of
the limitations of electronic controllers. It has been over a decade since the first articles
dealing with photonic control systems for phased array antennas were first published.
Since then, the field has been developing rapidly in various parts of the world, such as
Australia, Canada, Europe, Japan and the USA. Researchers come from different
organizations such as large industrial companies, small technology bussinesses, academia,
goverment research organizations, and corporate research labs. Photonic delay line
applications are spreading to other fields such as laser radars, ultrasound, optical
communications, optical memories and astronomy.
This dissertation proposes, studies, analyzes, and experimentally demonstrates the
use of ferroelectric liquid crystal (FLC) technology for wideband phased array control
applications. Two-dimensional (2-D) FLC devices are used as 2-D polarization switches in
photonic delay lines (PDLs) to control and process optical signals that drive the elements of
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a phased array antenna (PAA). This dissertation also studies, for the first time, fiber optic
delay lines based on polarization switching.
We begin in chapter 2 with a short introduction on phased array antennas and their
control techniques.
Chapter 3 deals with the implementation of true time delay control using photonics.
The advantages of photonics compared to electronics are presented. We review the history
of photonic delay lines and we address some of their problems. We introduce the concept
of polarization based switched photonic delay lines using FLC devices to obtain the high
switching speeds required for advanced phased array antenna applications. A novel
polarization leakage noise suppression scheme is proposed and experimentally
demonstrated to form record high signal-to-noise leakage (SNR) performance PDLs [1].
In chapter 4, we introduce a novel birefringence compensation technique for the
implementation of a reflective geometry polarization switched fiber-optic PDL. Using this
birefringence compensation technique it is possible to use single-mode (SM) optical fibers
even though our system is polarization dependent. This SM fiber based reflective PDL
architecture is capable of giving long and extra-long time delays using minimum physical
space. In this chapter, we first give the theoretical analysis of the birefringence
compensation technique and then we experimentally demonstrate its feasibility [1]. We also
experimentally demonstrate a 2-bit microwave band PDL based on this architecture for the
first time [2, 3]. This first experimental demonstration uses visible light for high observed
(to the PDL builder) accuracy of optical leakage noise, signal, crosstalk, and insertion loss
measurements.
Chapter 5 describes, for the first time in the literature, the use of FLC devices as
polarization switches for variable PDL applications [3]. The principle of operation of the
FLC device is described, and a single bit PDL experiment is demonstrated. Experiments
show, for the first time the limitations of FLC devices. The main limitation of these binary
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
mode FLC devices is that they do not perform equally well in both their states. This limits
the on/off performance of the switching fabrics formed using a FLC device and a
polarization beamsplitter, and thus the SNR performance of the PDL is also reduced. We
propose a solution to the above problem and we experimentally demonstrate that using the
novel active noise reduction scheme we can obtain high SNR PDLs using the low
performance FLC devices [4].
In chapter 6, two novel PDL architectures are proposed for obtaining the maximum
hardware compression reversible photonic beamformer ever proposed [5]. The first PDL
architecture uses ternary instead of binary designs, and the second PDL architecture makes
use of wavelength multiplexing to reduce the size, the number of PDL digits, and the
physical number of channels in a photonic beamformer. One-dimensional (1-D) and twodimensional (2-D) antenna steering are described.
One limitation of the previously proposed PDL architectures is that they can provide
a rather limited range of time delays. In chapter 7, we propose the first PDL architecture
that can give a wide range of time delays, from sub-nanoseconds up to several
nanoseconds, using the same basic structure and hardware [6]. This adaptable PDL
architecture also gives balanced loss and SNR performance. First the theory of the
adaptable PDL is presented and then proof of concept experiments are performed.
In chapter 8, we demonstrate for the first time a 3-bit PDL using FLC polarization
switching devices and directly modulated fiber-optic (FO) link feed [7]. Fiber-optics are
used for remoting. This is the first experimental demonstration at the near infrared (IR)
wavelength (A. = 1310 nm) that is popular for microwave band FO links. We test this
reversible FLC-based switched single-channel PDL at a 1 GHz FO-link modulation
frequency, and the overall laboratory system is characterized with respect to the system
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
parameters such as carrier-to-noise ratio (C/N), insertion loss, spatial interchannel
crosstalk, and time delays.
In chapter 9, we expand our 3-bit PDL to a 4-bit PDL, and we examine the several
PDL issues for an external modulation FO link feed [8]. For the first time multimode (MM)
fibers are proposed at the output of the PDL for improved coupling efficiency and an
improvement of 1.5 dB in optical insertion loss is demonstrated compared to a SM-fiber
output coupling system. This leads to a 3 dB improvement for the RF insertion loss for the
PDL which is an improvement of a factor of 2.
Chapter 10 describes the theory and a proof of concept experimental demonstration
of a synchronous amplitude and time control system for obtaining optimum dynamic range
variable photonic delay lines. Our approach is based on the high optical power available
from the optical source that can be adjusted and calibrated to obtain the output optical power
at levels required from the photodetector for optimum dynamic range. Record high
compression dynamic range (CDR) and spurious free dynamic range (SFDR) are obtained
for a switched PDL. The technique is tested at the 3-6 GHz frequency band showing a flat
performance for the gain of our PDL.
In chapter 11, we demonstrate the first ever 7-bit 16 active channel PDL for phased
array antenna control. We designed the multiple channel (33-pixel) FLC devices so that our
system is alignment tolerant. This is accomplished by designing the FLC pixels 1.65 mm in
diameter and with a 300 pm inter-pixel spacing. The controller is remotely fed by a unique
fiber array design that uses gradient index (GRIN) lens collimators for the input SM
polarization-maintaining (PM) fibers. The optical signal is collected by a similar fiber array
that uses MM-fibers for improved coupling efficiency. PDL issues such as within channel
leakage noise, interchannel crosstalk, time delays and insertion loss are examined.
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 12 describes a unique modular approach for our PDLs. This modular
approach gives small, robust, all fiber connectorized PDL modules useful for applications
where space and volume is limited. Two different approaches are presented, one based on
spherical microlenses [9] and the other on GRIN lenses [10]. Insertion loss, leakage noise,
and time delay analysis is performed, and compared with electro-mechanical and integratedoptic switches. A novel multichannel operation is also proposed using only one-physical
PM-fiber-channel with fiber Bragg gratings and multiple wavelengths.
Chapter 13 describes alternative technologies for the implementation of PDLs. The
first one is based on electrically switched holographic polymer-dispersed liquid crystal
(PDLQ devices that can perform both the optical switching and routing of the signal, thus
reducing the number of optical components in the PDL [11]. The second approach is based
on FLC polarization switching devices and polarization selective holograms (PSHs) [12].
Experiments are performed and leakage noise reduction schemes are developed to improve
system performance.
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2
PHASED ARRAY ANTENNAS
2.1 Introduction
Unlike large mechanically steered antennas, such as dish or parabolic antennas,
active phased array antennas (PAAs) offer many advantages, including beam steering
without physical movement, highly accurate beam pointing, and increased beam scan
flexibility in three dimensions (3-D). Presently, most PAAs are used for military
applications as electronically controlled PAAs are extremely expensive for large scale
commercial use. PAAs have features that are highly desirable for many emerging
commercial applications such as cellular communications, satellite communications, air
traffic control radars, and other mobile platform antenna systems.
PAAs can employ two types of control techniques for scanning an antenna beam.
The first technique is called phase-based scanning and uses modulo-27t phase shifters to set
the phase of the microwave signal that drives the antenna elements. This technique is
frequency sensitive, and causes beam squinting when the instantaneous bandwidth of the
signal exceeds a certain value while using a fixed phase setting [1]. The second technique is
called true-time delay steering and is frequency independent. In this technique, delay lines
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
arc used to give different time delays to the signals and allow wideband signals to be
radiated from the antenna elements without beam squinting [1].
2.2. The Phased Array Antenna
The PAA has an aperture that is assembled from a great many similar radiating
elements, such as slots, dipoles or printed circuit “patches” [1, 2]. Each element is
controlled individually in phase and amplitude. Accurately predictable radiation patterns and
beam directions can be achieved by varying the relative phase difference and the amplitude
of the signal that drives the radiators (antenna elements). Phased arrays have the potential
of operating over very wide bandwidths. The high end of the frequency band is limited by
the physical size of the elements, which must be placed close enough in the array to avoid
the generation of grating lobes [1]. A radar system that has the ability to change frequency
over a wide band can adapt its transmission to take into account frequency-dependent
multipath characteristics, target response, environmental conditions, interference and
jamming [1]. Moreover, wideband processing can give fine range resolution.
Phased array antennas break the conventional nexus between the aperture size and
the spatial resolution [2] and thus have been used for military applications as well as for
radioastronomy. Smaller overall size systems allow the fine spatial resolution required in
advanced radar applications, that would be impossible to be obtained using conventional
radar systems where the required aperture size would be limiting. Additionally phased array
antennas can track many targets simultaneously on a time-sharing basis. Thus, phased
array antennas can be used for air traffic control applications. Mechanically steered
antennas/radars waste time to point the beam from target to target. On the contrary, the
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
inertialess beam of a phased array antenna can “jump” from one point in space to another in
microsecond speeds, and it may widened or narrowed at microsecond speeds, thus
providing a great amount of agility [3].
2.3. Phased Array Antenna Control
The phenomenon of steering a phased array beam is the result of the energy from
each element adding in phase at some desired point. Thus, the desired scan direction can be
obtained by selecting a relative phase difference between the antenna elements. If the
phases of all antenna elements are equal, the resulting beam points in the direction of the
aperture’s boresight axis. Different beam directions can be obtained by applying the
appropriate phase sets to the antenna elements. Since the phases can be changed
electronically, and in our case photonically, an inertialess beam is formed that can be
directed at any direction within the field of view of the array aperture. Phased array
antennas can employ two types of control techniques for scanning an antenna beam. These
two techniques are described in the following sections.
2.3.1. Phase-based Steering
When energy is incident at a phased array at an angle other than broadside (Fig.
2.1), the incremental phase shift <t>required between adjacent antenna elements for a scan
angle 0 is given by [1]
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
<j>= -
dsin0,
(2.1)
a -r f
where A^F is the wavelength of the emitted RF signal, and d is the inter-antenna spacing
(Fig. 2.1).
Phase Front
Antenna
Element
0-271 Phase
Shifters
L
Figure 2.1: A typical phased array antenna with phase-based control. One dimension is shown for
simplicity.
This indicates that the required phase is frequency dependent. If the microwave
frequency is changed and the phase setting of the phase shifters is not changed the beam
will move. Thus, any change in frequency (instantaneous or tunable) will cause the beam to
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
deviate from the desired scan positions. This is called beam squinting and can be described
by [1]
A0 = -y- • tan 0,
(2.2)
where A0 (in radians) is the change in the scan direction, / is the frequency of operation,
Af is the change in frequency and 0 is the desired scan angle. Equation 2.2 indicates that
the degree of beam squint depends on the fractional bandwidth of the signal Aflf.
2.3.2. True Time Delay Steering
To prevent beam squinting while maintaining large instantaneous bandwidths, the
modulo-27E phase shifters must be replaced by time delay networks (Fig. 2.2). The total
delay path length that has to be provided amounts to L sin© ^, where 0 ^ is the maximum
scan angle for the aperture L. The incremental time delay is
T = —sin0,
c
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2.3)
where d is the inter-antenna spacing, c is the velocity of the electromagnetic radiation in air,
and 6 is the incremental scanning angle.
Phase From
Antenna
^E lem ent
Variable time delays
Figure 2.2: A typical M+\-elem ent phased array antenna with time delay based control. One dimension is
shown for simplicity.
A typical schematic diagram of a time delay network is shown in Fig. 2.3. The
smallest bit size is typically about XRF/2 or XRF, with the precise setting adjusted by an
additional phase shifter [1]. The signal is optionally routed via electronic switches through
the A/-delay paths whose length, and thus time of propagation, increase successively by a
power of 2. Since each switch allows the signal to either follow the delay path or the non­
delay path, a total delay T can be inserted. This delay T can take any value from 0 to (2N l) x, in increments of x. Note that binary algebra can be followed to calculate the obtained
delays. In general a time delay can be described by
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2.4)
T , = (b0i • 2° + bu • 2' + ... + bNj • 2 N) - t .
where b0i, b u, ..., bNi are switching factors that take the value of 0 or 1, depending on
whether or not the signal follows the non-delay or delay path respectively. Note that the
binary design significantly reduces the number of required delays, since the number of
delay bits N required to achieve M delays is given by the following relation
M = 2",or N = ^ ^ - = \og,M.
log2
52
(2.5)
(2N- i y z
2t
Variable
Phase
Shifter
Bit 1
Bit 2
Bit N
To the
antenna
element
Figure 2.3: A typical N-bit switched time delay line network that requires N-l 2x2 switches to implement
the 2 n different time delay settings. Signal in each bit can follow either the delay or non-delay path; a
microwave phase shifter provides the fine modulo-2ft phase control.
A 1° beam scanned to 60° can have 60 independent beam positions, requiring 60
different time delay settings. This can be achieved with a 6-bit delay line, that gives 64
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
different time delay settings. The least significant bit (LSB) will be equal to A.RF, and the
most significant bit (MSB) will be equal to 3 2 ^ [1].
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3
PHOTONIC DELAY LINES
3.1 Introduction
Multichannel photonic delay line (PDL) architectures have been proposed for
transmit/receive mode phased antenna applications. These architectures are based on two
dimensional spatial light modulators (SLMs) that act as optical polarization switching
elements. Such elements can be nematic liquid crystal (NLC) SLMs [1,2], ferroelectric
liquid crystal (FLC) SLMs [3], magnetooptic SLMs [3] or multiple quantum well
(MQW) SLMs [3]. Optical delay lines can be formed using free-space or solid optics
propagation delay.
PDLs are very useful signal processing tools. PDLs can be used for a wide range
o f optical signal processing applications including astronomy [5], laser radars [6], data
storage [7], ultrasound [8], and optical communications. PDLs are being studied to
develop a future, wide bandwidth, compact, lightweight and small size, PAA controller
[9]. The ability to use time delays to provide broadband antenna beam steering has been
known since World War II, but electrical time delay control of PAAs has been a complex
and hardware intensive option, particularly due to the frequency sensitive, heavy, and
power consuming microwave waveguide-type delay lines that are available today. These
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
electrical delay limitations can be overcome by PDLs that can potentially offer s ig n ific a n t
advantages compared to electronic delay lines. In particular, at high electrical signal
frequencies, such as the microwave and millimeter-wave bands, photonic processing can
offer significant advantages such as large instantaneous and tunable signal processing
bandwidths (several gigahertz), since optics is essentially transparent to the RF
modulation. Optics also provides parallel processing capabilities that can lead to compact
and lightweight processing modules (e.g., a set of 32 beams can propagate through a
system with an active optical area o f l " x 1” ). In a photonic approach, remote control of
the PAA can be accomplished using the low loss optical fibers to transmit the signal from
the controller site to the antenna elements. This gives an additional degree o f freedom of
positioning the controller at a remote location from the antenna, and can make the overall
system more lightweight (compare the mass o f a typical microwave cable 40 g/m, with
that of a space qualified fiber 0.8 g/m). Photonics also provides protection from
electromagnetic interference (EMI) and electromagnetic pulses (EMP) since free-space,
solid optics (e.g., glass), and fiber-optic cables are non-conductive dielectrics and so they
do not disturb the RF field.
3.2. Prior Work in Photonic Delay Lines
Over the past few years, several optoelectronic technologies have been proposed
for making variable PDLs. These techniques vary in their approaches and the optical
technologies [9].
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The first to propose a passive multichannel fiber-optic delay line based on
photodetector switching to select the desired time delay signal was Levine [10]. Instead
o f switching the detectors, Ng. et. al. used electrically switched semiconductor lasers to
implement the time delays [11]. The key limitation o f this technique is the noise
accumulation because o f cascaded active components, and a very large number o f lasers,
detectors, and associated hardware resulting in a relatively high-cost, hardware-intensive
system. An optical waveguide based switching network vising integrated electro-optic
switches for routing the optical signal into external single mode fibers was proposed by
Soref [12]. A single channel 6-bit waveguide switching network was later implemented
[13]. Other integrated optic approaches include electrostatically actuated metal membrane
optical waveguide switches [14], and micromachined meander-line thin-film piezoelectric
microactuators for switching the light in different fibers [15]. Arrayed optical waveguides
[16] have also been used to form a wavelength dependant PDL.
Fiber delays have also been used to form non-switched photonic time delay
networks. In this approach light is directed to predefined fixed length fibers to obtain the
desired time delay [17, 18]. A programmable binary fiber-optic (FO) delay line
architecture was proposed by Goutzoulis [19] based on GaAs MESFETs for electrically
switching paths between a non-delay electrical path and a delay FO path. The use of
switched FO delay lines [20], where 2x2 cross bar electro-optic (EO) switches were used
to switch paths of the optically modulated microwave signal was proposed. High
interchannel crosstalk levels is the key limitation of this approach.
Recently, the use o f independent (e.g., multichannel prism geometry) dispersive
fibers with a single high power tunable laser source has been proposed for making
continuously variable (non-binary switching) PDLs [21-23]. A similar approach has also
been proposed where instead of the fiber prism, a fiber with multiple Bragg gratings is
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
used [24, 25]. The above systems are limited only to transmit operation. In addition,
recently, the use o f wavelength multiplexing has been proposed to reduce hardware in a
photonically controlled phased array antenna that use fiber delays [26, 27]. Here,
hardware compression is achieved using a cascade of two wavelength independent,
multichannel, binary, switched PDLs.
Non-fiber based techniques have also been proposed for the implementation of
time delay lines. A two dimensional (2-D) coherent optical architecture for time-delaybased PAA beamforming using free space delay lines has been proposed [2, 28]. This
interferometric architecture is based on polarization switching by 2-D SLMs based on
NLC technology and free space propagation based delay lines using polarizing
beamsplitters (PBSs) and prisms. The key limitation of this approach is the transmit only
operation.
Another technique for implementing PDLs for both transmit and receive mode
antenna applications is an incoherent (non-interferometric) reversible optical architecture
also using 2-D polarization switching arrays [1, 29]. A single bit, 25-channel incoherent
beamformer has been demonstrated using nematic liquid crystal (NLC) polarization
switching devices at visible wavelength [30]. Based on the above incoherent beamformer,
an experimental demonstration of a three-delay unit two-antenna array receiver was also
implemented at an optical wavelength of 1.3 pm [31]. So far, only NLC arrays have been
demonstrated for these polarization SLMs [30, 32]. Compact solid optic based switched
binary PDLs have also been proposed based on the incoherent reversible architecture [7].
Yao and Maleki extended the incoherent beamformer to ultrashort delays for mm-wave
applications [33].
Other optoelectronic technologies proposed for making variable photonic delay
lines include acousto-optics (AOs) [34-36], serial feeding and optical gating [37, 38] and
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
coherent detection methods [39]. A collection o f the most important papers in the field
can be found in reference [40]
3.3.
The Motivation for using Ferroelectric Liquid Crystal Polarization Switching A rray s
The key motivation for using FLC polarization switching arrays is the high
switching times reported for these materials [41]. There are radar applications that require
very high beam scan rates [42] that can not be obtained using NLC devices [32]. Thus,
the FLC devices are currently an excellent candidate for optical switching elements.
Furthermore, our general motivation for using 2-D LC pixelated polarization switching
arrays (SA) is the maturity of the low cost, flat panel LC display technology. Mature LC
technology can lead to cost effective multichannel delay lines for large scale PAAs, since
large 2-D LC SLMs can be used to form the switching elements in parallel binary fireespace, solid-optics and/or fiber delay lines. These LC SLMs act as high quality
polarization rotators, with independent electronic control of each of the pixels. Each of
the pixels can have two states, the “ on” state and the “ o ff’ state. The “ on” state rotates
the incident polarization by 90°, and the “ o ff’ state leaves the polarization unchanged.
For an A/-element antenna array, these 3-D PDLs have M independent parallel
optical processing channels. Because we propose the use of the mature large area LC
technology to form the required 2-D pixelated polarization switching array devices, the
number of pixels in the device can easily match any large number M of antenna elements
required in advanced radar applications. For instance, for a M - 5,000 element radar,
5,000 pixel LC devices can be commercially fabricated for use in the 3-D PDLs.
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Additional advantages o f LC technology is the low crosstalk between neighboring
channels which along with the use o f imaging optics can give high interchannel isolation
[30]. Furthermore, PDLs based on polarization switching can use either direct modulation
o f semiconductor lasers or external intensity modulation with an array o f electro-optic
modulators. Thus, radar carriers into the millimeter-wave regime are possible.
3.4. The Proposed Ferroelectric LC-based Photonic Delay Line
The proposed PDL architectures using FLC optical polarization switching 2-D
pixelated SLMs are based on Thompson polarization beamsplitters (TBS) and polarizing
cube beamsplitters (PBSs) for both transmissive and reflective geometries [43]. The
choice o f any o f these architectures is based on the specific requirements o f the individual
bits [43]. Fig. 3.1 shows the proposed three-dimensional (3-D) PDL system. Two
different architectures are shown for the photonic delay modules. Bit 1 is the transmissive
feed-forward PBS-based delay line, and bit N is the reflective delay line. These
architectures will be discussed in more detail later in this chapter.
The schematic diagram of a jV-bit PDL is shown in Fig. 3.2. Light has two
possible paths to follow in every bit, the non-delay path and the delay path. The delay
paths have lengths that correspond to time delays o f t , 2t , ... 2^ 'x. There are 2N different
time delay settings for the PDL network. These 2N different time delay settings give 2N
different time delays that eventually drive the antenna elements. The optical signal is
optionally routed through the //-delay paths whose length increase succesively by a
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
power o f 2. This is accomplished using 2-D polarization switches (e.g., FLC devices) and
passive polarization-based routing optical elements (e.g.. PBS).
\
TIR
2-D Switch
A rtsy
PBS
Non-PM
lay!
M irror —
Configuration
„J1 R
2-D
>
M icro-Lens
A rray
P o la rize r,
PBS
C
0
Sw itch
Array
2-D Switch,
Array
■
n
Polarizer
2 D S w itch
M inor
R e v e rs ib le S y s te m
Light
M icrow ave
Band O ptical
Intetuity
M odulated
Light IN /O U T
Bit I
Photonic
Delay M odule
Bit N
Photonic
D elay Module
Bit 2
Photonic
Delay M odule
Light to/from
antenna array
tig h t
Figure 3.1: A A/-channel A-bit PDL network for the control of phased array antennas. (QWP: Quarter wave
plate; M: mirror; SLM: spatial light modulator; PBS: polarizing cube beamsplitter).
M SB
Delay path
LSB
Bit t
B it 2
Bit AT
Bit 3
Figure 3.2: The PDL schematic diagram. (LSB: Least significant bit; MSB: Most significant bit).
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5. Photonic Delay Line System Issues
3.5.1. Signal-to-Leakage Noise Ratio
A very important issue in the implementation of PDLs for wideband phased array
antennas is the signal-to-leakage noise ratio (SNR) of the system and whether the PDL
introduces additional noise to the RF signal. Optical SNR is defined as 101og(signal
power/leakage noise power). As signal, we define the optical power in the optical beam
o f the desired polarization that travels through the desired delay or non-delay path of the
bits; all other optical power measured at the output is regarded as leakage noise optical
power. The optical SNR is limited by the polarization leakage from the polarization
switches as well as the optical polarization components (e.g., PBS). This polarization
leakage has to be suppressed in order for the PDL network to meet the specifications set
by the antenna application. Often the term electrical SNR is mentioned. The electrical
SNR is important in phased array antenna applications, as the current (or voltage)
generated by an optical detector at the output of the delay line is proportional to the
incident light intensity, and this output eventually drives a signal processing component
such as an antenna element in a radar communication antenna array system. The electrical
SNR is related with the optical SNR by
Electrical SNR = 2 x Optical SNR,
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.1)
where the factor 2 is due to the photocurrent generated by the photodiode, which is
proportional to the square of the incident optical power.
3.5.2. Switching Speed
The switching speed o f the polarization switches is a very important issue and is
the key motivation for this dissertation. So far, work focused on the mature nematic
liquid crystal (NLC) technology for SLMs has demonstrated moderately fast switching
times of 1.5 ms by using the transient nematic effect, while maintaining high > 30 dB
switch on/off levels that are typically required for high performance phased array systems
[32]. For typical high performance radars, beam scanning rates of a 1,000 beams/s
corresponds to a 1 ms transmit/receive beam dwell time [44],
For higher beam scan rates, such as required in our advanced “ Aegis” radar
application [42], faster optical switching devices are required. FLCs with reported
switching times as low as 1 ps [41] are currently an excellent candidate optical switch
technology for higher beam scan rates. Nevertheless, there are currently limitations with
FLC devices. Unlike current state-of-the art NLC devices, FLC devices do not produce
very high optical polarization extinction ratios (ER) (e.g., > 40 dB) for both the “ on” and
“ o ff’ modes o f the device. The poor optical ER are due to the fact that the FLC devices
do not fully rotate the input light polarization by exactly 90°. This problem is not an issue
for strictly on/off optical shutter type FLC applications where the device is sandwiched
between two high extinction ratio fixed crossed polarizers. On the other hand for
switching fabric-type uses such as our PDL, this FLC extinction ratio problem leads to
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
high system leakage noise. This leakage noise from FLC devices can deteriorate the
overall system performance. These low on/off performance FLC devices need to be
combined with certain optical noise reduction schemes, that can give us the desired
advanced wideband radar high performance PDLs, where both fast switching speeds (e.g.,
< 3 5 ps) and very high optical on/off isolation (e.g., 45 dB) can be achieved at both
output ports of the switching fabric [45-47].
3.5.3. Fiber Optic Links
The current wavelength for high speed analog light modulation phased array
antenna applications is in the near infrared band (e.g., 1300 nm), mainly because o f recent
commercial developments in high performance analog fiber-optic links using near
infrared, semiconductor-based optical transmitter and receiver technology [48]. Fiber
optic links offer many advantages compared to coaxial cables and other metalic
waveguide systems; these include low attenuation at high data rates low susceptibility to
electromagnetic interference, small size, light weight, and compatibility with optical
processing schemes [49]. Furthermore, they can be closely confined with negligible
crosstalk. These fiber-optic links are being considered as economical and practical
solutions for RF signal transmission and distribution, particularly for microwave phased
array antennas/radars. Currently, there is a considerable research activity on the
development of FO links [48]. Thus, we will be looking into issues o f fiber optic links
and fiber interconnections for our PDL.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5.4. Phased Array Antenna Remote Control
There are PAA applications where the controller has to be located at a remote
location, such as the Aegis Shipboard application. Processing o f the microwave signal in
the optical domain offers the ability to use optical fibers for remote control of the PAA.
Fibers offer significantly better transmission loss compared to coaxial systems, especially
over appreciable distances. Silica-based single-mode optical fiber loss at 1300 nm is ~
0.47 dB/km [50]. On the other hand, RG-400 coaxial cable has losses o f 1115 dB/km and
a semirigid coaxial cable has losses o f 790 dB/km at 5.0 GHz [51]. This is a difference of
three orders of magnitude. Even with the inclusion of the electro-optic (EO) and opto­
electronic (OE) conversion losses in the system, and the optical interconnection losses,
the optical transmission system would provide lower losses for FO link lengths o f 50-100
m. Thus, the use o f low loss fibers allows remoting o f the photonic beamformer, along
with providing a compact, lightweight, and low EMI microwave frequency signal
interconnection and distribution method, such as needed for very large aperture wide
instantaneous bandwidth PAAs. However, there are losses associated with multiple fiber
interconnects that limit the maximum number of array channels in the systems. Thus,
accurate analysis o f such losses is crucial to the design o f an optimal photonic fiber-based
system.
3.5.5. Fiber Optic Interconnects and Interchannel Isolation
Fiber optic remoting brings up the issue of fiber to fiber interconnects between the
input and output planes o f the PDL. A 2-D fiber array can distribute the processed signals
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from the controller to the antenna elements. There is a one-to-one correspondence
between the input and output 2-D fiber arrays. Our approach is to use microlenses (e.g.,
gradient index or GRIN lenses) in combination with bulk imaging lenses to efficiently
couple light from the input to the output 2-D fiber array.
An important issue with our N bit, M channel, 3-D PDL architecture is also the
interhannel crosstalk. As interchannel crosstalk we define the optical power leaking from
one channel to adjacent ones. Since the signal of interest is the time delayed or non­
delayed signal, leakage from one channel to the adjacent ones, that do not carry the same
delay “ information” , will be translated to noise and will deteriorate the system
performance. The interchannel crosstalk is most crucial at the switching planes, since
these are the decision centers. At these planes, the state of polarization o f adjacent
channels should be totally independent of the neighboring channel switching array
settings. Thus, there is a need o f an effective way of limiting the interchannel crosstalk to
acceptable levels for PAA applications (i.e., < - 60 dB electrical). Imaging optics can be
used to meet the above requirement. A 1-to-1 imaging from polarization switching array
to polarization switching array can be obtained using bulk imaging optics. This can be
achieved using a 4-/imaging system consisting of two lenses of focal lengths f.
We see that both coupling fiber efficiency and interchannel crosstalk can be
optimized by the use of imaging optics. In the following paragraphs we show how this
imaging optics can benefit our PDL performance.
As mentioned earlier microlens-based (e.g., GRIN lens) fiber optic collimators are
use to couple the microwave modulated optical signal into the PDL. The optical wave
coming out o f such a FO-collimator has a Gaussian profile, and it has its beam waist at
the surface o f the GRIN lens. A similar, if not exactly the same, Gaussian profile is
required for the optical beam at the output plane, where a similar GRIN lens FO
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
collimator is used. That means that the optical beam needs to have its beam waist at the
front surface o f the GRIN lens. For a Gaussian beam imaged by a lens (Fig. 3.3) we can
write [52]
W
1.
T;
Lens
Figure 3.3: Gaussian beam transformation by a lens, (w„ vv2 are the input and output beam waists,
respectively)
/
2
(A~/)2-(W fxf
(3.1)
and
w,
w,
_h~ f
l\ ~ f
(3.2)
where /„ l2 are the distances of the input and output beam waists, respectively, / is the
focal length (FL) of the lens, w„ w2 are the input and output beam waist, respectively, and
X is the wavelength o f the light in vacuum. In our case w2 must be equal to w,. From Eq.
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.2), ,W\ = w2 for /, = l2. Thus, if /, = f l2 also equals f Since a 2-D beam array enters the
PDL system, an additional equal FL lens is required to image, in a 1-to-l correspondance,
the input 2-D beam array to an output 2-D beam array. This lens is positioned 2 / distance
from the first lens and the output plane is at a distance/from the second lens.
¥
GRIN
GRIN
Single
Mode
Fiber
Single
Mode
Fiber
1
2
8
Input Plane
9
Output Plane
Figure 3.4: The 4 /im aging system consisting o f two lenses of focal lengths f. The two glass plates simulate
additional optical components in the path. (ng : index of refraction of the glass plate).
Fig. 3.4 shows a 4 / imaging system for a GRIN to GRIN coupling system. Two
equal thickness glass plates (thickness = d) are positioned between the GRIN lens and the
conventional lens. These glass plates are used to simulate additional optical elements in
the system, and as will be shown later they can be PBSs, or glass plates to adjust the time
delay obtained by the PDL bit. Now the physical distance I between the GRIN lens and
the conventional lens is not equal to / The actual physical distance can be calculated
using ABCD optical matrix analysis for Gaussian beam propagation [52]. The ABCD
matrix for the entire system, from the input plane (plane 1) to the output plane (plane 9),
can be written as
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A
B
C
D
—M h x M x3x ... x M 2 x M x,
(3.3)
where
II
*
II
*
M s = M xo =
'1 d 1x1
"1 d
, a /3 = m x2 =
0 1
0 1
1 d2
0
1
, and M7 = Mg =
1
0
'
/"
1
(3.4)
represent the matrices of a Gaussian beam propagation for distance dx, d, d2, and /
respectively.
M 2 = M xx =
1
0
0 1In,
(3.5)
represents the matrix for an air-glass interface,
= M x3 =
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.6)
represents the matrix for a glass-air interface, and
M* = M a =
1
0
/ '
1
(3.7)
represents the matrix for a lens of focal length f Thus, the overall matrix can be written
as
'A
C
B'
D
-1
^ n t d\ + d + nsd 2 - n j
n
0
(3-8)
-1
The new Gaussian beam characteristics can be found from the initial beam using the q
parameter of the beam [52]
_ Ay.qx+B
q2 ~ C x q x+D
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.9)
where qu q2 are the beam parameters for the input and output beams, respectively. The
beam parameter q o f a Gaussian beam is related with the waist of the beam (w) and the
beam radius (R) by the following equation
!-!_
q
R
1
nw2n
(3.10)
where n is the index o f refraction o f the medium of propagation, and A. is the wavelength
o f the light in vacuum. Using Eq. (3.8) and (3.9) q2 can be expressed as
.„ 'V W
q 2 — H\ ^ A
+ 'V /2 - V
(3.11)
Since we said that the input Gaussian beam has its beam waist at the input plane (plane
1), this means that /?, is infinite, thus
. n w 2n
<1\ = J ~ y ~ ,
Thus, Eq. (3.11) can be written as
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.12)
Since the desired beam at the output plane (plane 9) needs to have its beam waist
at this plane, R2 should also be infinite, and the beam waist is also required to be equal to
wx. Thus, the real part of Eq. (3.13) has to be zero. This gives,
d l +d2 = f -
(3.14)
By adding d to both sides we obtain
di + d2+ d = f
n„
1-d .
(3.15)
Since d l+d2+d = /, where / is the physical distance between the lens and the GRIN lens.
Eq. (3.15) can be written as
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This is an important result for the implementation of our free space and solid optics PDL
designs. In general, in our PDL designs more than one glass plates exist. Thus, Eq. (3.16)
will have additional terms of the form dx(1-1/w;) for each additional glass plate.
The above analysis shows that the 4 / imaging system gives the appropriate
characteristics to the propagating beam such as the one required for optimized coupling
efficiency to the output GRIN lens. Furthermore, the 1-to-l imaging system can help to
keep low interchannel crosstalk in a multichannel PDL [30],
3.5.6. Insertion Loss
Another important issue of the PDL is the loss in the optical power introduced by
the optical components. SNR and optical power loss are closely related, since high optical
loss will lead to a lower signal power, and thus a lower SNR. Additionally the dynamic
range (DR) of the fiber optic link will decrease with increased insertion loss. The
dynamic range is extremely important for phased array antenna applications since it
characterizes the system in terms of the minimum and maximum signal it can transmit or
receive. The use o f antireflection (AR) coated optical elements is expected to reduce
reflection losses in the system.
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.6 Three Dimensional Photonic Delay Lines Based on Polarization Switching Arrays
3.6.1. Transmissive PDL Architectures
As transmissive PDL architectures we define those architectures that have both
delay and no-delay path propagation once through each component of the delay and non­
delay path, respectively. Fig. 3.5 depicts the feed-forward DPL architecture. Which is
based on polarization beamsplitters (PBSs). The system of Fig. 3.5 forms a PDL bit,
which consists of a non-delay (solid line), and a delay (dashed line) path. Horizontal or ppolarized light coming from a microlens array hits the switching array (SA1). Each pixel
of SA1 can be set either “ on” or “ o ff’, to either rotate the incident polarization by 90°,
or leave it unaffected. When the polarization changes to vertical or s-polarization, the
light follows the delay path, while when it remains horizontal follows the no-delay path.
Since PBSs as well as SAs can have polarization leakage, noise can built up, especially in
a cascade architecture of PDLs. Such an effect can be deleterious for the performance of
the PAA controller. Leakage noise is defined as the light of unwanted polarization
propagating through the system. For a certain delay setting the light coming from the
unwanted path is the noise, and the light coming from the desired path is the signal.
Improvement of the PDL performance is possible with the use of a noise reduction
technique [4, 45] that will be described in details in a later section. At the output of each
PDL bit, a noise filter can be used. It consists o f a SA and a p-polarizer (or a 5-polarizer,
depending on the desired input polarization o f the next bit). We will call this filter, the
active noise filter, since the SA is an active optical component. Signal and noise come
from different paths, and thus, they have orthogonal polarizations to each other. Using the
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
active noise filter, the noise can be rejected, and only the signal propagates towards the
second bit. For example, assume that the desired path is the no-delay path and due to
noise leakage we have some light coming from the delay path. The polarization o f the no­
delay path is p-polarized, while that from the delay path is ^-polarized. Setting the SA
“o ff’ the desired p-polarization remains unchanged and passes through the p output
polarizer, while the unwanted ^-polarization remains unchanged and is blocked by the
output polarizer. The PDL performance can be further improved by introducing a
polarizer after the deflected output port of the first PBS. This polarizer is aligned with its
polarization axis parallel to the ^-polarization. The deflected output port of the PBS has
poor polarization extinction ratio (ER) performance (ER < 50:1) due to the high ppolarization leakage from this port [53]. The ^-polarizer restrains the p-polarization
leakage of the PBS from propagating through the delay path. We will refer to the spolarizer as the passive noise filter, in contrary with the active noise filter, because a
polarizer is a passive optical component.
=»L2
=»L2
P assive
N oise F ilter
M icrolens A rray
N V
SA1
PBS
PBS
: : :
out
S A 5 \ Active
N oise Filter
Figure 3.5: The PBS transmissive feed-forward PDL architecture.
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As mentioned earlier, one of the PDL system requirements is the low interchannel
crosstalk. We propose a 1:1 imaging optics set-up in a 4•/ configuration to keepthe
interchannel crosstalk to low levels. SA1 is placed at the effective front focal plane o f the
first lens, and SA2 is placed at the effective rear focal plane of the second lens. As
effective focal length we define the physical distance between the lens when other optical
elements (e.g., PBS, glass plates) are inserted in the path, as was shown in section 3.3.5.
The separation between the two lenses is two FL.
The use o f imaging optics dictates the use o f mirrors for deflecting the light,
instead o f the other alternative, i.e., total internal reflection (TIR) prisms. Fig. 3.6 shows a
ray tracing diagram for a set of beams deflected by a TIR prism. The principle o f TIR is
based on the angle o f incidence of the light on the hypotenuse of the prism. There will be
an angle o f acceptance of incident beams that can be totally internally reflected. For a
typical TIR prism with an index of refraction o f
= 1.5, the critical angle is q>c = 41.8°.
Ray 1, in Fig. 3.6 is incident on the hypotenuse of the prism with an angle of incidence
always greater than 41.8° and it is totally reflected. On the other hand, ray 2 has to satisfy
the condition cp2 > 41.8°, and thus, cp2’ - 90° - cp2 = 48.2°, and in effect \j/2’ > 180° - (45° +
{p2’ ) = 86.8°. This leads to v|/2 < 90°- iy2’ = 3.2°. Thus, the angle of incidence o f ray 2
onto the TIR prism face has to be 0inc < arcsin(1.5 sinvj/j) = 4.8°. Such an angle can be
obtained for R/D ratios of at least 5.95, where R is the distance of the front face o f the
TIR prism from the lens and D is the distance between the two outermost beams incident
on the lens. For D«2 cm, the required length R will be at least 12 cm. This in effect sets a
limit on the use o f TIR prisms for only long enough delay paths that can satisfy the above
condition.
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I IK P rism
V2'
♦ l2 4l.8”
s 90-^2=48.2"
y j 2 I80°-(45"+4j")
y iS 9 0 * -< p i= 3 .r
01. S 4 8"
Ray I
Ray 2
C
Figure 3.6: Ray tracing diagram for total internal reflection from a prism of a pair of beams focused by a
lens.
The time delay t for this PDL architecture, taking into acount the results of section
3.3.5 is
c
(3.17)
where/, and/2are the focal lengths of lenses LI and L2, respectively.
In Fig. 3.7 and 3.8 two different configurations of the PBS PDL systems are
shown. The first consists of a sequence of PDL bits in cascade, while the second one has
the bits interchangeable at the two sides of a “sandwiched” design of two SAs and a
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
i*.*
s^L2
i.
Hi
M icrolens
A rray «
'i
Passive
N oise R Ite r
Active
] N oise R Ite r
a
1 Passive
, N oise R Ite r
’ SA
OUT
SA PBS
PBS t • p b s
Active
Noise R Ite r
PBS
SA
Figure 3.7: PBS based transmissive PDL cascade architecture.
Passive
N oise R Ite r
PBS
PBS
T o next bit
SA1
M icrolens
Array
SA2
PBS
PBS
Figure 3.8: PBS based transmissive PDL “sandwiched” architecture.
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
P
polarizer, which is the active noise filter of one bit and the input SA o f the following bit.
The advantage o f the second architecture over the first design is the use o f only a pair of
single-substrate SAs, that are possible because o f the mature large scale liquid crystal
(LC) display fabrication techniques. In addition, the optical alignment and assembly of
this PDL system becomes simpler and the overall cost can be lower because the SAs are
on the same single substrate, and no cutting and mounting of each individual array is
required.
A second PBS transmissive PDL architecture, the feed-back architecture, is shown
in Fig. 3.9. The principle o f operation is similar to the architecture discussed earlier. In
this design we introduce a second set of mirrors and imaging system in the delay path, but
only one PBS is used. The cost o f this architecture is almost the same as that for the
previous architecture. This is because the cost o f two high reflectivity mirrors and two
AR coated lenses is almost the same as that o f a commercially available PBS. The
advantage o f this architecture is the smaller size o f the module for the same time delays.
In the feed-forward architecture, the time delay is given by the propagation time
difference between the delay and no-delay path. In the feed-back architecture the no­
delay path is common in the delay path too, so the time delay is obtained by the actual
time it takes for the light to propagate through the delay loop (dashed line). Furthermore,
since in the feed-back architecture the no-delay path is not involved in the time delay, it
can be very small, thus minimizing the overall size o f the system. This can not be done in
feed-forward architecture, since time delay is dependent on the non-delay path length too.
Nevertheless, the size of the delay path in the feedback architecture is dependent on the
size o f the optical components.
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
M icrolens A rray
<z
OUT
SA M
SA 2
L2
A ctive
N o ise R Ite r
Rgure 3.9: The single PBS transmissive feed-back PDL architecture.
The feed-back architecture can be modified and used as a re-circulating
architecture. In this case, an additional SA is introduced in the delay path, before the
PBS. Then the light remains in the delay path, until the SA is switched to direct the light
out of the delay path. In the re-circulating architecture, pulsed light has to be used. The
pulse duration and separation should be such that there is no overlapping between
different settings of the PDL. This means that all the channels should first be clear of the
propagating light before the PDL setting is changed.
An alternative passive polarization router is the Thompson beamsplitter (TBS). A
PDL architecture based on TBSs is shown in Fig. 3.10. The much better performance of
the TBSs (ER > 10,000:1) for both ports, compared to that of cube PBSs (ER ~ 1,000:1
for the straight port, ER < 50:1 for the deflected port) improves the polarization ER of the
delay path, and results in a high signal-to-leakage noise for the PDL. Thus, in this
architecture, a passive noise filter is not necessary. On the other hand, the TBS cost
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
today, which is four times higher than the PBS cost, can increase the price of the system
significantly, thus limiting the use of TBSs in non-cost sensitive PDL applications.
L2
TB S
TB S
M icrolens A rray
SA
OUT
SA
A ctive Noise R Ite r
Figure 3.10: The TBS based transmissive PDL architecture.
3.6.2. The Symmetric PDL Architecture
There are applications, such as in mm-wave radars, that require very short time
delays, e.g., < 0.5 ns. The symmetric PDL architecture shown in Fig. 3.11 can give these
short time delays. We call this architecture symmetric because the two arms of the PDL
have exactly the same physical length. A relative time delay difference between the two
paths can be introduced by placing two high quality glass plates in each of the two arms.
The relative optical path length difference between the two arms can be controlled by an
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
appropriate choice o f the thickness of the two glass plates. We choose to use two glass
plates in each of the arms, instead of just one plate, mainly because of mechanical
stability issues that arise for thin plates. It is also beneficial for the PDL performance in
terms of aberrations, due to the symmetry of the design. As an example, for time delays
of 0.01 ns, the required thickness is 2 mm for a material with an index of refraction of
1.5. A thin plate of 2 mm thickness, placed in one of the arms can be susceptible to
thermally or mechanically induced material stresses, strains, and vibrations. This can alter
the uniformity of the thickness or index of refraction o f the plate, thus, affecting
differently the time delays o f the beams passing through the multichannel PDL. On the
other hand, a set of two glass plates with a thickness difference of 2 mm will be less
susceptible to such environmental and packaging related effects, giving a uniform time
delay to all the beams in the multichannel PDL. Moreover, controllable NLC array
devices can also be used to give the small time delays, as the NLC index of refraction is
voltage controllable.
This architecture has basically the same characteristics as the double PBS
transmissive architecture, as the same optical components are used. In effect, the cost of
the symmetric architecture is the same as that o f the PBS transmissive feed-forward
architecture.
For this PDL architecture the time delay is given by
c
c
ng
\
1
n
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.18)
OUT
G lass Plate
o f thickness c
ti
rz>
Active N oise F ilter
SA
PBS
L2
L2
P assiv e
N oise R Ite r j y
G lass Plate
o f thickness
d
J
M icrolens A rray
SA
PBS
Figure 3.11: The symmetric PDL architecture. Solid line represents the non-delay path and the dashed line
represents the delay path.
where nt is the index of refraction of the glass plates, dv d2 are the thicknesses of the glass
plates in the non-delay and the delay path, respectively. /,, / 2 are the FL of the lenses in
the non-delay and the delay path, respectively.
In this architecture is also important to select f x and f 2 such that not only Eq. (3.18)
is satisfied but also the total physical distance between the input and output planes is
equal for both optical paths. This guarantees that the 1:1 imaging will be satisfied
between the input and output plane for both PDL settings and paths. Using the approach
similar to the one in section 3.3.5 we find that this happens only when
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.6.3. The Reflective Geometry PDL Architecture
The PDL architectures discussed earlier can provide short time delays (<0.5 ns),
as well as moderate time delays (e.g., < 5 ns). When the required time delays are long
(e.g., > 5 ns), issues related to large free-space/solid optics systems such as degree of
mechanical stability and size problems o f the PDL arise. These problems can be
overcome with the use o f a reflective geometry PDL architecture. The reflective geometry
is defined as the PDL geometry where the delay and/or the no-delay path light can
propagate more than once through the optical components that compose the path.
The proposed compact reflecting geometry for a A-bit switched photonic delay
line for microwave signal processing and control, such as for phased array antenna
control is depicted in Fig. 3.12. All possible delay path configurations are shown, namely,
the free space delay, the solid optics delay, and the non-polarization maintaining (PM)
fiber delay. The advantage of this reflective configuration is that the light travels twice
through the same path acquiring time delays that with a transmissive architecture would
require longer delay paths. In conjunction with the use of imaging optics and the
possibility of multichannel operation, the reflective architecture can be used to provide
photonic delay line systems that are compact, small, and lightweight.
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Free Space Bit
Solid Optics Bit
IV(1
\G
Ll
= QWP
, no
IN PJ
1 ,*
1 SLM1
L2
Ll
P2
fl i1
uSLM2
= QWP
!
Faraday Rotator
™ QWP
Non-PM Fiber
3 PBS
Coupling Optics
r r
SLM3I
m P3
V ja III
l2
Non-PM Fiber Bit
SLM4
= QWP
M12
M4
BIT 1
BIT 2
> - OUT
SLM6
QWP
BITN
Figure 3.12: The proposed compact reflective geometry jY-bit switched photonic delay line for microwave
signal processing. This structure uses free space (bit 1), solid optics (bit 2), and non-PM fiber (bit N) delay
lines and a novel polarization noise reduction scheme. The fiber delay design passively compensates for all
externally or internally induced fiber birefringence effects, thus maintaining a high quality linear
polarization at the optical switching and redirection planes o f the switched delay line.
The free space delay line (bit 1, in Fig. 3.12) works as follows. Vertical or spolarized light passes through the first polarizer (vertical polarizer). SLM1 acts as a
switch to either change the polarization to horizontal or /7-polarization when it is set in its
“ on” state, or leaves the input polarization unchanged if it is set in the “ off” state. The
cube PBS (PBS1) acts as a path selector, directing light depending on the polarization of
the incident beam. When /7-polarized light hits the PBS1, it travels straight through
towards the SLM2. SLM2 is set in the “ on” state and thus changes the incident ppolarization to 5-polarization, which then passes through the second vertical polarizer
towards the second bit o f the PDL. On the other hand, when SLM1 is set in its “ off’
state, the light stays 5-polarized, and is deflected from PBS1 towards the quarter wave
plate (QWP), which has its axis at 45° with respect to the incident 5-polarization. When spolarized light hits the QWP, it changes to circular polarization. After reflection from the
mirror (M l), this light passes through the QWP again, changing to /7-polarized light.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Then this light passes through PBS1 and hits another QWP, that has its axis at 45° with
the incident /7-polarization. After reflection from the mirror (M2), the beam is redirected
through the QWP and becomes 5-polarized. This 5-polarized light is deflected by an angle
o f 90° from the PBS1 towards the SLM2. This SLM is set in the “ o ff’ state, which does
not change the incident polarization. 5-polarization passes through the vertical polarizer
(P2) towards the second bit o f the PDL.
The second bit in Fig. 3.12 consists of a solid optics delay line. The retracing
procedure o f the light is the same as for the first bit. The only difference now is that
instead o f free space propagation in the delay path, light propagates through a solid optics
medium, with an index o f refraction “ r i'. The higher the index o f refraction of the
medium compared to the index o f refraction of free space, the longer the time delay to the
propagating light for the same travel distance.
The iV-th bit o f Fig. 3.12 consists of a non-PM fiber delay line. The use of fibers is
necessary when long (e.g., > 5 ns) time delays are needed. The use of PM fibers is
dictated in order to keep the high state of polarization (SOP) of the propagating light in
polarization switching based systems that use fibers. It is obvious that any change in the
birefringence o f the fiber segment will degrade the system performance drastically, since
the architecture is based on maintaining the high SOP of the propagating light. Our
proposed design does not need to use PM fibers. Instead ordinary, less costly, non-PM
fibers can be used in the system. Note that instead of a QWP used in the free-space and
solid-optics bits, we use a Faraday rotator (FR), with power of 45°, right after the nonPM fiber, for the fiber delay line. The propagation of the light through the delay path is
similar to the other delay bits. Later, we analytically show that the use o f a Faraday
rotator after the non-PM fiber can compensate for the random induced birefringence of
the fiber, that occurs due to the changing environmental conditions or the changes in the
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
optical wavelength, among other factors [54]. In the next chapter, we will describe and
analyze in more detail the fiber-delay line. The non-delay path propagation in the PDL is
the same as for the two previous cases, that involve the free-space or solid-optics bits.
As mentioned earlier the three different delay lines can be used to give a different
range of time delays, and hence are equally important for most variable delay line
applications. Using examples, we will calculate and compare the time delays for the three
cases. In general, the time delay for any given bit can be approximately written as:
t=
2 n Lx +2■ L2 + 2 •d PBS •nPBS + 4• dQV/p •nQWP + 2 •d pR ■%
(3.20)
where L x, and Z,2 are the free-space distances in the delay path (Fig. 3.12); n, nPBS, «QWP,
and rtpR, are the indeces o f refraction of the medium used in the delay path, the PBS, the
QWP and the FR, respectively, c is the velocity of the light in vacuum, and dpBS, dQW?,
dpR, is the side length of the PBS, the QWP and the FR, respectively. Note that if fibers
are used as the medium o f propagation in the delay path, long time delays can be obtained
without increasing the actual dimensions of the PDL. This happens because for instance,
a fiber of any reasonably long length (> 10 cm) can be easily wound up on a small -1.5
cm diameter reel that can be fit in a small volume in the delay line. Thus, depending on
the delay required, one delay medium and method is more appropriate than the other.
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.7 Active Noise Filter Experimental Demonstration
In the previous sections we proposed and briefly described the use o f an
additional polarization switch and a polarizer to suppress noise leakage due to the
polarization switch or the polarization components. The active noise filter (or noise
reduction) experimental setup for a 1-bit time delay unit is depicted in Fig. 3.13. Two
TBSs are used to form a crossed polarization setup. The extinction ratio (ER) in units of
decibels is defined as 10 log (max/min), where max/min is the ratio o f the maximum
output power in one linear polarization versus the minimum output power corresponding
to the other orthogonal polarization. The ER for the crossed polarization TBS setup is
50.94 dB, whereas the NLC devices LC1, LC2, and LC3 have optical on/off ratios of
34.52, 38.09, and 35.44 dB, respectively, s-polarized light from a 13 mW He-Ne laser
passes through TBS1 to enter the PDL. All NLC devices are inserted with their nematic
director at 45° to the incident polarization. These NLC devices are parallel rub
birefringent-mode devices and are driven by 0-5V, 1kHz square-wave signal. LC1 and
LC2 act as optical switches, changing the linear SOP when they are “ on” while letting
the same polarization through when they are “ o ff’. LC3 is set to act as quarter wave
plate. Note that LC1 and LC2 always act out of phase with each other; that is, when LC1
is “ on” (rotating the input polarization) then LC2 is “ off’ (no rotation). When LC1 is
“ off” the beam enters the delay line, emerges in the s state, and then passes through the
LC2 (now “ on” ) to enter TBS2 in thep state. The optical power measurements for both
the non-delay and the delay paths of the PDL are made with a Newport Model 815 digital
power meter.
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LC3
—
-
I
f
He-Ne laser U
TBS1
P(v)
1
1
/
7
"
Detector
F BS
LC2
LCl
TBS2
P(h)
QWP
“ M2
Figure 3.13: The experimental setup for the PDL noise reduction scheme. The NLC devices are driven by
0-5 V, I KHz square wave signal. The “ on” or “ off’ operation can be selected by setting the Vp drive level.
The electrical SNR and optical ER are shown in Table 3.1. Very good results were
obtained for the electrical SNR and the optical polarization ER of the system using the
noise reduction technique. Remember that the electrical SNR is defined as 201og(signal
optical power/noise optical power). Where signal optical power is the power in the optical
beam o f the desired polarization that travels through the set delay or non-delay path of the
PDL, all other optical power measured at the output is regarded as the noise optical
power.
The key results from Table 3.1 is that, for the case in which LC2 and TBS2 are
not present, i.e., no noise removal/suppression is used, the PDL performance deteriorates
drastically. This is particularly severe for the delay path because the cube PBS
performance is poorer at its deflected port compared with its straight port, and so some
optical ER improvement technique must be used in the PDL to retain a high SNR. Note
that TBS2 by means of its deflected port does an excellent job of rejecting the s-polarized
noise optical power from the PDL, thus producing the high (>92 dB) electrical SNR
values.
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.1: Electrical signal-to-leakage noise ratio and optical extinction ratio measurements in dB for the
delay-path and the non-delay-path o f the optical delay line with and without the noise reduction scheme.
Electrical SNR (dB)
Optical ER (dB)
With Noise
Without Noise
With Noise
Without Noise
Reduction
Reduction
Reduction
Reduction
Non-Delay
96.90
49.40
50.75
24.22
Delay
92.40
28.88
50.43
14.77
Path
3.8. Conclusion
In conclusion, we have described different photonic delay lines that cover different time
delay ranges from ultra-short delays using the symmetric architecture to very long using
the reflective architecture. A polarization-based noise reduction technique for PDLs has
been introduced. The noise suppression technique was experimentally demonstrated with
a setup based on NLC optical switching devices, a cube PBS, and high-quality TBSs.
High electrical SNR (>92 dB) and optical ER (>50 dB) values were obtained with this
noise suppression technique.
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4
FIBER OPTICS-BASED PHOTONIC DELAY LINE
4.1 Introduction
Certain applications require long time delays that are impractical to achieve with free
space or solid optics. An important issue with our three-dimensional (3-D) photonic time
delay architectures is mechanical stability and size, particularly in cases where long time
delays (>5 ns) are required for the most significant bit (MSB) in the binary delay line
architecture. A shorter/compacter optical delay path for time delay implementation was
proposed and demonstrated in the previous chapter. This PDL architecture is based on a
novel reflective geometry photonic delay line design [1, 2], Using this reflective design,
with the special polarization noise reduction scheme and a fiber-bireffingence compensation
method [2], it becomes possible to use free-space, solid-optics, and non-PM fiber-based
optical delay paths that are smaller, lighter and more compressed (in volume) than previous
transmissive designs. As such, we can expect greater mechanical stability from our new
reflective designs. Such applications are radar simulation and testing, RF transversal
filtering and multipath cancelation, and true time delay phased array antenna control. Unlike
the free-space delay medium used in previously demonstrated polarization-based PDLs [3]
optical fibers must be used for providing the longer (> 5 ns) time delays. It is well known
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that systems based on polarization cannot tolerate changes in the state of polarization (SOP)
because this leads to unwanted effects such as signal fading, loss in optical heterodyning
efficiency in interferometric systems, an increase in unwanted noise sources, and overall
higher loss in the system. Systems that use fibers to carry the information coded optical
signals in the time delay paths, must use PM fibers to retain the high degree of SOP.
Nevertheless, any changes in the PM fiber’s optical birefringence caused by the changing
environment or even the change in the optical wavelength can degrade the desired SOP of
the travelling light. Moreover, PM fibers cost a lot more money than the regular telecom
grade single mode fibers. Multichannel delay line applications, that use many fiber
segments with cascading architecture, increasing the cost of the system as well as the noise
sources. Therefore we cannot tolerate any major changes in SOP caused by the fiber
segments.
The use of a passive technique for compensation of polarization changes induced by
any optical medium’s birefringence on a propagating light beam was proposed by Martinelli
[4]. In this passive technique an optical beam-retracing geometry coupled with a Faraday
rotator-mirror (FRM) setup was used to eliminate the effects of the changing material
birefringence on the input and output SOP. In fact, today many fiber-optic isolators use a
similar mechanism to prevent unwanted feedback in many optical systems. This chapter
describes how the birefringence technique proposed by Martineli can be incorporated to
form a high-SNR non-PM fiber-based birefringence-compensated photonic delay lines [2].
In the following sections the novel birefringence compensation technique theory
and its experimental demonstration is presented [2]. The “vital” noise reduction scheme is
also incorporated in the PDL for obtaining high performance multichanel polarization based
fiber delay lines. This birefringence compensation technique is based on the optical beam
retracing geometry coupled with the use of a Faraday rotator-mirror system which
eliminates the effects of the changing material birefringence on the input and output SOP.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2. The Fiber Optic Photonic Delay Line
Figure 4.1 shows the top view of the proposed W-bit M-channel switched fiber
delay line for the case of N = 3 bits and M = 16 channels. Here 16 linearly polarized and
collimated beams enter the system and pass through a 4 x 4 pixelated array of a
polarization-mode SLM such as NLC, or FLC SLM. When turned “on”, each SLM pixel
has the capability to rotate the input beam linear polarization by 90°. In the “off” state the
pixel does not rotate the input beam polarization. The SLM combined with the cube PBS
forms a multichannel optical switch, whereby, depending on the SLM setting, beams are
directed toward the fiber paths or straight through to the next PDL bit. For the case of an
input horizontal or p-polarized beam with the SLM pixel “on”, the vertical or ^-polarized
beam leaving the SLM is deflected 90° by the PBS into a fiber-coupling lens, such as a
gradient-index (GRIN) lens, that is connected to a single-mode non-PM fiber of a certain
desired length. The j-polarized light input into the fiber leaves the fiber through another
fiber lens. At this stage, depending on the fiber birefringence that the traveling light has
suffered (this depends on the individual fiber and the external conditions), the light leaving
the fiber is no longer s-polarized. This perturbed light passes through a Faraday rotator to
strike a mirror that reflects the light back through the Farraday rotator (FR) and the fiber. It
turns out that this beam-retracing operation, coupled with the FR and mirror, induces
symmetry properties on the Poincare sphere rotations, causing the fiber entrance and exit
SOPs to be always orthogonal points on the Poincare sphere [4]. Thus, if ^-polarized light
enters the non-PM fiber, then p-polarized light leaves the fiber on the return path,
regardless of the birefringence effects in the fiber and the fiber-coupling optics. This
preservation of the linear SOP is critical for operation of the proposed switched delay line.
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Note that the p-polarized light leaving the fiber after the mirror reflection travels through the
PBS to be retraced through a quarter-wave-plate (QWP) - mirror arrangement. The light
returning from this mirror is s-polarized and is reflected by 90° by the PBS to enter a SLM
high-extinction-polarizer combination. This combination forms the polarization noise
suppresion filter (active noise filter) discussed in the previous chapter. In this way the
unwanted noise signals can be rejected and suppressed in the system.
Bit 1
Bit 2
Bit 3
T s ss sss sss ss J 'y s's sss sss ss s>
J 's s s s s s s y s s s s s J ///J J /y 7 /‘ ‘S m
Fiber L ens
-Mirror
Tp*" Faraday Rotator
*"0 Q Q 0
Non-PM Fibers
Beam Array
QQ
16 delayed
signals
f
SLM
i
w
i
qtjt
\_____,
\
/
^ Mirroi
Mirror
Noise Reduction SLM and Polarizer
Figure 4.1: Top view of the proposed fiber birefringence-compensated N-bit Af-channel switched fiber PDL.
(QWP: quarter wave plate; P: polarizer; SLM: spatial light modulator)
4.3. Non-PM Fiber based Photonic Delay Line Theory
In this section, the theoretical analysis of the Faraday rotator-mirror optical setup
that can compensate for the random birefringence effects on the polarization state of a beam
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which retraces its path, such as in the reflective geometry fiber delay line, is performed.
Theoretical analysis will show that the entrance and exit polarizations for the delay segment
of the bit structure are always orthogonal to each other, regardless of the retardation the
beam suffers when traveling through a general retarder (e.g., a non-PM fiber). The
analysis to follow for our PDL is based on matrix operators, as the serial cascaded nature
of our photonic delay line architecture translates simply to basic matrix operations, i.e.,
product of Jones matrices, for convenient mathematical system analysis.
The matrix for a Faraday rotator can be written as [5]
KI=
~ cos0
—sin0
sin0'
cos0
(4.1)
where 0 is the power of the Faraday rotator. This holds true for the incoming beam from
the PBS. For the beam coming back, after reflection from the mirror, the Faraday matrix is
[«r]=
"cos0 -sin0"
sin0 cos0
(4.2)
Thus, for the Faraday rotator-mirror-Faraday rotator optical assembly with a
Faraday rotator power 0 = 45°, the Jone’s matrix is given by
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
-sin 0
cos0
i
sin0
1_
(4.3)
0
0
cos0
o
-s in 0
cos0
i
"cos0
sin0
o1
1
[FRM} =
A general optical retarder R with a retardation T given in radians rotated by an angle <|>
relative to the x-axis in the photonic delay line (see Fig. 4.2) can be expressed as [5]
*(<P) =
'*,.(0)
*«(♦)
*22(0).
cos20 eJ^ + sin20 e~j^
j sin 20 sin y
y sin 20 sin y
sin20 eJ^ + cos20 e~j^
(4.4)
Note that in our PDL case, where we are using a non-PM fiber that can undergo
various externally or internally induced optical birefringence effects that change the input
linear SOP, the general retarder R simulates this random non-PM fiber medium, with T and
0 a function of time t, i.e.,
T=r(f) and
0 = 0 (t). Thus, the operator [J(t)] for the
reversible optical delay path in the single bit delay structure that includes the random
retarder, the FR, the mirror, and back through the FR, the random retarder again, can be
expressed as
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Slow axis of the random birefiingent
medium perturbation
Fast axis of the random birefringent
medium perturbation
f
Cross-section of photonic delay path
that contains the fiber delay medium
Light returning after
double passage through .
the delay medium
S'
Temporally varying angle
.
Light into the
delay medium
Figure 4.2: The geometry of a general retardation plate, such as a non-PM fiber in the delay path of the
photonic delay line. Angle <|> is the angle between the fast axis “f” of the retarder and the x-axis that is
parallel to the horizontal or p-polarized light The y-axis is parallel to the vertical or ^-polarized light
[7 (f)] =
[*H>)] [FRM] [*(<!>)] =
A
B
C
D
(4.5)
Using the [FRM] matrix results in Eqn. 4.3 and the equality of the /?,,(<j>) and /?22(<|>)
elements with the /?,,(-<)>) and
elements, respectively, as well as the opposite signs
of the R n(<j>) and /?21(<t>) elements with the /?,2(-<t>) and /?21(-<{)) elements, respectively, we
obtain
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A = D = -ysin2<|)sinyCOS2<t>exp(yy) —ysin2<|>exp(—yy)sin2<|>siny +
+ ysin2<J>sinyCOS2<t>exp(yy) + ysin2<|>exp(—y'y)sin2<|>siny =
(4.6)
= 0,
B = C = (sin2<(> exp(y'y) + cos2<j) exp(-jy)) •(-c o s 2 <j) exp(y'y) - sin2<>exp(-jy)) +
+ (-y'sin2(j)siny) (—y'sin2<|)siny)=
^
^
= - s in 4<{>-cos4<j>-sin2<()cos2(|) ex p (y T )-sin 2<|>cos2<{> exp(yT )-sin22<j>sin2y =
= -
1.
Note that the trigonometric identities shown in Eqns. 4.8-4.11 were used for the analysis in
Eqns. 4.6 and 4.7.
exp(yT) - exp(-yT) = 2cos T ,
(4.8)
cos2<j>-t- sin2<|>= 1,
(4.9)
2 sin <f>cos <t>= sin2<j>,
(4.10)
sin2y =y(l-c°sr).
(4.11)
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Thus, the [/(t)] matrix for the reflective non-PM fiber delay path is given by
'
0
-1
-
1
0
'
(4.12)
First note that [/(/)]=[/!, or the matrix is fixed and independent of time. This
implies that the structure of this matrix guarantees that the exit polarization state is always
orthogonal to the entrance state, regardless of the birefringence effects the traveling beam
suffers at any given time. For instance, if the input to this delay-path is a vertically
polarized light beam (as in our non-PM fiber delay), then the returning output beam is
always horizontally polarized. This can be represented mathematically as
"0 - f "O'
-1 0 1
T
0
(4.13)
where [£out] and [Ein] are the output and input light field vectors, respectively. Thus, the
Faraday rotator-mirror arrangement can be used to provide the proposed birefringence
compensation vital for the high performance operation of PDL that includes non-PM fibers,
or for that matter any random birefringence perturbations. In our previous analysis, it is
assumed that the temporal random birefringence fluctuations happen over a time scale that
is much longer than the temporal delay introduced by the delay medium, such as the non58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PM fiber. In otherwords, both the input optical beam and the output optical beam from the
delay path must suffer the same birefringence perturbations for this passive birefringence
compensation technique to work. In our phased array antenna application case, it is
expected that these perturbations will occur at a slow rate (e.g., over several seconds), such
as slow variations in optoelectronic module package temperature and mechanical stress, or
slow changes in humidity. Thus, for our application, this passive compensation method is
adequate.
4.4. Birefringence Compensation Experimental Demonstration
As mentioned above, fibers change birefringence and hence optical polarization
properties of a traveling optical beam, leading to noise generation in a polarization-based
delay line. In this paragraph the birefringence noise compensation technique is
experimentally demonstrated. As mentioned earlier, after the fiber, depending on the fiber
birefringence the light has suffered, the fiber output light is no longer vertically polarized.
Fig. 4.3 shows the fiber birefringence-compensation experimental set-up for a single-bit
delay line. A FR with a rotation power of 45° at 633 nm, is inserted in the delay path.
Instead of using a non-PM single mode fiber with externally induced stress, for proof-ofprinciple results and accuracy of induced birefringence noise we perform an experiment
with a parallel-rub birefringence-mode NLC device LC3 that has an ellectrically controllable
birefringence. LC1 is set in the “o ff’ state, and LC2 in the “on” state. LC3 has its NLC
director at 45° with respect to the incident 5-polarization.
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
v? r/r/m m i
Faraday Rotator
P(j-polarization)
ULC3
Detector
He-Ne laser
0
LC1
PBS
LC2
P(p-polarization)
3 QWP
TZZZZZam
Figure 4.3: The birefringence compensation experimental setup. LC3 simulates a variable birefringence by
controling the driving voltage of the NLC device (LC3).
A 0-7C total (round-trip) birefringence corresponds to a voltage change from 4.11 V
to 5.77 V on LC3. A series of measurements of the signal and noise power were taken as
well as measurements of the p- and ^-polarization optical power for the birefringence
compensated delay path output beam. Electrical SNR measurements and optical ER were
obtained as a function of the total induced birefringence noise, and are shown in Figure
4.4. The average value of the electrical SNR was SNR = 98.35 dB, while the average
SNR variation was A(SNR) = ± 0.584 dB. The average optical ER was ER = 39.20 dB
and the average ER variation was A(ER) = 0.0481 dB. These results indeed show that the
birefringence-compensation technique based on a reflection geometry with a Faraday
rotator-mirror system negates the unwanted polarization effects of a variable birefringent
medium such as an optical fiber in the delay path.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
[02
S N R = 9 8 .3 5 dB
A SN R = ± 0.584 dB
O
Q
-a
o
•3
0a£
41
40
1
94
§
92
|
39
E x tin c tio n R a tio
38
90
E R = 3 9 .2 0 dB
A E R = ± 0.0 4 8 1 dB
37
88
86
36
35
84
3.1
3 .3
3 .5
3 .7
3.9
4.1
4 .3
4.5
4 .7
4 .9
5.1
5 .3
5 .5
5 .7
L iq u id C r y s ta l A pp lied V o ltag e V p (V )
-I
E q u iv a le n t T o ta l In d u c e d B lre frig e n c e
N o ise ( r a d ia n s )
Figure 4.4: Optical Extinction Ratio and Electrical Signal-to-Leakage Noise Ratio vs. the LC3 applied
voltage and the equivalent total induced birefringence noise.
The birefringence compensation technique was also tested using a non-PM single
mode fiber. The LC3 of the previous experiment was replaced by a 633 nm single mode
fiber of length 1.76 m (Figure 4.5). A GRIN collimator-lens is attached to one end of the
fiber, the other end is cleaved. The light coming from the PBS was coupled into the fiber
using an objective lens (40x, numerical aperture: NA = 0.65, focal length = 4.3 mm).
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
W Z gm Ml
Faraday Rotator
GRIN lens
^
Non-PM fiber
Objective
LC1
Detector
He-Ne laser
P(/7-polarization)
QWP
M2
Figure 4.5: The fiber birefringence compensation experimental setup. (QWP: quarter wave plate; PBS:
polarization beamsplitter, LC: liquid crystal switch; P: polarizer; M: mirror).
When stress is applied on the fiber, the output power is not affected. It remains at
the same level with a variation of ± 0.04 mW. Different kind of stresses and temperature
changes were tested, such as squeezing the fiber with fingers, pressing the fiber against the
optical table with hand or metallic plate, holding the fiber with hands that had been warmed
by rubbing, or just placing the fiber on the cold surface of the optical table. Table 4.1
shows the measured electrical SNR and its variation for the experimental PDL under the
above conditions. In order to show the importance of the Faraday rotator in the system, the
rotator was replaced by a NLC device, that acts as a QWP, with its NLC director at 45°
with the incident j-polarization. We immediately noticed that the QWP-mirror configuration
cannot compensate for the birefringence noise and the system performance deteriorates
drastically. Several stresses and temperature conditions were tested for this case as before.
The output power of the PDL drastically changed depending on the external conditions.
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The electrical SNR drops significantly and varies from 40 dB to 59 dB depending on the
induced fiber birefringence.
Table 4.1: SNR and average SNR variation of the output power for the 1-bit fiber delay line using the
birefringence compensation technique.
Stress and/or Temperature
change
SNR
(dB)
A(SNR)
Fiber hanging in air no
stress or temperature change
58.58
0.97
Hand stress
45.53
1.67
Pressing with metallic plate
59.07
1.35
47.92
1.59
(dB)
on the table
Fiber on the table-no stress
4.5. Microwave Band Demonstration of a Reflective Geometry Fiber and Free Space
Photonic Delay Line
In the previous sections, we studied and experimentally demonstrated a fiber-based
“single” bit photonic delay line that uses a novel birefringence compensation technique.
This experiment was performed using unmodulated light for aquiring appropriate optical
alignment experience and PDL characterization. In this section, a “2-bit” structure of the
compact reflective geometry delay line, using a 1 GHz microwave band optical modulation
of the input laser beam is demonstrated [6]. The use of microwave frequency optical
modulation with these time delay structures, gives us, for the first time, an inside look at
63
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
the issues and quality of RF signal processing provided by this reflective optical design.
Moreover, since these photonic delay lines are initially intended for microwave signal
processing and control applications such as phased array antenna control, we must study
system performance in the microwave domain, as this data is important information for an
antenna design engineer. Note that our demonstrated photonic time delay structure has both
a compact free-space delay bit and a non-PM fiber delay bit; thus giving us design insight
into two kinds of photonic delay structures.
Microwave frequency (i.e., 1 GHz) optical modulation via an acousto-optic (AO)
modulation system is introduced for characterization of the PDL in the RF domain using
RF spectrum analyzer measurements. This bulk-optics AO modulation system is
particularly suited for high optical power single input-single output photonic signal
processing systems such as transversal filters [7]. Another application using the high
power AO optical modulation system is the single input-multiple output system such as a
transmit-only phased array antenna used in broadcast-modes (e.g., satellite television
broadcast antennas). Fig. 4.6 shows how this AO modulation system can be combined
with our proposed PDL for use in these applications. Because AO devices can operate in
the RF and microwave band (e.g., < 3 GHz) domain, our AO modulation system using
two AO devices can provide an efficient means for microwave band optical modulation upto the lower C-radar band (i.e., < 6 GHz). Using either direct modulation of low power
(e.g., 4 mW) semiconductor lasers, or external modulation of moderate input power (< 100
mW) laser beams using integrated-optic Mach-Zehnder type modulators can be used with
our PDL to give higher than C-band optical modulation [8].
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
RF modulation at /
High Power Laser
CW or Pulsed
High Speed
Photodetector Array
Photomc
AO Modulation
System
-j-O
b-
RF Output
l:N
splitter
System
Amplifiers
Weighted
Summer
(a)
RF modulation at /
Light modulated at 2 /
High Power Laser
CW or Pulsed
AO Modulation
System
Broadcast Antenna
High Speed
Photodetector Array
Photonic
Time
Delay
System
l:N
splitter
TV
Microwave
Radiation
Amplifiers
Antenna Element
(b)
Figure 4.6: The high optical power AO modulation system being used for (a) single input-single output
microwave transversal filter and (b) single input-multiple output transmit/broadcast microwave phased array
antenna.
The experimental setup for the 2-bit 1-channel PDL is depicted in Fig. 4.7. Note
that as the first experiment for a cascaded 2-bit delay line, we choose to work with visible
632.8 nm light for ease in alignment and accuracy of optical signal, noise, and loss
measurements. The next question is, how do we optically modulate this visible light beam
at microwave frequencies. Recently, an interferometrically stable optical architecture that
can indirectly modulate visible light beams using AO devices has been recently developed.
This AO device-based optical architecture has been used to efficiently process RF signals
(e.g., 120 MHz) for a variety of applications that include phased array antenna control[911], signal correlation [12], convolution [13], spectrum analysis [14], and notch filtering
[15]. Fig. 4.8 shows a similar optical modulation architecture using two wide-band
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
acousto-optic modulators (AOMs) to obtain the desired 1 GHz microwave band optical
modulation of the visible light beam.
Ml
zzz
Faraday
Rotator
TIR prism /
22
M3
□Q W P (LC5)
LC1
PDL
Input Port
✓
TBS1
P(j-poIarizarion)
LC2
r. vertical polarization
pi horizontal polarization
GRIN lens
2 :1 imaging
LC3
High-Speed
Photodetector
12
V.
PBSl
r
Input light modulated
at/ = 1 GHz
Objective
TBS2
PCx-polarization)
PBS2
QW PC □
LC4
TBS3
PQ>-polarization)
~ QWP (LC6)
EZ222a
T O 7 //1
M4
M2
Bit 1
f=l GHz
L-band Time Delayed
Signal Out for driving
a Sub-array in a
phased array antenna
Bit 2
Figure 4.7: The experimentally demonstrated 2-bit, 1-channel switched photonic delay line setup using the
compact reflective geometry in the delay paths. The experiment is performed at a 1 GHz optical modulation
frequency using a 633 nm visible input light beam.
A 632.8 nm 10 mW continuous wave (CW) He-Ne laser is used as an input to the
AOM-based optical modulation system. To obtain the 1 GHz optical modulation that is
transmitted through the PDL, two AOMs are used for the experiment. As shown in Fig.
4.3, the He-Ne laser light is focused with a 10 cm focal length (FL) lens (LI) into the first
AOM1 that is driven by a 500 MHz RF signal. The input light to AOM1 is diffracted as a -1
order, negative Doppler shifted (i.e., -500 MHz) optical beam. A 15 cm FL lens (L2) then
collimates the two beams (DC and -1), which are focused into the second AOM2 using
another lens (L3) with a focal length of 15 cm. AOM2 is also driven by the same 500 MHz
RF signal. The DC light from AOM1 is diffracted as a +1 order, positive Doppler shifted
(+500 MHz) beam by AOM2. Part of the -1 order from AOM1 travels through AOM2,
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
becoming collinear with the +1 order from AOM2. The +1 and -1 orders remain collinear
after AOM2. The net Doppler shift that should result from the beating or heterodyne
detection of these +1 and -1 order beams is 500 MHz-(-500 MHz) or the required 1 GHz
frequency. A 20 cm FL lens L4 is used to collimate these +1 and -1 order light beams
before entering the PDL. A 1 GHz (frequency modulated) CW light beam enters the PDL,
as is required for this proof-of-principle PDL test experiment.
CW RF signal at
f=500 MHz v
DC
DC Block
He-Ne
Laser
+1 & -1
AOM2
DC
AOM1
L2
L3
Optical Beams
Input to PDL
LA
1 GHz Optical
Modulation
Figure 4.8: The acousto-optic modulator-based light modulation technique for obtaining the 1 GHz
modulated optical input signal for the PDL.
The components used in the experiment are described in detail, as their specific
performance will allow us to determine the overall performance of the PDL. These
components include (a) Parallel-rub birefringent mode NLC devices LCl, LC2, LC3, LC4,
LC5 and LC6 that are inserted with their NLC directors at 45° with the incident s or p linear
polarization. These devices act as high performance polarization rotators. LC5, LC6 act as
QWPs, rotating the polarization by 90° when light passes twice through them, (b) Two
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
broadband (400-700 nm) cube PBSs arc used from Newport Corp. with a specified
extinction ratio (p/s) for the non-delay port of >1000:1. The extinction ratio (s/p) for the
reflected port was measured as <50:1. (c) Three TBSs with a high extinction ratio of
>10,000:1 at both ports, (d) 40 cm and 20 cm focal length biconvex spherical lenses were
used for 2:1 imaging of the beam for better coupling efficiency into the cleaved fiber face.
(e) Objective lens (x40, NA = 0.65, FL = 4.3 mm) and a non-PM single mode fiber at
632.8 nm are used. A GRIN collimator-lens is attached to one end of the fiber and, the
other end is cleaved.
Based on the measured transmission/reflection efficiencies of the
optical
components listed in Table 4.2, the expected and measured optical losses for the four
settings of the PDL are shown in Table 4.3. For the delay paths, a higher loss is anticipated
due to the greater number of optical components used in this path, which introduce higher
attenuation.
The L-band signal modulation is photo-detected using a high speed photoreceiver.
Spectrum analyzer measurements (see Fig. 4.9) are used to determine electrical loss values
o f -9.9 dB, -21.0 dB, -40.3 dB, -41.7 dB, for the first, second, third and fourth setting,
respectively. Since the spectrum analyzer gives direct electrical power measurements, we
have to divide the above measurements by two to get the optical power values, that except
for the fourth case, are very close to the expected optical losses. All expected loss values
are calculated assuming that all the input light energy travels through the selected optical
paths in the delay line. In the experimental scenario, the finite on/off optical switching
ratios for the SLMs causes some input light to travel through the unselected paths. For the
four different delay line settings, this unwanted leakage light, that adds to the final output
desired light energy, can have different energy levels. In
our fourth case/setting, this
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
leakage light is relatively high enough compared to the desired path light, indicating a
higher observed optical power than expected. Hence the 4.6 dB difference.
Table 4.2: Measured transmission/reflection efficiencies of the optical components.
Optical component
Efficiency
Optical component
(%)
Efficiency
(%)
LC1
82.55
LC2
88.89
LC3
88.80
LC4
91.10
LC5
89.97
LC6
88.00
QWP
96.00
Fiber coupling system
TIR prism
92.97
1st bit non-delay
25.0
LI
90.80
1st bit delay
22.0
L2
92.20
Faraday Rotator Mirror
96.0
Lenses
PBS 1
PBS2
non-delay port
88.7
reflected port
97.0
non-delay port
82.0
reflected port
89.0
TBS
90.0
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4.3: Expected and measured optical loss for the four settings of the PDL.
PDL Settings
Expected Optical Loss*
(dB)
Measured Optical Loss**
(dB)
-4.56
-4.95
-7.08
-10.50
-20.77
-20.15
-24.43
-20.85
Is*Bit - non-delay
2ndBit -non-delay
1st Bit - delay
2nd Bit -non-delay
Is*Bit -non-delay
2nd Bit - delay
Is*Bit - delay
2nd Bit - delay
* Expected loss data were calculated using measured optical component efficiencies.
** Experimental data were calculated using RF spectrum analyzer measurements.
Optical SNR measurements were obtained for all the different settings of the PDL.
For the case when the bits are set for the non-delay path, the leakage noise comes from the
delay path. In order to separate the leakage noise from the signal, the leakage noise must be
deflected, so that the signal and noise do not overlap spatially on the high speed output
detector. By tilting the mirror (M2 for the first bit, or M4 for the second bit) the noise can
be deflected. An iris is then used to let only the noise hit the measurement detector. For the
case when the bits are set for delays, the noise comes from the non-delay paths. In this
case, the signal light travels in the delay path, and can be blocked, letting only the noise
power hit the detector. Subtracting the noise power from the total output power of the bit,
the signal
power can be calculated, and thus the value of the optical SNR can be
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
determined. The key optical/RF noise introduced by our PDL to the L band signal was the
optical switch leakage noise and the highest leakage noise was measured as -117 dBm,
when the first bit was set for delay and the second bit was set for the non-delay path.
O ✓S
1 0 0 . ft CS
«
\ . CMS !
i Y
(a)
D2/
(b)
Figure 4.9: (a) Spectrum analyzer trace of the 1 GHz signal modulation measured at the entrance of the
PDL, using the high speed photodetector and (b) spectrum analyzer trace at the output of the PDL for the
first setting (where both bits are set for the non-delay path setting), indicating the -9.9 dB electrical loss of
this setting. Notice the similar spectrum analyzer floor for both readings, indicating no extraneous frequency
pick-up.
Table 4.4 shows that the noise reduction scheme is “vital” for generating high
SNRs from the PDL. Moreover, the cascading architecture of our noise reduction scheme
reduces the noise transmitted from bit to bit in the N-bit delay line structure. Hence, noise
originating from the first bit time delay structure is rejected by the noise reduction scheme
and will not be transmitted towards the next time delay bit. In our experimental case, for the
no-delay case of the second bit, the SNR improvement is not as high because the noise
level is very low due to the high noise energy attenuation in the fiber delay path.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4.4: RF SNR for all the different settings of the PDL with and without the noise reduction scheme.
Electrical SNR at 1GHz
without noise reduction
(dB)
Electrical SNR at 1GHz
with noise reduction
(dB)
1“ Bit - non-delay,
49.30
89.80
A
1“ Bit - delay, A
24.80
75.40
(1st Bit - non-delay),
82.00
83.00
B
2nd Bit -non-delay
(1” Bit - delay), B
2ndBit - delay
(Is* Bit -non-delay),
64.00
62.50
23.68
48.40
49.40
52.33
PDL Setting
2nd Bit -non-delay
B
2nd Bit - delay
(1“ Bit - delay), B
A: Measured at the output of the first bit; B: Measured at the output of the second b it
4.6. Conclusion
A novel fiber birefringence-compensation technique for switched PDLs has been
proposed. Theoretical analysis was performed and showed that the birefringence
compensation technique can indeed compensate for the induced birefringence. A proof of
concept experiment was demonstrated for a 1-bit optical delay structure that yielded high
optical ER (> 98 dB) and electrical SNR (> 98 dB) values. We have also demonstrated at
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
L-band a 2-bit, 1-channel, NLC-based PDL, using the novel compact reflective geometry
for one free-space and one non-PM fiber delay path. The PDL had a base spectrum floor of
less than -128 dBm, and a highest leakage noise level of -117 dBm introduced by the NLC
optical switches and the PBSs. Measurements of SNRs were obtained for all the different
settings of the PDL, with and without the use of the novel and necessary noise reduction
scheme. Worst case RF SNRs of around 50 dB were obtained. High losses were present in
our test system due to the non-AR coated optical components, and more importantly, via
the objective-to-fiber coupling system. These losses can be minimized by the use of ARcoated optical components and better fiber coupling optics, giving higher signal power and
thus higher SNR. We have shown that using cube PBSs, it is possible to construct
compact, reflective geometry delay paths, provided the proposed polarization noise
reduction scheme is used at the output of each bit. We have shown how non-PM fibers can
be used for the longer time delays while maintaining linear SOP as required in a
polarization switching based delay line.
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5
CHARACTERIZATION OF A FERROELECTRIC LIQUID CRYSTAL
BASED TIME DELAY UNIT FOR PHASED ARRAY ANTENNA
APPLICATIONS
5.1. Introduction
Various PDL architectures based on optical polarization switching 2-D pixelated
SLMs for phased array antenna control were described in the previous chapters and in
references [1-3]. As we mentioned these PDLs need to satisfy both high optical on/off
isolation (e.g., > 30 dB) at both output ports of the switching planes, and high switching
speeds that meet the requirements for the beam scan rates of the radar. Our initial work was
focused on the mature NLC technology for SLMs in these time delay systems. The parallelrub birefringent mode NLC devices can provide a moderately fast switching time of 1.5 ms
by using the transient nematic effect, while maintaining high > 30 dB switch on/off levels
that are typically required for high performance phased array systems [4], For typical high
performance radars, beam scanning rates of a 1,000 beams/s corresponds to a 1 ms
transmit/receive beam dwell time [5]. Using a time-multiplexed antenna beam scanning
technique suggested in ref. 6, two independent system channels are used in the photonic
time delay control system, and only the beam dwell time limits the beam scan rate. Thus,
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
using NLC devices with at best a 1 ms switching time, up to 1,000 beams/s scan rates are
practical and achievable. For higher beam scan rates, such as required in advanced radar
applications [7], faster optical switching devices are required. Ferroelectric liquid crystals
(FLCs) with reported switching times as low as 1 |is [8] are currently an excellent
candidate optical switch technology for higher beam scan rates.
In this chapter, we show how FLC devices, when combined with certain optical
noise reduction schemes [9, 10], can give us the desired advanced wideband radar high
performance PDLs, where both fast switching speeds (e.g., < 35 ps) and very high optical
on/off isolation (e.g., 45 dB) can be achieved at both output ports of the switching fabric
[11, 12]. This is unlike all previous FLC switching fabric designs that have been unable to
simultaneously provide both very high (> 45 dB) optical on/off isolation and moderately
fast (< 35 ps) switching speeds at both output ports [13-19].
Specifically, we show that unlike current state-of-the art NLC devices, FLC devices
do not produce very high optical polarization extinction ratios (e.g., > 40 dB) for both the
“on” and “off” modes of the device. This problem is not an issue for strictly on/off optical
shutter type FLC applications where the device is sandwiched between two high extinction
ratio fixed crossed polarizers. On the other hand for switching fabric-type uses such as our
PDL, this FLC extinction ratio problem leads to high system leakage noise. Previously, for
a NLC-based PDL, a simple noise reduction technique was used to reduce leakage noise
caused by the relatively large output port extinction ratio non-uniformities in dielectric
multi-layer thin film cube PBS used to build the PDL [10, 20]. In this chapter, we use the
same filtering technique to suppress noise caused by the current state of the art FLC
devices. The rest of the chapter describes the theory and experiment for achieving the
desired very high on/off isolation and high speeds from the three dimensional PDL.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.2. The Ferroelectric Liquid Crystal Principle of Operation
Before we describe the FLC-based PDL, a description of the principle of operation
of the FLC devices will be presented.
Ferroelectricity in chiral smectic liquid crystal phases was first demonstrated by
Meyer et.al. in 1975 [21]. Molecules in smectic C* medium are periodically disposed along
one axis (e.g., axis z). Each molecule in the layer is tilted with an angle 0,, characteristic of
the FLC mixture. All the available positions of the molecules in one layer define a cone, as
shown in Fig. 5.1. Due to symmetry when the molecule is chiral, a spontaneous
polarization P exists [22]. This vector is perpendicular to the molecule and contained into
the layer plane. Thus, all the available directions of P are tangent to the circle of intersection
of the cone with the xy-plane. In the smectic chiral LC there is a precession of the
molecular director induced by the chirality of the medium. So from layer to layer the
molecular director, and hence the polarization, slightly rotate about axis z. Since the
molecules rotate about the helix with their polarization pointing in all directions, the dipole
moments cancel each other. Hence, a macroscopic sample of smectic chiral LC is not
ferroelectric. Clark and Lagerwall [8] fabricated the first macroscopic ferroelectric LC
switch by suppressing the chiral LC helix. This is accomblished by fabricating very thin
layers of smectic chiral LCs, which now due to the boundary conditions imposed by the
alignment layers are strong enough to supress the helix. This type of liquid crystal is called
surface-stabilized ferroelectric liquid crystal (SSFLC).
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cone of the available
molecular directions in the layer
f
-
Tilted molecule in the
Figure 5.1: Position of the smectic chiral LC molecule in the layer.
Glass Substrate
Electrodi
ElectrodeGlass Substrati
Smectic C* Layers
Figure 5.2: The surface-stabilized FLC configuration first demonstrated by Clark and Lagerwall. n is the
molecular director, P the ferroelectric polarization, and z the layer normal.
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 5.2 shows the SSFLC configuration. Smectic layers are perpendicular to the
substrates and the molecular director ( n) moves along the surface of a cone, whose axis is
normal to the layers and parallel to the cell substrates [23]. For a SSFLC cell the molecular
director can be switched electrically between two uniform orientations, both of which are in
the plane of the LC film. The switching of the director is due to the interaction of the
polarization P perpendicular to the director and the applied electric field E. These states are
commonly refer to as the up (+x direction in Fig. 5.2) and the down (-x direction in Fig.
5.2) states, according to the orientation of the spontaneous polarization P. The two states
of the molecular director are separated by an angle of 20,, where 0, is the tilt angle of the
molecules within the smectic layers.
If the tilt angle, 0,, is 22.5°, this gives an angle of 45° between the two molecular
director states (or the two optical axes positions). Since the FLC molecules are birefringent,
and because of the two possible states of the molecular director, an electrically switched
half wave plate (HWP) can be realized. This can be accomplished if the right cell thickness
is selected for a specific wavelength. The principle of operation of the electrically switched
FLC device used as polarization switches is shown in Figure 5.3. Fig. 5.3(a) shows the
“off’ state of the FLC device where the optical axis of the cell is alligned parallel to the
input polarization. For the “o ff’ state of the FLC device the input polarized light “sees” one
index of refraction and thus passes through the FLC device without any effect on the SOP.
For the “on” state of the FLC device, obtained for the reversed bias voltage, the optical axis
rotates by 45°. The input polarization is now considered to consist of two orthogonal
components. The first one parallel to the optical axis, and the other one perpendicular to the
optical axis. These two polarization components “see” two different indeces of refraction.
With the appropriate selection of the FLC cell thickness, the phase difference between the
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
two polarization components can be n. Thus, after propagation of the light through the
FLC cell the SOP changes by 90°. Typical voltage magnitudes required to electrically
switch the optical axis of the FLC are between ± 5 V [24] and ± 24 V [25].
3-D View
F r O I lt V i e w
Rubbed Optical Axis
Optical Axis when A
FLC is “off*
Optical Axis
*
^Rubbed Optical
t
IN
OUT
FLC: “off”
Rubbed Optical Axis /
Rubbed Optical Axis,
/ 22.5
Optical Axis
Optical Axis when
FLC is “on”
E oul
B o u tj
OUT
IN
■+5V
FLC: “on”
Figure 5.3: Principle of operation of the FLC polarization switch, (a) “off’ state, (b) “on” state.
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.3. The Ferroelectric Liquid Crvstal-based Photonic Delay Line
A PDL based on FLC devices and TBSs as well as an active noise filter is described
in this section. As mentioned earlier, the FLC devices are based on the electro-optic
principle of a switchable HWP. Each device consists of a thin layer (e.g., < 2 Jim) of FLC
material sandwiched between two glass plates. As mentioned earlier, the liquid crystal is a
uniaxial birefringent medium, with its optic axis oriented parallel to the liquid crystal
molecules. The optic axis of FLCs has two preferred directions that are separated by
approximately 45° and can be controlled by the polarity of the applied voltage across the
electrodes. Thus, the two states of the FLC device can be selected. The devices are driven
with a ± 5 V bipolar, DC balanced (i.e., a 10
square wave) waveform. A + 5 V with
respect to the reference will place the device in the “o ff’ state, and a -5 V with respect to the
reference will place the device in the “on” state [24].
The design and experimental setup of the FLC PDL is depicted in Fig. 5.4. Two
high contrast ratio polarizers PI and P2 (Melles Griot Model 03FPG 007, Dichroic sheet
polarizers) are used to form a crossed polarization setup. Horizontal or p-polarized light
from a 532 nm, 100 mW, diode pumped frequency doubled Nd:YAG laser, after
attenuation by a neutral density filter (ND = 20), passes through a HWP and changes to
vertical or ^-polarized light. The j-polarized light passes through the first polarizer PI
(vertical polarizer) and hits the first FLC device (FLC1). When FLC1 is set in the “on”
state, it changes the polarization from s to p-polarization, while when it is in the “o ff’ state,
it leaves the input polarization unchanged. The first TBS (TBS1) acts as a path selector.
When the light is s-polarized (FLC1: “o ff’), it passes straight through TBS1, towards
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TBS2 and FLC2 (straight path operation), which is set in its “on” state, and thus changes
the polarization to p-polarization. Then it passes through the output polarizer (P2), and hits
the photodetector. When the light is p-polarized (FLC1: “on”), it gets deflected by 45° by
TBS 1 (delay path operation). In the delay path operation, the light is retraced by a set of
two mirrors (Ml and M2) into TBS2, and after TBS2, follows the same path as in the first
case. FLC2 now needs to be set in the “o ff’ state so as not to change the incident ppolarization. Thus, for proper PDL operation, the two FLC devices operate in opposite
modes, i.e., when FLC1 is “on”, FLC2 is “o ff’ and vice versa. Note that if this single-bit
PDL is followed by a cascade of PDL, with increasingly larger delay paths by a factor of 2,
an N-bit photonic delay line system can be formed that can be used for phased array
control.
M2
Ml
Port 2
ND Filter
FLC2
Diode pumped
Nd:YAG laser
Photodetector
(s-polarization)
TBS1
P o rt 1
TBS2
(p-polarization)
Figure 5.4: The experimental setup for the photonic time delay unit using ferroelectric liquid crystal optical
switching devices. The dashed line represents the delay path, while the solid line represents the non-delay
path. (P: polarizer, TBS: Thompson polarization beamsplitters; M: mirror; HWP: half wave plate).
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.4. The FLC-based Photonic delay Line Experiment and System Issues
The FLC devices used in our PDL were Displaytech Inc., Model LV100 OEM
housing devices that are designed for shutter type applications where the device is placed
between two high contrast ratio polarizers. The FLCs used in the experiment were designed
for operating wavelength range of 400 - 700 nm, with center wavelength X = (510 ± 25)
nm, and are 25 mm active optical diameter devices [24]. In otherwords, the thickness of the
FLC material is chosen to give a retardation of K at the design center wavelength.
5.4.1. Optical Leakage Noise Reduction
For our PDL application, we need to use the FLC devices as polarization switches
and not as simple on/off light shutters. Thus, for our switching application, we are
interested in the polarization ER available from the FLC devices as well as the optical on/off
ratios at the output ports of the switching planes, as these numbers will ultimately limit the
performance of the PDL, particularly in terms of the switch-based leakage noise in the
system. The ER is defined as the ratio of the maximum output optical power in one linear
polarization (/„,„) versus the minimum optical power corresponding to the other orthogonal
polarization (/^ J . The ER in dB is defined as lOlogC/^/^,,). The optical on/off ratio is
defined as the ratio of the output optical power measured at the desired port when the
device is set for “on” operation versus the output optical power measured when the device
is set for “off’ operation.
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For our experimental PDL set-up, using a 532 nm CW laser, FLC1-TBS1 optical
switch had on/off ratios of 1:21 and 14,223:1 for port 1 and 2, respectively, while FLC2TBS1 optical switch had on/off ratios 1:20 and 12,000:1 for port 1 and 2, respectively (see
Fig. 5.1). Both device measurements are taken independently, by placing the test FLC
device at the position shown for FLC1 in Fig. 5.4. Thus, it is clear that these Displaytech
FLC shutter devices work adequately as switches for only one of the output ports of the
switch structure, namely port 2. The poor optical on/off ratios (-1:20) obtained at port 1 are
due to the fact that the FLC devices do not fully rotate the input light polarization by exactly
90°. Thus, when the FLC device is set for on operation, a strong component of the other
orthogonal polarization leaks towards port 1. In our experimental case, this makes the
polarization ER of the light after the FLC device to be as poor as 20:1. For the case of “o ff’
mode operation, the measured ER is as high as 10,000:1. Note that if the input light is ppolarized instead of s-polarized, port 2 (p-polarization port) will have the poor optical
on/off ratio and port 1 (^-polarization port) the high optical on/off ratio. In this case, when
the FLC is set in the “on” state, the p-polarized light will not be rotated exactly by 90°, and
a ^-polarization leakage from the FLC will be directed to port 2, giving a lower optical
on/off ratio for port 2.
The optical on/off ratio is similar to the contrast ratio used for the characterization of
the FLC devices. The contrast ratio is defined as the ratio of the maximum transmitted
intensity versus the minimum transmitted intensity through the FLC when it is placed
between two crossed polarizers [13]. We choose to talk about optical on/off ratio because
in our experimental case, and for the delay path operation, the FLC is between crossed
polarizers, while for the no-delay path operation the FLC is between parallel polarizers.
Previously reported contrast ratios cover a wide range of 20:1-100:1 [8, 14, 15] to 1000:11500:1 [13, 16]. We see that our results at port 2 are an order of magnitude greater than
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that in references 13 and 16. Nevertheless, the port 1 switch performance of —1:20 is
inadequate for many high performance applications, such as phased array radar true-time
delay control. Thus, we must use some type of switch leakage noise reduction scheme to
obtain high performance from our FLC-shutter device-based PDL. This improvement is
explained in the next section, where the use of an active noise filter can significantly
improve the system performance. This active noise filter simply consists of an additional
FLC device and a polarizer, shown as FLC2 and P2, respectively, in Fig. 5.4.
5.4.2. Signal-to-Leakage Noise Ratio and Polarization Extinction Ratio
Electrical SNR measurements and polarization ER measurements were obtained for
the two different settings of the PDL with and without the use of the active noise filter. The
optical power measurements for both the straight and the delay paths of the PDL are made
using a Newport Model 815 digital power meter. Table 5.1 shows the electrical SNR and
polarization ER measurements. The straight path operation occurs when FLC1 is “off” and
FLC2 is “on”. The delay path operation occurs when FLC1 is “on” and FLC2 is “off” .
From Table 5.1, we see that when FLC1 is “on” and the noise reduction scheme is
not used, the SNR and ER of the PDL unit are very low, mainly because of the high level
noise contribution coming from the straight path due to the low on/off ratio of port 1. In the
case when FLC1 is “off”, the noise coming from the delay path is quite low and thus the
SNR does not decrease as much as for the previous case. Table 5.1 shows a great
improvement of the SNR and ER measurements when the noise reduction scheme is used.
The output polarizer P2 suppresses the noise traveling through the PDL. Even for the
troublesome case when FLC1 is “on”, the SNR is 96 dB. P2 also improves the state of
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
polarization (SOP) of the PDL output beam, giving an increase of more than 25 dB to the
output ER.
Table 5.1: Electrical SNR and Polarization ER measurements for the delay path and straight path of the
PDL with and without the novel active noise filter
Electrical SNR (dB)
Polarization ER (dB)
PDL
With Noise
Without
With Noise
Without
Path
Filter
Noise Filter
Filter
Noise Filter
Non-Delay
103.9
77.34
39.89
12.36
Delay
96.7
24.37
39.73
12.15
Our experiments also showed that it is more beneficial for the SNR performance to
operate the two FLC devices in opposing states, i.e., when for example FLC1 is “on”, then
FLC2 is “o ff’, and vice versa. This happens because the FLC devices do not perform
equally well for both of their states. The out of phase operation leads to a balancing of the
polarization leakage noise between the two settings of a bit-module, and hence light lost
due to the active noise reduction filter is approximately the same for both settings.
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.4.3. FLC Switching Speed
The temporal response characteristics of these FLC devices were also tested, as the
higher switching speed promise of FLC devices compared to NLC devices is the key
motivation for using these FLC devices. A 50% duty cycle ± 5 V bipolar waveform of
frequency/ = 891 Hz, from a Wavetek Model 164 signal generator, were used to drive our
FLC devices. Fig. 5.5 shows oscilloscope traces of the driving square wave signal (bottom
trace) and the temporal response signal for FLC1 (top trace). A high speed detector, New
Focus Model 1801, was placed at port 1 to obtain the temporal response signal due to the
finite switching time of FLC 1. As observed from Fig. 5.5(a), it takes about 73 (is of time
delay for the FLC to respond to the new driving signal level. Also, the rise and fall time of
the FLC device defined as the time it takes for the FLC device to change from 10% to 90%
(or from 90% to 10%) of its final optical power value, was measured as 75.8 (is (Fig.
5.5b). Thus, the total time it takes for the FLC devices in the PDL to switch between the
delay and no delay modes of the PDL is ~ 150 (is.
This finite mode change time is due mainly to the finite time it takes for the FLC
molecules to start moving/rotating and following the fast drive voltage changes, physically
limited by the viscous forces in the FLC material [17]. The switching speed of the FLC
device is given by [18]
T=
P»E
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(5.1)
Figure S.S: Oscilloscope traces of the FLC switching time that is controlled by the ± 5 V bipolar
waveform. The bottom traces show the bipolar drive signal to FLC1, while the top traces shows the photo­
detected optical output at port 1; (a) 73 ps time delay before the FLC device starts responding to the applied
voltage, (b) 75.8 ps (10% to 90% or vice versa) rise or fall time.
where T) is the viscosity, P is the medium macroscopic polarization, and E is the applied
electric field. Rise (10% to 90%) and fall (90% to 10%) times are usually about 1.8-t [13].
For higher switching times, FLC materials with lower T| and higher P must be used, or a
higher voltage across the electrodes (E=V Id) must be applied. Nevertheless,
T|
and P are
fixed for a specific FLC material. As mentioned earlier the FLC molecules have two
preferred directions that are separated by approximately 45° and can be controlled by the
polarity of the applied voltage across the electrodes. When a given voltage threshold is
reached, the liquid crystal molecules orient themselves in one of their prefered directions,
and when the applied voltage is reversed, the molecules reorient themselves to the other
direction.
In order to facilitate faster FLC switching, higher electric field must be applied.
Nevertheless, high voltage above a certain limit can permanently change the macroscopic
polarization of the FLC material, as well as shorten the device lifetime. Hence, a different
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
technique for driving the FLC device must be utilized. Such a technique is based on short
duration high voltage transients. A specially optimized waveform that exhibits a ± 15 V
switching transient that quickly decays at -300 ps to a ± 5 V holding voltage can be used
[24, 19]. The ±15 V switching transient voltage must be short in duration to avoid
permanent change in the ferroelectric polarization. Once the FLC molecules are forced into
a given state, the FLC drive waveform can “relax” to the much lower holding voltage of ±
5 V.
Using this special electronic addressing, switching times of 35 |is were obtained. In
Fig. 5.6 the temporal response of a FLC device is shown, when the device is driven by a
waveform with a ± 15 V switching transient and a holding voltage of ± 5 V. Fig. 5.3 also
shows a -30 jxs delay of the FLC responce to the driving waveform. The same frequency
of 891 Hz, as in the case of Fig. 5.5, has been chosen for comparison. The driving
waveform was attenuated by 8 dB before being fed to the oscilloscope to avoid overdriving
the oscilloscope. Fig. 5.3 shows a ± 6 V switching transient and a ± 2 V holding voltage
that correspond to a ± 15 V switching transient and ± 5 V holding voltage, respectively,
from the driver when attenuated by 8 dB.
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
'J
Figure 5.6: shows oscilloscope traces of the FLC switching time that is controlled by the specially
optimized waveform with a ± 15 V switching transient voltage and a ± 5 V holding voltage (bottom trace).
A 35 ps switching time is shown (top trace). Note also the -30 ps delay of the FLC responce to the
applied waveform.
Note that in most phased array applications, the time sequenced antenna beam
positions (and hence the PDL delay/no-delay settings) are known a priori. This means that
we can overcome the finite delay time of the FLC response if we apply the FLC device
control signals a little earlier, i.e., 73 (is (or 35 (is - if the special transient waveform is
used) earlier in the case of our experimental PDL in this chapter.
5.5. Conclusion
In conclusion, it is the finite FLC rise/fall time that fundamentally limits our phased
array system scanning speed. For our experimental PDL with a 75 |is FLC rise/fall time, a
maximum beam scanning speed of -13,000 beams/s can be achieved using the dual system
channel time multiplexed antenna control mode [6]. Note that this is over an order of
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
magnitude improvement of the previously demonstrated NLC-device based PDL [4].
Furthermore, we have demonstrated high electrical SNRs (>96 dB) and polarization ERs
(-40 dB) for our PDL using a novel active noise filter. Higher FLC switching times of 35
ps were also obtained using a specially optimized waveform that uses a short duration (300
(is) high switching voltage (±15 V) transient followed by a holding voltage of ±5 V.
In the following chapters we will use this high speed and high SNR FLC based
PDLs to form N-bit microwave band optical delay lines using 1310 nm lasers using both
direct and external modulation techniques.
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 6
PHASED ARRAY ANTENNA MAXIMUM COMPRESSION
REVERSIBLE PHOTONIC BEAMFORMER USING TERNARY
DESIGNS AND MULTIPLE WAVELENGTHS
6.1. Introduction
Several three dimensional (3-D) PDL intensity modulation-based architectures for
transmit-receive mode phased array antenna control based on 2-D pixelated optical
polarization switching arrays have been proposed and/or demonstarted in the previous
chapters. These polarization switching arrays can be either NLC or FLC devices. These
reversible 3-D photonic systems can have free-space, solid-optics, and fiber-based optical
time delays [1-4]. For an M-element antenna array, these 3-D PDLs have M independent
parallel optical processing channels. Because we propose the use of the mature large area
LC technology to form the required 2-D pixelated polarization switching array devices, the
number of pixels in the device can easily match any large number M of antenna elements
required in advanced radar applications. For instance, for a M = 5,000 element radar,
5,000 pixel LC devices can be commercially fabricated for use in the 3-D PDLs.
Nevertheless, from a hardware assembly and maintenance cost point of view, it would be
beneficial to reduce the number of control modules or physically independent time delay
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
channels in the antenna array control system. Hence, as is being done today, most large
advanced wideband phased array radar designs use antenna array partitioning, where the
array is divided into smaller clusters of fewer antenna elements [5]. In this case,
independent signal time delay control via a multichannel time delay controller is provided to
the clusters for the appropriate inter-sub-array beamforming, while element level
microwave phase shifters provide the independent signal phase delay control within the
clusters for intra-subarray (or between the elements in each subarray) beamsteering.
Recently, the use of wavelength multiplexing has been proposed to reduce
hardware in a photonically controlled phased array antenna [6]. Here, hardware
compression is achieved using a cascade of two wavelength independent, multichannel,
binary, switched PDLs. In addition, recently, the use of independent (e.g., multichannel
prism geometry) dispersive fibers with a single high power tunable laser source has been
proposed for making continuously variable (non-binary switching) PDLs [7, 8]. In this
chapter, a novel photonic beamformer architecture that combines switched PDL ternary
designs, wavelength multiplexing, wavelength independent optical time delays, and
wavelength dependent optical time delays, to form a low interconnection complexity,
maximum hardware compression, reversible (transmit-receive mode) antenna control
architecture is described [9, 10]. Unlike previous beamformer designs, the basic
architecture is formed by a serial cascade of one single physical channel (not multichannel)
wavelength dependent (e.g., dispersive fiber) delay PDL and one multichannel wavelength
independent delay PDL that leads to antenna control architecture simplicity [9, 10]. Both
architectures for one dimensional (1-D) and 2-D antenna beam steering are described.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.2. Maximum Compression Reversible Photonic Beamformer System Architecture
6.2.1. Photonic Delay Line Ternary Designs
So far, all switched serial PDL architectures discussed in the literature were binary
optical layout architectures where PDL modules are serially connected along one packaging
dimension, and each PDL module has two specific light paths; one called the no-delay path
and the other called the delay path. Hence, to get 2s different time delay settings for phased
array control, N of these binary layout modules must be cascaded in one packaging
direction.
The first part of our proposed system compression technique relates to PDL
structure packaging. Specifically, we suggest an alternative to the classic binary serially
cascaded PDL design. More specifically, we suggest a ternary optical layout switched PDL
design which leads to a redistribution of PDL structure weight and physical volume. Fig.
6.1 shows a schematic diagram of this ternary optical design layout for making a variable
PDL, and also shows a table with the possible 3N different delay settings. In this ternary
architecture, each PDL module has three different optical paths that are selected by
choosing amongst the 0, 1, or 2 values allocated to the ternary digit. Thus, unlike the
binary switched PDL where module cascading happens in one dimension leading to a
“long” barrel-type physical structure, the ternary switched PDL design requires fewer
modules per cascading direction for the same amount of variable delays. This is because
optics allows us to easily add an additional orthogonal cascading dimension to the photonic
module physical structure, thus forming the ternary digit module. Hence, for applications
with particular weight and space constraints, this ternary layout adds increased flexibility
from a packaging point of view as the overall PDL weight and volume gets distributed in
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
two dimensions and hence leads to a more symmetrical and balanced overall variable PDL
design. Note that each ternary layout module requires that the time delays of the two upper
delay paths have a time delay ratio of 1:2. Also note that from a component count point of
view, both the binary and ternary module designs are equivalent; nevertheless, the number
of independent PDL modules cascaded per dimension is smaller for the ternary switched
PDL leading to a less elongated and perhaps mechanically more stable physical structure.
MSD
MSD
Total
Tim e Delay
?2
18-X
-‘-g-
6-x
3 -t
-*•
LSD
3*
f
0-x
0
0
1-x
0
0
1
2-x
0
0
2
0
0
3-x
0
I
4-x
0
1
1
5-X
0
I
2
6-x
0
2
0
7-x
0
2
I
8-x
0
2
2
9-x
I
0
0
10-x
I
0
1
0
8-x = Ox-32 + 2-X-3' + 2-x-3° = 0-x + 6-x + 2-x
LSD
2-x
=-g~
3
2
3'
->
24-x
2
2
25-x
2
2
1
26 -X
2
2
2
25-x = 2-X-32 + 2-X-3' + l-x-3° = 18-x + 6-x + 1-x
Figure 6.1: Schematic diagram of the 3-digit ternary optical layout switched PDL, and a table showing the
possible delay settings. MSD: Most Significant Digit, LSD: Least Significant Digit
Figure 6.2(a) and (b) shows two possible designs for multichannel single stage
ternary layout PDL modules using cube PBSs and TBSs, respectively, and 2-D
polarization switching optical arrays (SAs). Both PDL module designs use a free-space or
solid-optics delay path. Fiber options are also possible, and are shown in Fig. 3. Fig. 3(a)
shows the first stage of a PBS based multichannel, fiber delay, ternary, transmissive PDL
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Passive
Noise Filter
PBS
PBS
L2
SA
L2
Active
Noise Filter
Passive
Noise Filter
L2
OUT
/
PBS
M'rmlens Array
PBS
(a)
M
Microlens Array
Active
Noise Filter
(b )
Figure 6.2: (a) A single stage of a PBS based PDL ternary architecture; (b) A single stage of a TBS based
PDL ternary architecture. (TBS: Thompson polarizing beamsplitter, PBS: Cube polarizing beamsplitter,
Ls: Lenses, M: Mirror, SA: Switching Array, P: Polarizer).
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
architecture. The lengths of the fibers are selected to give the required time delays as shown
in Fig. 6.1. For this architecture, PM fibers need to be used in order to keep the high SOP.
When microlens (e.g., GRIN lens) fiber collimators are used for coupling light in and out
of the system, low coupling optical loss (e.g., < 1 dB) can be obtained for GRIN-to-GRIN
distances of less than 5 cm [11]. Thus, if the longest free space and solid optics
propagation in the PDL system is of the order of ~ 5 cm, there is no need for bulk imaging
optics. Note that when the light propagates in the fibers, it is guided. Thus, the architecture
can be very compact by placing the PBS as shown in Fig. 6.3(b) where a perspective view
is shown. Note that the input and output of the two sets of fiber delays lie on different
planes, hence making fiber assembly a simpler task.
6.2.2. Compressed Wavelength Multiplexed Reversible Photonic Control Architecture for
1-D Antenna Beam Steering
The second part of our system compression technique relies on combining
wavelength multiplexing with wavelength dependent delays such as dispersive fiber delays
to form a fully reversible array antenna controller that uses a simple cascade of a single
physical channel dispersive fiber PDL with a multichannel non-dispersive PDL. When this
novel architecture is combined with our previously mentioned ternary PDL designs, we can
get the highest proposed hardware reduction both in physical size and weight redistribution
and number of components for a photonically controlled phased array system. In general,
the wavelength multiplexing technique is based on partitioning the phased array into subarrays. For instance, a 2-D aperture antenna could be mechanically scanned in azimuth
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TOP VIEW:
3-D VIEW:
Set of equal
length fibers
Set of equal
length PM fibers
PM Fibei
PBS
PBS
SA
• OUT
SA
Set of equal
length PM fibers
PBS
Figure 6.3: (a) Top view of the first stage of a multichannel, fiber-delay, ternary transmissive PDL; (b) 3-D
view of the multichannel, fiber-delay, ternary transmissive PDL (PBS: Cube polarizing beamsplitter, SA:
Switching Array, P: Polarizer).
Height Direction
(Electronic Scan)
Antenna Site
l:Af
DEMUX/MUX
Sub-array 1
/-Fiber Bundle
Remote LINK
Azimuth Direction
(Mechanical Scan)
From/To
Controller Site l
Sub-array /
tm»(6)=rm(9)+t^0)
^M
Transmit Mode
(Tx)
1:M
DEMUX/MUX
Antenna Element
Optoelectronic T/R
Modules
Antenna
Element
Receive Mode
(Rx)
Figure 6.4: The geometry of a 2-D phased array antenna that is mechanically steered in azimuth and
electronically steered in height.
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
while electronically steered in height, with the antenna array divided along the height axis
into a total of / sub-arrays, with M antenna elements per sub-array (Fig. 6.4). The m-th
element of the (v+l)-th sub-array requires a time delay that is equal to the time delay of the
m-th element of the v-th sub-array plus a bias delay (v = 1,2,..., J). Thus, as indicated in
ref. 6, the photonic control system can be designed as a cascade of two multichannel binary
switched PDL subsystems.
The proposed wavelength multiplexed reversible photonic control system for 1-D
steering of phased array antennas is shown in Fig. 6.5. The first single channel switched
PDL sub-system will give the required time delays that control the M antenna elements in
each sub-array. The second /-channel switched PDL sub-system will give the additional
required bias delays for controlling each of the / sub-arrays. Unlike ref. 6, in our novel
system control architecture shown in Fig. 6.5, we are using a single physical channel,
wavelength dependent delay-based switched PDL in cascade with a /-channel, wavelength
independent optical delay-based switched PDL. Hence, the new design gives clear
hardware reduction by using only one physical multichannel PDL instead of two physical
multichannel PDLs. Furthermore, our PDLs are based on ternary and not binary optical
layout designs; hence a further balanced redistribution of weight and volune in the phased
array controller.
Single physical channel ternary dispersive fiber PDL architectures are shown in
Fig. 6.6. Fig. 6.6(a) shows a single stage of a TBS based, single physical channel,
dispersive fiber, ternary, transmissive PDL architecture. Because of the limits set by the
geometry and size of the TBS, imaging optics are required in the long ffee-space non-delay
path. Our system is based on polarization switching, thus dispersive PM fibers [12] are
required to keep the high SOP of the light. Fig 6.6(b) shows a single stage of a PBS based
single physical channel, dispersive fiber, ternary, reflective PDL architecture. In the
reflective architecture, non-PM fibers can be used with a FR-M configuration, where the
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Controller Site
To Post
Processing
System
I:M
RF Network
Rx Tx
A.-MUX/
DEMUX
Single Physical
Channel
Ternary
A.-Dependent*
F-Digit
Switched PDL
/-fib e r Bundle
Remote LINK
1
/
-
-
2 Xu *2.- Ai
j
Antenna Site
Controller
O ptoelectronic T/R
M odules
Rx
Tm(0)
Transmit
Signal
Generator
T/R
Electronic
Switch
Tx
I :J
Optical
Splitter
| Receiver |
A.i.A*..
IX
|
..u
Rx
/ Physical
Channel
Ternary
XIndependent
G-Digit
Switched
Bias PDL
A ntenna Elem ent
Optoelectronic T/R
M odules
Antenna
Element
l:Af
DEMUX/
MUX
W avefront
rmv(0)=Tm(0)+Xv(0)
I
DEMUX/
MUX
1
DEMUX/
MUX
•e.g.. dispersive fiber,
fiber with gratings
T ransm it Mode
(Tx)
Antenna
Sub-Array
Receive Mode
(Rx)
Figure fi.5: The novel basic wavelength multiplexed reversible photonic control system tor l-L ) steered
phased array antennas using a single physical channel F-digit switched wavelength dependent PDL in
cascade with a wavelength independent G-digit switched PDL with multiple physical channels.
retracing of the beam through the fiber compensates for any random induced birefringence
effects that the fiber may have suffered due to the changing environmental conditions or the
changes in the optical wavelength [4, 13]. The zero delay is obtained when the light
propagates through arms 1 and 3. The SA between the two PBSs is used to change the
polarization from horizontal to vertical in order for the light to go through arm 3. The time
delay x is obtained after propagation through arms 2 and 3. In this case SA is set to the
“off’ state so as to leave the polarization vertical. Time delay of 2x is obtained after light
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
propagates through arms 2 and 4. The length of the second fiber is such that the total
obtained time delay is 2x. This architecture is less costly when compared with the
transmissive architecture, since inexpensive non-PM fibers are used and shorter fiber
lengths are required.
T O P VIEW :
Single Stage M odule
D ispersive P M -Fiber
TBS
TBS
SA
SA
tb
SA
:
TBS
D ispersive
P M -F iber
SA
1:1 Im aging
(a)
TO P VIEW ;
Singlc Stage M odule
ARM !
F R - M a Dispersive
U N on-PM Fiber
SA
ARM2
OUT
SA
ARM3
QWP
ARM4
FR-M
D ispersive N on-PM Fiber
(b)
Figure 6.6: (a) TBS based single physical channel, dispersive PM-fiber, ternary PDL transmissive
architecture; (b) PBS based single physical channel, dispersive non-PM-fiber, ternary PDL reflective
architecture.
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A fiber with Bragg gratings has also been recently used to form a discretely variable
non-switched optical delay element using a single tunable laser source [14, 15]. Therefore,
we also propose a single channel switched fiber-grating-based PDL for forming the
required single channel wavelength dependent F-digit switched PDL needed for our PDL
system. Fig. 6.7 shows a typical single stage for a single physical channel switched fibergrating-based PDL. This system cannot utilize the single FR-M configuration discussed
earlier as each wavelength is reflected at a different position within the fiber. Polarization
maintaining fibers with Bragg gratings [16] are required to keep the high SOP of the light.
A QWP before the fiber is needed to change the polarization of the light from horizontal to
circular, and after reflection from the Bragg grating and propagation through the QWP for
second time, from circular to vertical.
TQP VIEW:
Single Stage Module
ARMl
SA
Light
^ QWP
Absorber/Block
S
ARM2
If
■d-HIIIMII 1H1111—I
SA
OUT
Fiber Bragg
Grating
ARM3
SA P
QWP
Fiber Bragg == N
Grating
"T
ARM4
Figure 6.7: PBS based single physical channel fiber grating PM-fiber, ternary PDL architecture. (SA:
switching array: QWP: quarter wave plate; P: polarizer, M: mirror).
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
So far we have discussed how the individual wavelength dependent and wavelength
independent PDLs work, that are the building blocks of the overall photonic control
system. Now, we discuss how our compressed reversible photonic control architecture
works. In the transmit mode, M high power or optically amplified light sources such as
high power diode lasers (H-LD) that operate at different wavelengths, A.,, X2, ... XM, are
modulated by the transmit signal. High power optical sources are needed in the transmit
operation because signals are split and distributed to the entire antenna array, and the optical
losses can add up quickly. A M:1 optical ^-multiplexer is used to couple the signals at
different wavelengths into a single physical channel wavelength dependent such as a
dispersive fiber based F-digit ternary layout switched PDL sub-system. The different
wavelengths “see” different indices of refraction in the dispersive fiber and thus travel with
different velocities, obtaining small time delay differences with respect to each other.
The individual time delays with reference to the antenna and control architectures
(Fig. 6.4 and Fig. 6.5) are given by
(6 . 1)
where m = 1,2,..., M, n(Xm) is the index of refraction that the m-th wavelength sees, Le is
the selected fiber length the light eventually travels through for obtaining the desired 0 far
field antenna beam height angle, c is the velocity of light in vacuum.
refraction can be expressed as
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The index of
( 6 .2 )
where D is the dispersion of the fiber, and Xm is the optical carrier. Hence, the time delay
difference Axm(0) between wavelength X.m and the reference wavelength X, due to the
selected fiber length dispersion is going to be
Axm(0)=x(n(0) - t ,(0) =
=L0 D ( X m-Xl)=Le- D A X m.
(6.3)
A 1:J optical splitter is used to split the M delayed signals into J sets of different
wavelength signals. These J sets of X.,, X^,... XMsignals then pass through the G-digit J physical channel ternary switched wavelength independent delay-based bias PDL sub­
system, obtaining the appropriate bias delays. The bias delay for the v-th sub-array where v
= 1,2,..., J can be expressed as
xv(0) = ^ . ,
c
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(6.4)
where
is the selected optical path length for the v-th subarray feed that the light travels
through for steering the antenna beam to a 0 beam height angle.
The general expression for the time delay of the m-th antenna element of the v-th
sub-array can be described by the following equation
U 0 ) = X m(0)+ T v(0),
(6.5)
where Tm(0), and xv(0) are given by equations (6.1) and (6.4). As an example, we show
the time delay of the M-th antenna element of the 7-th sub-array, which is also the longest
time delay, and is given by
U 0 ) = **,(0) + ^(0).
( 6 .6 )
where xM(0) is the longest time delay from the single physical channel /•’-digit wavelength
dependent PDL, and x /0 ) is the longest bias time delay from the G-digit wavelength
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
independent PDL. Finally, J 1:M optical ^-demultiplexers are used to separate the different
M signals that drive the elements of each sub-array.
In the transmit mode operation of the antenna array, the microwave signal for
transmission modulates the high-power laser diodes (H-LDs) of the controller
optoelectronic transmit/receive modules, and the required time delayed signals are detected
by the photodetectors at the antenna element optoelectronic transmit/receive module (See
Fig. 6.8). The signals from the photodetectors pass through the transmit/receive (T/R)
module electronic phase shifters that must be used to compensate for element level signal
path/channel phase or ultrashort time delay errors due to noise sources such as small
variations in the length of optical and/or electrical paths. Fortunately, because electronic
phase shifters already exist in the commonly used microwave monolithic integrated circuit
(MMIC) T/R modules for phased array antennas, it thus becomes simple to compensate for
vital element level phase errors that exist in all practical deployed antenna systems.
C ontroller O ptoelectronic T/R Module
MMIC
OHIO
H igh Power
L aser Diode
Tx
£
Electronic
T/R Switch
Tx
RF
LIGHT
P-Bit Electronic
Phase Shifter
Rx
Y-branch /
Optical Switch
Rx
T ransm it M ode
(Tx)
— » —o —
Phntodetector
Pow er A m plifier
(PA)
A ntenna Elem ent Optoelectronic T/R M odule
Low Noise Amplifier
(LNA)
O EIC: O ptoelectronic Integrated Circuit
MMIC
OE1C
R IN : Relative Intensity Noise
L o w R IN
L
s e r Diode
l / io o c
L aaser
LIGHT
*y
M M IC: M onolithic M icrow ave Integrated Circuit
^
I
r* H _
Y-branch f
Optical Switch
RF
A ntenna Element
™ rf-is
S L :• t5
Photodetector
**
Figure 6.R: (a) Controller optoelectronic transmit/receive module, and (b) antenna element optoelectronic
transmit/receive module used in our wavelength multiplexing photonic control systems.
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As mentioned before, our control system is reversible, as essentially the same
hardware is used for both the transmit and receive modes of the antenna. In the case of the
receive mode, each antenna element is connected to a low noise laser diode (LD). The M
outputs of each antenna sub-array modulates the M LDs which operate at the M different
wavelengths. Note, that the wavelength of the LD used for the m-th element for each of the
J sub-arrays is the same. Light propagates in the opposite direction in the system when
compared to the transmit mode. For each sub-array, the wavelengths are multiplexed and
pass through the bias wavelength independent PDL. After the bias PDL, the bias time
delays have been canceled. A J: 1 optical combiner is used to couple all the signals into the
single physical channel wavelength dependent PDL sub-system, where now the shorter
time delays corresponding to the individual elements of each sub-array are cancelled. The J
received optical signals all at a specific wavelength (see Fig. 6.5) are directed to their
detector in the controller optoelectronic transmit/receive module that completes a J signal
intensity sum. Then all M signals are electrically summed, and sent to the electronic
receiver for post-processing.
6.3. Compressed Reversible Photonic Control Architecture for 2-D Antenna Beam Steering
The previous section discussed compressed reversible photonic control for 1-D
antenna beamsteering. There are important radar applications where agile 2-D antenna
beamsteering is required. In this section, we discuss two different time delay architectures
suitable for 2-D antenna beamsteering.
The first architecture, shown in Fig. 6.9, uses one single physical channel ternary
^.-dependent F-digit fiber switched PDL and one N physical channel ^-independent ternary
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
G-digit switched PDL. The first PDL controls the relative time delays of the M antenna
elements in the antenna x-axis and hence controls scanning in azimuthal a direction. The
second PDL controls the bias time delays of the N rows of the antenna along the y-axis for
scanning in height (or 0). The m-th element (m = 1, 2, ..., M) of the (n+ 1)-th (n = 1, 2,
..., N) row has the same time delay as the m-th element of the n-th row plus a bias time
delay. As in the 1-D antenna beamsteering case in the transmit mode, M H-LDs that operate
at different wavelengths, X,, X ^ ... XM, are modulated by the transmit signal. A M : 1 optical
X-muItipIexer is used to couple the signals at different wavelengths into a single physical
channel X-dependent F-digit ternary switched PDL sub-system. The signals obtain relative
time delays depending on their wavelength and the total length of the fiber selected as in a
dispersion fiber PDL. As similarly shown in Eqn. 6.1, these delays can be given by
*.«*) =
n( K ) A*
(6.7)
where La is the selected fiber length for a far field antenna beam a azimuth angle.
A 1:N optical splitter is then used to split the M delayed signals into N sets of
different wavelength signals. These N sets of X,, X2, ... Xw signals will eventually be fed
to the N different antenna rows respectively to control the steering in height. After passing
through the G-digit Af-physical channel ternary switched X-independent bias delay PDL
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C o n tro ller Site
I:M
R F N etw ork
Tx
\
| Receiver
To Past
Processing
System
|
Rx
Transmit
Signal
Generator
T/R
Electronic
Switch
Tx
A ntenna Site
Controller Optoelectronic T/R
Modules
Rx
Single Physical
Channel Ternary
Dependent
F-D igit Switched
PDL
Antenna Element O ptoelectronic
T/R M odules
N-fiber Bundle
Rem ote LINK
Xm
N Physical
Channel Ternary
X- Independent
G-Digit Switched
Bias PDL
I:M
Rx
Transmit Mode
(Tx)
DEMUX/MUX
i
Figure 6.9: The novel wavelength multiplexing reversible photonic control system for 2-D beam steering
of phased array antennas using a single physical channel X-dependent ternary PDL and a multichannel Xindependent ternary PDL, for independent control in the two scan axes. (MUX: multiplexer; DEMUX:
demultiplexer).
sub-system, the n-th PDL physical channel M wavelength signal set (see Fig. 6.9) obtains
the appropriate delay that can be described by
Tn(0) = ^ - ,
c
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(6.8)
where n= 1,2, ..JV, L„e is the appropriate optical path length required for the n-th row time
delay associated with a 0 height angle for the far field antenna beam pattern. Finally, N I'M
optical ^.-demultiplexers are used to separate the different M signals that drive the elements
of each sub-array. Hence, the total time delay required for the mn-th element is given by the
sum of the time delay needed for azimuthal steering and the bias time delay required for
height steering, or
^ „ (a ,9 ) = t J a)+tn(Q),
(6.9)
where xm(a) and t„(0) are given in Eqn. 6.7 and Eqn. 6.8, respectively.
A different phased array antenna element distribution geometry is shown in Fig.
6.10, that is most suitable for very large (e.g., >10,000 element) phased arrays. In this
geometry, the 2-D antenna is divided into 2-D sub-arrays, with equal number of elements
per sub-array. Fig. 6.10 shows a 2-D antenna partitioned in H sub-arrays where H = /x j, I
is the number of columns, and J is the number of rows. Each sub-array consists of M x N
antenna elements. The sub-arrays are represented by the notation Sm where u = 1, 2, ..., I,
and v = 1,2,..., J. The antenna elements are given by the notation A mnuv where m = 1 ,2 ,
..., M, and n — 1,2, ..., N.
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
S\J
SiJ
5 12
N
521
5/1
2
1
M \
I 2
Sub-anay 11
Antenna Element A \fl\
I
1
Sub-array
mv
: Sttv
m = l,
2,.... M;
Antenna E lem ent: Amnuv u =1,2......./;
* = 1.2.......N
v =1, 2.......7
Figure 6 .1H; 2-D sub-array partitioning of a 2-D phased array antenna. The antenna is divided into H subarrays, with each sub-array containing MxN elements.
Fig. 6 .11(a) shows the photonic phased array antenna controller. First, a single
physical channel channel ternary ^.-dependent switched PDL and a N channel Xindependent PDL are used to control the time delays of the antenna elements A mnttV within
one sub-array. This procedure is exactly the same as discussed earlier for the 2-D antenna
with independent 1-D antenna beam steering in both azimuth and height directions (see Fig.
6.9). Hence, the two subsystems set the time delay of the individual elements, where the
delay for the mn-th element is described by
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Controller Site
____ _
l:M
X /. *2
X *f
l:N
Tm(o)
Optical
Splitter
t/.
RF N etw ork I Reeerverl
X /. X 2„ ... X *f
Single
Physical
Channel
Ternary XD ependent
f-D ig it
Sw itched
• X I,
PDL
M:I
l:tf
ical
Splitter
Electronic
Switch
Controller
Optoelectronic T/R
Modules
NxH
,
XI.X2.~~.XM
X /. X 2 - ... x * f
N th
2
H
Optical
Splitter
X /. X 2
X gf
I ://
Optical
Splitter
Tm(a)+t/(0)
1
2
1
"
AVI
Physical
Channel
Ternary
X-Independent
2nd C-Digit
Switched
Bias PDL
5 ./. x * ... xu
H
t
2
Nx 2
NxH
1
Tm(aHt2(e)
~
Receive
M ode
(Rx)
r m (a)+ T n (8 )+ T Bv (cr.0)
tfxff
2xH .
X /.T 2 —
»
«
Xu
X/. X2 Xm
To/From
Antenna Site
2x1
------
I
1st
A W M iber Bundle
Remote LINK
2x2
X /. X 2
Ixff ,
X/, X 2....
\M
rm(a)+t,,(0)+Wa,0)
Xu
1x2
Ixl
Rx
Transm it
Transmit
Signal
Generator
H=txJ
X *f
rm(a)+TN(9)
Rx Tx
X-MUX/
DEMUX
A/ Physical
Channel Ternary
X-Independent
(7*Digit Switched
Bias PTDL
'X / . X 2
T o Post
Processing
System
(a)
Tm(tx)+r«(0)
\:M
DEM UX/M UX
Antenna Site
Am
xW i(a»+ tM »H T /7/a,0)
**
h
*
—H j
\ ■“
*
To/From M h
row o f S /7
\
:
\:M
DEM UX/M UX
WxW-fiber Bundle
Remote LINK
NxH
!:Af
DEM UX/MUX
. j-2 ^
» ----------------------------------------U------- 1! Twi ► -
>«(a)+T M e>+ t | i ( o , 0 )
1 — «■
Xt
*
• '“ T " ' !
U
*
1 1 « f 1-
lU r: J m
T o /R o m M h
row o f S 21
To/From M h
row o f S 11
^
From Controller Site
To Controller Site
X |. X 2
XA/
rw(a)+ ri(0)+Tn(a.0)
l:Af
DEM UX/M UX
To/From 1st
" row o f S /7
tm(a)+t„(9)+r«»(a.8)
I:M
XT'
I:M
TGT r«(a)+t|(0)+tti(ot,0)
; »
——
tin(O)+t|(0)+t2i(a.0)
D EM UX/M UX
Transmit M ode
(Tx)
Receive Mode
(Rx)
-F ^ r
To/From 1st
ro w o f S 21
Xi
DEM UX/M UX
-
(b)
To/From 1st
row o f S 11
Antenna Element Optoelectronic
T/R M odules
Figure 6.11: The novel CREOL wavelength multiplexed reversible photonic control system for 2-D beam
steering of phased array antennas using 2-D sub-array partitioning, (a) controller site, and (b) antenna site.
Ill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
= Tm(a )+Tn(0)-
(6.10)
These N signals for sub-array 2-D control are further split into H channels and fed
into an //-channel ^-independent PDL where appropriate sub-array bias time delays
XBV(a,0) are added. The total time delay for the mn-th antenna element in the uv-th subarray can then be given by
^ ( a . e ) = **,(01,0) + 'c.w(a,0) =
= [xm(a)+x„(0)]+xBV(a,0).
(6.10)
These H x N signals then are directed to \:M ^.-demultiplexers to separate the different
wavelengths and then each one of the M xN xH signals is fed to its appropriate antenna
element in the array.
Note that all our control systems are reversible, as the same PDL hardware is used
for both the transmit and receive modes of the antenna. In the receive mode, light
propagates in the opposite direction in the system when compared to the transmit mode.
Hence, indeed maximum hardware compression is possible.
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.4. Photonic Control System Issues Related to the Phased Array Application
An important point to note in our beamformer design is how current fiber-optic
technology can be used to satisfy the different required functions of the two different PDLs
needed in our approach. In other words, the much longer antenna bias delays required for
inter-subarray antenna beamsteering are well handled by our ^-independent PDL design
based on solid-optics, free-space, and near-zero dispersion fiber [17, 18] optical delays,
that can easily provide long (e.g., > 1 ns) delays. On the other hand, for the much smaller
(e.g., < 1 ns) within antenna sub-array delays required for the within subarray antenna
beamsteering, we can effectively use current dispersive fibers to form the single physical
channel PDL. In this way, we do not require extremely long (e.g., 1 km) dispersive fiber
lengths, saving system cost in terms of fiber assembly time and packaging, overall
controller size and power budgets. Today, typical fiber dispersion magnitudes range from
20 ps/km-ns to 80 ps/km-ns, depending on the fiber and the operation wavelength [19].
Note that fiber dispersion can be either positive or negative. Since, in our system
we have the flexibility to select the short wavelengths to get the shorter delays and the long
wavelengths to get the longer delays or vice versa, we can drive the appropriate antenna
elements within the subarray to get the desired antenna beamsteering. Lately, high negative
dispersion fibers with dispersion as high as -134 ps/km-ns at a 1550 nm wavelength have
been specially designed and developed for dispersion-compensation in wavelength
multiplexed fiber systems [20]. We can use such a fiber to obtain the small time delays
required in our system.
Fig. 6.12 shows a diagram of the calculated absolute value of the time delays
obtained using such a fiber for a 3-digit dispersive fiber, ternary PDL; such as the one
shown in Figure 6.6(b). Short time delays under 1.1 ns are generated by this 10 channel
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
‘
■
'
— T ___ I_______
I_____ I
_
0
I
I
I
1
1
1
1
1
1
1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
Wavelength (nm)
Figure 6 .12: Diagram showing the calculated absolute value of the even time delays 2 t, 4 t ,
26 t of
the single channel dispersive fiber PDL for 10 wavelengths with spacing of I nm. The dispersion of the
fiber is - 134 ps/km nm at 1550 nm, the fiber increment is 30 m. Note that since the fiber has negative
dispersion, the long wavelengths travel faster in the fiber, and thus obtain shorter time delays than the short
wavelengths.
1546-1555 nm wavelength multiplexed design. Note that since the Fiber has negative
dispersion, the long wavelengths travel with faster velocities in the Fiber than the short
wavelengths, and thus they obtain shorter time delays. The 3-digit ternary PDL can give 27
different time delay settings. In Fig. 6.12, for simplicity in viewing, only the even time
delays, such as 2 x, 4 -t , ... 2 6 -t, are shown. The fiber length increment chosen is 15
meters. Note that with high dispersion fibers, shorter fiber length increments give the time
delay resolution required within the antenna subarray, since the delay depends on the
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
product of the dispersion with the length of the fiber. Hence, it is more beneficial from
packaging point of view to use short, high-dispersion fibers, than long, low-dispersion
fibers. An added benefit of the short-length fibers is their lower cost. Shorter fibers also
make the system less sensitive to environment-based gradient effects such as thermal and
stress gradients which can get amplified by the length of the fiber.
Thus,
our
antenna
control architecture using both long time delays, wavelength independent and short time
delays, wavelength dependent dispersive fiber PDLs matches the requirements of future
subarray control-based advanced large microwave phased array radars. Nevertheless, by
using additional ternary digits in a switched PDL with the added modules using km lengthtype dispersive fibers and wavelengths spread over a larger optical bandwidth, longer time
delays for inter-subarray beamforming can also be implemented for antenna control. This
is the case shown in Fig. 6.9 where for a 2-D antenna, independent 1-D antenna beam
steering for azimuth control is implemented via one physical channel switched wavelength
dependent PDL that provides a continuous-signal time delay ramp for both inter-subarray
and within-subarray beamforming. A second wavelength independent PDL with multiple
physical channels provides the independent time delays for steering in the orthogonal height
scan direction.
Note that for the PDLs in our controller, the optical SAs have to be equally effective
over the A.,, A^, ... XM wavelength range of operation. Although there are various options
to implement the optical switching in a multiwavelength PDL such as active wavelength
optical filters [21], we are focusing on 2-D LC polarization switching devices that can
provide both low cost and high performance, i.e., large pixel count, high polarization
extinction ratios, and moderately fast switching speeds, particularly for the large channel
count antenna array control application. Polarization switching is based on retardation
effects on the propagating light and is wavelength dependent. Our interest is to obtain high
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SNR and fast switching times. We have shown, in chapter 4, that FLC devices can give
relatively fast 35 fis switching times while maintaining very high > 95 dB electrical SNRs
[22]. FLC SAs can be designed to act as half-wave retarders at a designed center
wavelength. An external voltage signal is used to change the retardation of the device pixel
between two states which either give a 90° rotation, or leave the incident light polarization
unaffected [23]. For our wavelength multiplexed system, it is important that we examine
the effects of wavelength diversity on the performance of the controller using FLC SAs.
We can calculate the theoretical retardation T for the different wavelengths from the
well known expression [24],
T(\) = ^ A n - d ,
(6.12)
A,
where d is the FLC device thickness calculated to give a retardation of n at the center
wavelength, and An is the birefringence of the standard FLCs fabricated by Displaytech,
and is given by [25],
a
40.1,19692.5
An(A.) = 0.142-----— I---- — .
K
A,
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(6.13)
For a FLC device designed to act as a half wave retarder at 1310 nm, and for a set
of wavelengths from 1290 nm to 1330 nm, the retardation obtained is between 1.014 7t and
0.9859 it. This is a maximum deviation of -1.4 % from the desired retardation of it
obtained for the center wavelength. Note that using a higher wavelength such as 1550 nm,
the maximum deviation will be smaller (-1.3%) since the retardation is inversely
proportional to the wavelength. The above result means that when the FLC devices are set
to rotate the polarization by 90°, the polarization does not fully rotate by 90° for all the
wavelengths in the system. This, in effect, will give unwanted polarization components
propagating through the system that can add to overall system noise. We have
demonstrated the use of active noise filters in PDLs that efficiently reject these unwanted
polarization components, thus improving the system performance in terms of SNR [4].
Nevertheless, the rejection of the unwanted wavelength dependent polarization
noise by the active noise filters will eventually result in different optical power levels for the
different wavelengths.
This optical power variation across the wavelength multiplexed channels can be
calculated. This maximum theoretical intensity variation occurs because the FLC devices
cannot fully rotate the polarization for all the wavelengths when they are set “on”. The
analysis to follow explain this calculation. The FLC device can be represented as a wave
retarder with retardation T. The matrix for a retarder with retardation T can be written as
[24],
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 "i
V2 i
-r
i
’exp(-yT /2)
1
' c o s (r/2 )
-jsin fT /2)
1
1
exp(yT /2) 'V 2
i
"
t
r
i
-ysin(IV2)"
cos(r /2)
If the input polarization is horizontal, the retarder will change the polarization to one
that can be described by the following matrix,
E=
' c o s(r/2 )
—/s in (r /2)
-y s in (r/2 /
cos(r /2)
T
0
’ c o s(r/2 )
-y'sin(r /2)
where we have supposed unit amplitude. We see that if T =
t z
,
(6.15)
the state of the new
polarization will be vertical. But since the retardation depends on the wavelength, the
polarization will not be vertical for all the wavelengths. The output electric field will thus
be,
"0 0 T cos(T/2) "
E
^OUt = _0 lj[-y s in (r/2 )
0
-ysin(T /2)
and thus the intensity after the polarizer can be written as,
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(6.16)
=sin2(r/2).
(6.17)
The output intensity for the center wavelength is going to be
/_. = sin (7t/2) = l,
(6.18)
and thus the variation of the intensity for all the wavelengths will be
AI = 1- s i n 2(7t/2).
(6.19)
For the wavelengths mentioned earlier and for the specific Displaytech FLC
devices, the maximum intensity variation is less than 0.05%, or a maximum -0.0023 dB
optical gain variation, as shown in Fig. 6.13. This small variation in the optical power for
the different wavelengths will eventually be translated into small variations in
radiofrequency (RF) losses of -0.0045 dB for the element level
antenna channels. If
needed, in our system this very small per channel variation can be eliminated
in the
controller optoelectronic transmit/receive module (Fig. 6.8), where we can make use of the
variable gain amplifiers (VGA) in the individual module. Fig. 6.14 shows similar results
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.0000 -
-0.0005 -
c
o
2
i
>
_c
3
O
”3CJ
•a
a.
O
-0.0010 -0.0015 -
-
0.0020
-
jo b
f in e
-0.0025
1290
1300
1310
1320
1330
Wavelength (nm)
Figure 6 13- Calculated optical gain variation across a 40 nm optical bandwidth for the wavelength
multiplexed optical signals that pass through a FLC device polarization rotator designed for 1310 nm center
frequency. Both the 3 dB type plot and the high resolution/detail plot are shown, indicating essentially no
optical signal variation across the X-bandwidth.
0.0000 ■
-
0.0002 -
/
c
o
ca -0.0004 ‘i
>
_c -0.0006 '3
O
"3
o -0.0008 ■0
CL
O
-0.0010-
J
/
/
J
\
/
j.
\
\m
\
I.,
1546
M v d o fit(im )
\
T
1530
1535
1540
1545
1550
1555
1560
Wavelength (nm)
Figure 6.14: Calculated optical gain variation across a 40 nm optical bandwidth for the wavelength
multiplexed optical signals that pass through a FLC device polarization rotator designed for 1550 nm center
frequency. Both the 3 dB type plot and the high resolution/detail plot are shown, indicating essentially no
optical signal variation across the X-bandwidth.
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
using the 1550 nm band. Thus, these FLC-based optical switches and active noise filters
can be used in our wavelength multiplexed photonic controller with essentially no effect on
the overall antenna system performance.
Recall that the photodetector in each controller site optoelectronic T/R module adds
light beams at the same wavelength. This can give rise to a coherent beat note problem that
can be reduced by (a) randomizing the optical phase of the different beams with the same
wavelengths (i.e., making the light beams temporally mutually incoherent) using a moving
random phase mask screen and/or (b) using carefully designed laser wavelengths that have
slight optical frequency offsets with respect to each other that result in out-of-receiver filter
bandwidth spurious signals that can be electronically suppressed. These wavelength offsets
are designed to be small enough that PDL operations are essentially not effected and can be
compensated in array optoelectronics. These wavelength dependent system issues will be
dealt with in later experimental studies.
Finally, we look at typical radar applications where such controllers can be
deployed. For a typical 2-D antenna aperture with azimuthal mechanical steering and
electronic 1-D height steering, the antenna array design parameters might be M = 50
elements per subarray, with J — 25 subarrays along one axis of the antenna aperture. This
gives a total of 1,250 individually fed antenna elements in the radar. This design implies
that we need 50 separate wavelengths for lasers used in the system. Today, commercial
available distributed feedback (DFB) LDs have narrow spectral widths (Full Width Half
Maximum <0.1 nm) and LD-to-LD wavelength separation of 1 nm [26]. Thus, DFB LDs
can be used in the proposed 1-D steered wavelength multiplexed photonic reversible time
delay control system (see Fig. 6.5) to provide the 50 different wavelengths. The above LD
specifications also meet the resolution of commercially available wavelength multiplexers
and demultiplexers [27]. hence, the component technology exist to put together such a
photonic control system.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For the 2-D steering cases shown in Fig. 6.9 and Fig. 6.11, typical numbers are as
follows. For the case of 2-D steering with independent 1-D beam steering in the two
orthogonal directions, the typical antenna array design parameters can be M = 50 elements
in the x-direction and N = 50 in y-direction of the 2-D antenna face. This gives a total of
2,500 antenna elements in the radar. This design also needs 50 separate wavelengths for
lasers used in the system, as the previous design indicated for 1-D beam steering only.
Note that km-length type dispersive fibers are needed to make the dispersive fiber PDL
modules for obtaining longer time delays (>1 ns). For the 2-D phased array antenna with 2D sub-array partitioning, typical number of sub-arrays can be 10 x 10 with 10 x 10 subarray elements. This gives a total number of 10,000 antenna elements. Note that the
requirements for the LDs and the wavelength multiplexers and demultiplexers are less
stringent than before, since only 10 different wavelengths are used. Designs that use higher
number of wavelengths can be used to reduce the number of sub-arrays or to obtain phased
array antennas with even a higher number of elements.
6.5. Conclusion
In conclusion, a novel ternary PDL optical layout design that has improved weight
and volume distribution flexibility when compared to previous “long” binary PDL
structures has been introduced. A basic wavelength multiplexed reversible photonic time
delay control system is introduced that combines a single physical channel ^.-dependent
fiber switched PDL in cascade with a multiple physical channel ^.-independent switched
PDL. This wideband phased array control architecture leads to a reduction in the number of
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
control modules used for beam steering and control. For the multi-channel ^-independent
PDL, a solid optics and/or free space propagation design is prefered since the use of fibers
in every PDL digit would increase the interconnection complexity. On the other hand, fiber
delays are an excellent choice for the single physical channel X-dependent PDL. Two
different options of the single physical channel wavelength multiplexed PDL are proposed.
The first one uses dispersive fiber delays, while the second one is formed using fibers with
Bragg gratings. The single physical channel dispersive fiber (or fiber grating) switched
PDL can provide the short time delays required for the within antenna subarray
beamforming. On the other hand, the multichannel ^.-independent PDL can provide the
larger time delay inter-subarray antenna beamsteering. The use of short fiber lengths is
preferable, both from a cost and a packaging point of view, and the environmental effect
such as thermal and stress gradient fiber sensitivity issue, that can impact final antenna
system performance. Thus, our antenna control architecture perfectly matches the needs
and requirements of future sub-array control-based advanced large microwave phased array
radars. Nevertheless, km-length wavelength dependent fibers can also be used for cases
where longer delays are required, although at a higher cost and packaging complexity.
Both 1-D and 2-D antenna steering is possible with are proposed systems.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 7
ADAPTABLE-DELAY B ALANCED-LOSS BINARY PHOTONIC DELAY
LINE ARCHITECTURES USING POLARIZATION SWITCHING
7.1. Introduction
So far, no PDL architecture has been described that can be adapted to provide the
wide range of time delays, i.e., subnanoseconds to a few tens of nanoseconds, that are
required in a variety of phased array antenna applications ranging from large military radars
to smaller base station antennas for cellular communications. A combination of different
PDL modules based on a variety of design geometries and polarization components have
been proposed to obtain the required long [1-6], moderate [1, 5, 7] and ultra-short time
delays [5, 7-11]. The variable Af-bit PDL control system is formed by cascading N
independent binary PDL modules. Thus, from an assembly and packaging point of view, it
would be beneficial to use the same PDL module for a wide range of time delays with
minimum changes in hardware. For multichannel or array type applications, it is also
important to maintain balanced optical signal loss between the two settings of each single
bit PDL module. Generally, the different performance (loss) numbers of the optical
components as well as the different number of optical components in the two possible paths
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of a binary PDL module can lead to different optical losses between the two settings. In
general, for large phased array applications, large variations in the signal amplitude across
the array channels is not desirable as it leads to wastage of excess optical power in the
channels that suffer less optical losses. Hence, a PDL that maintains a balanced loss
between the two settings of each PDL module would be beneficial from optical power
budget point of view.
For the first time, a balanced loss PDL module design that can give a wide range of
time delays, e.g., a tenth of a picosecond to tens of nanoseconds, is described [12, 13].
Theoretical analysis and experimental results obtained for this PDL architecture are
compared with the previously proposed reflective architecture (Chapter 3). In addition, two
hardware compression techniques based on wavelength multiplexing and polarization
multiplexing are proposed that can be used with the adaptable PDL architecture to realize
multichannel PDLs.
7.2. The Adaptable Delay Reflective-Symmetric PDL Architecture
A variety of switching array (SA)-based PDL architectures can satisfy the different
time delay requirements for the phased array antenna application. Table 7.1 shows the
obtainable range of time delays for each of these previously proposed PDL architectures
including fiber [1-6], free space [1, 5-7], and solid optics [3, 8-11]. From Table 7.1, it can
be seen that the most versatile architecture in terms of possible acquired time delays is the
reflective PDL architecture. Nevertheless, there are applications, such as in millimeter wave
radars, where time delays shorter than 0.1 ns are required. If the symmetric and the
reflective PDL architectures were to be combined, then the full time delay range would be
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7.1: Time delay range for the different PDL architectures.
Time Delays
PDL
Ultra-
Short
Moderate
Long
Extra
Architecture
short
0.1ns<t<0.5ns
0.5ns<x<5ns
5ns<t<10ns
Long
x<0. Ins
Symmetric
Reflective
T>10ns
V
• V
Transmissive
V
V
V
V
V
covered by using just only one PDL architecture. We propose a reflective-symmetric
architecture which combines the characteristics of the two PDLs in one, and thus gives the
desired time delays. Specifically, this new adaptable reflective-symmetric architecture
borrows elements from the previously proposed reflective architecture, presented in chapter
3, where light travels twice through the delay path [4]. Using this reflective design with a
special polarization scheme and the birefringence compensation technique, it becomes
possible to use free-space, solid-optics, and non- PM fiber-based optical delay paths, that
are smaller, lighter and more compressed than other transmissive PDL designs. When nonPM fibers are utilized to obtain the long time delays, the birefringence compensation
technique is used which consists of a Faraday rotator-mirror (FR-M) module [4, 5]. The
propagation of the light twice through the fiber delay path and the Faraday rotator leads to
the compensation of any random induced birefringence effects that the fiber may have
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
suffered due to the changing environmental conditions or the changes in the optical
wavelength. Nevertheless, this reflective PDL design is limited to moderate and long time
delays mainly because of limitations set by the size of the cube PBS, as indicated in Table
7.1.
QWP
M
Glass of
thickness d
IN
Microlens
Array _
L
QWP
c
c
M
Microlens .*<
Array
'ST
A^
IN
PBS
i ii i □
SA
P QWP
OUT
OUT
(a)
(b)
Faraday RotatorMirrors
Non-PM fibers
Microlens
Array -
QWP
SA
OUT
(c)
Figure 7.1: Three options of the proposed PDL, using (a) solid optics, (b) free space, and (c) fiber delay
paths for ultra short, moderate, and long time delays, respectively. (P: Polarizer, QWP: Quarter wave plate,
M: Mirror, SA: Switching Array).
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Our new basic reflective-symmetric architecture design for the adaptable delay PDL
single bit module is depicted in Fig 7.1. Fig. 7.1(a) shows the solid optics option for ultrashort time delays. Fig. 7.1(b) shows the moderate time delay option, where free-space
imaging optics are used for low interchannel crosstalk. The design of Fig. 7.1(b) can also
give ultra-short time delays if the relative difference of the focal lengths / , , and f 2 is small
enough. Fig. 7.1(c) shows the non-PM fiber version for long time delays.
The general module operation is described as follows. Linearly polarized (i.e.,
vertical or 5-polarized) light enters the PDL system in the form of a two-dimensional (2-D)
optical beam array. The polarization switching array (SA) independently controls the state
of polarization of each of these incident beams. When the SA is set “on”, it rotates the
incident polarization by 90°, and when the SA is set “o ff’, the incident polarization is
essentially unaffected. Specifically, when the SA is set in its “on” state, light changes to
horizontal or p-polarization and travels straight through the PBS towards the non delay
path. Light travels through the QWP which has its optical axis at 45° with the incident ppolarization. After reflection from the mirror (M) the light passes one more time through the
QWP and changes to s-polarization. Then it is deflected by 90° from the PBS towards the
output port of the PDL. On the other hand, when the SA is set “off”, the ^-polarized light is
deflected by the PBS towards the delay path. Light travels twice through the QWP (for the
solid optics and free space case), changes to p-polarized light and travels through the PBS
towards the output of the PDL bit. At the output of the PDL, an additional SA and polarizer
form the active noise filter to suppress any leakage noise coming from the first SA or the
PBS. As mentioned earlier, for the fiber delay PDL, the light follows a similar path; in this
case, a Faraday-rotator with a power of 45° is used instead of a QWP.
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.3. Adaptable Delay PPL Time Delay Analysis
As mentioned earlier, the proposed PDL architecture can generate a wide range of
time delays ranging from the ultra-short (e.g., 0.1 ps) time delay to extra long (> 10 ns)
time delay. The delay range depends on which hardware option of those shown in Fig. 7.1
is implemented.
Using the previously proposed reflective PDL architecture (Fig. 7.2), the possible
time delays are limited by the dimensions of the PBS. Note that in the delay path, the light
travels thrice through the PBS, while in the non-delay path the light travels through the
PBS only once. Thus, the minimum possible relative path difference between the two paths
is equal to two PBS lengths. For a typical 25.4 mm cube PBS and for a zero distance
between the PBS, the QWP and the mirrors, the minimum obtainable time delay is
M
QWP
Microlens
Array '
SA
PBS
IN
OUT
SA
" QWP
M
Figure 7.2: The previously proposed reflective PDL architecture. (P: Polarizer, QWP: Quarter wave plate,
M: Mirror, SA: Switching Array).
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t=
n x d _ 1.5 x (2 x 25.4 x 10
c
3 x l 0 8 m/s
m)
= 0.254 ns,
(7.1)
where n = 1.5 is the index of refraction of the PBS, d is the path difference between the
delay and the non-delay path (two PBS lengths), and c is the speed of light in air.
Using our reflective-symmetric PDL architecture (Fig. 7.1) where the delay and
non-delay paths are independent of each other, the relative path length difference can near
the zero mark. Thus, by having a very small optical path length difference between the two
paths, ultra short time delays (e.g., 0.1 ps) can be obtained using either the solid-optics or
ffee-space option. Millimeter wave antenna applications require time delays as short as a
tenth of a picosecond. These time delays can be realized by placing one of the mirrors
(e.g., the delay path mirror) on a micrometer resolution translation stage and then adjusting
the relative optical path length difference between the two paths to sub-millimeter resolution
(Fig. 7.3(a)). Note that in order to achieve a 0.1 ps time delay resolution for the PDL
module, a relative optical path length difference of d = 0.015 mm (or 15 pm) is required
between the two PDL paths, as shown in Eq. (7.2).
t=
n x d _ l x (2 x 0 .0 1 5 x 1 0 3 m)
= 0.1 ps,
c
3 x 108 m/s
(7.2)
This resolution is easily achievable with today’s high resolution (1 pm or better)
micropositioning systems.
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
M icro m eter resolution
M
translation stage
Glass plate of
thickness d\
QWP
Microlens
Array
M icrolens
Array
PBS
N
QWP
PBS
IN
SA
SA'
OUT
OUT
(a)
Glass plate of
thickness dn
(b)
M
I W W /W A W I
Microlens
Array
PBS
QWP
NLC device
SA
QWP
OUT
(c)
Figure 7.3: Sub-picosecond time delay option of our proposed adaptable PDL architecture using (a) a
micrometer resolution translation stage, (b) two glass plates of different thickness, and (c) a birefringentmode electrically controlled NLC device, whose index of refraction can be precisely controlled.
A different approach is to use two glass plates of different thickness; one in each of
the paths of the PDL, where the delay and non-delay path lengths in free space are equal
(Fig. 7.3(b)). The small difference between the thickness of the two glasses will give the
desired time delay. In order to obtain a 0.1 ps time delay resolution and for glasses with
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
index of refraction of 1.5, the required relative thickness is 0.01 mm (or 10 pm), as shown
in Eq. (7.3).
nxd
1 .5 x (2 x 0 .0 1 x l0 3 m)
= 0.1 ps,
3x10 m/s
(7.3)
It is also possible to use a birefringent mode NLC device in the delay path with its
NLC molecular director aligned with the vertical (or horizontal) polarization (Fig. 7.3(c)).
The refractive index of the NLC device can be controlled using a low voltage electrical
signal, and thus the effective optical path length of the delay path can be set to the desired
magnitude. In addition to the sub-picosecond time delays, extra long time delays (>10 ns)
are also possible using fiber optic cables in the delay path. Thus, the adaptable reflectivesymmetric PDL architecture is capable of giving the required wide range of time delays.
7.4. Balanced Loss Performance of the Proposed Adaptive Delay PDL - Theory
Another very important attribute of the proposed adaptive PDL design is that light
travels the same number of times through the PBS for both the delay and the no-delay path.
Thus, light suffers the same loss through either path. This results in a balanced loss for
both states of the switched module. For most high power phased array applications, the
signals driving the antenna elements are required to be of equivalent amplitudes, providing
maximum transmit power efficiency. Thus the balanced loss performance between the two
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
settings of the PDL is a critical system design need. Note that if variable gain control
amplifiers (VGCAs) are used to equalize the amplitude of the signals, this leads to power
wastage for channels that have less losses compared to channels with higher losses. Thus,
from optical power budget point of view, the proposed reflective-symmetric PDL offers the
great advantage of efficient usage of optical power. The section to follow quantifies the
performance of our PDL.
Using AR coated optical components, the total insertion optical loss for any of the
two paths in the single bit PDL structure is theoretically estimated at 1.94 dB. This is the
worst case scenario, and uses a maximum number of optical components in each of the two
PDL paths i.e., glass plates, QWPs and mirrors. The expected loss number for the PDL
has been calculated based on the optical component loss numbers given by the
manufacturers and the measured FLC insertion losses which at present are 0.7 dB (or
15%). This high insertion loss of the FLC device is present because three cells stacked
together are used to achieve the required
k
radians birefringence at 1310 nm. At present a
near infrared (IR) single FLC cell, unlike a visible-light single cell, does not possess
enough birefringence to function as a halfwave plate. Hence, the 0.7 dB insertion loss. The
total insertion loss for our 7-bit PDL, such as those required for advanced phased array
antenna applications, is expected to be -13.65 dB. This number is feasible provided that
ffee-space and/or solid optics PDL bit structures are used. If fiber delays are used in the
most significant bit of our PDL system, an additional 1.6 dB optical insertion loss is
expected. This loss is due to the coupling loss of the fiber and the insertion loss of the
FRM. It is expected that better FLC materials can lead to lower insertion loss (e.g., 0.45
dB, or 10%) FLC single cell devices. This can lead to a total optical insertion loss of 1.45
dB per PDL bit, and a total insertion loss of 9.8 dB for the 7-bit PDL. It should also be
noted that the use of fibers in the delay path will not give a balanced loss performance for
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
both paths of the PDL bit, due to the higher insertion loss of the fiber delay path. Hence, a
passive optical attenuator can be used in the other path to balance performance.
The balance loss for both of the PDL settings also gives a balanced SNR for both
switched states. Remember that optical SNR is defined as 101og(signal power/noise
power), where signal is the optical power in the optical beam of the desired polarization that
travels through the desired delay or non-delay path of the module, and all other optical
power measured at the output is regarded as noise optical power. Using the commercially
available PBS transmission and reflection characteristics, Tp = 95%, Rp = 5%, Ts = 0.1%
and Rx = 99.9% (see Fig. 7.4), we can calculate the optical SNR of the previously
proposed symmetric PDL. T and R stand for the optical power transmission and reflection
coefficient of the incident light, respectively, and p and s refer to the horizontal and vertical
polarized light.
Rp =5%
Rs = 99.9%
Figure 7.4: The transmission (T), and reflection (R) intensity coefficients of a typical commercial cube
PBS. s: vertical polarization; p: horizontal polarization.
For p-polarized light and the non-delay setting, the SNR is a 55.80 dB, as depicted in Fig.
7.5(a) and derived in Eq. (7.4).
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SNR = lOlog
_T£b x S0
E_
Rp x
rtx R,xS,p /
= 55.80 dB.
(7.4)
The optical power flow diagram of Fig. 7.5(a) shows the signal and noise optical
power as they propagate through the system. For input optical power of Sp, the optical
signal at the output is Tp x Sp while the noise output power is Rp x Ts x Rs x Sp. Note also
that the SNR numbers have been calculated assuming that the SAs do not degrade the
extinction ratio, (i.e., p/s or s/p ratios) of the incident beam. Only the cube PBS
performance is taken into account due to its highly asymmetric performance for its two
output ports.
M
QWP'
M
<'TpRpSp
Q W P'
Signal: Tp-Sp
In
I -1
RcS
s>>s
Noise:Ts-Ss
In
Out
^ ”
Noise:Rp Ts Rp Sp
Signal: Rs Tp Rs -Ss
Ts-RpSp
sQ** ^
Out
TpRs^s
QWP
QWP
M1
(a)
(b)
Figure 7.5: SNR analysis of (a) the non-delay mode and (b) the delay mode of a previously proposed
symmetric PDL. Signal is shown with a single arrow, while noise is shown with a double arrow. (M:
Mirror, QWP: Quarter Wave Plate).
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The optical SNR for s-polarized light and the delay setting is a low 29.77 dB, as
depicted in Fig. 7.5(b) and derived in Eq. (7.5). The input optical power is Ss and the
optical signal at the output is R s x T Px Rs x Ss while the noise output power is Ts x S s.
SNR = lOlog
( Rs x T p x R s x S ^
TxSr
= 29.77 dB.
(7.5)
Unlike the PDL in Fig. 7.5, Fig. 7.6 shows the two settings of the proposed
adaptable PDL where the optical SNR for the non-delay (and p-polarized input light) and
the delay (and ^-polarized input light) is
SNR = lOlog
' R .xT 'X S,'
= 42.78 dB,
(7.6)
and
SNR = lOlog
rT,xR ,xS,'
= 42.78 dB
Rp x T , x S ,
(7.7)
respectively. Hence, the PDL design proposed in this paper gives the desired balanced
SNR performance for the two settings of the module.
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
M
-QWP
p RP'SP
Rs Ssf
I
1
P<
In
M
Tg-Rp-Sp
Ts-Sg
V.
Noise
Signal
Signal
Out Rs'Tp-Sp
Tp-Rs-Ss
(a)
Noise
Out
RpTs Ss
(b)
Figure 7.6: SNR analysis of the (a) non-delay mode and (b) delay mode of the proposed PDL. Signal is
shown with a single arrow, while noise is shown with a double arrow. (M: Mirror, QWP: Quarter Wave
Plate).
7.5. Experimental Verification of the Balanced Loss PDL Architecture
The experimental set-up for the proposed adaptable balanced loss PDL architecture
is shown in Fig. 7.7(a). In this experiment, we are interested in characterizing the PDL in
terms of losses and SNR. Thus, continuous wave (CW) unmodulated light is used as input
to the PDL. CW light from a diode pumped Nd:YAG laser (^=1319 nm) is fed into the
PDL bit using a GRIN lens connectorized PM fiber. A high extinction vertical (or s)
polarizer is used to clean the beam from any horizontal (or p) polarization component. The
measured polarization extinction ratio at the input of the PDL was 10,000:1 (or 40 dB). The
polarization switches used in the experiment are ferroelectric liquid crystal (FLC) devices
that act as programmable half wave plates. Each PDL bit has two FLC polarization
switching devices as described earlier. The first one is used for optical path switching. This
FLC device acts as a programmable half wave plate that either rotates the incident
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
polarization by 90° or leaves the polarization unaffected. Depending on the state of
polarization of the light, the switched optical beam can follow either the delay or non-delay
path. The second FLC device along with a high extinction ratio polarizer form the active
noise filter that suppresses any noise leakage from the first FLC device and the PBS. The
two FLC devices are chosen to operate out of phase. That means that whenever one is set
“on” and thus rotates the incident polarization, the other one is set “off” and leaves the
polarization unaffected. This out of phase operation has been found to give lower leakage
noise levels for our PDL modules [14]. For example, when FLC 1 is “on”, it rotates the
polarization to p and the light goes through the non-delay path. FLC2 is set “o ff’, and it
leaves the polarization unaffected, to go through the 5-polarizer. On the other hand, when
FLC1 is set “o ff’, the light is deflected by 90° from the PBS and follows the delay path. In
this case FLC2 is set “on” and rotates the p-polarization to 5-polarization, and the light exits
the PDL module.
The optical insertion loss for the delay and non-delay path was measured at 1.80 dB
and 1.85 dB, respectively. These numbers are slightly lower than the theoretically expected
ones, mentioned in the previous section. This is because of the tolerance in the insertion
loss of the AR-coated components, and because the 1.95 dB estimated loss is the one
associated with the worst case scenario, which considers the maximum number of optical
components in either of the two paths.
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Delay Path
Delay Path
M
, QWP
QWP
FLC1
FLC1
\
PBS
IN
/
/
Non-Delay
Path
FLC2
.^ P (p-polarization)
PBS
OUT
IN
Non-Delay Path
FLC2-
^ -P (^-polarization)
QWP
OUT
M
(a)
(b)
Figure 7.7: (a) The experimental set-up for the proposed adaptable PDL architecture; (b) The experimental
set-up for the previously proposed symmetric PDL architecture. Delay paths are shown with double arrows,
and non-delay paths with single arrows.
Optical SNR measurements were obtained using a power meter at the output of the
PDL module. For each of the two settings, the signal propagates through the desired delay
or non-delay path, and noise propagates through the non-desired non-delay or delay path,
respectively. Signal and leakage noise optical power measurements are easily obtained by
just blocking the light traveling through the noise (or signal) path and measuring the optical
power at the output of the module that corresponds to the signal (or noise). Table 7.2
shows the optical and electrical SNR for both settings of the proposed PDL. For
comparison, the old reflective architecture was also tested. This experimental set-up is
shown in Fig. 7.6(b). The delay path operation occurs when FLC1 is “off” and FLC2 is
“on”. The non-delay path operation occurs when FLC1 is “on” and FLC2 is “o f f ’. SNR
measurements for this PDL are also shown in Table 7.2. The results in Table 7.2
experimentally verify that the proposed adaptable reflective-symmetric PDL architecture
gives a balanced SNR performance for both settings of the PDL.
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7.2: Optical and Electrical Signal-to-Noise Ratio for both settings of the reflective and the new
reflective-symmetric adaptable architecture.
PDL
Optical Signal-to-Noise Ratio
Electrical Signal-to-Noise Ratio
__________ (dB)__________
__________ (dB)__________
Refective
Adaptable
Reflective
Adaptable
Delay
23.46
41.32
46.92
82.64
Non-Delay
36.32
41.82
72.64
82.64
Setting
7.6. Hardware Compressed Versions of the Adaptable Balanced Loss PDL Architecture
7.6.1. Wavelength Multiplexing-based PDL
Our proposed adaptable balanced loss PDL can be used for the implementation of a
single physical channel wavelength dependent PDL [16]. Fig. 7.8 shows two possible
configurations of our adaptable PDL architecture using wavelength multiplexing. Fig.
7.8(a) shows a non-PM dispersive fiber based design. M different wavelengths are
combined into a single PM-fiber and fed to the PDL. When the light goes through the delay
path (dispersive fiber), the different wavelengths “see” different indices of refraction and
thus travel with different velocities. A cascade of such PDL modules will give the required
time delays to the different wavelengths that will eventually drive the antenna elements.
Fig. 7.8(b) shows a fiber Bragg grating based adaptable balanced loss PDL design. The
delay path consists of PM-fiber with Bragg gratings. The PM-fiber is required to keep the
high state of polarization of the light. A QWP before the fiber is needed to change the
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
polarization of the light from horizontal to circular, and after reflection from the Bragg
grating and propagation through the QWP for second time, from circular to vertical. Each
wavelength is reflected from a different Bragg grating. Thus, each wavelength travels
through a different fiber length and obtains the appropriate time delay.
PM-Fiber with
Bragg Gratings
Non-PM
Dispersive Fiber
QWP
Microlens
Microlens
QWP
IN
IN
M
Polarization
Switch
Polarization
Switch
OUT
OUT
(a)
(b)
Figure 7.8: (a) Wavelength multiplexing technique using dispersive fibers, (b) Wavelength multiplexing
technique using fiber Bragg gratings. QWP: quarter wave plate; P: polarizer; M: mirror.
7.6.2. Polarization Multiplexing based PDL
Another hardware compressed version is based on polarization multiplexing. Fig.
7.9 shows a possible polarization multiplexing scheme. In this design, two adjacent PDL
channels (channel 1 and 2) are combined in one path, and thus a reduction by a factor of
two is possible in the number of fiber loops. After propagation through the fiber the two
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
channels are separated again before they are sent to the next PDL module in the cascade of
PDL bits. When channels 1 and 2 are vertically or s-polarized they are deflected by 90°
from the PBS. Channel 2 travels through a HWP and its polarization changes to horizontal
(or p). Channel 1 travels through a compensation plate that is designed to give equivalent
time delay to channel 1. The p- and s-polarized beams propagate through a beam-displacing
prism (BDP). This BDP has the appropriate thickness so that the two beams after the BDP
are collinear. The two orthogonal polarized beams are coupled into a PM fiber. The slow
axis of the fiber can be chosen to be aligned along one of the two orthogonal polarizations
(i.e., along the p-polarization, Fig. 7.9). The output of the fiber can be set so that its slow
axis is aligned along the other orthogonal polarization (i.e., along the 5-polarization, Fig.
7.9). Thus, channel 1 is now p-polarized and channel 2 is 5-polarized. The two orthogonal
polarized beams are separated after propagating through a BDP. Channel 1 goes through a
compensation plate and channel 2 goes through a HWP. Thus, both beams are p-polarized
and travel through the PBS towards the output of the PDL module. Note that the
compensation plates have to be designed such that both beams obtain the same time delay.
When the two incident beams are p-polarized they go through the non-delay path that
consists of a QWP and a total internal reflection prism (T1R). When the two incident beams
are of different polarization, one will travel through the delay path and the other through the
non-delay path.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PM Fibers
slow axis
slow axis
GRIN Lens
BDP
Microlens Array
p
■ HWP
[~~1 Compensation Plate
□to
top
^ P ' " « Is
PBS
TIR
HWP
SA1
OUT
Figure 7.9: Polarization multiplexing technique using beam-displacing prisms. (HWP: half wave plate; P:
polarizer, TIR: total internal reflection prism; BDP: beam dispacing prism).
1.1. Conclusion
A new binary PDL architecture that allows a wide range of time delays as well as
provides a balanced loss performance between the two settings of the PDL has been
proposed and experimentally demonstrated. The propagation of the optical signal through
the same number of optical components leads to the equivalent optical loss that is beneficial
both from an optical power budget point of view and a signal processing flow point of
view. Furthermore, the proposed adaptable PDL architecture gives optical SNRs > 40 dB,
which leads to electrical SNRs > 80 dB; a number that is highly desirable for advanced
signal processing such as phased array antenna control. Experiments conducted with the
proposed adaptable PDL and our previous reflective PDL verify the versatility and
improved performance of this new PDL. Two hardware compression techniques were also
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
discussed. The first one uses wavelength multiplexing in conjunction with dispersive fibers
or fiber Bragg gratings to reduce the physical PDL channels to one. The second technique
uses polarization multiplexing that causes a hardware reduction by a factor of two. This is
accomplished by multiplexing two adjacent channels in the PDL into a single physical delay
path.
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 8
DIRECTLY MODULATED SEMICONDUCTOR LASER FED
PHOTONIC DELAY LINE USING FERROELECTRIC LIQUID
CRYSTALS
8.1. Introduction
Our initial experimental NLC and ferroelectric liquid crystal FLC-based PDL
research used visible light for acquiring appropriate optical alignment experience and
procedures, and for high observed (to the PDL builder) accuracy of optical leakage noise,
signal, crosstalk, and loss measurements. The current preferred wavelength for high speed
analog light modulation applications is in the near infrared band (e.g., 1300 nm), mainly
because of recent commercial developments in high performance analog fiber-optic (FO)
links using near infrared, semiconductor-based optical transmitter and receiver technology
[1]. These FO-links are being considered as economical and practical solutions for RF
signal
transmission and distribution,
particularly
for
microwave
phased
array
antennas/radars. Hence, PDLs for the antenna application must take advantage of the lowloss and cable flexibility of fiber-optics to remote antenna control systems to an
environmentally friendlier site. Furthermore, as mentioned in earlier chapters, FLC devices
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
are an excellent candidate optical polarization switch technology for faster switching times.
Visible light CW experiment demonstrated 35 (is switching speeds while achieving 40 dB
optical on/off isolation ratios at both ports of the switching fabric [2,3].
Hence, in this chapter, we demonstrate for the first time, a PDL that uses the
desired fiber-optic remoting via a directly modulated 1310 nm semiconductor laser link,
and FLC devices for higher speed optical path switching [4, 5]. We test this reversible
FLC-based switched single-channel PDL at a I GHz FO-link modulation frequency, and
the overall laboratory system is characterized with respect to the system parameters such as
carrier-to-noise ratio (C/N), insertion loss, spatial interchannel crosstalk, and time delays.
Large area single pixel FLC devices are used; hence the single channel PDL. Note that our
experimental set-up is readily expandable to a multichannel PDL system if multi-pixel FLC
devices and multiple input/output fiber assemblies are used.
8.2. 3-Bit 3-D PDL using FLC Devices. Imaging, and Remoting
8.2.1. Experimental Set-up
The experimental setup of our 3-bit PDL is shown in Fig. 8.1. A Lasertron model
QLINK1-051 microwave fiber-optic transmitter (X=1310 nm) and receiver are used to form
the remote fiber link. The Lasertron transmitter is designed to directly convert RFmodulated electrical signals to optical signals. The modulated light is coupled into the PDL
by a single mode (SM)-fiber which is connected to the transmitter module by a FC-PC (flat
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
connections-physical connection) fiber-optic connector. A GRIN lens is connectorized at
the other end of the several meters length single-mode fiber, and forms the input port to the
PDL. The GRIN-lens based FO collimators were purchased from Wave Optics, Inc. The
light power output directly from this GRIN lens for our system was typically 1.8 mW.
From
Tnuttmtnrr
B IT 1
BIT 2
BIT 3
Figure 8.1: The experimental 3-bit PDL system using FLC devices, imaging optics, and fiber-optic
remoting. The non-delay paths are represented with solid lines while the delay paths are represented with
dashed lines. (TIR: Total internal reflection prism; PBS: Polarizing beamsplitter cube; L: Lenses, M:
Mirror).
Our PDL system is based on polarization switching, where switching between the
two orthogonal polarizations, i.e., vertical (or ^-polarization) and horizontal (or ppolarization), occurs. The light transmitted from our commercial microwave fiber-optic
transmitter to the PDL is for the moment, through a non-PM fiber. This makes the
polarization of the input light to be in general not linear, mainly because of the randomly
induced fiber birefringence due to external factors such as environmental changes. This
polarization fluctuation, although varying slowly, would adversely affect our overall
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
system performance, particularly in terms of noise leakage through the FLCs and the
PBSs. In order to temporarily overcome this problem for our laboratory system, a parallelrub birefringent-mode NLC device and a polarizer are placed at the input of the PDL system
to control the input polarization. The state of polarization of the input light can then be
controlled by choosing both the appropriate angle of the NLC molecular director relative to
the vertical or horizontal axis, and the NLC cell applied voltage. In this way, the magnitude
of the vertical component of the input light can be maximized while minimizing the
horizontal component. Then a high extinction ratio polarizer is used to block the unwanted
horizontal polarization. The NLC-based polarization controller helps in maintaining a
maximum of 1.6 mW of high linear polarized light. Of course, a more appropriate fiber
feed to our PDL would be a PM fiber feed connectorized to an optical transmitter module.
8.2.2. The PDL Free-Space Feedback, Feed-Forward and Symmetric Delay Architectures
We choose three different PDL bit architectures for our system to demonstrate ultrashort, moderate, and long time delays [6]. 3-D refractive imaging optics are used within the
PDL to minimize interchannel crosstalk [7]. A 4 •/ imaging system is formed to image the
input (object plane) of each bit at its output (image plane), where this output is the input for
the next bit. In general, / refers to the focal length of the spherical plano-convex lenses
used in the PDL.
The first bit is based on the single cube PBS, feedback circular delay path geometry
design, and can provide the long time delays (> 5 ns). A 3-D imaging system in the delay
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
path consists of two spherical lenses L2’s of focal length f 2 = 40 cm, and four total TIRs.
The time delay for this first bit is given by
Ar = — + 4 ^ tir
c
PBS
°T IR
/ I T IR >
n PBS
PBS
( 8 . 1)
PBS
where ( n ^ , n?BS) and (<fnR>^PBs) are the index of refraction and the side physical size of the
(TIR, PBS) pair, respectively, c is the velocity of light in vacuum. For our case,
= 1.5 and dPBS =
= nPBS
= 2.54 cm, and hence the designed time delay for bit 1 is 5.69 ns.
The second bit is based on the transmissive feed-forward, two cube PBS
architecture, and gives the moderate time delay (e.g., < 5 ns) in our 3-bit PDL. For the
straight (non-delay) path, the imaging lenses have focal lengths/3 = 12.5 cm, and for the
delay path the imaging lenses have focal lengths of/4 = 25 cm. The designed time delay for
the second bit is given by
A/ = 4
( 8 .2 )
For our given components, this is a designed time delay of 1.67 ns.
The third bit is a symmetric PDL design that provides the ultra-short time delay
(<0.1 ns). The two paths in the bit have the same physical length, and imaging optics using
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lenses L5 and L6 with focal lengths f s and f 6 respectively are used (/j = f 6). A very short
optical path-length difference between the two paths can be introduced by placing two glass
plates; one plate in each switched path in the bit. The ultra-short time delay can thus be
generated by controlling the relative difference in the thickness of the two glass plates. In
our experimental case, the two glass plates have thicknesses of d { = 0.635 cm and d2 =
0.953 cm for the non-delay and the delay paths, respectively. For bit 3, this ultra-short time
delay can be calculated from the expression
(8.3)
where n = 1.5 is the index of refraction of the glass plates. In our case this delay is
calculated to be 8.8 ps.
8.2.3. The Ferroelectric Liquid Crystal based Polarization Switches
The optical switches used in the PDL are FLC devices that are based on the electro-optic
principle of a switchable HWP. Each device consists of three FLC cells in cascade (Fig.
8.2). These cells consist of a thin layer (e.g., < 2 |im) of FLC material (Displaytech:
MX8068) sandwiched between two glass plates (1.09 mm thick) and two Indium Tin
Oxide (ITO) electrodes (30 nm thick) used for the electronic addressing [8]. The liquid
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
crystal is a uniaxial birefringent medium, with its optic axis oriented parallel to the liquid
crystal molecules. The optic axis of the FLCs has two preferred directions that are
separated by approximately 45°, and can be controlled by the polarity of the applied voltage
across the electrodes. Thus, the two states of the FLC device can be selected. The three
cells that form each FLC switching device are aligned with their axis parallel to each other.
Three cells are used to achieve the required 7t radians birefringence at 1310 nm as at
present, a single cell (unlike at visible light) does not possess enough birefringence to
function as a halfwave plate in the near infrared [9].Remember that these devices are driven
with a specially optimized waveform that exhibits a ± 15 V switching transient that quickly
decays at -300 ps to a ± 5 V holding voltage [8].
Indium Tin Oxide
Electrodes
C Material
Glass
Driving Voltage (V)
+15 V
V
1
\
_
- 5V r
-15 V 11
~
~
—
300 |is
Figure 8.2: Our infrared 1310 nm ferroelectric liquid crystal (FLC) polarization switching device. Each
device consists of three FLC cells in cascade.
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.3. Ferroelectric Liquid Crystal PDL Demonstration and System Issues
The following sections describe experimental data from our 3-bit 3-D FLC PDL,
and important system issues related to this data are discussed.
8.3.1. Insertion Loss
An important issue in our PDL system is the optical loss introduced by the optical
components. Our attempt is to use AR coated optical components to minimize the optical
losses in the system. We use off-the-shelf optical components that have been AR coated for
1310 nm. Component losses are less than 0.25% for every glass surface, and the
reflectivity of the mirrors is greater than 99%. The two lenses (L2) in the delay path of the
first bit were without AR coatings because of the current non-availability from commercial
sources for our specific design at the time of the experiment. The non-AR coated lens pair
had measured optical losses of 14% and 16%. Table 8.1 shows the measured and expected
optical losses for each bit and for the entire PDL system. Note that all experimental
insertion loss data are less than the expected insertion loss values calculated using
manufacturer listed specifications. For bit 2 and bit 3 modules, the optical insertion losses
are = 1.5 dB, a highly desirable number when using a cascaded N-bit PDL design. Only
the bit 1 module showed a higher loss, and there are three reasons for this occurrence.
First, unlike all the other FLC devices that had a measured transmission loss of = 16 %,
FLC1 used in bit 1 had a measured transmission loss of 26 %. Second, the L2 lens pair has
no 1310 nm AR coating, hence an additional * 8 % loss per lens. Third, the polarizer after
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 8.1. Measured and designed optical losses for each bit and for the overall PDL.
Optical Loss (dB)
PDL Settings
1
Measured
Expected
2
Measured
Expected
3
Measured
Expected
4
Bit 1
Bit 2
Bit 3
Non-delay
Non-delay
2.40
<2.50
1.67
< 1.87
Non-delay
1.49
Non-delay
Non-delay
1.67
2.40
<2.50
Non-delay
2.40
<2.50
Non-delay
< 1.87
Delay
1.61
<2.00
Delay
< 2.00
Delay
1.37
< 2.00
Non-delay
1.49
< 2.00
Delay
3-D PDL
5.56
<6.47
5.44
<6.47
5.50
<6.47
Measured
2.40
1.61
1.37
5.38
Expected
<2.50
<2.00
<2.00
<6.47
Measured
Delay
5.5
Non-delay
1.67
Expected
< 6.9
Delay
Measured
5.5
< 1.87
Non-delay
1.67
Non-delay
1.49
< 2.00
Delay
1.37
Expected
< 6.9
< 1.87
Delay
Delay
Measured
5.5
1.61
1.49
Expected
< 6.9
Delay
5.5
< 6.9
<2.00
Delay
< 2.00
Delay
< 10.87
1.61
<2
1.37
<2
8.48
< 10.87
5
6
7
8
Measured
Expected
<2.00
Non-delay
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.66
< 10.87
8.54
< 10.87
8.6
FLC2 also did not have a 1310 nm AR coating; so another 8 % transmission loss. Thus,
unlike bit 2 and 3, bit 1 suffered the higher optical insertion loss due to four nonoptimized
components. Finally, for the overall PDL, two other components did not have 1310 nm AR
coatings. These are the NLC device at the input of the PDL and the last polarizer (after
FLC6). Hence, if the five non-AR coated components were also AR coated, the overall
PDL will have lower insertion loss than demonstrated in our experiment. Hence, a
laboratory system does demonstrate the power of AR coated optical components,
particularly in our case of cascaded optical components.
8.3.2. Interchannel Crosstalk
Interchannel crosstalk is also a significant factor for analyzing overall system
performance. Light from one optical channel in our 3-D multichannel PDL system can leak
to the adjacent channels and this effect will be translated to interchannel crosstalk or noise
in the system. The interchannel crosstalk was accurately measured with an infrared detector
(3 mm in diameter). Optical power measurements were performed every 1.8 mm in the
orthogonal directions, namely, the x and y directions of the center active optical channel.
We selected the distance of 1.8 mm because this is the diameter for the commercial GRIN
lenses at 1310 nm [10]. Because we are using 25.4 mm side cube PBSs, for a 1.8 mm
interchannel distance, our 3-D PDL can pack about 196 independent time delay optical
channels based on the GRIN-lens design based on a compact, high packing density,
hexagonal GRIN-lens array geometry design for the 2-D output coupling optics of
multichannel PDL systems [11]. This fiber-coupling optics design consists of a 2-D array
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-2 0 -
-20 -
-30 -
S -40-
•10
0
s
•to
10
Lateral O ffset in x (m m )
0
s
10
Lateral offset in y (mm)
(a)
(b)
Figure 8.3: Optical interchannel crosstalk relative to the center active channel of the 3-D PDL, with
measurements taken along the (a) x and (b) y directions at the PDL output plane. A maximum optical
crosstalk of -27.47 dB (or - 54.94 dB RF) is measured at the nearest to center channel in the x-direction.
These optical power measurements are directly taken from the PDL output plane, before the GRIN-lensfiber assembly. The 3-D PDL has a channel capacity of 196 channels.
of GRIN lenses stacked side-by-side and top-to-bottom, with GRIN center-to-center
distances of 1.8 mm. Hence, the output port interchannel measurement distance is 1.8 mm.
Figure 8.3 shows the measured interchannel crosstalk for our PDL at the output plane,
where data is taken for 11 channels in the x-direction and 11 channels in the y-direction, all
relative to the center active optical channel. A highest -27.47 dB optical crosstalk level was
measured for the nearest to center channel in the ^-direction. Hence, the highest RF
interchannel crosstalk for our experimental PDL was -54.94 dB. Note that our crosstalk
measurements were direct optical power measurements taken at the PDL output plane
before the output GRIN-lens. Because each output GRIN-lens is positioned such that it has
on axis alignment with its respective channel, light coupling is maximized for the on-axis
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
GRIN-lens. This means that the off-axis GRIN-Ienses by nature of their positions are not
optimized for high off-axis optical coupling efficiency. In particular, because GRIN lenses
coupled to single mode fibers have small numerical apertures (i.e., tight light acceptance
angles), it is expected that GRIN-fiber coupled RF crosstalk levels will be much better than
reported in this experiment This effect has been experimentally demonstrated for the
visible spectrum where using 633 nm light and 633 nm GRIN-lens FO-collimators, with a
measured nearest channel RF crosstalk levels of <-120 dB [11].
8.3.3. PDL RF Signal Measurements
Other important issues of any PDL system are its C/N performance, and whether it
introduces additional noise to the directly modulated fiber-optic link RF signal. Fig. 8.4a
shows the signal taken direcdy from the Lasertron microwave fiber-optic link when the
transmitter and receiver fiber-optic modules are directly connected with a FC/PC
connectorized single mode optical fiber. The transmitter is driven by a 15 dBm, I GHz
signal from a Hewlett Packard synthesizer. The -28.33 dBm output signal is generated by
the Lasertron fiber-optic receiver module. This also indicates a 43.33 dB RF link loss.
Today, fiber-optic links with 30 dB loss are commercially available from Lasertron Inc.
[12]. Lower RF fiber-optic link losses approaching 20 dB have also been reported using
down-con version of microwave signals [13]. Recent developments on direct-modulation
FO-links can be found in reference 1.
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 8.4: C/N measurements of the 1 GHz modulation Lasertron QLINK1-051 fiber-optic link (a)
without (103 dB/Hz) and (b) with (7934 dB/Hz) the PDL. The noise marker is placed at a 100 kHz offset
and the analyzer RBW = 1.0 kHz.
Using a RF spectrum analyzer, a RF noise floor of -131.33 dBm is measured at a
100 kHz offset from the 1 GHz fiber-optic link modulation frequency, for both when the
link is not connected and when it is connected to the PDL. A 1 kHz analyzer resolution
bandwidth (RBW) is used. A C/N of 103.00 dB/Hz at 100 kHz is measured for the link
without the PDL (see Fig. 8.4a), while a C/N of better than 79 dB/Hz is measured at
100kHz offset when the PDL (setting 3) is connected to the directly modulated fiber-optic
link (see Fig. 8.4b). Table 8.2 shows the C/N for five different settings of the PDL,
indicating similar C/N’s, except for setting 5 of the PDL that incurs higher insertion loss
due to four unoptimized (no AR-coated) components in bit 1. The noise floor of the fiber
link-PDL system remains below -130 dBm for all settings of the PDL, as indicated for
setting 3 in Fig. 8.4(b). Note the flat noise floor around the 1 GHz spectral peak for the
entire 250 kHz analyzer bandwidth. This implies that our PDL C/N measurements are RF
analyzer noise floor limited and the “noise” in the C/N refers to this analyzer noise floor.
Nevertheless, these C/N measurements help us to accurately determine the RF insertion
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
loss values via the RF power or dBm analyzer measurements. This also indicates that our
PDL noise floor is equal to or lesser than the RF analyzer floor.
Table 8.2: RF analyzer-limited C/N ratio measurements for five different time delay setting of the PDL.
These analyzer noise floor limited measurements were obtained over a 250 kHz analyzer bandwidth using an
analyzer resolution bandwidth of 1 kHz.
PDL Setting
Carrier-to-Noise ratio
Bit 1: Non-Delay
Bit 2: Non-Delay
(Setting 1)
78.97 dB/Hz
(Setting 2)
77.17 dB/Hz
(Setting 3)
79.34 dB/Hz
(Setting 4)
76.16 dB/Hz
(Setting 5)
67.17 dB/Hz
Bit 3: Non-Delay
Bit 1: Non-Delay
Bit 2: Non-Delay
Bit 3: Delay
Bit 1: Non-Delay
Bit 2: Delay
Bit 3: Non-Delay
Bit 1: Non-Delay
Bit 2: Delay
Bit 3: Delay
Bit 1: Delay
Bit 2: Non-Delay
Bit 3: Non-Delay
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The signal level detected after propagation through the PDL and fiber link is -54.66
dBm, indicating a 26.33 dB PDL RF insertion loss. If we treat the PDL as an attenuator,
then an equivalent C/N can be calculated and compared to the original link C/N. This
equivalent C/N can be calculated by taking a sum of the measured link-PDL C/N plus the
attenuation (or negative gain) of the PDL. In this case, we get 79.34 + 26.33= 105.67
dB/Hz, which is similar to the C/N available form the Lasertron fiber-optic link. Hence,
our PDL does essentially act as an attenuator to the microwave fiber-optic link, with
minimal link C/N reduction due to increased RF noise.
The measured PDL setting 3 RF insertion loss of 26.33 dB is equivalent to a 13.17
dB optical insertion loss. From Table 8.1, we note that a 5.5 dB optical insertion loss is
measured for the PDL setting 3 from the bit 1 input port (after the polarizer that follows the
NLC device) to the bit 3 output port (just before output GRIN). Compared to the direct
fiber connection between the fiber-optic transmitter and fiber-optic receiver, the remote
fiber link connection to the PDL uses one extra FC/PC connector, a PDL input light
polarization controller (PC), and free-space light coupling to the output GRIN-lens. All
these items cause the extra 13.17-5.5=7.7 dB optical insertion loss in the system. We
measured a 4.3 dB (or 63%) optical insertion loss in the free-space-to-output GRIN-lens
coupling optics. Hence, the remaining 7.7-4.3=3.4 dB optical insertion loss occurs due to
the PC and the additional FC/PC connector. With proper design, these losses can be greatly
reduced.
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.3.4. Time Delay Measurements and PDL Switching Speed
Fig. 8.5 shows the time delayed signal for three different PDL settings. The top
trace (trace A) shows the signal when light goes through the non-delay setting for all the
bits (setting 1). The middle trace (trace B) is obtained when bit 2 and bit 3 are set for delay
and bit 1 for non-delay (setting 4). The designed time delay is 1.67 ns. The delay measured
using a Tektronix digital oscilloscope is 1.66 ns (Fig. 8.5a). The bottom trace (trace C) is
obtained for the PDL setting in which the first bit is set for delay and the other two for non
delay (setting 5). The designed time delay was 5.67 ns. The oscilloscope trace measures a
time delay of 5.72 ns as shown in Fig. 8.5b (Fig. 8.5b shows a 0.72 ns time difference
between traces A and C, because the signal has been delayed by 5 cycles plus the 0.72 ns.)
The slight discrepancy in the designed and measured time delays for our PDL occurs due to
the variable and finite manufacturer specified error tolerances of the optical components
used in the PDL, as well as due to the tolerance of the optical path length difference
between the non-delay and the delay path, plus the measurement errors due to the use of a
digital oscilloscope to measure time delays. The resolution of the oscilloscope
measurements was 1 ps.
The temporal response characteristics of the FLC devices are also tested, as the
higher switching speed of FLCs compared to NLCs is the key motivation for using these
FLC devices. The temporal response is measured at the output of the first bit. The FLC
devices are driven with the specially optimized waveform mentioned in Chapter 4. This
waveform exhibits a ±15 V switching transient and a ±5 V holding voltage. In order to
observe the switching time of the FLC devices, either the non-delayed or delayed signal has
to be isolated. In our measurements the non-delayed signal is isolated by blocking the delay
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 8.5: Oscilloscope traces showing the time delayed signals for three different time delay settings of
the PDL. The top trace (trace A) corresponds to the measured zero delayed signal, the middle trace (trace B)
corresponds to a measured 1.66 ns delayed signal, and the bottom trace (trace C) corresponds to a 5.720 ns
measured delayed signal. The arrows show the points between which the time delay measurement was done.
Scope resolution is 1 ps.
path. A photodetector (with a rise time of < 2 (is) is placed at the output of the bit and
detects the temporal response signal due to the finite switching time of the two FLC
devices. Fig. 8.6 shows oscilloscope traces of the FLC device driving signal (bottom trace)
and the overall temporal response signal for the FLC device-switched optical signal (top
trace). The driving waveform was attenuated by 10 dB before being fed to the oscilloscope
to avoid over driving the oscilloscope. Measurment showed that it takes about 33 (is of
time delay for the FLC device to respond to the new driving signal level. Also, the rise and
fall time of the FLC device defined as the time it takes for the FLC device to change from
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10% to 90% (or from 90% to 10%) of its final optical power value, was measured as 35
Its. Thus, the total time it takes for the FLC devices in the PDL to switch between the delay
and no delay modes of the PDL is ~ 68 (is. Note that in most phased array applications, the
time sequenced antenna beam positions (and hence PDL delay/no-delay settings) are known
a priori. This means that we can overcome the finite delay time of the FLC response if we
apply the FLC device control signals a bit earlier, i.e., 33 (is earlier in the case of our
experimental PDL.
(a)
(b)
Figure 8.6: Oscilloscope traces (a) the 33 (is time delay before the FLC device starts responding to the
applied voltage and (b) the 35 (is (10% to 90% or vice versa) rise time or fall time. (Top trace: the
photodetected optical output showing the FLC-device time response. Bottom trace: the specially optimized
waveform with a ± 15 V switching transient voltage and a ± 5 V holding voltage. Note that the driving
voltage observed on the oscilloscope has been attenuated by 10 dB).
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.4. Conclusion
In conclusion, a 3-bit PDL using FLC devices was demonstrated for the first time.
Although current FLC devices have limited performance in terms of output beam
polarization extinction ratios, we have used our second cascaded FLC device based active
noise filter to demonstrate high isolation optical switching with a switching time of 35
jisec, over an order of magnitude improvement over our previously demonstrated NLC
switches. This 3-bit PDL also uses imaging optics and system remoting via a directly
modulated 1310 nm semiconductor laser fiber-optic link, thus demonstrating low
interchannel crosstalk and fiber-based remote control, respectively. Low -54.94 dB RF
interchannel crosstalk was directly measured in the nearest adjacent PDL output channel.
Spectrum analyzer-limited data indicates that our PDL system essentially does not raise the
RF noise floor. In essence, the PDL acts as an RF attenuator, providing additional insertion
loss to the fiber link-PDL system. PDL measured insertion loss has been characterized
indicating that a near 1.5 dB optical loss is achievable per bit if optical components are AR
coated. Further insertion loss improvements are possible by reducing the FLC device
insertion losses, improving output free space-to-GRIN lens optical coupling efficiency, and
replacing the five non-AR coated optical components.
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 9
SWITCHED PHOTONIC DELAY LINE FOR PHASED ARRAY
ANTENNA CONTROL USING EXTERNALLY MODULATED
MICROWAVE FIBER-OPTIC LINK
9.1. Introduction
In the previous chapter we demonstrated a 3-bit PDL that was fed by a direct
modulation fiber-optic link. Although direct modulation fiber-optic links are smaller in size
and low in cost, they presently have limited dynamic range and frequency response when
compared to extemaly modulated links [1]. On the other hand, the higher RF gain and
lower noise figure of external modulation analog fiber optic links can offer better dynamic
range [2]. In this chapter, a 4-bit PDL using FLC devices and an externally modulated
fiber-optic link [3] is demonstrated for the first time. A Mach-Zehnder integrated-optic
modulator is used to modulate the light. Three dimensional imaging optics and
antireflection coated optics are used to minimize PDL insertion losses and interchannel
crosstalk.
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.2. The External modulation Fiber-optic link fed photonic delay line
The experimental set-up of the external modulation fiber-optic link fed PDL is
depicted in Fig. 9.1. A high optical power (250 mW) CW dioded pumped Nd:YAG laser
(A^=1319 nm) is used as the laser source. The laser is connectorized with a PM-fiber. A
variable optical attenuator is used to contol the optical power of the light before going
through the Mach-Zender electro-optic modulator. Then light is launched via a GRIN-lens
pigtailed PM fiber into the PDL. All fiber to fiber connections are done using FC/PC
connectors. The polarization ER after the GRIN lens is 30 dB. A high extinction ratio
polarizer is used at the input of the PDL. To improve the polarization ER of 40 dB. A
GRIN-lens pigtail single-mode fiber is used at the output of the PDL to collect and direct
the light to the detector.
The experimental set-up of the 4-bit PDL line is shown in Fig. 9.2. The first bit is
based on the adaptable PDL design [4, 5]. The delay path consists of a single mode fiber
with a Faraday rotator mirror (FR-M) with a total fiber length of 3.15 m. Lenses LI and L2
PM-Rber
Variable
Optical
Attenuator
PM-fiber
PM-Rber
^
PM-Rber
Bectro-Opdc
Modulator
R Fln
CW High
Power Lsser
PM-Rber
GRIN-lens
Signal
RFOut
H ighspeed
Figure 9.1: The experimental set-up of the external modulation fiber-optic link fed photonic delay line.
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
are of the same focal lengths. The light travels twice through the fiber and thus gets a
relative time delay o f 31.5 ns compared with the signal traveling through the other non­
fiber path. The second PDL bit is also based on the adaptable PDL architecture. The focal
length of lens L3 is 128.032 mm and of lens L4 is 130.8 mm, a 14.95 mm thick glass
(OHARA: LAH-58) is used to adjust the delay for the desirable 0.2 ns time delay. The third
bit has been designed for a 0.1 ns time delay and is based on the symmetric architecture,
where the delay and the non-delay paths have the same physical length. Two thick (17.33
mm) glass plates (OHARA: LAH-58) are placed in the delay path to make the optical path
longer, and thus obtain the desired time delay. L5 and L6 have focal lengths of 260.040
mm and 256.064 mm respectively. Two HWPs are placed before the output PBS to rotate
the polarization of each of the paths by 90°. Hence, each of the two possible signals
undergo one reflection and one transmission through the PBSs, thus giving a balanced loss
PDL bit performance. The final bit is based on the adaptable PDL architecture and has been
designed for 0.4 ns time delay. L7 and L8 have focal lengths of 256.064 mm and 260.04
mm respectively. A glass plate of 39.59 mm thickness is placed in the delay path to create
the appropriate optical path length difference between the two paths and give the desired
delay of 0.4 ns.
Each PDL bit has two FLC polarization switching devices. The first one is used for
the optical path switching, and the second FLC device along with a high extinction ratio
polarizer form the active noise filter that supresses any noise leakage from the first FLC
device and the PBS. The two FLC devices are chosen to operate out of phase. That means
that whenever one is set “on” and thus rotates the incident polarization, the other one is set
“off’ and leaves the polarization unaffected. This out of phase operation has been found to
give lower leakage noise levels for our PDL modules [6].
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.51 m
M
QWP
BIT 1
0.8 ns
From Modulator
PM-Fiber
FLC1
0.9 m
JM
PM-Fiber
FLC3
o\
r
f
-
i
. pDCC ZZ 1l * FLC2
'M
-
I
M
qw p
* 1
y*
' PFLC4
tra.
Glass
Plate
BIT 2
0.27 ns
Glass Plate 2
L5-
L4»'
M
BIT 3
0.1 ns
BIT 4
0.42 ns
M
m
L5
t
L7
QWP
z
pgg FLC6 P JFLC7
L8
*
FLC8
Glass
Plate 3
M
mbbbbi
UGRIN-lens
SM-Fiber
To Detector
Figure 9.2: The experimental 4-bit photonic delay line using FLC devices, imaging optics, and fiber-optic remoting. The non-delay paths are
represented with solid lines while the delay paths are represented with dashed lines.
9.3. Photonic Delay Line Optical Loss Budget
One of the main issues of any PDL is the optical insertion loss. We have proposed
the use of AR coated optical components to reduce the optical insertion loss of our PDL.
Table 9.1 shows the expected and measured optical loss for both settings of each of the
PDL bits. The expected optical insertion loss was calculated based on the optical
component loss values given by the manufacturers, and the measured optical losses for the
FLC devices.
From Table 9.1 we can see that the losses are at the expected levels. Overall, our
optical insertion loss numbers are limited from the rather high loss of the FLC devices. The
measured typical optical loss for our FLC devices is 0.7 dB. Considering that each PDL bit
has two FLC devices, a 1.4 dB loss is due to the FLC devices. This high loss is due to the
fact that the FLC devices consist of a cascade of three FLC cells, each sandwiched between
two glass plates. Triple cells are currently used because of the current limitation of the
manufacturer to provide thick enough single cell FLC devices that can rotate the
polarization by 90° at the required 1319 nm wavelength. It is anticipated that as technology
develops and new FLC materials are synthesized, double cell or single cell FLC devices
with lower insertion losses (~ 0.3 dB) will become available. This will further reduce the
optical insertion loss of our PDL by ~ 0.8 dB per bit, implying a typical 1 dB optical
insertion loss per bit. The measured optical loss for the 16 settings of our 4-bit PDL are
shown in table 9.2. These measurements were obtained using an optical power meter at the
output of the PDL.
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9.1: Expected and measured optical insertion loss for the binary settings of each of the four PDL
bits.
Non-Delay
Delay
Expected
Measured
Expected
Measured
Optical Loss
Optical Loss
Optical Loss
Optical Loss
(dB)
(dB)
(dB)
(dB)
1
1.72
1.66
4.28
4.50
2
2.02
1.90
2.04
1.98
3
1.99
1.91
2.01
1.94
4
1.81
1.90
1.96
2.00
#B it
Table 9.2: Average PDL optical loss
PDL Setting
Measured Optical
PDL Setting
Measured Optical
Bit 1 Bit 2►Bit 3 Bit 4
Loss (dB)
Bit 1 Bit 2» Bit 3 Bit 4
Loss (dB)
N
N
N
N
7.35
D
N
N
N
10.71
N
N
N
D
7.46
D
N
N
D
10.77
N
N
D
N
7.38
D
N
D
N
10.79
N
N
D
D
7.15
D
N
D
D
10.86
N
D
N
N
7.29
D
D
N
N
10.62
N
D
N
D
7.41
D
D
N
D
10.73
N
D
D
N
7.35
D
D
D
N
10.64
N
D
D
D
7.49
D
D
D
D
10.77
D: Delay setting; N: Non-delay setting
169
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.4. Improved Output Coupling Efficiency using Multimode Fibers
Another important issue in our PDL is the output optical coupling system. The PDL
consists of 4 bits, and thus has sixteen different settings. The signals of those settings
travel through sixteen different path combinations. All these different signals have to be
effectively collected and detected. Since fiber remoting is proposed to distribute the
processed optical signals to the photodetectors, a large area (1.8 mm in diameter) GRINlens pigtailed fiber can be used to maximize the optical coupling efficiency. GRIN-lens
pigtailed SM- fiber as the collecting element has been proposed [7]. Nevertheless, the
coupling losses associated with the GRIN-lens pigtailed single-mode fiber make this
approach rather intensive since all sixteen settings have to be collinear at the output of the
PDL. Hence,a significant amount of this insertion loss is due to the limited microlens to
SM-fiber coupling efficiency at the output plane of the PDL. In our previous 3-bit PDL
experimental demonstration we measured a 4.3 dB loss due to the SM-fiber coupling
efficiency. Note that GRIN lens manufacturers normally specify a < 2 dB optical loss for
two GRIN lenses placed with a free-space gap of < 2.5 cm [8]. This coupling loss is due to
the fact that even the slightest free-space beam expansion results in a change in the beam
wavefront which in turn limits coupling efficiency into the SM-fiber. In our PDL, the
distances are larger than 2.5 cm, hence we use imaging optics. Nevertheless, effects such
as aberrations from various optical elements degrade the quality of the PDL laser beam at
the PDL output plane, which in turn further decreases the GRIN-to-SMF coupling
efficiency.
Ideally, in order for the laser beam to be coupled efficiently into a GRIN lens
pigtailed single mode fiber, the beam diameter (twice the beam waist) has to be of the order
of 0.4 mm [8]. The GRIN lens receives the collimated beam and focuses the light into the
170
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
single mode fiber. If the beam is not perfectly collimated or its size is larger than 0.4 mm
on the front surface of the GRIN lens, the GRIN lens does not focus the light to a small
spot required for coupling the light into the fiber. Note, that the core diameter of a single
mode fiber at 1310 nm is 9 pm, and the numerical aperture (NA) is 0.11, which makes this
SM-fiber coupling a rather difficult issue. One approach to solve this problem in our PDL
is to use multimode fibers instead of single mode fibers at the PDL output plane. Typical
core diameters of multimode (MM) fibers range from 50 pm to 200 pm, with NA from 0.2
to 0.37 respectively. The larger fiber size of the MM-fiber can give greater collecting power
and can thus make the system less sensitive to small optical misalignments, making the
coupling much easier, even for cases where the beam has suffered diffraction and
aberrations. Key parameters of the multimode fiber are the ease of connectability, modal
dispersion, and modal noise. For our phased array antenna control application, the
multimode fibers will be used to transfer the optical signals from the controller site to the
antenna site. For ship board applications, this is a rather small distance of under 100 m.
Thus, no significant effects due to modal dispersion are expected due to the short haul fiber
delivery distance.
Two sets of measurements were performed, one using a GRIN-lens pigtailed
single-mode fiber (9 pm core diameter) and another using a GRIN-lens pigtailed multimode
fiber (50 pm core diameter). The measurements were performed using an optical power
meter to detect the light at the output end of the fiber that would eventually connect to a
photodetector. The average coupling loss for the 16 settings of the PDL when the single­
mode fiber was used was 2.2 dB with a standard deviation of 0.5 dB. When the multimode fiber was used, the coupling loss reduced to 0.5 dB, with a standard deviation of
0.32 dB. This 1.5 dB optical coupling efficiency improvement will eventually give a 3.0
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9.3: Coupling loss for the SM-fiber and the MM-fiber FO collimator system.
PDL Setting
SM-fiber
PDL Setting
MM-fiber
B iti Bit 2 Bit 3 Bit 4
Coupling Loss
Bit 1 Bit 2 Bit 3 Bit 4
Coupling Loss
(dB)
(dB)
N
N
N
N
1.67
N
N
N
N
0.16
N
N
N
D
1.64
N
N
N
D
0.17
N
N
D
N
2.01
N
N
D
N
0.23
N
N
D
D
2.28
N
N
D
D
1.14
N
D
N
N
1.73
N
D
N
N
0.20
N
D
N
D
1.78
N
D
N
D
0.41
N
D
D
N
1.94
N
D
D
N
0.21
N
D
D
D
1.93
N
D
D
D
0.18
D
N
N
N
2.19
D
N
N
N
0.81
D
N
N
D
2.22
D
N
N
D
0.84
D
N
D
N
3.12
D
N
D
N
0.70
D
N
D
D
3.2
D
N
D
D
0.75
D
D
N
N
1.96
D
D
N
N
0.82
D
D
N
D
1.91
D
D
N
D
0.82
D
D
D
N
2.6
D
D
D
N
0.73
D
D
D
D
2.61
D
D
D
D
0.69
D: Delay setting; N: Non-delay setting
172
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dB RF gain to the signal when a multi-mode fiber is used. Experiments were also
performed using a 200 |im core diameter multi-mode fiber. No improvement was observed
in the optical coupling efficiency of the output port fiber coupling system compared with
the 50 |im fiber.
9.5. The Leakage Noise Issue
Another system issue of the PDL is the within channel leakage noise, or the SNR
issue. Table 9.4 shows the RF SNR obtainedvalues for the 16 settings of the PDL. Both a
single-mode (9 pm core diameter) and a multi-mode (50 pm core diameter) fiber were used
at the PDL output port to obtain the measurements. From Table 9.3 it can be seen that the
PDL electrical leakage noise is less than -70 dB for all 16 time delay settings, using either
of the output port coupling systems. The use of the multi-mode fiber does not degrade the
SNR performance of the PDL. Thus, in conjunction with the higher optical coupling
efficiency that provided by the multi-mode output fiber, it is the preferred coupling output
fiber. This results also show that cascading binary PDL modules does not sigbnificantly
degrade the SNR performance. Remember that in chapter 6 a 40 dB optical (or 80 dB
electrical) SNR value was obtained for a single bit adaptable PDL. The SNR obtained for
the 4bit PDL is close to the 80 dB electrical SNR value of the single bit PDL.
173
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9.4: Electrical SNR values of the PDL using (a) GRIN lens pigtailed single-mode fiber and (b) GRIN
lens pigtailed multi-mode fiber coupling system.
Electrical Leakage Noise-based SNR (dB)
PDL Setting
Single-mode Fiber
Multi-mode Fiber
Coupling System
Coupling System
Bit 1 Bit 2
Bit 3
Bit 4
N
N
N
N
79.18
78.80
N
N
N
D
79.03
78.67
N
N
D
N
90.68
88.09
N
N
D
D
89.82
86.00
N
D
N
N
74.15
73.78
N
D
N
D
73.98
74.08
N
D
D
N
79.46
78.70
N
D
D
D
79.68
80.11
D
N
N
N
75.41
76.18
D
N
N
D
75.51
77.67
D
N
D
N
74.02
76.90
D
N
D
D
73.94
77.95
D
D
N
N
71.48
70.29
D
D
N
D
71.95
70.46
D
D
D
N
73.82
74.08
D
D
D
D
74.15
73.64
D: Delay setting; N: Non-delay setting
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.6. RF Power Measurements
The electro-optic modulator (UTP, APE MZM-1.3-8-T) was fed by a 6 GHz,
11.98 dBm RF power level signal from our HP 83752A signal generator. The C/N of this
input RF signal measured at 100 kHz offset was 107.0 dB/Hz (Fig. 9.3).
T
(a)
(b)
Figure 9.3: RF spectrum analyzer traces at 6 GHz showing (a) the 11.98 dBm RF signal fed to the electooptic modulator; (b) its 107 dB/Hz C/N measured at 100kHz offset
First the performance of the externally modulated fiber-optic link without the PDL
was tested. The optical attenuator in the fiber optic link was adjusted so that a 3 mW optical
power level is incident on the photodetector. This optical power is selected because the
photodetector cannot tolerate higher optical powers. The RF gain of the externally
modulated FO link is shown in Fig. 9.4. The fiber-optic link exibits a -42 dB RF gain (or a
42 dB RF loss). The C/N remains near the same level, i.e., at 104.2 dB/Hz (Fig. 9.4 (b)).
175
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 9.4: RF spectrum analyzer traces showing the 42 dB RF loss of the externally modulated 6 GHz
fiber-optic link, (a) RF power drops at -30.01 dB, (b) a 104.2 dB/Hz C/N is measured at 100kHz offset.
When the PDL is inserted into the externally modulated link path via the use of
fiber-to-free space coupling optics an extra RF loss is expected. A single-mode fiber was
used at the output of the PDL because a single mode connectorized photodetector was
available. Fig. 9.5 shows an additional RF loss of 19.16 dB of the externally modulated
fiber-optic link due to the PDL (PDL setting: zero delay). Note that this RF loss also causes
a C/N degradation of about ~ 19 dB.
The C/N of the system should be kept at the same level after propagation of the
signal through the PDL. This is particularly important in the receive mode of the photonic
beamformer as signals coming from the antenna elements are lower in power.
The external modulation FO link gives the extra degree of freedom of adjusting the
input optical power to the PDL. This will have as a result higher optical power impinging
on the photodetector. Thus, by using higher optical powers the insertion loss of the PDL
can be compensated. This is valid only because the photodetector is capable of handling
only low optical powers (e.g., -3 mW) and because the laser can have much higher optical
power (e.g., 200 mW). In our experiment, the optical attenuation of the laser light was
reduced by 9.5 dB using the variable attenuator. The RF power detected reached the initial
176
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
value of -30.00 dBm and the C/N also got higher and approached its original value at
104.00 dB, as shown in Fig. 9.6.
Note a -1 9 dB
C/N degradation
'-':K:l*
PDL Setting Optical Loss: 135 dB (measured)
Input Optical Fiber and Input Polarizer Optical Loss: 0.1 dB (measured)
Output SM-Fiber Coupling Loss: 1.8 dB (measured)
One additional FC/PC Connector 035 dB (typical)
{
ExPcctcd RF L®** 19 00 dB
Figure 9.5: Spectrum analyzer trace showing a 19.16 dB RF loss o f the 6 GHz fiber-optic link with the
PDL set for zero delay. This 19.60 dB RF loss is consistent with the expected RF loss based on the optical
losses due to the PDL setting, the input polarizer, output single-mode fiber and the additional FC/PC
connector.
I ''’
(a)
' ''W V ’
(b)
Figure 9.6: RF spectrum analyzer traces showing the dynamic range loss compensated 6 GHz (a) RF power
at -30.00, and the (b) 104.0 dB/Hz C/N measured at 100 kHz offseL
177
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Note that the PDL does not degrade the C/N of the signal. As C/N degradation we
define the difference between the C/N when the PDL is not included in the link path and
when the link is included in the link path. Overall the C/N degradation for our PDL over the
16 time delay settings is 2 dB.
Post-amplification of the photodetected signal can also be used. Thus, the RF
power level can reach higher values such as those required for phased array antenna
applications. Again the optical attenuation of the laser light was reduced by 9.5 dB. A 22
dB RF amplifier (Mini Circuit: Z-RON8) is used to amplify the photodetected signal. This
time, the RF signal reaches -8.33 dBm (Fig. 9.7(a)), and the C/N rises close to its original
value at 103.8 dB/Hz (Fig. 9.7 (b)). Note that in this case the use of a post amplifier will
probably deteriorate the dynamic range (DR) of the FO link - PDL system [9]. The effect of
the PDL on the DR of the system will be examined in the next chapter.
.Yi!"
oV )
1
Figure 9.7: RF spectrum analyzer traces 6 GHz showing (a) the RF power at -8.33 dBm, and (b) the 103.8
dB/Hz C/N measured at 100 kHz offset
178
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.7. S-Band Operation of the Externally Modulated PPL
One of the advantages of photonic beamforming is the ability to change RF
frequency of operation over wide bandwidths without essentially changing any of the
hardware. All previous experimental demonstration was performed at 6 GHz. Fig. 9.8
shows operation at 3 GHz. Fig 9.8 shows a 12.17 dBm, 108 dB/Hz C/N signal that is fed
into the electro-optic modulator. Fig 9.9 shows the post amplified photodetected signal.
The PDL was set for zero time delay and the optical attenuator was set so that the same
optical power is impinging on the photodetector as for the FO-Iink without PDL case.a 9.5
dB gain higher optical power compared to the link without the PDL.
, »
■
‘i u h
(a)
..I
r
'T
•
lirM.
■ ’' • w
.i!
•
(b)
Figure 9.8: RF spectrum analyzer traces of the 3 GHz signal fed into the electro-optic modulator (a) RF
power. 12.17 dBm, and (b) C/N: 108.3 dB/Hz measured at 100 kHz offset
179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 9.9: RF spectrum analyzer traces of the 3 GHz photodetected signal after post amplification (a) RF
power -7.50 dBm, and (b) C/N: 105.0 dB/Hz measured at 100 kHz offset
9.8. Time Delay Measurements
Time delay measurements were also obtained for the 4-Bit PDL. As mentioned
earlier the second, third and fourth bits in the PDL have been designed for a 0.2 ns, 0.1 ns,
and 0.4 ns time delay, respectively. Fig. 9.10 shows, the measured time delays obtained
from the second, third and fourth bits respectively. The top traces show the non-delayed 6
GHz signal and the bottom traces show the 6 GHz delayed signals. The measured values
are shown in Fig. 9.10; (a) 0.270 ns, (b) 0.105, and (c) 0.425 ns. This difference from the
designed time delays is due to tolerances due to the lens and glass fabrication process as
well as due to small displacement errors of the optical components in the delay or non-delay
paths. The laser source used in the experiment had multiple longitudinal modes, with a
longitudinal mode spacing of 4.5 GHz. Thus, there were RF noise spikes at 4.5 GHz
intervals. This is the reason why the 6 GHz signal on the oscilloscope has some ripples.
The solution to this problem is to use a single frequency laser as an optical sourse.
180
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(C)
Figure 9.10: Oscilloscope traces at 6 GHz showing a (a) 0.270 ns, (b) 0.105 ns, and (c) 0.425 ns time delay
for the second, third and fourth PDL bits. (Top traces: the non-delayed signal, bottom traces: the delayed
signal).
181
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.9. Interchannel Crosstalk
Interchannel crosstalk measurements were also obtained for our PDL. As
interchannel crosstalk, we define the optical power leaking from one channel to the adjacent
ones. In the experimentally demonstrated PDL, there is only one active channel. Thus, in
order to take the interchannel crosstalk measurements, the output GRIN lens of the multimode fiber coupling system was translated in the two orthogonal directions of the output
plane (-x and -y) in steps of 1.8 mm, which is the typical diameter of the GRIN lens. The
total translation in each direction was 7.2 mm which corresponds to five GRIN-Iens
positions. One position is at the center, and the others are at 1.8 mm and 3.6 mm from each
side of the center GRIN-lens. This 7.2 mm span was used because in our system we used
10 mm active area polarizers. Fig. 8.11 shows typical interchannel crosstalk values. Note
that a worst case leakage noise of -42 dB was measured in the adjacent channel, which
leads to an RF interchannel crosstalk of - 84 dB.
9-* r . -
-30.
1r*t* .
B4ah»CB>lNTrmpn«l>ihg» i(ijM }.
JM rtwQElW J ^ p jjriHb«>»7(pp)L,-
(a)
(b)
Figure 9.11: Optical interchannel crosstalk relative to the center active channel when the PDL is for
maximum delay setting, with measurements taken along the (a) x and (b) y directions at the PDL output
plane. A maximum optical interchannel crosstalk o f -42.1 dB (or -84.2 dB RF) is measured at the nearest
to center channel in the y-direcdon.
182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9.10. Conclusion
A 4-bit PDL using ferroelectric liquid crystal devices and externally modulated S/C
band FO-link for fiber remoting was demonstrated for the first time. An average low PDL
RF leakage noise (<-77 dB) over the 16 time delay settings was demonstrated. Low 1.9
dB/bit average optical loss was measured, which can be further reduced to -1 dB using
lower loss (0.3 dB) FLC devices. A RF gain of 3 dB with a multimode fiber output port
versus a single-mode fiber output port was also demonstarted. A -105 dB/Hz C/N was
also maintained by our PDL system, which is consistent with the generator signal quality
driving the integrated-optic modulator. The C/N degradation due to the PDL in the
externally modulated fiber-optic link path was estimated at 2 dB. Low interchannel RF
crosstalk (averaged < - 84 dB) was measured relative to the center active channel of the
PDL.
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 10
SYNCHRONOUS AMPLITUDE AND TIME CONTROL FOR AN
OPTIMUM DYNAMIC RANGE VARIABLE PHOTONIC DELAY LINE
10.1. Introduction
As we have seen, one of the main issues with variable PDLs and their use in
practical phased array antenna applications is their insertion loss. RF insertion loss
numbers of 3.6 dB/bit have been demonstrated [1] and are expected to be reduced at < 2.8
dB/bit with improved FLC devices. Nevertheless, this PDL loss, in conjunction with the
limited FO-link gain, especially at high RF frequencies (e.g., > 2 GHz) is sometimes the
limiting factor in terms of compression dynamic range (CDR), spurious free dynamic range
(SFDR), and noise figure (NF) of the photonic system. These factors are important for
phased array antennas applications, and hence are the subject of this chapter with respect to
PDLs.
Over the last few years, the use of analog FO links for phased array antennas has
been explored [2]. Currently, external modulation FO links give more degrees of freedom
to the optical engineer to obtain optimized performance. Thus, an external modulation FOlink gives better performance characteristics for analog applications compared to direct
modulation FO-links [3,4]. Externally modulated FO-links compared to directly modulated
184
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
links have demonstrated higher gains [4, 5], higher modulation frequencies [6] and
bandwidths [5, 6], and higher dynamic range [7]. Specifically, a gain of + 11 dB has been
reported using a 22 MHz bandwidth [4]. Frequencies of 20 GHz with a 6.6 GHz
bandwidth [6] have been demonstrated, and maximum CDR of 160 dB-Hz and SFDR of
117 dB-H z273 has been reported for an external modulation link at 870-930 MHz [7].
Using a balanced detection scheme for a 3 GHz FO link, a CDR of 168.4 dB -Hz and a
SFDR of 119.5 dB-Hz273 has also been reported [8]. Thus, the preferred choice at present
for phased array antenna applications is external modulation FO links.
Externally modulated FO-links typically include a diode-pumped solid state laser
with an optical output power of -200 mW or more, a Mach-Zehnder integrated electro-optic
modulator capable of handling 200 mW or more of CW optical power, and a fast
photodetector that usually has a maximum acceptable optical power of 3-5 mW for
operation at frequencies > 1 GHz. This 3-5 mW limitation is due to the non-linear response
of the photodiode at high optical power densities. The low optical input power
photodetector is currently the limiting factor for low or negative FO link gains operating at
frequency bands of a few gigahertz. Gain is considered in the general sense where negative
gain means loss. The lower levels of acceptable optical power on the photodetector is due
to the small photodetector area (e.g., - 10 - 20 Jim in diameter) required so that the
corresponding capacitance does not limit the link operational bandwidth [5]. Thus, the
available optical power from the laser needs to be attenuated before reaching the receiver.
This can be done by using a FO attenuator as shown in Fig. 10. i. There are two possible
FO attenuator positions in the FO link. The first is before the electro-optic modulator and
the second is after it. Since the modulator is capable of handling the laser optical power, we
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
select to place the FO attenuator after the modulator. For reasons that will be discussed later
(section 10.4) this FO attenuator position gives a better performance to the FO-link in terms
of noise issues. Typical FO link gain numbers obtained using a diode pumped Nd:YAG
laser (X = 1319 nm), a UTP Mach Zehnder analog modulator, and a Lasertron fast
photodetector for operation in the 3-6 GHz band is — 30 dB. Preamplifier and/or post­
amplifiers can be used to improve the FO link gain, but this may affect the NF and/or the
dynamic range of the link [9].
Light
PM-Fiber
PM-Fiber PM-Fiber
Electro-Optic
Modulator
PM-Fiber
FC/PC - FC/PC
CWHigh
Power Laser
PM-Fiber
Variable
Optical
Attenuator
Highspeed
Photodetector
FC/PC-FC/PC
RFIn
RFOut
Figure 10.1: Typical experimental set-up for the externally modulated fiber-optic link. A fiber-optic
attenuator is used to adjust the optical power impinging on the photodetector. (FC/PC: flat
connection/physical connector type FO-connector)
To use a FO link for phased array antenna applications, a variable PDL is inserted
after the FO attenuator. The optical insertion loss of the PDL will lower the optical power
detected at the photodetector and thus will further reduce the FO-link gain. Thus, a limited
DR will be obtained for the FO link - PDL system. This is particularly important in the
receive mode of the photonic beamformer as signals coming from the antenna elements are
lower in power. Note also that for most advanced phased array applications, the signals
driving the antenna elements are required to be of equivalent amplitudes. Thus a balanced
loss performance between the different settings of a PDL is also a critical system design
need. We have proposed a balanced insertion loss PDL bit architecture that has equal loss
186
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for each PDL bit setting [10]. Nevertheless, small insertion loss variation due to the
different optical losses of the optical components or due to the loss variation throughout
their entire active area may lead to an insertion loss variation for the different PDL channels
and/or settings. Thus, the use of the optical attenuator is required not only to provide the
optimum optical power for maximum dynamic range but also to provide equal optical signal
amplitudes for all different PDL settings and channels.
In this chapter, we describe an optical amplitude control system that operates in
synchronism with the PDL to provide the necessary signal amplitude calibration and the
important dynamic range loss compensation that gives optimized RF performance of our
FO link-PDL system [1]. We also demonstrate how this optical attenuation system can be
used to obtain maximum RF dynamic range for a FO link-PDL system composed of
commercially available
components. The proposed signal amplitude calibration and
dynamic range loss compensation technique is not limited to our PDL design and
technology, and can also be used with alternative PDL approaches. To our knowledge, our
amplitude controlled approach gives the highest FO link dynamic range ever reported when
using a switched PDL system [11-16]. The previously reported high dynamic range (160
dB-Hz) has been demonstrated using a non-switched fiber-optic PDLs [17].
10.2. The Synchronous Signal Calibration and Dynamic Range Loss Compensation
Technique
The proposed synchronous signal calibration and DR loss compensation method is
based on the high optical power available from the laser source and an electrically
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
controlled optical attenuator after the electro-optic modulator. This attenuator is set such that
the incident optical power after propagation and attenuation through the PDL is at levels
required from the detector for maximum photodetected dynamic range. Fig. 10.2 shows
such a configuration where the optical attenuator is computer controlled. For PDL
calibration purposes on a day-to-day basis, the RF power splitter sends a small portion of
the RF power to the power meter and with the help of a computer and a data base, the
appropriate feedback signal is sent to the attenuator for fine control. Because the PDL loss
for each bit setting is known, the attenuator settings are also known a priori, and thus the
appropriate attenuator setting is applied via an electronic signal controlling the high speed
optical attenuator. Advanced phased array antenna applications require switching times of
few microseconds. Currently, the fiber-optic attenuators available in the market are limited
to slower switching times (e.g., 200 ms full span scanning for a 30 dB attenuation range)
[18]. Thus, a faster switching speed optical attenuator is required.
PM-Fiber
PM-Fiber
Light
_____
PM-Fiber PM-Fiber
Qectro-Opuc
Modulator
Q e^Q
FC/PC-FC/PC
PM-Fiber
PM-Fiber
Variable
Optical
Attenuator
Photonic
GRIN-lens
RF In
Control Signal
Depending on the
PDL Setting
CW High
ower Laser
GRIN-lens
SM-Fiber
RF Power Meter
Control Computer
RF
Splitter RF Out
.
f lu
High Speed
Photodetector
RF to Antenna
Figure 10.2: Dynamic range loss compensation method based on high speed electronic control of the
variable optical attenuator in synchronous control with the PDL settings.
188
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 10.3 shows the basic structure of a digital control, high speed, binary design,
variable optical attenuator that operates in synchronism with the variable PDL. Each
attenuator module has two possible settings; the zero attenuation (A0) setting and the A,,
where i = 1, 2,... N, attenuation setting. This is achieved by having two light propagation
paths with different optical losses in the attenuator module. One path gives the zero
attenuation (AJ and the other one gives the Ai attenuation. The attenuation modules are
arranged in a cascade switched configuration, with each module having twice the
attenuation of the previous one, i.e., AI+I = 2 x A r Thus, the cascade of these binary
attenuation modules provides the desired attenuation range. Note that since PM fibers are
used in the system configuration (see Fig. 10.2), a polarization dependent high speed
optical attenuator can be used. Similar to our PDL designs, one polarization based
attenuator uses high speed (e.g., 35 (is) binary FLC polarization switching devices and
PBSs to route the optical signal to one of the two possible paths in the attenuator module
[19].
Attenuator
Module 1
Attenuator
Module 2
Attenuator
Module N
Out
Figure 10.3: Basic design of a high speed, variable optical attenuator that operates in synchronism with the
variable PDL. The variable optical attenuator consists of a cascade of binary attenuator modules.
Fig. 10.4 shows a multichannel design using attenuation plates that can be simple
neutral density filters designed to provide the desired attenuation. Another option is to use
fixed power polarization rotators such as half wave retarders, or voltage controlled
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
birefringent mode NLC devices. These polarization rotators are set such that they rotate the
incident linear polarization by a predefined angle. Then due to the output PBS, only one
polarization component (e.g., horizontal polarization) of the linearly polarized light will be
transmitted through the PBS towards the output of the attenuator module. Thus by
adjusting the degree of rotation of the incident polarized light, the desired attenuation can be
realized. The first polarization switch labelled S in Fig. 10.4 is used to control the SOP of
the input light to the single bit module. Depending on this SOP, the light can follow either
the zero attenuation path A a or the A, path. The second switch S in the module and the
polarizer in the next module are used to supress any polarization leakage due to the
switches and the PBSs. Note that the polarization rotator approach gives greater flexibility
to the optical engineer compared with the neutral density filter approach to obtain the
desired attenuation, since the attenuation can be fine tuned to the desired level by either
adjusting the voltage applied to the NLC device or the orientation of the optical axis of the
half wave retarder.
A bsorber
PBS
T IR Prism
In
PBS
A,,
TIR Prism
Figure 10.4: The high speed, variable optical attenuator based on fast FLC polarization switching devices
and polarization beamsplitter cubes (TIK: total internal reflection; P: polarizer, S: polarization switch; A,:
attenuation plate with attenuation value A(; A„: attenuation plate with zero attenuation)
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Another FLC-based optical attenuation system is depicted in Fig. 10.5. In this
approach, FLC devices operate in the binary phase mode and act as beam profile spoilers.
This can be achieved by placing the FLC device such that the optical axis is switching
symmetrically around the output polarizer axis, and thus the output signal is phase
modulated with a phase factor of k [20, 21]. By individually controlling the axis orientation
of the 2-D FLC-pixelated array, areas with alternating 0 and n phases can be realized. Thus
a variable fringe spacing 0-k phase 2-D grating can be realized that can cause phase
perturbations on the beam phasefront. In general, the 0-7E phase distribution can have any
configuration and not necessarily a “grating-like” one. Single-mode fiber coupling
efficiency using a FO-collimator approach is highly dependent on the wavefront
characteristics of the incident wave [22, 23]. Thus, the 2-D phase perturbation can affect
the coupling eficiency of the beam into the fiber-optic collimator. Hence, an attenuated
optical signal can be obtained by properly adjusting the phase perturbation distribution
across the incident beam.
An alternative gray scale optical attenuator option is shown in Fig. 10.6. This
design is based on holographic polymer dispersed liquid crystal (PDLC) devices. PDLC
devices can be used as variable diffraction efficiency voltage controlled gratings, and have
been reported to have switching speeds of < 50 (is [24]. The excess light in a PDLC device
can be rejected into the first diffraction order. Thus, the right amount of optical attenuation
can be accomplished. Compared to the binary design, this approach gives a reduced size
and reduced excess optical loss attenuator as it has only one attenuation stage.
Nevertheless, it requires a precise analog voltage controller for driving the PDLC device.
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FRONT VIEW:
T O P VIEW:
2-D Pixelated FL C Phase Cell Array
I
2
/
0- Jt phase based
C array
To IxJ
—
pixel Driver
P(p)
Linear
Polarized
Light In
.
P-
P(s)
FLC l
GRIN
_
— CZb SM F
FLC 2
polarization
Individual pixel
with electrodes
it phase pixel
□ 0 phase pixel
FL C /
- 'i
m
Perturbed Beam
! / 1m
'
BBS
Figure 10.5: Phase perturbation-based optical attenuator with IxJ independently, 0-7t phase, controlled FLC
arrays (SMF: Single mode fiber, GRIN: gradient index lens)
TO P VIEW:
FRONT VIEW:
2-D Pixelated PD LC Array
I
2
/
1jh
2
To IxJ ____
pixel Driver
J
he u p DP HE 1(1
i n ip m i
__ PDLC Grating
Linear
Polarized
Light In
Rejected Optical Pow er
in Diffracted Beam
—lJ
i
p-polarization
Linear
Polarized
Light Out
Individual
■— pixel/electrode
Attenuated
Undiffracted Beam
Figure 10.6: Single stage gray scale optical attenuator based on a holographic polymer dispersed liquid
crystal device with IxJ independently controlled variable diffraction efficiency programmable gratings (PGs).
The mentioned optical attenuator architectures perform RF signal attenuation in the
optical domain, as the optical signal propagates through the PDL system. A different optical
architecture for obtaining the important signal calibration is based on an N-bit optically
controlled microwave photoconductive attenuator [25,26]. This approach is based on the
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
RF in/out
Port 1
PC substrate
PCS contacts
Signal from
Computer
Si:CPW PCS
VCSEL
Driver
Port 2
VCSEL
Array
RF out/inl
I
Figure 10.7: The jV-bit electro-optic attenuator based on the photoconductive effect Two-dimensional
VCSEL array is used to activate each photoconductive bit. (SirCPW PCS: coplanar microwave waveguide
on a photoconductive silicon substrate).
photoconductive effect in coplanar waveguide (CPW) microwave transmission lines, and
the attenuation is applied directly on the microwave signal via optical means. The
photodetected microwave signal is transmitted to the antenna element via a microwave
transmission line fabricated on a silicon photoconductive substrate. When optical beams hit
the transmission line, hole-electron pairs are generated. Thus, a solid state plasma is created
in the semiconductor. The interaction of this plasma with the propagating microwave signal
provides the RF attenuation. It has been shown that if the transmission lines are illuminated
by N control beams having different optical powers that follow a binary pattern, then a Nbit attenuator can be implemented. Fig.
photoconductive-effect-based
attenuator.
10.7 shows a possible design for the
The
optical
beams
incident
on
the
photoconductive material can be from a vertical cavity surface emitting laser (VCSEL) array
[25]. The VCSEL structure contains many individual laser diodes that can be independently
addressed using a two dimensional VCSEL driver and interface. The individual lasers can
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
be computer controlled to be either “on” or “off’ depending on the desired attenuation level.
In the “on” case the individual laser can be arranged to have a binary power pattern. Thus,
by independently turning “on” and “o ff’ the lasers aiV-bit RF attenuator can be realized.
In the following paragraphs, we describe and test a method for dynamic range loss
compensation in a FO link without an RF amplifier.
10.3. Experimental Demonstration of the Signal Calibration and Dynamic Ranee Loss
Compensation Technique
The optical source used in our experiments was a diode pumped Nd: YAG laser at
1319 nm from ATX Telecom Systems, Inc. The electro-optic modulator is a UTP LiNbOj
Mach Zehnder analog modulator biased at quadrature, and the detector is a Lasertron QRX51-053 receiver. A manually controlled OZ-Optics FO-attenuator is used to simulate an
electronically controlled high speed optical attenuator. First the performance of the
externally modulated FO link without the PDL at 6 GHz is tested. The opdcal attenuator in
the FO-link was adjusted so that an optical power of 3 mW is incident on the photodetector.
This 3 mW optical power is the optimum optical power for our Lasertron receiver. The FOlink exhibits a -32.16 dB RF gain (or a 32.16 dB RF loss) at 6 GHz. Fig. 10.8 shows the
measured FO link fundamental-output versus the link fundamental-input plot at a frequency
of 6 GHz. The 1 dB compression output power is -11.26 dBm at a +20.9 dBm input
power. As 1 dB compression point we define the point where the conversion loss is
increased by 1 dB. The noise floor power spectral density of the FO link was measured at 139.2 dBm/kHz (or -169.2 dBm/Hz). Thus, a GDR of 127.8 dB at 1 kHz bandwidth (or
157.8 dB-Hz) is calculated. The CDR is a measure of the variation of the signal levels that
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
can be carried by the link. Typically it is defined as the difference (in dB) between the
maximum detectable power in the linear regime (defined 1 dB higher than the 1 dB
compression output power) to the minimum detectable power (usually defined equal to the
noise floor) [9]. The CDR is also often called the signal-to-noise ratio of the FO-link. The 1
dB input power compression point is found to be close to the theoretically expected one
given by UTP for their modulator [27]. The rms VK of the UTP modulator is 4.3 V that
corresponds to a PKof 25.7 dBm. Px is defined as the RF power required to generate VK
[29]. The 1 dB input power compression point (P^., dB) is then estimated by [27, 28]
n,.—
,dB = ^ -3 .9 dB= 21.6 dBm.
( 10. 1)
The third order intermodulation (TOI) point, defined as the point where the
fundamental signal and third-order intermodulation product powers are equal, can also be
calculated using the following equation to be [27,28]
Pin.joi = P* +5.1 dB= 30.7 dBm.
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
( 10.2 )
FO Link without PDL (f = 6 GHz)
Thermal Noise Floor
-
20-
I dB compression
at I KHz B W :-144 dBm
-
output power = -11.26 dBm
20 9 dBm
-4 0 -
-6 0 -
3
-8°
-
a
e§ -100-
120Noise Floor
N F = 3 6 .9 dB,
-139 2 dBm
-140
-15 2 dBm
-107 04 dBm
-160
140
-120
-100
-80
-60
-40
-20
0
20
40
RFt> (dBm)
Figure 10.8: FO link fundamental and two-tone intermodulation distortion output versus the link
fundamental input at 6 GHz (resolution bandwidth I kHz).
Using the PmT0( and the fact that the two tone intermodulation distortion plot has a slope of
3, the expected two tone intermodulation distortion link output versus the link fundamental
input power can be plotted (Fig. 10.8). Hence, a potential spurious free dynamic range
(SFDR) of 91.8 dB-kH z273 or (111.8 dB-Hz273) is estimated. SFDR is defined as the
maximum difference between the noise floor and the fundamental output which produces
distortion terms of equal amplitude to the noise floor. The noise figure (NF) of the FO link
is also measured at 36.9 dB. The NF is the degradation in the signal-to-noise ratio caused
by transmission through the link when the input noise is thermal noise at 290 K and is
defined as [29]
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
NF = 10 • log ^p u t signal-,o-noise ratio
output signal -to - noise ratio
where
(10.3)
Sa / Na
St, is the output and input signal power, respectively. N a is the available output
noise power and N , = K -T-B is the available input noise power from the generator, at a
temperature T = 290 K and bandwidth B. K is the Boltzmann’s constant. Rearranging
equation (10.3) the NF can also be expressed as
NF = PNoisc Floor - [ - 174 + 10 • log(B)] - G,
(10.4)
where PNoiseFloor is the noise floor power measured in dBm at a specific bandwidth B, -174
dBm is the available thermal noise at the input of the link at a temperature of 290 K and a 1
Hz bandwidth, and G is the gain of the FO link in dB.
When the PDL is inserted into the externally modulated link path via the use of
fiber-to-ffee space coupling optics (e.g., GRIN lens FO-collimators), an extra RF loss is
expected. The FO attenuator remains in its previous setting, hence the input optical power
to the PDL was 3 mW. The 3-bit PDL had a worst 5.7 dB and a best 5.5 dB optical
insertion loss. Using the optical attenuator we equalized the optical insertion loss for all
different settings at a 5.5 dB. Thus, the RF gain of the FO link is now expected to be 32.16 + 2 (-5.5) = - 43.16 dB. Indeed Fig. 10.9 shows a measured RF FO-link gain of 43.17 dB at 6 GHz. Fig. 10.9 also shows the measured FO-link-PDL system fundamental
197
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FO Link with the PDL (f = 6 GHz)
0
T herm al Noise Floor
-20
TOT,
at Ik H z BW : -144 dBm
-
-40 -
20.4 dBm
43.17 dB
-60-80-
« -100-120
-
Noise F lo o r
-140 -160
-142.8 dBm
' i
-140
.
■
-120
-99.6 dBm
KTB = -144 dBm
-12.8 dBm
-100
RF(n(dBm)
Figure 10.9: FO link fundamental and two-tone intermodulation distortion output versus the link
fundamental input at 6 GHz when the PDL is inserted in the optical path (resolution bandwidth 1 kHz).
output and the expected two tone intermodulation distortion link output versus the link
fundamental input. We assume that the effect of the PDL in the two tone intermodulation
distortion would be just an attenuation of the output signal. This is assumed because the
third order intermodulation products are generated by the non-linearities in the external
modulator. Thus, the PDL just attenuates the third order intermodulation product power by
the same amount as the fundamental, i.e., by the amount that the gain decreases. The 1 dB
compression output power is -22.8 dBm at +20.4 dBm input power. We measured a noise
floor power spectral density of -142.8 dBm/kHz (or -172.8 dBm/Hz). Thus, the CDR is
120 dB-kHz (or 150 dB -Hz). This lower, by -11.01 dB FO link gain affects the CDR,
which now is 8 dB lower than in the previous case. The lower noise floor observed in this
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
case is due to the lower optical power impinging onto the photodetector and thus the lower
photodetector shot noise which is the dominant source of noise in our externally modulated
FO link. The potential SFDR can also be estimated as before at 86.8 dB -kHz273 (or 106.8
dB -Hz273). The NF of the system is calculated using Eqn. (10.4) at 44.3 dB, a 7.4 dB
degradation compared to the previous case.
The next step was to reduce the optical attenuator setting by 5.5 dB so that the PDL
optical insertion loss is balanced, and the light incident on the photodetector is again 3 mW.
The measured FO link-PDL system fundamental output and the expected two tone
intermodulation distortion link output versus the fundamental link input are plotted in Fig.
10.10. The compensation of the PDL optical insertion loss by an equal amount of reduction
in the optical attenuation of input optical power to the PDL has an effect such that the FO
link gain and the CDR reach their link without PDL values of -32.16 dB and -127.8 dB at 1
kHz bandwidth (or 157.8 dB-Hz). The 1 dB compression output power is -11.4 dBm at a
20.8 dBm input power. The noise floor power spectral density is -139.2 dBm/kHz (or 169.2 dBm/Hz), and the NF and the SFDR also obtain their link without PDL values.
The experiment was repeated at 3 GHz to test the wideband capability of the FO
link. Fig. 10.11 shows the fundamental output and the expected two tone intermodulation
distortion link output versus the fundamental link input of the FO link at 3 GHz. A CDR of
123.05 dB-kHz (or 153.05 dB-Hz), and a SFDR of 88 dB-kHz273 (or 108 dB-kHz273)
was obtained. Fig. 10.12 shows the linear and third order intermodulation distortion
response of the FO link with the PDL. The insertion loss of the PDL reduces both the CDR
and the SFDR at 116.5 dB-kHz (or 146.5 dB-Hz), and 83.5 dB-kHz273 (or 103.5
dB-kHz273) respectively. Fig. 10.13 shows the linear and third order intermodulation disto-
199
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
F O L ink w ith the PDL and the
C om pression D y n am ic R an g e L oss C om pensation T echnique ( f = 6 G H z)
TO I
T h e rm a l N o ise F lo o r
-
20-
-I I 4 dB m
at I k H z B W :-1 4 4 dB m
—
2 0 .8 dB m
-4 0 32 16 dB
-6 0 -
E
C
■Q
a
a
^
-
100 -
-
120N o ise F loor:
N F=36 9 dB
-139 2 dB m
-1 4 0 -1 0 7 0 4 d B m
-160
-140
-1 2 0
-1 0 0
-80
K T B = -!4 4 dBm
-60
-40
R F |o (dB m )
-IS 2 dB m
-20
0
20
40
Figure 10.10: FO link fundamental and two-tone intermodulation distortion output versus the link
fundamental input at 6 GHz when the dynamic range loss recovery technique is used (resolution bandwidth I
kHz).
rtion response of the FO link with the PDL when the dynamic range compensation
technique is used. Again full dynamic range compensation is obtained using our technique.
A CDR of 123 dB-kHz (or 153dB-Hz) and a SFDR of 88 dB-kHz2* (or 108 dB -kH z2*)
was obtained. The optical attenuator settings used for this experiment was the same as the
one used in the 6 GHz case. Note that the acquired CDR and SFDR are ~ 5 dB and ~3 dB
respectively, lower than in the 6 GHz case. This is due to the lower PKof the modulator at
3 GHz, and thus the lower 1 dB compression point at this frequency. Our experiments
200
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FO Link without PDL (f = 3 GHz)
Thermal Noise Floor
-
20-
TOI
at IkHz BW: -144 dBm
-16.17 dBm
-4 0 -
17 dBm
-6 0 S
2
33.17
-80 3
e
u
-,00-l
-120
-
Noise Floor:
NF = 38 dB
-.139,22 dBm
-140
-106 dBm
-160
-140
-120
KTB = -144 dBi
-100
-18 28 dBm
0
-20
-4 0
20
40
RF,In (dBm)
Figure 10.11: FO link fundamental and two lone intermodulation distortion output versus the link
fundamental input at 3 GHz (resolution bandwidth I kHz).
FO Link with the PDL (f = 3 GHz)
Thermal Noise Floor
-
20-
TOI
at IkHz B W :-144 dBm
•26.4 dBm
-4 0 16.9 dBm
-6 0 -
B
2
43.3 dB
-80 -
3
S
-loo-
120-
Noise Floor
NF = 44.4 dB
-142.9 dBm
-1 4 0 -99.6 dBm L~KTB— 144 d B m / |6 j3
-160
140
-120
-100
- 80
-60
-40
RF|i( (dBm)
-20
0
20
40
Figure 10.12: FO link fundamental and two-tone intermodulation distortion output versus the link
fundamental input at 3 GHz when the PDL is inserted in the optical path (resolution bandwidth 1 kHz).
201
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FO Link with the PDL and the
Dynamic Range Loss Compensation Technique (f£ = 3 GHz)
-
20-
Thermal Noise Floor
at IkHz BW :-144 dBm
TOI
-16.0 dBm
17.2 dBm
-4033.18 dl
-60S
tt
•a
-80-
-120
-
-140 -■
Noise Floor
-139.8 dBm
NF = 37.4 dB,
-106.6 dBm '-KTB =
-160
140
-120
-100
-80
-60
-144
dBm /
-40
-18.47
-20
dBm
0
20
40
RF(n (dBm)
Figure 10.13: FO link fundamental and two-tone intermodulation distortion output versus the link
fundamental input at 3 GHz when the dynamic range loss recovery technique is used (resolution bandwidth 1
kHz).
showed a 1 dB compression input power of 17 dBm, which is ~ 4 dB lower than the 1 dB
compression input power at 6 GHz.
The gain of the FO link was also tested for the 3 GHz to 6 GHz band. A HewlettPackard RF network analyzer (HP-8720-D) was used. Fig. 10.14 shows the gain for the 3
to 6 GHz band for the three different experimental setups. The RF input power of the
network analyzer was set at 0 dBm. Fig. 10.14(a) shows the gain for the FO link with the
optical attenuator set to give 3 mW of optical power incident onto the photodetector. Fig.
10.14(b) shows the FO link with the PDL and the same setting for the attenuator, while
Fig. 10.14 (c) shows the FO link with the PDL and the attenuator set so that the optical
power impinging onto the photodetector is again 3 mW. In Fig. 10.14, it can be seen that
the PDL acts only as an attenuator to the FO-link. It can be seen from the network analyzer
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
plots that the loss due to the PDL is -0.6 dB lower for the 3 GHz case than for the 6 GHz
case.
a M ttM U tttt
■*« la*M
CH I
M l IM S «M ONE
M
N.1MI
mtt
• „
INi
td
■
SCA .E
1 0 dB/dh
MU
t. 42.71111QM
S
ass m s
MU
SCA .E
10 dB/dh
w e
JU N i
OKI
j. •UM<IOKI
---
----Ai
ft
i.
A
i
9
---- ---- ----
------- ----
IU R T U N M M O H l
-
---- -----
IT O f U N M W Q H l
STM T U N M W S H l
(b)
fai
i ifcajt
l.;4 U } l 4
M
C M IM S M S M b
1.
4U S31 \
QHB
SCA -E
1 0 dB/dh
0*U
1. 4 2 . IS I d
1
am
ft
f t'
ITOf UNMMOMi
srjurr ummmomi
(c)
Figure 10.14: Network analyzer plots showing the RF gain of the FO link (aj without the PDL, (b) with
the PDL, and (c) with the PDL and the dynamic range compensation technique.
203
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10.4. Discussion of FO Link-PDL System Performance
The dynamic range recovery technique proposed is easily accomplished since the
available optical power from the diode pumped Nd: YAG laser is high enough so that after
propagation of the optical signal through the FO link and the PDL, the optical power
impinging onto the fast photodetector is more than what is required from the photodetector
for optimum dynamic range. The optical attenuator is used to adjust the optical power at
levels acceptable from the photodetector for operation below the saturation and non-linear
regimes. For example, for a 7-bit PDL with an optical insertion loss of 1.3 dB/bit, the total
optical insertion loss of the PDL is 7 x 1.3 dB = 9.1 dB. The optical insertion loss of an
external modulator is ~ 7 dB (~ 4 dB due to the device loss and 3 dB due to the quadrature
bias operation). Thus, the total optical insertion loss is 16.1 dB. Additional optical losses
due to the fibers interconnecting the laser to the modulator, the modulator to the PDL, and
the PDL to the photodetector, and the fiber connectors can be < 1 dB. If a 200 mW diode
pumped laser is used, the optical power available for delivery at the photodetector is 4
mW. This is more than what is required by our photodetector for optimum dynamic range.
Thus, the FO attenuator can be used to adjust the optical power. Note that if our
photodetector could handle higher optical powers, the dynamic range would have been
higher. Currently, the limited optical power capacity of the photodetectors in the S or C
band is the limiting factor for much higher dynamic range in these bands.
The proposed optical attenuation system does not require a wide range of
attenuation levels (i.e., from 0 dB to >10 dB). In the previous chapter, experiments
showed that the within channel PDL bit insertion loss variation due to the optical insertion
losses of the optical components is ~ ± 0.05 dB [1]. For the 4 bit PDL system we
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measured a ± 0.2 dB insertion loss variation [1]. Thus, it is expected that for a 7-bit PDL,
the total insertion loss variation will be - ± 0.4 dB. Insertion loss variation due to the non­
uniformities throughout the entire active area for the optical components are expected to be
at the worst case within a ± 5 % of their specifications. This leads to a less than ± 0.3 dB
variation. So the worst case scenario variation is expected to be ± 0.7 dB. Hence, the
amplitude control system would need to have the capability of 1.4 dB of maximum
attenuation. To obtain this 1.4 dB optical attenuation with a resolution of 0.1 dB, a 4 bit
binary optical attenuator can be used. Note also that the RF characteristics of the modulator
and the photodetector affect the overall gain of the system and may limit the performance
(Fig. 10.14(a)). For a completely balanced, flat gain response throughout the entire
operational band (e.g., 3 GHz to 6 GHz) the different RF performance can be calibrated
and balanced out using the optical attenuation system.
Note that the dominant noise in our FO link is the shot noise, since the relative
intensity noise (RIN) of the diode pumped laser is — 170 dBm/Hz. In our experiment only
the input power impinging on the photodetector changes, thus the noise floor variation in
the different experimental cases is dependent on the effect of the optical power on the shot
noise. The modulator thermal noise is proportional to the microwave-modulated optical
power and becomes significant as the input optical power is increased. Since in our case the
optical input power to the modulator is the same, the modulator thermal noise contribution
is the same in all the cases. This is the reason why we have chosen to put the FO attenuator
after the modulator and not before it. The rest of the noise contribution comes from the
photodetector dark current and the thermal noise that are independent of the optical power.
From the above discussion and the fact that the FO link gain increases proportional
to the square of the optical power [7], while the noise floor increases proportional to the
205
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
optical power [7], the FO link dynamic range can be optimized by adjusting the optical
power. Currently the limit to further maximization of the CDR and the SFDR in the GHz
regime is the inability of the fast photodetectors to handle larger optical powers that could
eventually lead to an improved FO link-PDL system dynamic range and may be in the 3 dB
NF limit [30]. Nevertheless, recent results have shown that photodiode non-linearities can
be reduced by increasing the bias [31], thus promising higher dynamic range. A balanced
high power photodetection scheme has also demonstrated increased dynamic range [8].
Increased dynamic range can also be achieved by low biasing the Mach-Zehnder modulator
with the penalty of reduced gain, provided that single-octave operation is needed [32].
Optical carrier suppression has also been proposed for improved dynamic range [33]. It has
been proposed that this optical carrier suppression can be accomplished entirely within the
modulator waveguide circuitry [34].
10.5. Conclusion
In conclusion, a dynamic range recovery technique for switched PDLs used for
phased array antenna control applications has been proposed and experimentally
demonstrated. This method employs a high speed variable optical attenuator that operates
synchronously with the variable PDL to maintain an optimized optical power level on the
output photodetector. Various optical designs to implement this high speed attenuator have
been proposed. Dynamic range recovery was demonstrated for a 3-bit PDL with an optical
insertion loss of 5.5 dB, fed by a FO-link consisting of commercially available
components. A compression dynamic range of 158 dB-Hz and noise figure of 36.9 dB
206
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were measured for the FO link-PDL system. This is the highest CDR reported today for a
switched PDL fed by a FO-Iink. The spurious free dynamic range was estimated to be >
111 dB-Hz273. The use of the high speed optical attenuator also provides the necessary
signal calibration for the different PDL settings and channels. Improved overall system
dynamic range can be accomplished with lower insertion loss PDLs, optical carrier
suppression, low biasing of the electro-optic modulator, and/or photodetectors that can
handle larger optical powers.
207
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 11
THE FIRST DEMONSTRATION OF A 7-BIT 33-CHANNEL PHOTONIC
DELAY LINE FOR PHASED ARRAY RADARS
11.1. Introduction
In the previous chapters we investigated different PDL structures and we tested
the RF performance o f these PDLs under different FO link feeding systems. Nevertheless,
all o f our previous experimental demonstrations were single channel demonstrations and
issues related with interchannel cross-talk were simulated by translating the output fibers.
Typical delay line systems for advanced phased array radars require multiple channel
operation and more than 4 bits. For example, for scanning a radar beam in a ± 60° angle
with a resolution of 1°, a 7-bit PDL, with the capability of 128 settings, is required.
In this chapter, we demonstrate the first 7-bit 33-channel PDL system designed to
meet specific phased array antenna requirements as set by the Lockheed-Martin Company
radar engineers. This is the first demonstration of a 7-bit 16 active channel PDL system
using any photonic technology. Previous systems include the Hughes Corporation 3-bit,
8-channel true-time-delay network demonstration [1]. This approach uses optoelectronic
switching of 8 lasers for each channel. Each laser is coupled to an optical fiber of a
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
different length, that are coupled to the same photodetector. By switching on only the
proper laser the required time delay is obtained. The Hughes group also demonstrated a
4-bit 8-channel time delay network by a combination o f electro-optic switching of four
lasers and four photodetectors [2]. Nevertheless, the Hughes Corporation approach
requires many lasers and/or photodetectors for each channel that leads to excess optical
power wastage and/or expenses. A 16-channel 5-bit time delay network was
demonstrated by Thomson-CSF-France [3]. This is a visible band transmit-only
heterodyne detection based system that uses slow (e.g., 100 ms) nematic liquid crystal
polarization switching devices. The Naval Research Laboratory (NRL) has recently
demonstrated fiber-optic true time steering systems for a 16-antenna element transmit
only [4], and 8-element receive [5] only operations. Their systems are based on
wavelength tuning and a prism geometry dispersive optical fiber system. Nevertheless,
these systems are currently limited by the slow tuning speed (e.g., 12.5 ms/nm) of the
laser source. Westinghouse/Northrup Grumman Corporation recently performed a field
demonstration of an eight-element receive 6-bit time delay steering system [6]. This
system architecture is based on a hardware-compressive scheme that uses optical
wavelength multiplexing to reduce the overall system complexity and volume.
Nevertheless, this system uses two different switchable optical networks, one for the
receive mode and one for the transmit mode. This is done to avoid excess optical losses
due to the fiber-optic combiners/dividers, especially for the receive mode.
In this chapter, we first describe the design of the input and output fiber arrays
used to feed our PDL, and then the design of the 33 pixel FLC switching arrays. PDL
issues such as time delays, insertion loss, within channel leakage noise and interchannel
crosstalk are discussed.
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.2 Delay Line Requirements
Before we describe our PDL system, we will present the delay line requirements
as set by the radar engineers for advanced phased array antenna applications such as the
navy Aegis radar system. Table 1 shows these requirements. The PDL is required to have
7 bits, with a LSB of 0.1 ns, and MSB of 6.4 ns. Numbers o f < - 60 dB are desirable for
both the within channel leakage noise and the RF interchannel crosstalk. The within
channel leakage noise is equivalent to the SNR o f the PDL, and is defined as
201og(leakage noise/signal). The RF crosstalk between delay channels relates with the
amount of the optical signal traveling through the desired path that leaks to the adjacent
channels. The requirements for the switching times after the command signal has been
applied at the delay-line connector are shown in Table 11.2. The transmit-to-receive
switching speed relates to the electronic switching of the phased array from the transmit
mode to the receive mode and vice versa. On the other hand, the time delay setting is the
important switching requirement for the time delay network and thus the phased array
antenna. Between two consecutive beam position settings for the delay network, the
phased array neither receives nor transmits. This is called the dead time, and the longer it
is, the higher the probability for the phased array antenna not to detect a target. Thus, the
dead time related to the device switching time should be as small as possible, and should
be a very small portion o f the dwell time of the antenna.
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 11.1: Delay line requirements.
Number o f bits
7
RF Loss/bit
2.86 dB
Least Significant Bit (LSB)
0.1 ns
Most Significant Bit (LSB)
6.4 ns
Maximum Time Delay Error
0.05 ns
Frequency Range
3 - 6 GHz
Within Channel Leakage Noise
< - 60 dB
RF Interchannel Crosstalk
< - 60 dB
Table 11.2: Switching time requirements for the delay line, after
command signal has been applied to the PDL connectors.
Transmit to receive
0.5 ps
Time delay setting
0.5
j is
11.3 Multiple Channel Photonic Delay Line Design Issues
Our multichannel fiber remoted PDL requires multiple input and output fiber
arrays and multi-pixel FLC polarization switching arrays. In this section we will describe
the design and the implementation o f these multiple channel components in our system.
The first issue under consideration is the choice o f the optimum FO-collimator
technology, in terms o f performance, beam characteristics, and size. GRIN-lens FOcollimators are an excellent candidate for fiber-to-free space interconnections because o f
211
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the mature technology that can give robust and compact designs. Furthermore, low
insertion loss numbers have been quoted for GRIN-to-GRIN coupling [7, 8]. Typical
GRIN-lens diameter is 1.8 mm. The GRIN-lens is mounted in a ferrule o f 2.4 mm outer
diameter [7]. This is currently the smallest size GRIN-lens based FO-collimator design.
The ferrule size limits the closest distance between two GRIN-lenses to be 2.4 mm as
shown in Fig. 11.1, when GRIN-lens FO-collimators are stacked in a hexagonal
configuration as the one proposed in reference [9].
2.4 mm
2.4 m m
GRIN-lens
1.8 mm
Ferrule
Figure 11.1: High density packing hexagonal configuration for a GRIN-lens FO-collimator array using
physical contact of the GRIN-lens ferrules.
212
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SM-fiber
GRIN-lens
__________
•"
Collimated Beam
..
(a)
Shifted SM-fiber
Tilted SM-fiber
GRIN-lens
Tilted Collimated Beam
GRIN-lens
Shifted Collimated Beam
(c)
Figure 112: Shift and tilt effects on the output collimated beam of a GRIN-lens FO-collimator when the
SM-fiber has a tilt or shift from the optimum position on the surface o f the GRIN-lens. (a) optimum SMfiber position, (b) shift o f the SM-fiber from the optical axis of the GRIN-lens causes tilt of the output
collimated beam, (c) tilt o f the SM-fiber from the optical axis o f the GRIN-lens causes shift of the output
collimated beam.
Nevertheless, such a design is not suitable for applications where the optical
beams travel long distances in free-space. That is because there are tolerances to the
GRIN lens size, ferrule size and position of the GRIN lenses in the hexagonal packing
configuration o f Fig. 11.1. Additionally, there are tolerances with the position of the
optical fiber exactly at the center and normal to the GRIN lens surface. In an ideal
situation, the output beam from the GRIN-lens FO-collimator comes out normal to the
GRIN lens surface, as shown in Fig. 11.2(a). Nevertheless, a small shift of the optical
fiber from the center of the GRIN lens causes a tilt in the direction of the collimated
output beam (Fig. 11.2(b)). On the other hand, if the fiber is not normal to the surface of
the GRIN-lens then the output collimated beam is shifted from the central output position
as shown in Fig. 11.2(c). In the case where there is both shift and tilt of the optical fiber,
213
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the output beam will be tilted and shifted from the optimum central direction. Thus, the
optical beams will not come out from the GRIN-lens array parallel to each other. This is
the limiting factor for stacking many GRIN-Ienses FO-collimators to form FO-collimator
arrays. So there is the need for external control for each FO-collimator to compensate for
the shift and tilt o f the collimated beams caused by fabrication and fiber-to-GRIN
alignment tolerances. Theory and experiments have shown that the most critical
alignment for GRIN to GRIN coupling via free space propagation is the GRIN lens tilt
and not the lateral translation [9].
Our effort was concentrated in finding a fiber-port technology that would allow
independent tilt control of the FO-collimators to form a multichannel FO collimator
array. After investigating different approaches we selected the OZ-Optics FO-flange as
the building block for our multichannel FO-collimator array. The OZ optics FO-flange is
shown in Fig. 11.3. It is a circular mount that can accommodate the GRIN-lens ferrule at
its center. To obtain the micro-tilt control, the flange is mounted on a reference block
using three screws. A plastic flexible O-ring is placed between the two parts to allow
small physical movement of the flange when it is pushed against the block by tightening
the adjustment screws. Loosening the adjustment screws causes the O-ring to relax and
moves the FO-flange backwards. Thus, by proper adjustments of the three screws the
desired beam direction can be accomplished. The three locking screws are used to lock
the flange to the desired position.
In order to obtain the highest possible packing density for our FO-collimator
arrays we chose a hexagonal configuration for our fiber array design, as shown in Fig
11.4. The distance between adjacent GRIN-lenses is 13 mm and the overall size o f a 37channel fiber array is 91 mm x 78.8 mm. It is obvious that using this approach the FOcollimator array becomes big and in order to use it for PDL applications we need to
214
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reduce its size. This is because typical optical element (e.g., lens) dimensions are < 50
mm. In our system the limiting aperture is the high extinction ratio Polarcor polarizer,
which is 15 mm x 15 mm.
O -ring
GRIN-lens
Ferrule
Fiber
11.5 mm
T ilt A djustm ent S crew
L ocking S crew
T h ree th read ed holes 120*
a p a rt fo r th e locking screw s
Three clear holes 120° apart
for the tilt adjustm ent screw s
Figure 11.3: The OZ-Optics FO-flange used as the building block for our input and output fiber-optic
arrays.
91 m m
FO-flange !
78.8 mm
13 mm
13 mm
Figure 11.4: The fiber array design based on the OZ-Optics FO-flange.
215
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A way o f reducing the size o f the beam array is imaging the fiber array and thus
the beam array using a telescope with magnification of < 1 to a smaller overall size array.
A telescope can be built using two plano-convex lenses as shown in Fig. 11.5. Several
sets of “ off-the-shelf’ plano-convex lenses were considered. We chose two plano-convex
lenses with radius o f curvature (R) 171.18 mm and 25.936 mm. This is equivalent to focal
lengths of 340.318 mm and 51.56 mm for wavelength of 1319 nm. Thus, the telescope
magnification is - 0.1515. This magnification is suitable for imaging in a diameter of <
15 mm up to 37 channels, while keeping the interhannel (center-to-center) distance at
1.97 mm. An interchannel separation o f ~ 2 mm, as the one accomplished by our
telescope, is also suitable for our FLC array design as will be shown later.
Input Plane
Output Plane
f
f
f
Figure 11.5: Telescope design using a set o f two plano-convex lenses.
216
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
f
Knowing the original size o f the beam-array and the m a g n ific a tio n o f our input
telescope, we simulated the beam array propagation through the telescope using the
optical system design software CODE V. Fig. 11.6(a) shows the position of the input
beams, while Fig. 11.6(b) shows the position o f the output beams as simulated by CODE
V. The intersections o f the grid-lines show the theoretical non-realistic optimum position
o f the output beams for aberration free thin lenses. The output beams are slightly shifted
towards the center of the array. The maximum shift is observed for the outermost
channels and does not exceed 70 pm. The small shift observed in the optical beams at the
output plane of the telescope raise questions for the design of the FLC multi-pixel arrays,
in terms of the pixel size, and inter-pixel distance. Inter-pixel distance also relates with
limitations set by the FLC array fabrication procedure. This is because the indium tin
oxide (ITO) electrodes that run between the pixels have to be far enough from each other
and from the pixels to avoid electrical interference. Manufacturer specifications for the
inter-pixel distance, and the pixel-to-ITO electrode spacing are 100 pm, with a 100 pm
wide electrode. The optimum design for the highest density FLC array is a 3 3-pixel
design as shown in Fig. 11.7. In this design only one electrode runs between any two
pixels. The FLC pixel size is 1.65 mm. This pixel size allows a 320 pm spacing between
the FLC pixels, which is enough to satisfy the pixel-to-ITO electrode spacing. The 1.65
mm pixel size makes our system alignment tolerant since this pixel size allows for small
beam misalignment from the center of the pixel due to the imaging from the telescope or
due to misplacement o f the FLC array from its optimum position in the system. The FLC
arrays used in our PDL system were fabricated by Boulder Nonlinear Systems (BNS),
Inc., Boulder, CO., and are shown in Fig. 11.8. The FLC material used is BNS, Inc.,
proprietary technology. Fig.l 1. 8(a) shows all 33 pixels set “ o ff’, while Fig. 11.8(b) and
217
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Input Fiber Assembly
Position o f the input Gaussian beams
33.774
£
22.516
CO
11.258
X
cs
■
0.000
E,
c
o
-11258
co
-22.516
O
CL,
-33.774.
— i—
-39
—i ■--- 1—
-26
-13
— t—
— i—
— i—
13
26
39
Position in x-axis (m m)
(a)
Output Fiber Assembly
5.118
B
B
3.412
1.706
x
CG
i
C
c
o
co
O
0.000
-1.706
-3.412
CL,
-5.118
-5.91
-3.94
-1.97
0.00
1.97
3.94
5.91
Position in x-axis (m m)
(b)
Figure 11.6: (a) Position o f the optical beams at the input plane of the telescope, (b) position of the optical
beams at the output plane o f the telescope.
218
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(c) shows two complimentary states of the FLC array with some pixels “ on” and some
“ o ff’ . The FLC arrays were single cell devices, where only one thick layer o f FLC
material has been sandwiched between two glass plates. The outside surface of the grass
windows was AR-coated for 1319 nm. Thus, unlike our previous FLC devices from
Displaytech, Inc., that were triple cell devices, these BNS FLC devices have a lower
optical insertion loss (e.g., 0.45 dB). The limitation of these FLC arrays is a small tilt o f
the optical axis in respect to the array axis (line across the centers of the seven central
pixels) as shown in Fig. 11.9. This tilt is ~ 6°, and it has an effect on the optical
polarization purity o f the optical signal propagating through the device. This polarization
purity problem does not significantly affect the within channel leakage noise performance
of our PDL, because our active noise filter suppresses any polarization leakage noise.
Nevertheless, the polarization purity degradation o f the optical signal increases the
insertion loss o f the PDL bit. This FLC optical axis tilt in respect to the array axis is due
to a fabrication limitation and is expected to be overcome by proper fabrication
techniques. Another way o f overcoming this problem is to use two HWPs, one at each
side o f the FLC array, to rotate the incident polarization from the laboratory coordinate
system to the FLC material optical axis coordinate system. Then after propagation
through the FLC device, the polarization changes back to the laboratory coordinate
system. This approach will increase the optical insertion loss of the FLC device by 0.08
dB due to the reflection loss from the two HWP, but the improvement due to the
polarization purity improvement is expected to be much higher.
219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1650 pm
320.42 nm
100 (im
100 (im
(a)
Figure 11.7: The FLC polarization switching array consisting o f 33 pixels, (a) FLC pixel array layout, (b)
detail of the pixels.
(a)
(b)
(c)
Figure 11.8: The ferroelectric liquid crystal polarization switching arrays, (a) all 33 pixels “ o ff’, (b) 9
pixels “ off’, (c) 24 pixels “ off” .
FLC optical axis
Amy Axis
OOQ OOO
O pO pO O
OOOOO
OO
Figure 11.9: The 6° tilt in the FLC optical axis in respect to the array axis.
220
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to deliver the RF signal to the phased array antenna elements, the
optically processed signal needs to be collected by optical fibers and sent to the antenna
site for detection. Fast photodetectors transform the optical encoded RF modulation to an
electrical signal. A design similar to the input fiber array design is used for the output
fiber ports. In this case multi-mode (MM) FO-collimators are used for ease in coupling.
Note, that a 7-bit PDL has 128 different settings. This means that the optical signal of
each channel travels through 128 different optical paths. Thus, the position o f the output
beam is almost impossible to be the same for all 128 different settings. Moreover, the
propagation of the optical beam through different optical elements may have imposed
different optical aberrations and wavefront changes to the optical beam for the different
settings. Thus, the use o f MM-fiber is preferred of the SM-fibers since the larger core
diameter (e.g., 50 pm) MM-fibers makes the coupling an easier task. We have previously
shown that the use of MM-fibers at the output of the PDL can lead to a 1.5 dB optical
insertion loss improvement compared to a SM-fiber output coupling system [10]. Note
also that we can use this MM-fiber output coupling system because in PDLs for phased
array antenna applications, remoting fiber lengths are less than 100-150 m, and thus
frequency and modal dispersion effects are negligible. In other words, we can use MMfibers because the photonic controller is located within 100-150 m from the antenna site.
For example the photonic controller can be located in the haul of a ship, while the radar is
at the top of the ship. Fig. 11.10 shows photographs o f the input SM polarizationmaintaining fiber array, and the output MM-fiber array, with 16 fibers each. A telescope,
similar to the one used at the input o f the PDL was used at the output o f the PDL. This
time the short focal length lens was placed closer to the output o f the PDL. This telescope
with a magnification o f -1/0.1515 = - 6.6, brings back the beam array to its original size.
221
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The remote control configuration o f our photonic controller is shown in Fig.
11.11. Note that the same optical hardware is used for both transmit and receive mode.
Electronic transmit/receive (T/R) modules are required to switch the system between
transmit and receive mode of the phased array antenna.
(a)
(b)
Figure 11.10: (a) The input SM polarization-maintaining fiber array, (b) the output MM fiber array.
P h o to n ic C o n tr o lle r S i t e
GRIN-lens
7-Bit
33-Channel
Photonic
Delay Line
Equal Length
PM-Fibers
GRIN-lens Array
Equal Length
MM-Fibers
A n te n n a S ite
Variable Optical
Attenuators for
Signal Calibration
RF Signal
for Transmit
Highspeed
Photodetectors
^Transmit
Electro-Optic
Modulators
Receive
Electronic
T/R modules
Electronic T/R modules
A A A A A n te n n a Elements
Figure 11.11: The fiber-optic remoted photonic controller for phased array antennas
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.4. The 7-Bit Photonic Delay Line System Design
As mentioned earlier, our variable binary PDL needs to have 7-bits with a LSB of
0.1 ns, and a MSB of 6.4 ns. Our design is based on “ off-the-shelf’ optical components.
Only the FLC switching arrays are custom made. We also use imaging optics in our PDL
for interchannel isolation and low insertion loss. Thus, appropriate focal length lenses are
required to obtain the specified delay lines.
In order to built the entire PDL system on a 5 ft x 8 ft optical table, different PDL
bit layouts were examined and designed. Our final PDL design consists o f five bits o f the
adaptable PDL architecture, one symmetric PDL architecture and one feed-forward
transmissive PDL architecture. Due to the unavailability of lenses that could give us the
exact desired time delays, one (or two, depending on the architecture) glass plate have to
be used to adjust the time delay to the desired value. The glass we selected is OHARA
LAH-58 [11]. We choose this glass for the low loss at 1319 nm, its good optical and
thermal properties, as well as its high index o f refraction (n = 1.854 at 1319 nm). The
high index of refraction leads to smaller glass thickness compared to a lower index of
refraction glass for a specific time delay adjustment. Table 11.3 shows the design o f each
PDL bit, with the radius of curvature (R) and focal length (/) of the lenses used, and the
required glass thickness (d).
All the lenses used were AR-coated for 1319 nm. BK7 glass lenses were used,
except from the two lenses in the non-delay path of the 3.2 ns PDL bit that were fused
silica. All lenses were 50.8 mm in diameter. This allows the optical beams to hit the lens
close to the optical axis limiting off-axis aberrations, such as coma and astigmatism. Fig.
11.12 shows the PDL system layout. All the PBSs in the adaptable PDL bits have a QWP
223
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
attached at their surfaces towards the two PDL paths. The QWP are necessary to rotate
the incident input polarization by 90° after propagation through the delay or non delay
path. In the symmetric and in the feed-forward PDL bit a HWP has been used in each o f
the paths. This is done so that the light undergoes one reflection and one transmission
through the two PBSs in the bit, and thus a balanced insertion loss and SNR performance
is obtained for these bits too, as for the case of the adaptable PDL [12].
Table 11.3: The 7-bit PDL design characteristics.
Designed
PDL
Delay
Architecture
(ns)
Non-delay Path
Delay Path
Glass Thickness
R
/
R'
/
D
(mm)
(mm)
(mm)
(mm)
(mm)
0.1
Symmetric
130.8
260.040
0.2
Adaptable
64.4
128.032
0.4
Adaptable
128.8
0.8
Adaptable
1.6
128.8 256.064
17.263
133.201
14.956
256.064
130.8 260.040
39.591
103.0
204.771
128.8 256.064
13.246
Adaptable
77.3
153.678
128.8 256.064
26.797
3.2
Transmissive
77.3
172.969
206.0 409.543
No glass
6.4
Adaptable
67.0
133.201
309.1
No glass
67.0
614.513
224
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIB
225
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 11.12: The experimental set-up of our 7-bil 33-channcl photonic delay line.
BIB
11.5. Time Delay Measurements
Time delay measurements were obtained for each bit of our PDL. A Hewlett
Packard (HP 8720D) network analyzer was used to obtain the time delay measurements.
Table 11.4 shows the desired and the obtained time delays for our system. There is an
excellent match between the desired and the obtained time delays for the five adaptable
PDL bits, and a very close match for the other two. Two network analyzer measurements
are shown in Fig 11.13 as an example. The perfect match obtained for the adaptable bits
is because in these bits the delay and non-delay paths are independent with each other.
Thus, small tolerance errors due to the lens or glass fabrication or the exact position of
the optical components can be corrected by slightly adjusting the physical length o f one
o f the paths. This may cause a slight defocus in the optical beams, that does not
significantly affect the coupling efficiency at the output of our PDL due to our MM-fiber
output coupling system. The overall maximum time delay error in the PDL is 0.05 ns
which is within the allowable time delay error set by the delay requirements. This error
can be corrected with proper PDL bit design or using the electronic phase shifters, that
already exist in a phased array antenna.
Another important issue with the time delay characteristics o f our PDL is the
length o f the optical fibers. All of the optical fibers should have the same length so that
for the same delay setting, two different channels would give the same time delay. This is
not a trivial task, especially when FC/PC connectorized fibers are used. The typical FOcollimator FC/PC connectorized patchcord tolerance length is ± 1 cm, which leads to a ±
0.033 ns time delay error for each of the two channels. Nevertheless, this error is constant
and known for each channel. Hence, it can be corrected by calibrating the individual
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
photonic delay lines for each of the fibers. This time delay calibration or correction does
not need to be fast; so one way of doing this is to use piezoelectric transducers to stress
and elongate the fiber to obtain the required time delay correction.
Table 11.4: The desired and experimentally obtained time delays for the 7-bit PDL.
Bit Number
Desired Delay (ns)
Obtained Delay (ns)
1
0.20
0.20
2
6.40
6.40
3
0.10
0.11
4
1.60
1.60
5
0.80
0.80
6
3.20
3.16
7
0.40
0.40
I
i
!
mx
Ell
m
A L D ELAY
i
,
1
!
'
,
!
'
!
:
i
'
i
0,1
!m !
i !
--T
im
:
n r m i r * t n CLAY
aw
turn
1M
i
;
;
t?m\
1l
!
!
:
i
1
i
A
|
ii
I
1 1
i
i
!
!i
i
(a)
I
(b)
Figure 11.13: Network analyzer measurements showing the (a) 0.2 ns time delay and (b) the 0.8 ns time
delay.
227
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.6. 7-bit Photonic Delay Line Insertion Loss Performance
The insertion loss o f our PDL was measured. Table 11.5 shows the insertion loss
for each of the seven bits. The total loss from our PDL system is 11.6 dB. This leads to an
average optical insertion loss o f 1.65 dB, or a 3.3 dB RF insertion loss. This insertion loss
is mainly due to the insertion loss of the FLC devices and can be reduced with proper
FLC design. Note that there are two bits where the insertion loss is 1.5 dB; thus lower
insertion loss numbers are feasible. This 1.5 dB insertion loss comes from the two bits
that the lower loss FLC arrays were used. These numbers can go even lower by
improving the polarization purity problem of the FLC arrays as was described in section
11.2 .
Table 11.5: Optical Insertion loss for each bit o f the 7-bit PDL system.
Bit Number
Desired Delay (ns)
Optical Insertion Loss (dB)
1
0.20
1.7
2
6.40
1.5
3
0.10
1.8
4
. 1.60
1.7
5
0.80
1.7
6
3.20
1.7
7
0.40
1.5
228
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Additional loss in our system is due to the reflection loss o f the lenses in the two
telescopes, which is minimal (< 0.08 dB), and the fiber coupling loss at the output o f our
PDL. Currently the average output coupling loss for the central channels was measured at
1.7 dB. For the outer channels the coupling loss was higher at 3.5 dB. The higher loss for
the outer channels is due to the astigmatism of the beams caused by the small diameter,
short focal length lenses o f our telescopes. This can be improved by using larger diameter
lenses so that all the beams hit the lens close to the axis, and thus no significant
astigmatism affects the beams. Moreover, more careful alignment would improve the
overall insertion loss of the system.
11.7. Photonic Delay Line Switching Time Response
The switching time between different PDL settings is determined by the switching
response o f the FLC devices. The waveform required to drive the BNS multipixel single
FLC cell devices is a 50 % duty cycle ± 24 V square wave. Fig. 11.14 shows oscilloscope
traces with the switching response o f the BNS FLC devices. Fig. 11.14(a) shows the rise
time, while Fig. 11.14(b) shows the fall time. As rise time (fall time) we define the time it
takes for the signal to rise (fall) from the 10 % (90%) to the 90% (10%) o f the maximum
signal power. Note, that the rise and fall times are not the same because o f a hysterisis
effect observed for the FLC material. Note also that there is a finite delay time before the
signal starts rising from zero, this is —17 ps. Similarly, there is a ~ 30 ps delay time
before the signal starts falling from the maximum value. The asymmetrical performance
in the rise and fall time is currently a material characteristic, and is due to a preferential
229
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
molecular anchoring [13]. Nevertheless, in most scanning phased array antenna systems,
the time sequenced antenna beam positions in space are known a priori. Thus, the finite
delay time can be overcome if the FLC pixels are addressed 17 ps (or 30 ps) earlier. As
was shown in section 11.2, the time delay switching requirement is 0.5 ps. Nevertheless,
FLC materials have shown switching speeds as low as 0.5 ps [14] and thus future
developments in FLC technology can lead to 0.5 ps switching times. At present, the time
delay switching limitation can be overcome using time multiplexing for our PDL [15]. In
this case, two pixels o f the FLC switching array will be used for each PDL channel.
During the dwell time o f one beam of the phased array antenna, the second pixel is set for
the next PDL setting which corresponds to a different beam position. So when the dwell
time ends, the second set o f pixels assigned to the individual antenna elements will
operate, overcoming in that way the 100 ps switching time of our FLC devices. In order
to set the second set of pixels to the next PDL setting, the antenna beam position must be
known. As mentioned earlier when the phased array is in the scanning mode, the time
sequenced antenna beam positions in space are known a priori and thus the PDL settings
are known too.
(a)
(b)
Figure 11.14: Oscilloscope traces o f the switching time o f the 33 pixel BNS FLC devices (a) a 37.6 ps rise
time and (b) 100.4 ps fall time. Note also the finite delay times o f 17 ps and 30 ps o f the FLC response to
the applied waveform.
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.8. Wideband Operation o f the 7-bit PDL system
One of the main motivations for using photonics for the implementation o f delay
lines is to obtain wideband operation without any change in the hardware. We tested our
system for the entire band of 3-6 GHz. Fig. 11.15(a) shows the RF spectrum analyzer
measurement at 3 GHz for the FO-link without the PDL. Fig. 11.15 (b) shows the RF
spectrum analyzer measurement at 3 GHz for the FO-link with the PDL. In the second
case we have used gain balancing to compensate for the insertion loss o f our PDL. Gain
balancing is accomplished by using higher optical power at the input to the PDL. This
technique was described in chapter 10, where we demonstrated that an optical attenuator
system can be used for dynamic range compensation and signal calibration. Note, that the
noise floor for both measurements is the same. This means that the PDL does not
introduce any extraneous noise to the FO-link. The measurement was repeated at 4 GHz,
5 GHz, and 6 GHz, where similar results were obtained. Fig. 11.15 shows the 6 GHz RF
spectrum analyzer measurements.
A
,
, • 'V' ■/
(a)
(b)
Figure 11.15 (a) RF spectrum analyzer trace showing the 3 GHz signal o f the FO-link without the PDL, (b)
RF spectrum analyzer trace showing the 3 GHz signal o f the FO-link with the PDL using gain balancing.
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(b)
Figure 11.16: (a) RF spectrum analyzer trace showing the 6 GHz signal o f the FO-link without the PDL,
(b) RF spectrum analyzer trace showing the 6 GHz signal o f the FO-link with the PDL using gain
balancing.
Note also that the use of gain balancing allows our system to perform at its
maximum dynamic range. This is because, as we showed in chapter 10, external
modulation links have high optical power available and by using the amplitude control
system we can increase the input optical power to the PDL. Thus, the optical signal after
propagation through the system can be at levels required from the photodetector for
maximum dynamic range. Thus, results similar to the ones presented in chapter 10 for the
compression dynamic range and the spurious-free dynamic range can be obtained for our
7-bit 16 active channel PDL system.
11.9. RF Leakage Noise Measurements for the 7-bit PDL
Leakage noise measurements were also obtained for the 7-bit 16-channel PDL
system. Leakage noise equals the negative value o f the SNR, where SNR = 20 log
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(signal/leakage noise). As signal we define the light at the output o f the PDL that
propagates through the desired path, all other light at the output of the PDL is considered
as leakage noise. The average RF leakage noise, for the 128 different PDL settings, was 77.2 dB, with a standard deviation of 14 dB. This is better than the - 60 dB leakage noise
specification required for our project.
11.10. RF Interchannel Crosstalk for the 7-bit 16-channel PDL System
Interchannel crosstalk measurements were also obtained for our PDL. The worst
case is for the immediate neighboring channels. Fig. 11.17 shows the interchannel
crosstalk measurement for the central channel. The measurements were obtained using a
power meter. The optical power coming out o f the MM-fibers was measured when only
the central input fiber launches light into the PDL. A RF interchannel cross-talk < - 90 dB
was obtained. This is more than 30 dB better than the delay line requirements.
o.
9
3to
-2 0 •4 0 .
8
-6 0 O
-C
i
e
Radar Requirement:
-60 dB
-8 0 -1 0 0 -
I
_L 1
Fiber
Figure 11.17: Interchannel crosstalk measurements for the central channel.
233
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.11 Conclusion
In conclusion, we have demonstrated the first ever 7-bit 16 active channel
photonic delay line. This system is also readily available for extension to 33 channels by
inserting additional FO-collimators, since the FLC devices have 33-pixels. We have met
the within channel RF polarization leakage noise specification o f < - 60 dB for all 128
different settings using our novel active noise filter at the output o f each individual bit.
We have demonstrated RF interchannel crosstalk numbers o f < -90 dB for the immediate
neighboring channels. This is over 30 dB better than the required delay line interchannel
crosstalk. Using our modular adaptable PDL architecture we accomplished exact match
for five bits of time delays. Due to space limitations (i.e., optical table o f 5 ft x 8 ft in
dimensions), we have also used a symmetric architecture and a feed-forward architecture
PDL bit with similar RF performance to the adaptable PDL bits. We have designed and
built a SM polarization-maintaining fiber array, with independent micro-tilt control for
each o f the fibers to feed the optical signal to the system. A similar MM-fiber array was
built to collect the optical signals and deliver it to the fast photodetectors. This MM-fiber
approach can be used in our system because the distance between the photonic controller
and the antenna site is less than 100-150 m and thus frequency and modal dispersion is
negligible. We have also designed the FLC arrays with large pixels (e.g., 1.65 mm),
which makes our system alignment tolerant.
Future development o f this system includes improved on/off isolation FLC
performance obtained by using two HWPs, one in front and one after the FLC arrays to
compensate for the small FLC optical axis tilt with respect to the array axis. Improved
alignment procedure will also be followed to further reduce optical insertion loss for the
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
system, and smaller optical insertion variation between the different PDL settings.
Finally, a better telescope design can be used to reduce astigmatism on the outer beams,
which in effect will reduce the excess optical insertion loss for the outer channels.
235
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 12
ALL-FIBER CONNECTORIZED FIBER-OPTIC DELAY MODULES
USING 3-D POLARIZATION OPTICS
12.1. Introduction
An important issue for the implementation of PDLs that use 2-D polarization
switches is packaging. Compact PDL modules are necessary for fielded systems where
space and volume on the application platform (e.g., aircraft) are limited. So far, no ultra­
compact PDL design based on polarization bulk-optics has been proposed. In addition, no
polarization-based fiber PDL has been proposed that is easily adaptable to generate any
user desired optical delay. Current fiber optic (FO) cable and connector technology is an
excellent candidate for interconnecting environmentally robust and well engineered delay
control modules. In this chapter, a PDL module that meets the above mentioned
requirements i.e., compactness and time delay adaptability is proposed and demonstrated.
Our approach for achieving this PDL involves the use of microlens FO-collimators and
FC/PC connectorized fiber optics. Two different FO-collimator technologies are tested.
The first one is based on spherical microlens-based fiber ports and the second one is
based on GRIN-Iens collimators. A compact single micro-optical bench PDL is
experimentally demonstrated for each o f these technologies [1,2].
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12.2. Alternative Compact PDL Module Designs
A fiber connectorized A-bit PDL approach includes ease in assembly of Ambits,
ease in maintenance and repair, and compact size via fiber delay paths. A typical fiberoptically interconnected PDL system can be formed by a cascade o f single bit compact
PDLs (Fig. 12.1). This modular approach gives the flexibility to the system designer to
choose from a wide range o f available time delays, and their combinations.
Bit Structure
Out
1:2 FO
Switch
2.1
Combiner
Fiber-Optic
FC/PC
Connector
Photonic
Single Bit
Delay
Line
Bit 1
r Cable
FC/PC
Connectors
Photonic
Single Bit
Delay
Line
Bit 2
FC/PC
Connectois
Photonic
Single Bit
Delay
Line
Bit N
FC/PC
Connector
Figure 12.1: A fiber optically interconnected PDL system using a cascade o f single bit compact PDL
modules.
A typical fiber-optically connectorized PDL module can be implemented via a 1:2
FO switch that routes the signal to either the delay or non-delay path and a FO-combiner
(or 2:1 FO-switch) that combines the two independent channels into one before the signal
propagates to the next bit. Possible compact PDL module approaches include electro­
mechanical FO-switches and integrated electro-optic switches. Two possible PDL
configurations based on the above switching technologies are shown in Figure 12.2. The
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Electronic Contol Ports
FC/PC
connectors
In/Out
Electromechanical
FO Switch
85 mm
Reversible Design
Out/In
Electromechanical
FO Switch
SMF/MMF
0.25 dB
85 mm (a)
FC /PC
connectors
Electronic Contol Ports
Integrated ElectroOptic Switch
N on -R eversib le
D esign
Optical Combiner
(b)
Figure 12.2: (a) Electro-mechanical based PDL, and (b) Integrated electro-optic switch based PDL.
first technology can provide very good optical isolation (e.g., 60 dB) and low insertion
loss (~ 0.5 dB), but rather slow switching speeds o f > 7 ms [3]. On the other hand, the
integrated electro-optic switches offer faster switching speeds (e.g., 100 ps), but they
have poor optical isolation o f ~ 22 dB, and inherently high insertion loss (~ 4 dB) [4].
This high loss in conjunction with the ~3.5 dB loss o f a 2:1 FO-combiner leads to a PDL
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
module with insertion loss as o f > 7.5 dB for both PDL settings. Our compact PDL
module using bulk polarization optics and FLC polarization switches provides a system
trade off with < 10 j is switching speeds and > 42 dB optical isolations [5].
12.3. The Bulk Optics Ferroelectric Liquid Crystal based Compact PDL Module
Fig. 12.3 shows the proposed bulk optics FLC based PDL module. Polarization
maintaining fibers (PMF) must be used at the input and output ports since the module is
based on polarization switching. On the other hand, a non-PMF is used at the delay path.
The PDL is based on a reflective design that allows the use o f Faraday rotator-mirror
(FRM) connectorized SMF in the delay path with no signal degradation [6]. Thus, shorter
fiber lengths are required. Note that this easy to connect/disconnect option using FC/PC
connectors allows the user to achieve different time delays by simply changing the SMF
lengths. The PDL module using fiber micro lenses is expected to have optical losses of 3.2
dB for the non-delay path, and 5.4 dB for the delay path. The higher insertion loss for the
delay path is due to the insertion loss of the FRM and the additional FC/PC connectors.
These numbers have been estimated based on optical component loss numbers given by
the manufacturers, and expected FLC device losses of 0.45 dB (or 10 %).
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Electronic Contol Ports
4
»
In/Out
Reversible Design
FC/PC
connectors
Bulk Optics
FLC based
Switching
Fabric
85 mm
Out/In
85 m m
Figure 123 : The reversible bulk optics FLC based compact PDL module.
12.4. Spherical Microlens FO-Collimator based Switched Compact PDL
The experimental set-up of the bulk optics FLC based switching fabric is shown
in Fig. 12.4. An externally modulated FO-link with a diode pumped Nd:YAG laser at
1319 nm and a Mach-Zehnder electro-optic modulator is used. The modulated light is
coupled into the PDL via a PMF. This light is collimated using a microlens positioned in
an OFR fiber-port (PAF-X) which has three translation (x, y, z) and two tilt micro­
controls [7]. A linear polarizer is attached at the input port and is set to let only horizontal
or p-polarized light into the system. FLC devices are used as polarization switches (PSs).
PS1 acts as a switch to either change the polarization to vertical or 5-polarization when it
is set in its “ on” state, or leaves the input polarization unchanged if it is set in the “ o ff’
state. When p-polarized light hits the PBS, it travels through the fiber delay path. On the
other hand, when PS1 is set in its “ o ff’ state, the s-polarized light travels through the
240
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
non-delay path. The polarization of the light rotates by 90° after propagation through any
o f the two paths, and is directed through the PBS towards the output port. PS2 and an
additional polarizer forms the active noise filter to suppress any leakage noise coming
from PS1 or the PBS [5]. Output light is coupled via the PMF jumper to the
photoreceiver. The delay path consists of a FC/PC connectorized 10 cm length SMF
connected with a FC/PC connectorized 3 m long SMF terminated with a FRM. Fig. 12.5
shows photographs o f our compact PDL module built in the laboratory. The compact
PDL was built on a 40 mm x 40 mm aluminum optical bench. The OFR fiber-ports were
mounted at its sides. The overall size of the compact PDL, including the FC/PC
connectors attached on the fiber-ports, is 80 mm x 80 mm.
In
Light
Out
Input/Output
Port
Microlens
pMF
FC/PC
Jumper
FC/PC
SMF
Jumper
FC/PC
Microlens
Fiber-port
Micro lens
Output/Input
Light
In
Out
Port
FC/PC
PMF Jumper
SMF
FC/PC
FR-M
Figure 12.4: The experimental set-up o f the compact PDL module based on spherical microlens fiber-ports.
241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
OUT/IN f
II
c
S\ \ i l cl l Cs
l i her I )el a\
Por i
IN/OUT
<-------
OWP
PBS
M i r r o r Port
(a)
(b)
Figure 12.5: Photographs o f the compact photonic delay line based on spherical microlens fiber-ports (a)
Top view, (b) perspective view.
Table 12.1 shows the measured optical losses for the two settings as well as the
optimum projected loss. The expected loss numbers for our experimental PDL have been
calculated based on the optical component loss numbers given by the manufacturer and
FLC insertion losses of 0.45 dB (or 10%). The difference between the optimum loss and
the experimentally measured data is due to the higher insertion loss of the FLC devices,
242
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.96 dB and 0.75 dB (20% and 16%) used in our experiment This higher FLC device
loss leads to a 0.7 dB higher insertion loss for our prototype. Additional losses (~0.7 dB)
are due to alignment
Table 12.1: Expected and measured optical insertion loss.
PDL Setting
Expected Optical
Measured Optical
Difference
Loss (dB)
Loss (dB)
(dB)
Delay
4.9
6.2
1.3
Non-Delay
3.1
4.6
1.5
So far, researchers have used costly and sensitive microwave band test
instrumentation to characterize the time delays obtained from PDLs designed for radar
and antenna control applications. As we show perhaps for the first time, time delay
measurements of a PDL can also be obtained using a low frequency technique that does
not require costly microwave test equipment [8]. Time delay measurements of our PDL
were also obtained using this low frequency. Fig. 12.6 shows the low frequency square
wave signal that are used to drive the microwave analog intensity modulator (top traces).
Fig 12.6 also shows the photodetected signal at the output of the PDL module for both of
its settings (bottom trace). We can conclude that the obtained time delay is 54.90 ns 23.90 ns = 31 ns which is in agreement with the expected time delay from our 3.1m SMF
jumper.
243
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(b)
Figure 12.6: Oscilloscope traces showing (a) the non-delayed and (b) the delayed signal. Top traces: signal
driving the external modulator; Bottom traces: photodetected signal at the output o f the PDL module. The
markers have been positioned at the on-set o f the pulse, where the pulse gets to 10% o f its maximum value.
The optical SNR was also measured for our compact PDL module. For the non­
delay setting the optical SNR was measured at 39.7 dB, while for the delay setting it was
38.0 dB. The lower SNR observed for the delay path is due to the lower insertion loss of
the non-delay path, which allows higher leakage noise to reach the output o f the PDL,
compared to the delay setting case. However, the 38 dB SNR value is highly desirable for
phased array antenna applications.
12.5. GRIN Lens FO-Collimator based Switched Compact PDL
Fig. 12.7 shows the GRIN lens FO-collimator based PDL module. The design of
Fig. 12.7 is similar to the one presented in the previous section. Light from an externally
modulated FO-link is fed via a PMF into the PDL. This light is collimated using a GRIN-
244
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
lens FO-collimator from OZ-Optics. The GRIN lens FO collimator is positioned in an
OZ-Optics fiber collimator flange for tilt micro-controls [9], and on a standard x-y
translation micropositioner from OptoSigma. A linear polarizer is attached at the input
port and is set to let only horizontally polarized light into the PDL. FLC devices are used
as polarization switches (PSs) that either rotate the incident polarization by 90° when they
are set “ on” , or leave the input polarization unchanged when they are set “ o ff’.
Depending on the polarization, the light can either follow the non-delay path or the delay
(fiber) path. The output GRIN lens FO collimator collects the optical signal and delivers
it to the photoreceiver.
FO Flange
In
Light
In p u t/O u tp u t
p i^ p
^ Port
Jumper
Out
GRIN
F c /p c
Jumper
/
F ib e r-p o rt
FC/PC
GRIN
Output/Input
Port
In
Light
Out
FO Flange
PMF Jumper
FR-M
Figure 12.7: The experimental set-up of the compact PDL module based on GRIN lens collimators.
The optical loss for the two settings of our PDL was measured at 7.0 dB and 5.7
dB for the delay and non-delay paths respectively. The expected optical loss for the delay
245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
path is 6.5 dB while for the non-delay path is 5.5 dB. This expected loss has been
calculated based on the optical component loss numbers given by the manufacturers, the
measured FLC insertion losses o f 0.9 dB (or 19 %), and the expected FO coupling
efficiency using GRIN-to-GRIN distances compatible with our PDL. Additional losses (~
0.5 dB) are due to alignment. Again the insertion loss o f our PDL is mainly due to the
triple-cell FLC devices and the GRIN lens FO collimator coupling efficiency. As
mentioned earlier, the optical insertion loss of the PDL can be improved by ~ 0.9 dB if
single cell geometry FLC devices with higher birefringence FLC material are used. The
coupling efficiency o f the GRIN lens collimator assemblies can also be improved.
Although GRIN lens collimator technology is rather mature, there are tolerances related
with the fabrication and alignment procedures o f the fiber to the GRIN lens interconnects
that limit the coupling efficiency, especially at separations o f > 5 cm. We have
experimentally found that these loss numbers are highly dependent on the specific pair of
fiber collimators used. Experiments have shown that maximum coupling efficiency varies
with different collimators and for each pair occurs at a different GRIN-to-GRIN
separation. Fig. 12.8 shows the experimental data for optical insertion loss versus the
GRIN-to-GRIN distance. This optical insertion loss can be found within a band of ~ 0.7
dB. Note, that the beam propagates as a Gaussian beam with a specific beam divergence,
thus expanding as it propagates in free space. A GRIN lens FO collimator requires a 0.25
pitch GRIN lens. Often a smaller length GRIN lens is used, leaving an air gap between
the lens and the fiber, in order to reduce the back-reflection levels [10]. This air gap can
also be used to set the beam waist at a specific distance from the face o f the GRIN lens
[10] as shown in Fig. 12.9. This can result in a maximum coupling efficiency for larger
GRIN-to-GRIN separations. For example, the experimental data represented by circles in
Fig. 12.8 shows that the best coupling efficiency occurs for a 5.5 cm GRIN-to-GRIN
246
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
distance. Thus, if two similarly fabricated GRIN lens FO collimators with their waists at
a distance d from the GRIN lens are used, maximum or optimized coupling efficiency
will be obtained for a GRIN-to-GRIN separation o f 2d (Fig. 12.9(b)). Hence, it is possible
to design and manufacture GRIN lens FO collimators optimized for the free space
propagation o f our PDL paths and thus to obtain improved insertion loss numbers.
10-
Insertion Loss Band
otn
C
(0
o
a.
O
0
50
100
150
200
250
300
350
400
450
500
GRIN-to-GRIN distance (mm)
Figure 12.8: Measured optical insertion loss data between a pair o f GRIN lens FO collimators as a function
o f their distance from each other.
247
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Beam waist
i ___________ _
GRIN
0.25 gitch
_
(a)
GRIN
Beam waist
GRIN
~ r
-x -
<(b)
< 0.25 pitch
<r >
GRIN
Figure 12.9: (a) 0.25 pitch GRIN lens, (b) < 0.25 pitch GRIN lens with an air gap between the GRIN lens
and the fiber.
Another version o f our compact PDL can be realized if a PMF with Bragg
gratings is used in the fiber delay path as the one proposed in chapter 7 [2]. This approach
makes our single physical channel PDL module capable for multichannel operation. In
this PDL version, a quarter wave plate needs to be used in the fiber delay path. Fig. 12.10
shows such a design. The fiber Bragg gratings (FBGs) have to be recorded at specific
distances from each other such that the desired time delays are obtained for the different
wavelengths. An alternative approach for the delay path is one that consists o f long length
PMFs (with no FBGs) and small length PMFs with FBGs (Fig. 12.11). All the short
length PMFs are o f the same length. Hence, the desired time delays can be obtained with
appropriate choice of the long PMFs. This approach gives greater time delay flexibility to
the end user, since the time delays depend mainly on the length of the PMFs. Typical
dimensions of a PMF pigtailed FBG is 15-20 cm from each side [11]. The grating size is
~ 5 mm, and depends on the specification of the wavelength, the FWHM bandwidth, and
the desired reflectivity. Important issues with the FBGs based fiber delay are the
interchannel isolation and the effect of the distributed reflection of the wavelengths from
the FBGs. The interchannel isolation depends on the reflectivity of each FBG for the
248
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
desired wavelength, and its transmission characteristics for the wavelengths outside its
FWHM bandwidth. The distributed reflection happens because the FBG is not a thin
interface mirror, but a 3-D structure and it is based on constructive or destructive multiple
beam interference. Nevertheless, it is expected that the distributed reflection will not
cause significant time jitter errors.
PSl
QWP
M icrolens
Input/O utput
PM F Jum per
PM F
Jum per
Port
FC /PC
/
L ight
O ut
FC/PC
FC/PC
Fiber-port
O utput/Input
PS2
FC/PC
P ort
FC/PC
M icrolens
PM F Jum per
L ight
O ut
PM F w ith
FBG s
Figure 12.10: Single physical channel, wavelength dependent, compact PDL for multichannel operation
using FBGs. (P: polarizer; M: mirror; PMF: polarization maintaining fiber; QWP: quarter wave plate; FBG:
fiber Bragg grating; PS: polarization switch).
PM F
-rrpm
w -rr
|
pm f
W M l efe
FC/PC
,
connectors
pM F
-r
IIIIH
absorber
Figure 12.11: Delay path option based on alternating long length PMFs without FBGs, and short length
PMFs with FBGs.
249
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12.6. Ferroelectric Liquid Crystal Switching Speed
In the previous sections we demonstrated two compact PDL modules based on
two different FO-collimator technologies. Optical isolation of ~40 dB and optical
insertion loss numbers o f < 7 dB were obtained. These insertion loss numbers are much
better than the projected numbers for a compact inertialess switching PDL based on
integrated electro-optic switches. We have also shown that these insertion loss numbers
can be further reduced by proper design. The optical isolation performance of our PDL is
20 dB better than the integrated electro-optic approach but not as good as the electro­
mechanical switch option. Nevertheless, electromechanical switches are rather slow (e.g.,
> 7 ms) [3]. Thus the challenge is to obtain high switching speeds at 1319 nm for our
FLC-based PDL modules. Figure 12.12 shows the switching response o f our FLC
devices. Fig. 12.12 shows oscilloscope traces of the photodetected output signal from our
PDL when one o f the PDL paths has been blocked. A 9.84 ps rise time and a 10 ps fall
time is observed. As rise time (fall time) we define the time it takes for the signal to rise
(fall) from the 10 % (90%) to the 90% (10%) of the maximum signal power. The FLC
devices used were fabricated by Displaytech Inc., Longmont, CO, and the material is
Displaytech proprietary technology. The FLC device consists of three cells stacked
together. This is because the FLC material used does not have enough birefringence for
the 1319 nm band, and the preparation of a thick enough single cell is at present not
feasible. Thus, the FLC-based compact PDL gives a trade off between switching speed
and optical isolation and loss. In general, our approach gives a better overall performance
250
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
since it combines fast response time (10 ps) and high optical isolation (e.g., 40 dB) as the
ones required for phased array antennas.
(a)
(b)
Figure 12.12: FLC switching response at 1319 nm (a) a 9.84 ps rise time and (b) a 10 ps fall time.
12.7. Alternative Compact PDL Module based on Microelectromechanical System
Technology
An alternative approach for the implementation o f ultra-compact PDL modules is
based on the microelectromechanical system (MEMS) technology. The optical elements
can be positioned on top o f a' silicon surface-micromachined optical bench. These
mechanical structures have the functionality, stability and accuracy required for active
micro-positioning control o f the optical elements via electromechanical microvibromotor
actuators [12]. This technology can lead to very small size (e.g., 25 mm x 10 mm x 5
mm) and lightweight modules (e.g., ^ 0.3 Lb.). Figure 12.12 shows a possible design for
251
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
an ultra-compact PDL. The micromirrors can be moved along the direction of the optical
beams and thus a relative time delay between the two optical paths can be obtained. If
one of the micromirrors is replaced by a GRIN-lens based FO-collimator, as shown in
Fig. 12.13, extra-long delays can be acquired.
F ib e r d e la y o p tio n fo r
e x tra lo n g d e la y s
SM-Fiber*
Hinge
M icrom irror
7 1
Slider
PM Fiber
IN/OUT
M ia o m m o r
G : G R IN L e n s
P: P o la riz e r
S: P o la riz a tio n S w itc h
L I : S p h e ric a l L en s I, F I
L 2 : S p h e rica l L en s 2 , F 2
P B S : P o la riz a tio n B e a m sp litte r
M : M ic ro v ib ro m o to r a c tu a to r fo r tran sla tio n
N : M ic ro m a c h in e d A lig n m e n t G ro o v e /S lo t
L: M ic ro v ib ro m o to r a c tu a to r fo r ro tatio n
PM Fiber
IN/OUT
Figure 12.13: The ultra-compact PDL module based on MEMS technology. Both ultra-short time delay and
extra-long time delay options are shown.
12.7. Conclusion
In conclusion, in this chapter, we have proposed and experimentally demonstrated
for the first time, a single microbench fiber connectorized compact PDL using spherical-
252
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
microlens and GRIN lens FO-collimators, 3-D polarization optics, and FLC devices.
Insertion losses of < 7 dB were measured for both PDL modules. Important GRIN-toGRIN coupling loss data is presented. An approach using special design non-quarter pitch
FO-collimators is suggested for the large GRIN-to-GRIN distances in our PDL. A novel
low frequency technique was used to make time delay measurements at the high RF
frequency that does not require costly microwave test equipment. Future work relates to
further PDL miniaturization and multi-wavelength multichannel tests. Future work also
includes implementations using silicon micro-machined optical benches based on MEMS
technology.
253
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 13:
PHOTONIC DELAY LINES USING POLYMER DISPERSED LIQUID
CRYSTAL TECHNOLOGY
13.1. Introduction
In the previous chapters we described and demonstrated several PDLs using flat
planar liquid crystal technology such as NLC and FLC polarizations switching devices.
Nevertheless,
we are
currently exploring new photonic technologies for the
implementation of high performance optical switches, routers and combiners, that will
form a high performance switching fabric to be used in different PDL architectures. Such
technologies are based on polymer dispersed liquid crystals (PDLCs) and polarizaion
selective holograms (PSHs).
First, in this chapter, unlike our previous NLC and FLC based PDLs, we propose
and demonstrate for the first time, a PDL that does not require the use of passive
polarization optics such as cube polarization beam-splitters and beam-combiners for
optical signal routing [1]. Instead, the proposed PDL uses electrically switchable gratings
based on PDLC to do both the active optical signal switching and the passive optical
signal routing. Thus, a lower component count and potentially lower cost PDL is feasible.
254
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These PDLC devices can also form two-dimensional (2-D) switches [2] that can be used
as parallel processing elements to implement a multichannel PDL.
A new promising technology for optical signal routing is polarization-selective
holograms (PSHs). PSHs are optical elements that have a different phase function
depending on the state o f polarization of the incident light, and can be used to route an
optical beam at different angles depending on the polarization. Thus, in this chapter we
also propose and experimentally demonstrate a PSH-based PDLs. In our experiments,
PDLC devices are used as fixed PSHs to form beam routers and combiners. FLC devices
are used as polarization switches.
13.2. Basic Characteristics of the Polymer Dispersed Liquid Crystal Devices
Researchers have recently been studying the electro-optic characteristics and
applications of PDLCs [2 - 10]. The basic PDLC device is as an active element that can
be electrically switched from a light scattering state to a transparent state without
polarizers and alignment layers. Hence, these PDLC devices have been used as light
valves for displays [5]. Recently, holographic elements with diffractive powers that can
be electrically modulated have been demonstrated using PDLCs [4, 7]. These holographic
PDLCs diffract light because liquid crystal droplets are arranged in periodic channels,
thus creating an index modulation. Thus, PDLC-based switchable diffractive elements for
optical beam steering, holographic read only memories (ROMs), and switchable focus
lens applications can be realised [2,9].
255
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The PDLC electrically switched gratings are based on a photopolymer
holographic recording material (e.g., Polaroid DMP-128) whose microstructure after
processing contains a distribution o f submicron pores corresponding to a variation of
refractive index [2, 3, 11]. The porosity permits birefringent liquid crystals to be infused
into the material, filling the empty spaces. Thus an index modulation can be realized. By
selecting the appropriate liquid crystal material, the right index modulation can be
achieved which in conjunction with the proper grating thickness and period, and incident
linear polarization, results in a thick grating that can strongly diffract light into the first
order for the correct Bragg angle [12]. By reorienting the liquid crystal molecules in an
electrical field, the effective index o f the nematic liquid crystal filled pore obtains a value
that is close to the index o f refraction of the adjacent polymer areas and thus the index
modulation vanishes and therefore the diffraction grating vanishes. This applied electrical
field is of the order o f ~200-400 V across a 8 pm thick material. This is a rather high
voltage and it is currently limited due to the large surface interaction of liquid crystal
molecules in the sub-micron pores [2]. Another important class o f holographic PDLC
gratings are based on a simple single step fabrication technique, the polymerizationinduced-phase separation (or PIPS) process. In the PIPS process the grating is formed by
phase separation of the liquid crystal during holographic curing o f the photopolymer.
This is done using a homogeneous mixture of a photopolymer and a liquid crystal [13].
The chemical potential of the system changes as the photopolymer cures. This increases
the miscibility gap between the liquid crystal and its host [14]. The liquid crystal
therefore separates as a microdroplet phase and an index modulation between the polymer
and the liquid crystal droplets can be realized. Thus a grating that strongly diffracts into
the first order can be formed if the right combination of grating thickness, period, and
state o f polarization of the input light are selected.
256
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The ordinary refractive index (/70) of the liquid crystal is chosen to match the
refractive index of the surrounding medium («p). The extraordinary index ( n j differs
significantly [14]. When the electrically controlled refractive index matches the index of
refraction of the host polymer, the index modulation vanishes and thus the grating also
vanishes. Light then travels straight through the device without any deflection. When the
applied voltage is removed, the liquid crystal molecules reorient back to their initial state,
again creating an effective index modulation that significantly differs from the polymer
host index of refraction. For a particular polarization and direction o f light incident at the
Bragg angle the PDLC device diffracts light into the first order for one o f its states and
leaves the light unaffected for the other state.
TOP VIEW:
Polymer host
Pores infused with LC
3
s-polarization
s-polarization
Out
Out
npolymer host = nLC
Grating does not exist
npolymer host £ nt_c
Grating exists
V= 0
(a)
(b)
Figure 13.1: The principle o f operation of a polymer dispersed liquid crystal device. (wpt,lrn,trh~ .= index of
refraction of the polymer host, nLC= the eiectricaly controlled index of refraction o f the pore infused with
liquid crystal, LC: liquid crystal).
257
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The switching time o f PDLCs depends on the polymer and liquid crystal material
used, the morphology o f the PDLC, and in particular, parameters such as the liquid
crystal droplet size and shape and the PDLC channel width and liquid crystal fill factors
[15, 16]. The driving electrical modulation also plays a role in the switching speed.
Switching times of < 50 ps have been reported [6] using NLCs, which in general have
switching times of ms. This fast switching time is due to the small droplet size < 1 pm.
Another important issue with the switching response of PDLC devices is that the
relaxation time is two times slower than the switching time. Detailed analysis of the
switching characteristics o f the PDLC devices can be found in references [15] and [16].
13.3. The electrically Switched PDLC-based Photonic Delay Line
The experimental set-up for the proposed PDLC-based PDL is shown in Fig. 13.2.
A single AOM system is used to modulate the input light so time delay measurements
can be taken from our PDL. The AOM basically acts as an ON/OFF shutter for the input
light. Hence, pulses of light are sent through the PDL. Vertically polarized light from a
30 mW He-Ne laser is focused onto the AOM. The AOM is driven by the output signal
from a Mini Circuit Model ZAD 1-1 electronic mixer. The mixer inputs are a 500 MHz
carrier signal (7 dBm) and a square wave pulse (1 dBm). The pulse duration was 194.4 ns
and the pulse period was 779 ns. The DC beam of the AOM is blocked and the +1 order
beam is used as the input to the PDLC based PDL system. A Thompson polarizer (noted
as P in Fig. 13.2) is used to provide high extinction ratio vertical polarized light. In order
for the PDLC grating to work effectively, the incident polarization needs to be along the
258
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
grating fringe direction thus a HWP is used to rotate the incident vertical polarization to a
horizontal polarization. Note that in any single bit PDL module, the two paths have to be
combined before the beam exits the module. This is because single bit modules are
cascaded to form the W-bit PDL. As seen in Fig 13.2, we use a PDLC device for this
important 2:1 recombination operation. When these PDLC devices are set “ o ff’ (grating
exists), the diffraction efficiency is measured as 95.5% and 62.5% for PDLC1 and
PDLC2 respectively. On the other hand, when the PDLC devices were set “ on” (no
grating exists), 95.4% and 99.0% of the light remains on the zero order beam for PDLC1
and PDLC2, respectively. We are interested in the optical on/off ratio at the output ports
of the device switching planes. These on/off device numbers will ultimately limit the
performance of the PDL single bit module, particularly in terms of the switch-based
leakage noise in the module. In our experiments, we used two PDLC based switchable
gratings fabricated by Foster-Miller, Inc. The polymer host is DMP-128, with index of
refraction 1.56 [2]. The liquid crystal material is E7, with ordinary and extraordinary
indices of refraction of n0 = 1.54 and ne = 1.74 respectively [2]. If the electrically
controlled pore refractive index matches the index of refraction of the host polymer, the
index modulation vanishes and thus the grating also vanishes. Light then travels straight
through the device without any deflection (Fig. 13.1a). When the applied voltage is
removed, the liquid crystal molecules reorient back to their initial state, again creating an
index modulation and thus a grating (Fig. 13.1b). Hence, the PDLC device diffracts light
to a predefined direction for one of its states and leaves the light unaffected for the other
state. Thus, input light can be routed to one o f two independent optical beam paths. In
this way a binary state photonic switch is realized that forms the basis o f our PDL
structure. Note that turning a PDL device “ on” electrically turns the diffractive power of
the device “ o ff’, and vice versa. The optical on/off ratio is defined as the ratio of the
259
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
output optical power measured at the desired device output port when the device is set for
“ on” operation versus the output optical power measured when the device is set for
“ o ff’ operation. PDLC1 had measured on/off ratios of 21/1 and 1/21 for the nondiffracted and diffracted device ports, respectively. PDLC2 had measured on/off ratios o f
19/1 and 1/2 for the non-difffacted and diffracted device ports, respectively. Switching
time measurements were not obtained due to the non-availability of the proper driving
circuit, but as mentioned earlier they can be of the order of 50 ps or better.
He-Ne
Laser
Delay path
0
Vertical Polarization
| Horizontal Polarization
Spatial
Block
AOM
PDLC2
PDLC1
62°
DC
Block
52
Noise
Non-delay path
“ HWPr
//■w
Input Port
O utput
Port
Detector
Figure 13.2: The experimental set-up of the proposed PDLC-based photonic delay line, shown m a single
bit configuration. (AOM: acousto-optic modulator; L: lenses; P: polarizer; HWP: half wave plate; M:
mirrors, 0Bl: Bragg angle for PDLC1; 0B2: Bragg angle for PDLC2).
When no voltage is applied to PDLC1, light is diffracted at an angle of 62° from
the zero order beam and follows the delay path as indicated in Fig 13.2. After reflection
from mirrors Ml and M2, this light hits PDLC2 that is set “ on” . Thus the light
propagates straight through the PDLC2 device without any deflection. For the non-delay
260
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
path operation, PDLC1 is set “ on” and the grating does not exist. The light propagates
straight through PDLC1 towards PDLC2 which is set “ off” and hence PDLC2 diffracts
the light at an angle o f -52° with respect to the zero order beam. The PDL is designed
such that both PDL settings generate an output beam that follows the same output path, as
required for cascading PDL modules. Note that we drive the two PDLC devices out o f
phase, i.e., when PDLC1 device is set “ on” the PDLC2 device is set “ off” , and vice
versa. We do this because the two PDLC devices do not perform equally well in both o f
their states. We call this out of phase operation the “ orthogonal drive” configuration
[17]. This “ orthogonal drive” device configuration was first proposed and demonstrated
in chapter 4, to compensate for the poorer performance of the FLC devices when they are
set in their “ on” state compared with when they are set in their “ off” state.
13.4. PDLC-based Photonic Delay Line Issues
13.4.1. Time Delay Measurements
Time delay measurements were obtained for our PDL. Fig. 13.3 shows the square
wave envelope that was used to drive the AOM (top traces). Fig. 13.3 also shows the
photodetected signal (bottom trace) at the output of the PDL module for both the (a) non­
delayed and (b) delayed settings. The time markers have been positioned at the on-set o f
the traces. As on-set time we define the time when the pulse gets to 5% o f its maximum
value. A high speed photodetector (New Focus, model: 1801) was used to detect light at
the output o f the PDL. From Fig. 13.3, we can conclude that the obtained time delay is
261
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
126.55 ns - 121.85 ns = 4.70 ns, which is in agreement with the expected time delay
based on the optical layout. The total length of the non-delay path is 16.1 cm while the
length of the delay path is 159.2 cm. Thus, a time delay o f 4.77 ns is expected.
OOmU
200ml
*
Figure 13.3: Oscilloscope traces showing (a) the non-delayed and (b) the delayed photodetected RF signal,
indicating a 4.7 ns relative time delay. Top traces: signal driving the AOM; Bottom traces: Photodetected
signal at the output of the POL module.
Our PDL measured optical loss is ~2 dB for the delay path, and ~3.9 for the non­
delay path. These numbers include the reflection, scattering and absorption losses o f our
specific PDLC devices (PDLC1 loss: 20%, PDLC2 loss: 14%), as well as the diffraction
efficiencies o f each PDLC device for each of their settings. The higher loss in the delay
setting is due to the low diffraction efficiency (62.5%) o f our specific PDLC2 device
when is set “ o ff’. The use o f AR coatings combined with our PDLC1 type high
diffraction efficiency (i.e., > 95%) devices for all PDLC components can result in a low
loss (< 1.5 dB) PDL.
262
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.4.2. Leakage Noise in the PDLC-based Photonic Delay Line
The important issue o f the electrical SNR was also examined. Note that the SNR
is a measure o f the leakage noise in the system due to the non-optimized PDLC
diffraction efficiency. Using the measured characteristics of our PDLC devices, the
expected electrical SNR can be calculated and is shown in Table 13.1. Note that the non­
delay setting electrical SNR is ~30 dB lower than that o f the delay setting due to the
much lower (i.e., 62.5%) diffraction efficiency o f PDLC2 when it is in its “ off” setting.
Table 13.1 also shows our experimental electrical SNR measurements for both PDL
settings. In order to take optical power measurements, the amplitude modulation o f the
optical beam was removed. A CW sinusoidal signal at 500 MHz was used to drive the
AOM, and the +1 diffracted order travels through the PDL. The measurements were
obtained by measuring the optical power at the designed output port o f the PDL as shown
in Fig. 13.2. For each PDL setting, the signal measurements were obtained by spatially
blocking the leakage noise coming from the undesired path. Similarly, leakage noise
measurements were obtained by blocking the signal traveling through the desired path.
For the theoretically expected results, diffraction and scattering effects on the noise
beams are not taken into account. This explains the 8.4 dB difference between the
estimated and measured leakage of the non-delay case, which is due to the higher
diffraction loss that the noise beam suffers through the delay path. When PDLC 1 is “ on” ,
the noise beam acquires a higher degree of spatial divergence compared to the signal
beam. This lowers the level o f the photodetected noise at the output of the PDL. Hence, a
higher experimental SNR for the PDL. Note that in the delay case, the leakage noise does
not travel a very long distance before detection, and the divergence effects are less
prominent. Hence, the SNR is closer to the theoretical calculations.
263
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 13.1: Electrical signal-to-Ieakage noise ratio of the proposed PDLC based PDL.
Electrical Signal-to-Leakage Noise (dB)
PDL Setting
Measured
Expected
Delay
66.58
65.40
Non-delay
39.20
30.80
From Table 13.1 it is clear that the optical leakage noise values obtained via our
present PDLC devices are not adequate for applications that require optical leakage noise
levels of < -30 dB. These leakage noise numbers are mainly limited because of the
characteristics o f the PDLC devices presently available for our experiment. An obvious
solution to this limitation, is to use PDLC devices, with higher on/off ratios. Sutherland
et. al. have suggested that the on/off ratios can be improved by appropriate PDLC device
engineering. This can be accomplished by adjusting PDLC parameters such as droplet
size, volume fraction o f droplets, and device thickness by using material additives to the
initial polymer syrup [16].
The other technique that we propose is to use active noise filtering in this PDLCbased PDL. These noise filters can be realized by using an additional PDLC device per
PDLC switched path. Fig. 13.4 shows this novel PDL design for the PDLC device
settings for the non-delay operation. For the non-delay setting, noise travels through the
delay path. Thus, a PDLC device (PDLC4) can be used in its “ off” setting to deflect the
noise out o f the path. On the other hand, the signal light travels through the non-delay
path and PDLC3 should be set “ on” so that the signal light travels unaffected. In the
delay path setting, all PDLC devices are set in their other state (e.g., PDLC3: “ off” ;
264
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PDLC4: “ on” ). Using the above active noise filter technique for our PDLC-based PDL,
and a 95% diffraction efficiency for the “ off” state o f the devices, and a 99% zero order
efficiency for the “ on” state of the devices (note: these device numbers are in agreement
with the ones reported in our experiment), the PDL electrical SNR is estimated to be > 90
dB for both PDL settings. This is an improvement of over 20 dB for the PDLC PDL not
using active noise reduction. Fig. 13.4 shows how this SNR number was calculated based
on an optical flow diagram. Putting additional PDLC devices in the optical paths will
increase the optical loss to ~2 dB per PDL bit, provided that AR coated PDLC devices
with high diffraction efficiencies (e.g., > 95%) are used. The use of additional PDLC
devices as active noise filters does increase the component count. Nevertheless, increase
in PDL cost is minimal as PDLC device cost is expected to be low compared to the optics
used in the PDL. Specifically, PDLC cost are low due to the present large scale batch
fabrication methods o f liquid crystal technology, and the rather simple, single step PDLC
fabrication process based on the in situ formation of the hologram by the PIPS fabrication
process.
Noise Filter ,Block *
! PDLC4
Noise: 0.01 x ?•
Signal In
Optical
Power is
P
PDLC1
“on”
R e m a in in a n o ise : 0 .0 5 X 0.01 X P
PDLC3 “on'
Noise
Block
Signal:
0 .9 9 x P
J|
Signal: 0 .9 9 x 0 .9 9 x P
Optical Signal Out &
Leakage Noise
Figure 13.4: The proposed POLC-based PDL, set for no-delay, with active noise niters to improve the
electrical SNR to > 90 dB. Optical signal out = 0.95 x 0.99 x 0.99 x P, optical leakage noise = 0.05 x 0.05
x 0.01 x P (“ x” stands for multiplication).
265
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.5. Polarization Selective Hologram-based Photonic Delay Lines
As mentioned earlier, PSHs are optical elements that have a different phase
function depending on the state of polarization o f the incident light. Fixed PSHs have
been recorded in organic dyes for applications in holographic interferometry [18], in
dichromated gelatins for substrate mode holographic interconnects [19], and in
photorefractive crystals for free space optical interconnects [20]. Birefringent computer
generated holograms have also been demonstrated for the implementation o f 2x2 optical
switches for free-space multistage interconnection networks [21]. Other polarization
dependent beam routing elements can be implemented using NLC technology such as
birefringent-mode NLC devices previously used for active optical beam focusing [22],
and combining [23].
The following paragraphs discuss the details related to the theory o f the PSHbased switched PDL and its experimental verification [24, 25]. A PSH device can be
fabricated using any of the technologies mentioned in the previous paragraph. A proof of
concept experiment for the characterization o f such a PDL is also presented. The PSHs
used in our experiment are based on holographic polymer dispersed liquid crystals
(PDLCs). In the proposed PSH-based PDL, we do not electrically modulate the PDLC
diffractive power to form a polarization switch. Instead, the PDLC devices are used as
fixed PSHs to form beam routers and combiners. Fig. 13.5 shows the proposed PDL
based on polarization switching devices (e.g., FLC) and PSHs. A feature of our proposed
PDL is that it uses 2-D liquid crystal polarization switching arrays and single large area
PSHs. This makes the physical PDL implementation a somewhat simpler assembly
operation. Note that current liquid crystal technology can give high pixel count, low cost,
2-D flat panel switching arrays. In addition, the large area, single pixel polarization
266
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dependent optical routers/combiners required in our PDL can be fabricated using low
cost, batch production, standard holographic techniques.
Delay Path
1
2
O ut
N
PSH
PSH
Figure 13.5: A single bit o f the proposed photonic delay line based on polarization switching devices and
polarization selective holograms. (Dashed lines: delay path; Solid lines: non-delay path; PS: polarization
switch; PSH: polarization selective hologram; M: mirror; L: lens).
As mentioned earlier thick gratings are in general polarization dependent [12]. In
the case of PDLC gratings the polarization dependence is even more evident due to the
liquid crystal birefringence. This is possible when the liquid crystal is infused into the
photopolymer pores with the molecular director aligned such that with no electric field
applied across the PDLC device, the PDLC acts as a grating for one polarization, and as a
glass plate for the other orthogonal polarization. This can be obtained if the liquid crystal
molecular directors are aligned parallel to the fringe direction (Fig. 13.6). Thus, light with
polarization parallel to the fringe direction (i.e., horizontal) will “ see” an index
modulation created by the index of refraction of the polymer host (Wp) and the
267
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ya
y
PSH
'out.
First Order
M olecular
Director
Zero Order
(Leakage)
(a)
PDLC
First Order
(Leakage)
■out,||
Molecular
Director
▼x
Zero Order
(b)
Figure 13.6: The PDLC as a polarization selective hologram (a) horizontally polarized input “ sees” the
grating and is deflected into the first order, (b) vertically polarized input does not “ see” the grating and
passes through unaffected.
268
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
extraordinary index o f refraction (n j of the crystal (Fig. 13.6a). On the other hand,
vertically polarized light does not “ see” any index modulation. This is because vertically
polarized light “ sees” the ordinary index of refraction o f the liquid crystal which almost
matches the index of refraction of the polymer host (Fig. 13.6b). In our experiments, we
used two PDLC based gratings fabricated by Foster-Miller, Inc. The polymer host was
Polaroid DMP-128 and the liquid crystal was E7. With no electric field applied and for
incident light at Bragg angle (-30°) and polarization along the grating fringe direction,
the PDLC devices diffract the light at an angle o f 60° with respect to the zero order beam
(for our PDLC devices this happens for horizontal polarized light) (see Fig. 13.6). On the
other hand, if the incident polarization is perpendicular to the grating fringe direction
(i.e., vertical polarization), the light passes through the PDLC unaffected (Fig. 13.6).
Note that due to the small mismatch between «p and n0, there is some small vertically
polarized leakage in the first order. There is also a horizontally polarized leakage in the
zero order beam, due to the non-optimized index difference between the polymer host and
the extraordinary index o f refraction of the liquid crystal.
13.6. Experimental Demonstration of the PSH-based Photonic Delay Line
Fig. 13.7 depicts the experimental set-up for our proposed PSH-based PDL. A
Lasertron model QLINK1-051 microwave fiber-optic transmitter (X.=1310 nm) is used as
the laser source. The modulated light is coupled into the PDL via a SM-fiber
connectorized to a GRIN lens. A mechanical fiber-optic polarization controller is used to
make the input polarization horizontal. A high extinction ratio (10,000:1) polarizer is also
269
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
used to suppress any polarization leakage at the other orthogonal state. A FLC device is
used as a polarization switch. This switch acts as an electrically controlled half wave
plate. When the switch is set “ off” , the SOP of the light does not change. As mentioned
earlier, the PDLC device diffracts the light when the incident polarization is horizontal
since the light “ sees” the grating. The first diffracted order is at 60° with respect to the
zero order beam. After reflection from the mirrors, the light hits the second PDLC device
with an angle of 60° with respect to the non-delay path. The light is then diffracted
towards the output of the PDL. On the other hand, when the FLC polarization switch is
set “ on” , the polarization of the light is rotated by 90°. Hence, the light does not “ see”
the grating in PDLC1 device and passes straight through towards the PDLC2 device that
also acts as a flat glass plate for the vertically polarized light. Thus, the input light
propagates towards the output port o f the PDL.
RF In
D elay P ath'
Mechanical
Polarization
Controller
PDLC1
sm fI
PDLC2
GRIN
O ut
60°
Non-Delay
Path
Polarize FLC
Detector
O Vertical Polarization | Horizontal Polarization
Figure 13.7: The experimental set-up of the proposed PDL using a FLC device as a polarization switch and
PDLC devices as polarization selective holograms. (Dashed lines: delay path; Solid lines: non-delay path;
SMF: Single mode fiber; RF: radio-frequency).
270
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.7. Leakage Noise Measurements and System Improvements
Optical SNR measurements were obtained by independently measuring the signal
and leakage noise o f each PDL setting at the output o f the PDL bit. For example, for the
delay setting, the noise was measured by physically blocking the signal traveling through
the delay path, while the signal was measured by physically blocking the noise traveling
through the non-delay path. The optical signal was detected at the output of the PDL bit
using a large area detector (1 mm in diameter) and a power meter. Table 13.2 shows the
optical SNR for the non-delay and delay setting o f our PDL.
Table 13.2: Optical SNR measurements for the two PDL settings (without noise reduction).
PDL Setting
Optical SNR (dB)
Non-delay
22.0
Delay
16.7
These optical SNR numbers are rather limited. There are two reasons for this
limited PDL performance. The first reason is that PDLC devices are not 100% efficient.
For example, the PDLC1 device diffracts some part o f the vertically polarized signal into
the delay path, and some o f the horizontally polarized light into the non-delay path. This
ends up to be part o f the leakage noise at the output o f our PDL. We will be referring to
this noise as PDLC-based leakage noise. The second source o f noise in our system is the
FLC device and its limited on/off performance. We have seen that today’s FLC devices
271
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
do not fully rotate the incident polarization by 90° when they are set “ on” [26]. This
means that when the FLC device is set “ on” , unwanted horizontal polarization leaks
through the delay path and contributes to the output leakage noise. Moreover, when the
FLC device is set “ off” it does degrade the SOP of the incident polarization and thus
there is a vertical component which
leaks through the non-delay path. We will be
referring to this noise as FLC switch-based leakage noise. Optical SNR numbers of > 30
dB are required for most PDL applications. Thus, a way o f improving the system
performance is necessary. Two different ways of improving the system performance were
tested and are discussed in the following paragraphs.
13.7.1. Passive Leakage Noise Filter
As mentioned earlier, one source of the leakage noise is the PDLC devices. This
limitation is related with not getting high enough diffraction efficiency from the PDLC
devices for horizontal polarized light, as well as the non-zero diffraction efficiency for
vertically polarized light. This is due to the non-optimized index modulation. This is a
fabrication process limitation, and as mentioned earkier, it can be improved by careful
fabrication techniques. Sutherland et. al. have suggested that the diffraction efficiency
and leakage noise can be improved by appropriate PDLC device engineering, by
adjusting PDLC parameters such as droplet size, volume fraction of droplets, and device
thickness by using material additives to the initial polymer syrup [16].
For our PDLC1 device and for vertically polarized input incident light onto the
device, all the light is expected to pass through the device with no diffraction in the first
272
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
order beam. Nevertheless, a 3% diffraction is observed into the first order beam, which
translates into PDLC-based leakage noise in our PDL. Note that this leakage noise is
vertically polarized. Thus, the use of a horizontal polarizer can block the leakage
traveling through the delay path. In a similar way, for horizontally polarized light incident
onto PDLC1, a 100% diffraction efficiency in the first order is expected. In this case,
89% of the input light is diffracted and 11% stays in the zero order beam. This rather
large leakage noise travels through the non-delay path and significantly affects the PDL
SNR performance. Note that the leakage noise is horizontal and thus a vertical polarizer
would block the leakage. Note also that in both cases, the polarization of the signal
traveling through the desired path is parallel to the axis of the polarizer placed in the path
and thus will not be affected by the polarizers. Thus, a vertical and a horizontal polarizer
were positioned along the non-delay and the delay paths, respectively. Table 13.3 shows
the optical SNR measurements obtained using this passive noise filter. From Table 13.3
we can conclude that the use of the passive noise filter suppresses the noise leakage in the
delay setting by ~ 10 dB. In the non-delay setting we do not observe such an
improvement basically because the optical SNR number is already > 20 dB. Based on the
measured PDLC diffraction efficiency on the first order and the bypass efficiency on the
zero order beam, the theoretically expected optical SNR, without taking into account the
FLC switch-based leakage noise, is 26 dB and 17 dB for the non-delay and the delay
settings, respectively. These numbers are close to our experimentally obtained ones. The
passive noise filter gives optical SNR numbers of 23 dB and 25 dB respectively. This is
still not adequate for advanced phased array antenna applications or other PDL
applications that require optical SNRs of > 30 dB.
273
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 13.3: Optical SNR for all the different leakage noise filtering approaches. The SNR without any
noise filter is also shown for comparison.
Optical SNR (dB)
PDL
Without Noise
Passive Noise
Active Noise
With Passive and
Setting
Filter
Filter
Filter
Active Noise Filter
Non-delay
22.0
23.3
33.8
44.0
Delay
16.7
25.1
17.0
48.0
13.7.2. Active Noise Filter
The other source o f leakage noise in our PDL is the FLC polarization switch. The
FLC switch does not fully rotate the polarization of the incident light when it is set “ on”
[26]. This causes the horizontally polarized leakage noise to be deflected by the PDLC1
device in the delay path. Moreover, when the FLC switch is set “ o ff’, the SOP o f the
light in not maintained at the high input polarization extinction ratio. This vertically
polarized FLC-based leakage does not “ see” the grating on the PDLC1 device and
travels through the non-delay path unaffected. In both cases, signal and FLC-switch based
leakage noise are of orthogonal polarizations at the output o f the PDL. Thus, by using an
additional FLC device and a vertical polarizer at the output o f the PDL, we can suppress
this FLC switch-based leakage noise. For example for the delay setting, horizontally
polarized signal travels through the delay path, and vertically polarized FLC switch-based
leakage noise travels through the non-delay path. The output FLC is set “ on” and thus
rotates the polarization of both the signal and the leakage noise. The vertical polarizer is
274
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
then used to block the leakage noise and pass through the signal. In the non-delay setting,
FLC2 is set “ o ff’ and thus the vertically polarized signal from the non-delay path
remains unaffected and passes through the polarizer, while the horizontally polarized
FLC switch-based leakage-noise is blocked. Note that the two FLC devices operate in
opposite modes, i.e., when FLC1 is “ on” , FLC2 is “ o ff’ and vice versa. This is because
the two states of the FLC devices do not perform equally well [26]. The SNR
measurements of the PDL system using the active noise filter are shown in Table 13.3.
Using the active noise filter we improved the non-delay setting by >10 dB, but the delay
setting SNR remains at low levels (e.g., < 20 dB). This limited improvement for the delay
setting is due to the vertically polarized PDLC-based leakage noise that eventually passes
through the active noise filter and contributes to a low SNR. The FLC2 switch also
contributes to this noise since it is set in its “ on” state and there is some FLC switchbased leakage.
13.7.3. Combination o f the Active and Passive Noise Filter
From the PDL optical SNR results obtained from the passive and active noise
filters, we see that each approach improves the SNR for only one of the two settings. If
we were to use both o f the filtering methods simultaneously, we would obtain higher
optical SNR numbers for both PDL settings. The experimental set-up showing the
combination o f the two noise filters is depicted in Fig. 13.8. The passive noise filter
suppresses the PDLC-based leakage noise, while the active noise filter suppresses the
FLC switch-based leakage noise. Optical SNR measurements o f > 40 dB were obtained
275
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for both PDL settings and are shown in Table 13.3. These optical SNR numbers are
highly desirable for PDL applications such as phased array antennas.
Delay Path
M/ ;
— ► @1300 nm
\
y\
x
M
0 Vertical Polarization
1 Horizontal Polarization
H orizontal
Mechanical
Polarization
Controller
Active Noise
Polarizer
ra te r
\
P a s s iv e N o i s e
_
PDLC1
F itte r
PDLC2
GRIN
P o la riz er
60 °
FLC
Vertical
Polarizer
jr
Detector
Ll
Figure 13.8: The PDL experimental set-up showing the passive and active noise filters. P: polarizer, GRIN:
gradient index lens; SMF: single mode fiber; L: lenses; LD: semiconductor laser.
13.7.4. “ Orthogonal Drive” PDLC Device Configuration
The current limitation in our PDL system is mainly the low diffraction efficiency
o f our PDLC devices in the first order (e.g., -89%). Note also that in any of the two
PDL settings, both o f the PDLC devices either deflect the signal or bypass it. The
difference between the bypass efficiency in the zero-order beam
(—97%) and the
diffraction efficiency (-89% ) in the first-order beam also leads to an unbalanced SNR
performance for the two PDL settings.
276
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
We again propose the use of an “ orthogonal drive” [17] PDLC device
configuration to compensate for that unbalanced performance. For our case, this
“ orthogonal drive” PDLC device configuration is obtained by setting one of the PDLC
devices to diffract in the first order and the other device to bypass in the zero order for
each o f the PDL settings. This can be accomplished by fabricating the two PDLC devices
such that one device diffracts the vertical polarization and the other device diffracts the
horizontal polarization. Nevertheless, this approach will lead to increased expense and
fabrication time, since two different sets of PDLC devices will have to be fabricated. We
choose to use a simpler approach, where we place a half wave plate in each of the PDL
paths. Thus, the polarization of the light in any of the paths is rotated before it reaches the
PDLC2 device. In this case, a horizontal polarization signal coming from PDLC1 device
through the delay path is rotated to vertically polarized light, and does not “ see” the
grating in the PDLC2 device. Thus, it passes through the PDLC2 device unaffected.
Similarly, vertically polarized light coming from PDLC1 device through the non-delay
path is rotated to horizontally polarized light, and is deflected by the PDLC2 device.
Optical SNR measurements were obtained for this PDL that makes use o f the
“ orthogonal drive” configuration and are shown in Table 13.4. Note that we actually use
this orthogonal drive configuration twice in our system; once for the PDLC devices and
once for the FLC devices. From Table 13.4, we can conclude that the “ orthogonal drive”
configuration improves the overall SNR performance for our PDL even when no noise
filter technique is used. Note that the SNR numbers o f our PDL are still not high enough
when only one of the noise filter techniques is used. This is mainly due to the low
diffraction efficiency into the first order beam o f the PDLC devices. Note that if the
performance of the PDLC devices is improved such that they can give 99% diffraction in
the first order for the horizontally polarized light and a 99% bypass efficiency in the zero
277
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
order for the vertically polarized light, then the theoretical expected SNR would approach
40 dB with only one o f the noise filters.
Table 13.4: Optical SNR for the PDL with the orthogonal drive configuration and for all the different
leakage noise filter approaches. The SNR without any noise filter is also shown for comparison.
Optical SNR (dB)
PDL
Without Noise
Passive Noise
Active Noise
With Passive and
Setting
Filter
Filter
Filter
Active Noise Filter
Non-delay
21.3
22.7
26.0
46.3
Delay
22.7
26.4
24.9
48.0
13.8. Time Delay Measurements
Time delay measurements were also obtained for the single bit PSH-based PDL.
Fig. 13.9 shows oscilloscope traces of the non-delayed and the delayed signal (bottom
traces). The top trace represents the reference signal from the signal generator. A fiber
pigtailed fast photodetector (New Focus, Model: 1414-50) was used to detect the
modulated optical signal. The fiber used was a multi-mode fiber (50 pm core diameter)
connectorized to a GRIN lens. Fig. 13.9(a) shows the non-delayed signal, with a relative
delay from the reference signal of 36.28 ns. The time markers have been positioned at the
on-set of the traces. As on-set time we define the time when the pulse gets to 10% of its
maximum value. Fig. 13.9(b) shows the delayed signal, with a relative delay from the
278
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reference signal o f 40.34 ns. The relative time delay between the two photodetected
signals is the time delay obtained from our PDL bit and is calculated to be 40.34 ns 36.28 ns = 4.08 ns. The expected time delay can be estimated from the optical path length
difference between the two paths. The PDLC-to-PDLC distance for the non-delay path is
29 cm. The PDLC-to-PDLC distance for the delay path is 152 cm. Thus the expected
time delay can be found from the following equation
Ax
(Delay Path)- (Non-delay Path)
c
(153x10 —29x10 )m „ 1A
=---------------;----------- -— = 4.10ns
3x10 m/s
(a)
(13.1)
(b)
Figure 13.9: Oscilloscope traces showing (a) the non-delayed photodetected signal and (b) the delayed
photodetected signal. (Top traces: reference signal from the oscilloscope; Bottom traces: the photodetected
output signal).
279
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The 0.02 ns difference between the expected and the measured time delay is due
to measurement errors in the path lengths as well as the tolerance in the position o f the
time markers on the oscilloscope screen.
13.9. Insertion Loss o f the PSH-based Photonic Delay Line
An important issue of the PDL is the system insertion loss. Recently, we have
discussed insertion loss issues for our previously proposed PDL systems due to the bulk
optical components in PDL structures as well as the effect of the FLC devices in the loss
[27]. The main contributor to the insertion loss is the FLC devices that are currently
limited to an average o f 85% transmission efficiency (or 15% optical insertion loss) for
operation at 1300 nm. The new element in the PSH-based PDL is the PDLC devices. The
device insertion losses are due to reflection, absorption and scattering. Reflection losses
can be minimized using anti-reflection coatings on the glass substrates. Absorption loss is
rather minimal and the primary contributor to the PDLC loss is scattering [16]. Studies of
the liquid crystal droplet size have shown that scattering can be reduced using small
droplets (e.g., droplet diameter 0.04 pm) [16].
Our PDLC1 and PDLC2 devices had insertion loss of 0.8 dB (or 17%) and 0.7 dB
(or 15%) respectively. The overall insertion loss of the PDL when both noise filters are
used was measured at 2.9 dB. The use of the active noise filter contributes a 0.7 dB to the
total PDL insertion loss, while the passive noise filter contributes only a 0.07 dB to the
PDL insertion loss. However, the great improvement obtained for the optical SNR when
using the noise filters justifies the use of the noise filters eventhough the PDL shows an
280
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increased insertion loss. Nevertheless, as new FLC materials are developed and studied,
reduced insertion loss numbers for the FLC switches are expected. Improved insertion
loss is also expected from the PDLC devices as the PDLC technology matures.
13.10. A Compact Photonic Delay Line Architecture based on Polarization Selective
Holograms
13.10.1. Experimental Set-up
The overall size of the PSH-based PDL bit can be reduced if a reflective design is
used. Fig. 13.10 shows such a reflective design. A mirror is positioned in each of the two
PDL paths. These mirrors reflect back the light onto the PDLC device through exactly the
same path. A QWP is also used in each path so that the polarization of the light is rotated
by 90° after passing through the QWP twice. Thus light is directed towards the output of
the PDL bit. This reflective design uses half the optical path compared with the
transmissive design discussed earlier. It also uses fewer optical components since it
reuses the lenses and the PSH (in our case the PDLC device).
The compact reflective PSH-based PDL works as follows. When PS1 is set “ on” ,
it rotates the incident horizontal polarization by 90°. This vertically polarized light passes
through the PSH unaffected. After going through QWP1 twice, it is rotated back to
horizontally polarized light and this time “ sees” the grating in the PSH. Thus, it is
diffracted towards the output port of the PDL. When the PS1 is set in its “ o ff’ state it
leaves the incident polarization unaffected. The horizontally polarized light is then
281
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
diffracted in the first order beam and after passing through the QWP2 twice, it changes to
vertically polarized light. This vertically polarized light does not “ see” the grating in the
PSH and thus passes through it unaffected towards the output o f the PDL. Note that the
“ orthogonal drive” configuration comes naturally in the reflective PDL design because it
is needed to separate the output from the input port.
O diy Path
QWP
PSI
QWP
Active Noise
Filler
Out
Figure 13.10: The compact reflective photonic delay line architecture based on polarization selective
holograms.
13.10.2. Leakage Noise Measurements
A limitation of this reflective PSH-based PDL architecture is that no passive noise filter
can be used since the polarization o f the signal traveling in any of the two paths exists in
both orthogonal polarizations. This limits the optical SNR performance for our PDL
using the PDLC devices available in our lab. Nevertheless as mentioned earlier, higher
diffraction efficiency PDLC devices can be obtained and thus better SNR numbers can be
282
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reached. However, the active noise filter can be used to further suppress the FLC1 switchbased leakage noise.
Optical SNR measurements were obtained for this compact reflective PDLCbased PDL and are shown in Table 13.5. Two sets of measurements were obtained, one
without any noise filter and one with the active noise filter. Optical SNRs > 27 dB are
obtained for both settings. This close to 30 dB numbers can potentially exceed the 30 dB
level by using improved PDLC devices with higher diffraction efficiency.
Table 13.5: Optical SNR for the compact reflective PDL with and without the active noise filter.
Optical SNR
PDL
Without Noise
Active Noise
Setting
Filter
Filter
Non-delay
19.4
28.7
Delay
20.1
27.4
13.10.3. Time Delay Measurements
Time delay measurements were also obtained for the compact reflective single bit
PSH-based PDL. Fig. 13.11 shows oscilloscope traces of the non-delayed and the delayed
signal (bottom traces). The same technique as in the case of the transmissive PSH-based
PDL was used. Fig. 13.11(a) shows the non-delayed signal, with a relative delay from the
reference signal o f 34.28 ns. Fig. 13.11(b) shows the delayed signal, with a relative delay
283
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from the reference signal of 38.36 ns. The relative time delay between the two
photodetected signals is the time delay obtained from our PDL bit and is calculated to be
38.36 ns - 34.28 ns = 4.06 ns. The expected time delay is again 4.10 ns.
(a)
(b)
Figure 13.11: Oscilloscope traces showing (a) the non-delayed photodetected signal and (b) the delayed
photodetected signal for the compact reflective PDLC-based PDL. (Top traces: reference signal from the
oscilloscope; Bottom traces: the photodetected output signal).
13.11. Alternative Polarization Selective Hologram Designs
A unique feature of our PSH-based PDL is that it uses the mature, low cost liquid
crystal technology for the implementation of the polarization switches and standard
holographic techniques for the PSH devices. In our experimental demonstration we used
PDLC devices as PSH. An alternative PSH technique is based on computer generated
holograms (CGH) [28]. The maximum possible diffraction efficiency from a CGH
depends on the accuracy with which the theoretically continuous phase values are
actually fabricated. Theoretically, diffraction efficiencies o f 98.7% can be obtained for a
284
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16 level phase CGH, and 99.7% for a 32-level phase CGH [28]. Standard lithographic
techniques used for microelectronic fabrication can be used to create an accurate
multilevel phase profile with high spatial resolution. These CGHs are generally
polarization independent but a technique of making polarization dependent CGH has
been reported and demonstrated using a birefringent substrate to make a four level
birefringent CGH with a diffraction efficiency of 60% [28]. 16 or 32 level phase
birefringent CGHs that show higher diffraction efficiencies can also be used to implement
a PSH-based PDL.
As mentioned earlier thick gratings are polarization sensitive [12]. Thus, thick
gratings can also be used as PSH. For a transmission type o f phase volume hologram, it
has been shown that there exists a condition where the diffraction efficiency of one
polarization (s or p) is 100 % and the diffraction efficiency of the orthogonal polarization
(p or j) is 0 % [29]. It has been reported that a polarization selective hologram has been
recorded in DMP-128 photopolymer with a normalized diffraction efficiency of 99 % for
s-polarized light and 1 % for /^-polarized light [30]. The diffraction efficiency of the
volume holograms depends on the maximum index modulation that can be achieved for a
material, the wavelength, the grating period and the thickness. Usually the thickness is
material dependent and the wavelength is set by the system requirements and use. Thus,
the two factors important for a PSH are the index modulation that can be controlled by
the recording condition, and the possible Bragg angle and diffraction angles.
Another technique for implementing polarization selective beam routing elements
is based on the mature nematic liquid crystal technology. Due to the capability of NLC
materials to be controlled by low electrical voltages, reconfigurable beam routing
elements can also be formed. Liquid crystal based phase gratings for high efficiency light
valves has been proposed as early as 1979 [31]. Programmable liquid crystal devices for
285
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
variable focal-length lenses [22, 23] and for adaptive optical interconnects [32] and
alignment [33] applications have been proposed and demonstrated. For our PDL
application these birefringent-mode NLC devices can be set such that the liquid crystal
nematic director is along one of the two orthogonal polarizations. Hence, the index of
refraction has a specific phase profile for one o f the polarizations and a uniform index of
refraction (no index modulation) for the other polarization. Thus, only one o f the incident
linear polarization will be deflected to a predefined direction. These birefringent-mode
NLC-based beam routing elements can be either active or passive.
Figure 13.12 shows a birefringent-mode NLC device that acts as a polarization
dependent grating that strongly diffracts light in the first order for one polarization and
does not diffract the light for the other orthogonal polarization. The liquid crystal
molecules are oriented in layers such that horizontally polarized light “ sees” a maximum
index modulation ne - nQ while vertically polarized light “ sees” a uniform index o f
refraction (nQ). Transparent ITO electrodes are properly spaced to give the required
grating spacing for maximum diffraction efficiency.
An alternative birefringent-mode NLC based routing element will be one based on
a ramp like index o f refraction profile and hence an induced ramp like optical phase shift
to the incident beam. This approach can make use of a thin film-resistor network [34] on
the device substrate layer to control the voltages o f the independent electrodes by the use
o f only one external driver. This technique gives a near continuous index pertubation to
match the prism like phase pertubation required for the beam deflection [34], Figure
13.13 shows a possible configuration for the this birefringent-mode NLC device using
on-chip resistor-based control electronics. A gradual tilt of the liquid crystal molecules
gives the required index o f refraction profile to the device and thus for horizontal light the
devices act as a beam deflector.
286
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Eh
Ei
Horizontally
Polarized
Vertically
Polarized
Cover
NLC —;
Molecules —I
NLC
Molecules
Z
Electrod
Substrate
Light “sees
only ii,,
Light “sees” index
modulation n -
(a)
(b)
Figure 13.12: A PSH via a birefringent mode nematic liquid crystal based device used as a polarization
dependent diffraction grating.
Ein
Vertically
Polarized
NLC
Molecules
.Cover Glass
m
——
^ ^
--------------- \
Light “sees”
only n„
Horizontally
Polarized
NLC
Molecules
Resistor Bias—:
Network
Substrate
jfTj
Light “sees” index
distribution from n„ to n,
Eout
Figure 13.13: Top view of a PSH formed with a thin-film-resistor network based NLC deflector.
287
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13.12. Conclusion
In conclusion, we have proposed and demonstrated for the first time a photonic
delay line based on electrically switched gratings. For our experimental verification, we
used switched gratings based on PDLCs. Single-bit single-channel PDL SNR and time
delay measurements were presented, and were in agreement with the theoretically
expected results. The presently limited PDLC device on/off ratios limited our PDL
performance to at best a 65 dB electrical SNR. A PDLC-based, PDL module is proposed,
that uses two additional PDLC devices as active noise filters that can lead to much
higher, e.g., 90 dB electrical SNR for the PDL. Future work relates to improving our
present PDLC-based PDL performance.
We have also proposed and experimentally demonstrated a photonic delay line
based on FLC devices for polarization switching and PDLC devices as polarization
selective optical path routing components. Extensive investigation o f the leakage noise in
the system was performed, and two different leakage noise filters were investigated for
improving the SNR numbers. Improved SNR performance (> 45 dB) was obtained by
combining both noise filters and using the “ orthogonal drive” configuration. Time delay
measurements were also performed for our single bit single channel PDL. An alternative
reflective architecture was proposed. This reflective architecture uses fewer number of
optical components. In addition, the propagation of light twice through the two PDL
paths also gives a reduced size for our reflective PDL architecture. PSH based on
birefringent CGHs, phase volume holograms, and birefringent-mode NLC devices for
PDL applications were also proposed.
288
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF REFERENCES
Chapter 1
1. N. A. Riza, and N. Madamopoulos, “High signal-to-noise ratio birefringence
compensated optical delay line based on a noise reduction scheme,” Optics Letters, Vol.
20, No. 22, pp. 2351-2353, 1995.
2. World News in Laser Focus World, Vol. 32, No. 9, pp. 28-32, 1996.
3. N. A. Riza, and N. Madamopoulos, “Microwave band demonstration of a reflective
geometry fiber and free-space binary photonic delay line,” IEEE Microwave & Guided
Wave Letters, Vol. 7, No. 4, pp. 103-105, 1997.
4. N. A. Riza, N. Madamopoulos, “Characterization of a FLC-based time delay unit for
phased array antennas,” IEEE/OSA Journal o f Lightwave Technology, Vol. 15, No. 7,
pp. 1088-1094, 1997.
5. N. A. Riza, and N. Madamopoulos, “Phased array antenna maximum compression
reversible photonic beamformer using ternary designs and multiple wavelengths,”
Applied Optics, Vol. 36, No. 5, pp. 983-996, 1997.
6. N. Madamopoulos, and N. A. Riza, “Adaptable-delay balanced-loss binary photonic
delay line architectures using polarization switching,” Optics Communications, Vol.
152, pp. 135-143, 1998.
289
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7. N. Madamopoulos, and N. A. Riza, “Directly modulated semiconductor laser fed
photonic delay line using ferroelectric liquid crystals,” Applied Optics, Vol. 37, No. 8,
pp. 1407-1416, 1998.
8. N. Madamopoulos and N. A. Riza, “Switched photonic delay line for phased array
antenna control using externally modulated microwave fiber-optic link,” in Optical
Technology fo r Microwave Applications VIII, A. Goutzoulis, Ed., SPIE Proc. 3160,
pp. 45-54, August 1997.
9. N. A. Riza and N. Madamopoulos, “All-fiber connectorized fiber-optic delay module
using 3-D polarization optics,” LEOS’97 Conference Proc., 10th Annual Meeting
(TEEE-lasers and Electro-Optics Society, San Francisco, 1997), Vol. 2, ThW2, pp.
472-473.
10 N. Madamopoulos and N. A. Riza, “Reversible fiber-optic switched delay module
using GRIN lens fiber-optic collimators and ferroelectric liquid crystals,” LEOS ‘98
Conference Proc., 11th Annual Meeting (lEEE-lasers and Electro-Optics Society,
Orlando, 1998), ThFF4.
11. N. A. Riza and N. Madamopoulos, “Photonic delay line using electrically switched
gratings in polymer dispersed liquid crystals,” Optical Engineering, Vol. 37, No. 11,
pp. 3061-3065, 1998.
12.N. Madamopoulos, and N. A. Riza, “Polarization Selective Hologram-based Photonic
Delay Lines,” Optics Communications, Vol. _, pp.
1998.
290
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
1. T. C. Cheston and J. Frank, “Phased array radar antennas,” in Radar Handbook, 2nd
Ed., Chapter 7, M. I. Scholnik, Ed., (New York: McGraw-Hill, 1990).
2. N. Fourikis, Phased Array Based Systems and Applications, (New York: John Wiley
and Sons, 1997).
3. L. Stark, “Theory of phased arrays,” Proc. IEEE, Vol. 62, No. 12, pp. 1661, 1974.
Chapter 3
1. N. A. Riza, “Liquid crystal-based optical control of phased array antennas,” IEEEJOSA
Journal o f Lightwave Technology, Vol. 10, No. 12, pp. 1974-1984, 1992.
2. D. Dolfi, P. Joffre, J. Antoine, J.-P. Huignard, D. Philippet, P. Granger,
“Experimental demonstration of a phased-array antenna optically controlled with phase
and time technology,” Applied Optics, Vol. 35, No. 26, pp. 5293-5300, 1996.
3. N. A. Riza and N. Madamopoulos, “Characterization of a FLC-based time delay unit
for phased array antennas,” IEEE/OSA Journal o f Lightwave Technology, Vol. 15,
No. 7, pp. 1088-1094, 1997.
4. N. A. Riza, “Polarization based fiber optic delay lines,” in Optical Technology fo r
Microwave Applications VII, A. P. Goutzoulis, SPIE Proc. 2560, 120-129, 1995.
5. N. A. Riza, “Advanced novel photonic instrumentation for adaptive and interferometric
astronomy,” in Space Telescopes and Instruments IV, P. Y. Bely; J. B. Breckinridge;
Ed., SPIE Proc. 2807, pp. 335-342, 1996.
291
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6. N. A. Riza, “Multichannel variable optical control systems for large coherent optical
arrays,” in Laser Radar Technology and Applications, G. W. Kamerman, Ed., SPIE
Proc., Vol. 2748, pp. 333-339, 1996.
7. J. Fu, M. Schamschula, and H. J. Caulfield, “Modular solid optic tiine delay system,”
Optics Communication, Vol. 121, No. 1-3, pp. 8-12, 1S95.
8. N. A. Riza, “Photonically controlled ultrasonic arrays: scenarios and systems,” IEEE
Ultrasonics Symposium Proceedings, M. Levy; S. C. Schneider; B. R. McAvoy; Eds.,
Vol. 2, pp. 1545-1550, 1996.
9. N. A. Riza, Selected Papers on Photonic Control Systems fo r Phased Array Antennas,
MS 136 of SPIE Milestone Series (SPIE Press, Bellingham, Wash., 1998).
10. A. M. Levine, “Use of fiber optic frequency and phase determining elements in radar,”
Proceedings o f the 33rdAnnual Symposium on frequency Control, pp. 436-443, 1979.
11. W. Ng, A. A. Waltson, G. L. Tangonan, J. J. Lee, I. L. Newberg, and N. Bemstein,
“The first demonstration of an optically steered microwave antenna using true-timedelay,” IEEE/OSA Journal o f Lightwave Technology, Vol. 9, No. 9, pp. 1124-1131,
1991.
12.R. A. Soref, “Programmable time-delay devices,” Applied Optics, Vol. 23, No. 21,
pp. 3736-3737, 1984.
13.E. Ackerman, S. Wanuga, D. Kasemset, W. Minford, N. Thorsten, and J. Watson,
“Integrated 6-bit photonic true-time-delay unit for lightweight 3-6 GHz radar
beamformer,” IEEE International Microwave Symposium Digest, Part 2, pp. 681-684,
1992.
14. G. A. Magel, T.-H. Lin, L. Y. Pang, and W.-R. Wu, “Integrated optic switches for
phased-array applications based on electrostatic actuation on metallic membranes,”
Optoelectronic Signal Processing fo r Phased Array Antennas IV, B. M. Hendrickson,
Ed., SPIE Vol. 2155, pp. 107-113, 1994.
292
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15.N. A. Riza and D. L. Polla, “Micromechanical fiber-optic switches for optical
networks,” in Integrated Optics and Microstructures, M. Tabib-Azar; D. L. Polla, Eds.,
SPIE Proc. 1973, pp. 108-126, 1992.
16. S. Yegnanarayanan, P. D. Trinh, and B. Jalali, “Recirculating photonic filter a
wavelength-selective time delay for phased-array antennas and wavelength code­
division multiple access,” Optics Letters, Vol. 21, No. 10, pp. 740-742, 1996.
17. L. Cardone, “Ultra-wideband microwave beamforming technique,” Microwave
Journal, Vol. 28, No. 4, pp. 121-131, 1985.
18.E. N. Toughlian and H. Zmuda, “Variable time-delay system for broadband phased
array and other transversal filtering applications,” Optical Engineering, Vol. 32, No. 3,
pp. 613-617, 1993.
19. A. P. Goutzoulis, D. K. Davies, J. M. Zomp, “Prototype fiber optic delay line,”
Optical Engineeering, Vol. 28, No. 11, pp. 1193-1202, 1989.
20. J. J. Pan, “Fiber optics for wideband extra high frequency (EHF) phased array,” in
Optoelectronic signal processing for phased-array antennas, SPIE Proc. 886, pp. 6069, 1988.
21. R. Soref, “Optical dispersion technique for time-delay beam steering,” Applied Optics,
Vol. 31, No. 35, pp. 7395-7397, 1992.
22. R. D. Esman, M. Y. Frankel, J. L. Dexter, L. Goldberg, M. G. Parent, D. Stilwell,
and D. G. Cooper, “Fiber-optic prism true time-delay antenna feed,” IEEE Photonics
Technology Letters, Vol. 5, No. 11, pp. 1347-1349, 1993.
23 M. Y. Frankel and R. D. Esman, ‘True time-delay fiber-optic control for ultrawideband
array transmitter/receiver with multibeam capability,” IEEE Transactions on Microwave
Theory and Techniques, Vol. 43, No. 9, 1995.
24. L. J. Lembo, T. Holcomb, M. Wickham, P. Wisseman, and J. C. Brock, “Low loss
fiber optic time-delay element for phased-array antennas,” in Optoelectronic Signal
293
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Processing fo r Phased-Array Antennas IV, Brian M. Hendrickson; Ed., SPIE Proc.
2551, pp. 13-23, 1994.
25. G. A. Ball, W. H. Glenn, and W. W. Morey, “Programmable fiber optic delay line,”
IEEE Photonics Technology Letters, Vol. 6, No. 6, pp. 741-743, 1994.
26. A. P. Goutzoulis and D. K. Davies, “Hardware-compressive 2-D fiber optic delay line
architecture for time steering of phased-array antennas,” Applied Optics, Vol. 29, No.
36, pp. 5353-5359, 1990.
27. A. P. Goutzoulis, D. K. Davies, and J. M. Zomp, “Hybrid electronic fiber optic
wavelength-multiplexed system for hue time-delay steering of phased array antennas,”
Optical Engineering, Vol. 31, No. 11, pp. 2312-2322, 1992.
28. D. Dolfi, P. Joffre, J. Antoine, J.-P. Huignard, D. Philippet, P. Granger, J. Chazelas,
“Photonics for phased array radars,” in Optical Technology fo r Microwave
Applications VII, SPIE Proceedings., Vol. 2560, pp. 158-165, 1995.
29. N. A. Riza, “A transmit/receive time delay optical beamforming architecture for phased
array antennas,” Applied Optics, Vol. 30, No. 32, pp. 4593-4596, 1991.
30. N. A. Riza, “25-Channel nematic liquid crystal optical time-delay unit characterization,”
IEEE Photonics Technology Letters, Vol. 7, No 11, pp. 1285-1287, 1995.
31.H. R. Fetterman, Y. Chang, D. C. Scot, S. R. Forrest, F. M. Espiau, M. Wu, D. V.
Plant, J. R. Kelly, A. Mather, W. H. Steier, R. M. Osgood, Jr., H. A. Haus, J.
Simonis, “Optically controlled phased array radar receiver using SLM switched real
time delays,” IEEE Microwave & Guided Wave Letters, Vol. 5, No. 11, pp. 414-416,
1995.
32. N. A. Riza, “High-optical-isolation Iow-loss moderate-switching-speed nematic liquidcrystal optical switch,” Optics Letters, Vol. 19, No. 21, pp. 1780-1782, 1994.
294
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33.X. S. Yao and L. Maleki, “A novel 2-D programmable photonic time-delay device for
millimeter-wave signal processing applications,” IEEE Photonic Technology Letters,
Vol. 6, No. 12, pp. 1463-1465, 1994.
34. S.-C. Lin, R. S. Boughton, “Acousto-optic multichannel programmable true time delay
lines,” in Optical Technology for Microwave Applications, S.-K. Yao, Ed., SPIE
Proc. 1102, pp. 162-173, 1989.
35. E. N. Toughlian and H. Zmuda, “A photonic variable RF delay line for phased array
antennas,” Journal o f Lightwave Technology, Vol. 8, No. 12, pp. 1824-1828, 1990.
36. W. D. Jemison, and P. R. Herczfeld, “Acoustooptically controlled true time delays,”
IEEE Microwave & Guided Wave Letters, Vol. 3, No. 3, pp. 72-74, 1993.
37. P. A. Cohen, Y. Chang, A. F. J. Levi, H. R. Fetterman, I. L. Newberg, “Optically
controlled serially fed phased array sensor,” IEEE Photonic Technology Letters, Vol.
8, No. 12, pp. 1683-1685, 1996.
38. Y. Chang, B. Tsap, H. R. Fetterman, P. A. Cohen, A. F. J. Levi, and I. L. Newberg,
“Optically controlled serially fed phased array transmitter,” IEEE Microwave & Guided
Wave U tters, Vol. 7, No. 3, pp. 69-71, 1997.
39. P. M. Freitag and S. R. Forrest, “A coherent optically controlled phased array antenna
system,” IEEE Microwave & Guided Wave U tters, Vol. 3, No. 9, pp. 293-295, 1993.
40. N. A. Riza, Editor, Selected Papers on Photonic Control Systems fo r Phased Array
Antennas, MS 136 of SPIE Milestone Series (SPIE Press, Bellingham, Wash., 1998).
41.N. A. Clark and S. T. Lagerwall, “Submicrosecond bistable electro-optic switching in
liquid crystals,” Applied Physics Utters, Vol. 36, No. 11, pp. 899-901, 1990.
42. J. Adam, “Pinning defense hopes on AEGIS,” IEEE Spectrum, Vol. 26, No. 6, pp.
24-27, June 1988.
295
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43. N. A. Riza, and N. Madamopoulos, “Photonic time delay beamforming architectures
using polarization switching arrays,” in Advances in Optical Information Processing
VII, D. R. Pape, Ed. , SPIE Proc. 2754, No. 21, pp. 186-197, 1996.
44. T. C. Cheston and J. Frank, “Phased array radar antennas,” in Radar Handbook,
Chapter 7, 2nd Edition, M. I. Scholnik, Ed. (New York: McGraw-Hill, 1990).
45. N. A. Riza and N. Madamopoulos, “High signal-to-noise ratio birefringence
compensated optical delay line based on a noise reduction scheme,” Optics Letters, Vol.
20, No. 22, pp. 2351-2353, 1995.
46. N. Madamopoulos, New Focus Student Prize Winning Essay: “Photonics for Antenna
Applications,” Optics and Photonics News, OS A Magazine, Vol. 7, No. 11, 1996.
47. N. A. Riza and N. Madamopoulos, “Microwave Band Demonstration of a Reflective
Geometry Fiber and Free-space Binary Photonic Delay Line” IEEE Microwave &
Guided Wave Letters, Vol. 7, No. 4, 1997
48. E. Ackerman, C. Cox, and N. A. Riza, Editors, Selected Papers on Analog Fiber-Optic
Links,. MS 149 o f SPIE Milestone Series (SPIE Press, Bellingham, Wash., 1998).
49. E. Ackerman, S. Wanuga, D. Kasemset, A. S. Daryoush, N. R. Samant, “Maximum
dynamic range operation of microwave eaxtemal modulation fiber-optic link,” IEEE
Transactions on Microwave Theory and Techniques, Vol. 41, No. 8, pp. 1299-1306,
1993.
50. Fiber Optic Assemblies and Components Catalog, Wave Optics , Inc., Mountain View,
CA, USA, 1989.
51. United Microwave Prods lnc.Cata.log, United Microwave Prods Inc., Torrance, CA,
USA, 1982.
52. A. Yariv, P. Yeh, Optical Waves in Crystals, Propagation and Control o f Laser
Radiation (New York: John Wiley & Sons , 1984).
296
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53. N. A. Riza, “Liquid crystal-based optical time delay units for phased array antennas,”
IEEE/OSA Journal o f Lightwave Technology, Vol. 12, No 8, pp. 1440-1447, 1994.
54. M. Martinelli, “A universal compensator for polarization changes induced by
birefringence on a retracing beam,” Optics Communications, Vol. 72, No. 6, pp. 341344, 1989.
Chapter 4
1. N. A. Riza, “Polarization-based fiber optic delay lines,” in Optical Technology for
Microwave Applications VII, A. P. Goutzoulis, Ed., SPIE Proc. 2560, pp. 120-129,
1995.
2. N. A. Riza and N. Madamopoulos, “High signal-to-noise ratio birefringence
compensated optical delay line based on a noise reduction scheme,” Optics Letters, Vol.
20, No. 22, 2351-2353, 1995.
3. N. A. Riza, “25-Channel nematic liquid crystal optical time-delay unit characterization,”
IEEE Photonics Technology Letters, Vol. 7, No. 11, pp. 1285-1287, 1995.
4. M. Martinelli, “A universal compensator for polarization changes induced by
birefringence on a retracing beam,” Optics Communications, Vol. 72, No. 6, pp. 341344, 1989.
5. A. Yariv, “Operator algebra for propagation problems involving phase conjugation and
nonreciprocal elements,” Applied Optics, Vol. 26, No. 21, pp. 4538-4540, 1987.
6. N. A. Riza and N. Madamopoulos, “Microwave band demonstration of a reflective
geometry fiber and ftee-space binary photonic delay line,” Microwave & Guided Wave
Letters, Vol. 7, No. 4, pp. 103-105, 1997.
297
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7. N. A. Riza, “Optical transversal filter”. United States Patent, No. 5,329,118, July 12,
1994.
8. C. H. Cox, G. E. Betts, and L. M. Johnson, “An analytic and experimental
comparison of direct and external modulation in analog fiber-optic links,” IEEE
Transactions on Microwave Theory and Techniques, Vol. 38, No. 5, pp. 501-509,
1990.
9. N. A. Riza, “Novel acousto-optic systems for spectrum analysis and phased array radar
signal processing,” Ph.D. Dissertation (California Institute of Technology, Pasadena,
Calif., October 1989), available from University Microfilms, Ann Arbor, Michigan.
10. N. A. Riza, “A compact high performance optical control system for phased array
radars,” IEEE Photonics Technology Letters, Vol. 4, No. 9, pp. 1073-1076, 1992.
11. N. A. Riza, “Acousto-optic liquid-crystal analog beamformer for phased-array
antennas,” Applied Optics, Vol. 33, No. 17, pp. 3712-3724, 1994.
12. N. A. Riza, “In-line interferometric time-integrating acousto-optic correlator,” Applied
Optics, Vol. 33, No. 14, pp. 3060-3069, 1994.
13.N. A. Riza, “Space integrating interferometric acousto-optic convolver,” IEEE
Photonics Technology Letters, Vol. 7, No. 3, pp. 339-341, 1995.
14. N. A. Riza, “Optically efficient interferometric acousto-optic architecture for spectrum
analysis,” Applied Optics, Vol. 31, No. 17, pp. 3194-3196, 1992.
15.N. A. Riza, “Novel space/time integrating acoustooptic architectures for radar signal
processing,” in Optoelectronic Signal Processing fo r Phased-Array Antennas IV, SPIE
Proc. 2155, pp. 413-419, 1994.
298
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 5
1. N. A. Riza, “Liquid crystal-based optical time delay units for phased array antennas,”
IEEE/OSA Journal o f Lightwave Technology, Vol. 12, No. 8, pp. 1440-1446, 1994.
2. D. Dolfi, P. Joffre, J. Antoine, J. P. Huignard, J. Roger, P. Granger, ‘Tw o
dimensional optical beam-forming networks,” in Optoelectronic Signal Processing fo r
Phased-Array Antennas IV, B. M. Hendrickson, Ed., SPIE Proc. 2155, pp. 205-217,
1994.
3. N. A. Riza and N. Madamopoulos, “Photonic time delay beamforming architectures
using polarization switching arrays,” in Advances in Optical Information Processing
VII, D. R. Pape, Ed. , SPIE Proc. 2754, No. 21, pp. 186-197, Orlando, 1996.
4. N. A. Riza, “High-optical-isolation low-loss moderate-switching-speed nematic liquidcrystal optical switch,” Optics Letters, Vol. 19, No. 21, pp. 1780-1782, Nov. 1, 1994.
5. T. C. Cheston and J. Frank, “Phased array radar antennas,” in Radar Handbook, 2nd
Ed., Chapter 7, M. I. Scholnik, Ed., (New York: McGraw-Hill, 1990).
6. N. A. Riza, “Liquid crystal-based optical control of phased array antennas,” IEEE/OSA
Journal o f Lightwave Technol., Vol. 10, No. 12, pp. 1974-1984, 1992.
7. J. Adam, “Pinning defense hopes on AEGIS,” IEEE Spectrum, Vol. 26, No. 6, pp.
24-27, 1988.
8. N. A. Clark and S. T. Lagerwall, “Submicrosecond bistable electro-optic switching in
liquid crystals,” Applied Physics Letters, Vol. 36, No. 11, pp. 899-901, 1990.
9. N. A. Riza, “Polarization-based fiber optic delay lines,” in Optical Technology fo r
Microwave Applications VII, A. P. Goutzoulis, Ed., SPIE Proc. 2560, pp. 120-129,
1995.
299
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10. N. A. Riza and N. Madamopoulos “High signal-to-noise ratio birefringencecompensated optical delay line based on a noise-reduction scheme,” Optics Letters,
Vol. 20, No. 22, pp. 2351-2353, 1995.
11. N. A. Riza, and N. Madamopoulos, “Phased array radar control using ferroelectric
liquid crystal devices,” in LEOS ‘96 Conf. Proc.: 9th Annual Meeting (IEEE-Lasers
and Electro-Optics Society, Boston, MA, 1996) Vol. 2, WG2, pp. 52-53.
12. N. A. Riza and N. Madamopoulos, “Characterization of a FLC-based time delay unit
for phased array antennas,” IEEE/OSA Journal o f Lightwave Technology, Vol. 15,
No. 7, pp. 1088-1094, 1997.
13. D. J. McKnight, K. M. Johnson, and R. A. Serati, “Optical computing and image
processing with ferroelectric liquid crystals,” Applied Optics, Vol. 33, No. 14, pp.
2775-2784, 1994.
14. N. Collins, J. Gourlay, D. G. Vass, H. J. White, C. Stace, and G. M. Proudley,
“Measurements on ferroelectric liquid-crystal spatial light modulators: contrast ratio and
speed,” Applied Optics, Vol. 34, No. 26, pp. 5928-5931, 1995.
15. S. Fukushima, T. Kurokawa, and M. Ohno, “Ferroelectric liquid-crystal spatial light
modulator achieving bipolar image operation and cascadability,” Applied Optics, Vol.
31, No. 32, pp. 6859-6868, 1992.
16.R. Biirkle, K. Schweikert, and E. Liider, “High-performance FLC-display as a spatial
light modulator for optical signal processing,” in Advances in Optical Information
Processing VII, D. R. Pape, Ed., SPIE Proc. 2754, No. 19, pp. 154-165, Apr. 1996.
17. J. S. Patel and J. W. Goodby, “Properties and applications of ferroelectric liquid
crystals,” Optical Engineering, Vol. 26, No. 5, pp. 373-384, 1987.
18.M. G. Roe, K. L. Schehrer, “High-speed and high-contrast operation of ferroelectric
liquid crystal optically addressed spatial light modulators,” Optical Engineering, Vol.
32, No. 7, pp. 1662-1667, 1993.
300
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19.X. Lin, J. Ohstubo, and T. Takemori, “Real-time optical image subtraction and edge
enhancement using ferroelectric liquid-crystal devices based on speckle modulation,”
Applied Optics, Vol. 35, No. 17, pp. 3148-31554, 1996.
20. N. A. Riza, and N. Madamopoulos, “Microwave band demonstration of a reflective
geometry fiber and free-space binary photonic delay line,” Microwave & Guided Wave
Letters, Vol. 7, No. 4, pp. 103-105, 1997.
21. R. B. Meyer , L. Liebert, L. Strzelecki, and P. Keller, ‘Terroelectric Liquid Crystals,”
Journal de Physique - Letters, Vol. 36, pp. L-69, 1975.
22. J. Dijon, “Ferroelectric LCDs”, in Liquid Crystals, Applications and Uses, Vol. 1,
Chapter 13, B. Bahadur, Ed., (New Jersey: World Scientific Publishing, 1990).
23. L. M. Blinov, “Electric field effects in liquid crystals”, in Handbook o f Liquid Crystal
Research, Chapter 5, P. J. Collins, J. S. Patel, Eds., (Oxford University Press, New
York, 1997).
24. Displaytech Shutters User’s manual, Version 1.1, February, 1994, Displaytech, Inc.,
Boulder, CO, USA.
25.Boulder Nonlinear Systems Product Catalog, Boulder Nonlinear Systems, Inc.,
Lafayette, CO, USA, 1998.
Chapter 6
1. N. A. Riza, “A transmit/receive time delay optical beamforming architecture for phased
array antennas,” Applied Optics, Vol. 30, No. 32, pp. 4593-4596, 1991.
2. N. A. Riza, “Liquid crystal-based optical time delay units for phased array antennas,”
IEEE/OSA Journal o f Lightwave Technology, Vol. 12, No 8, pp. 1440-1447, 1994.
301
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3. N. A. Riza, “25-Channel nematic liquid crystal optical time-delay unit characterization,”
IEEE Photonics Technology Letters, Vol. 7, No 11, pp. 1285-1287, 1995.
4. N. A. Riza and N. Madamopoulos, “High signal-to-noise ratio birefringence
compensated optical delay line based on a noise reduction scheme,” Optics Letters, Vol.
20, No. 22, pp. 2351-2353, 1995.
5. N. Fourikis, Phased Array Based Systems and Applications, (New York: John Wiley
and Sons, 1997).
6. A. P. Goutzoulis, D. K. Davies, J. M. Zomp, P. Hrycak, and A. Johnson,
“Development and field demonstration of a hardware compressive fiber-optic true time
delay steering system for phased array antennas,” Applied Optics, Vol. 33, pp. 81738185, 1994.
7. S. T. Johns, D. A. Norton, C. W. Keefer, R. Erdmann, and R. A. Soref, “Variable
time delay of microwave signals using high dispersion fibre,” Electronics Letters, Vol.
29, No. 6, pp. 555-556, 1993.
8. M. Frankel, R. D. Esman, and M. G. Parent, “Array transmitter/receiver controlled by
a true time-delay fiber-optic beamformer,” IEEE Photonics Technology Letters, Vol. 7,
No. 10, pp. 1216-1218, 1995.
9. N. A. Riza and N. Madamopoulos, “Photonic time delay beamforming architectures
using polarization switching arrays,” in Advances in Optical Information Processing
VII, D. R. Pape, Ed., SPIE Proc. 2754, pp. 186-197, Orlando, 1996.
10. N. A. Riza, and N. Madamopoulos, “Phased array antenna maximum compression
reversible photonic beamformer using ternary designs and multiple wavelengths,”
Applied Optics, Vol. 36, No. 5, pp. 983-996, 1997.
l l .S . Yuan and N. A. Riza, “Robust packaging of photonic RF modules using ultra-thin
adaptive optical interconnect devices,” in Optical Technology fo r Microwave
Applications VIII, A. P. Goutzoulis, Ed, SPIE Proc. 3160, pp. 170-177, 1997.
302
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12. G. Hills, Private communication, Global Fiber Optics, Mississauga, Ontario, Canada.
13. M. Martinelli, “A universal compensator for polarization changes induced by
birefringence on a retracing beam,” Optics Communications, Vol. 72, No 6, pp. 341344, 1989.
14. L. J. Lembo, T. Holcomb, M. Wickham, P. Wisseman, and J. C. Brock, “Low loss
fiber optic time-delay element for phased-array antennas,” in Optoelectronic Signal
Processing fo r Phased-Array Antennas IV, B. M. Hendrickson, Ed., SPIE Proc.
2155, No. 13, 1994.
15. G. A. Ball, W. H. Glenn, and W. W. Morey, “Programmable fiber optic delay line,”
IEEE Photonics Technology Letters, Vol. 6, No. 6, 1994.
16. K. Brundin, Private communication, 3M Specialty Optical Fibers, West Haven,
Connecticut, USA.
17. R. Lundin, “Dispersion flattening in a W fiber,” Applied Optics, Vol. 33, No. 6, pp.
1011-1014, 1994.
18. S. P. Survaiya and R. K. Shevgaonkar, “Design of subpicosecond dispersion-flattened
fibers,” IEEE Photonics technology Letters, Vol. 8, No. 6, pp. 803-805, 1996.
19. SMF-28 Single-Mode Optical Fiber, Coming Inc., Coming, New York, USA.
20. M. Onishi, H. Kanamori, T. Kato, and M. Nishimura, “Optimization of dispersioncompensating fibers considering self-phase modulation suppression,”
OFC’96
Technical Digest, Vol. 2, pp. 200-201, February 1996.
21. Integrated Photonics Research Digest, Vol. 6, OSA Technical Digest Series (Optical
Society of America, Washington, D.C., May 1996).
22. N. A. Riza and N. Madamopoulos, “Characterization of a ferroelectric liquid crystal
based time delay unit for phased array antenna applications,” IEEE/OSA Journal o f
Lightwave Technology, Vol. 15, No. 7, pp. 1088-1094, 1997.
303
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23. Displaytech Shutters User’s manual, Version 1.1, February, 1994, Displaytech, Inc.,
Boulder, Colorado, USA.
24. A. Yariv, P. Yeh, Optical Waves in Crystals (New York: John Wiley & Sons, 1984).
25. C. Crandall, Private communication, Displaytech, Inc., Boulder, Colorado, USA.
26. David Samoff Research Center, Princeton, New Jersey, USA.
27. Photonic Integration Research Inc. (PERI). Piriodical No. II, Columbus, Ohio, USA.
Chapter 7
1. N. A. Riza, “Liquid crystal-based optical time delay control system for wideband
phased arrays,” in Analog Photonics, A. R. Pirich; P. Sierak; Eds., SPIE Proc. 1790,
pp. 171-183, 1992.
2. N. A. Riza, “Liquid crystal-based optical time delay control units for phased array
antennas,” IEEE/OSA Journal o f Lightwave Technology, Vol. 12, No. 8, pp. 14401447, 1994.
3. M. C. DeJule, T. L. Credelle, N. A. Riza, and D. E. Castleberry, “Compact
polarization dependent optical switching units,” United States Patent, Patent Number:
5,345,321, Issued: September 6, 1994.
4. N. A. Riza and N. Madamopoulos, “High signal-to-noise ratio birefringence
compensated optical delay line based on a noise reduction scheme,” Optics Letters, Vol.
20, No. 22, pp. 2351-2353, 1995.
5. N. A. Riza and N. Madamopoulos, “Photonic time delay beamforming architectures
using polarization switching arrays,” in Advances in Optical Information Processing
VII, D. R. Pape, Ed., SPIE Proc. 2754, No. 21, pp. 186-197, 1996.
304
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6. N. A. Riza and N. Madamopoulos, “Microwave band demonstration of a reflective
geometry fiber and free-space binary photonic delay line,” Microwave & Guided Wave
Letters, Vol. 7, No. 4, pp. 103-105, 1997.
7. D. Dolfi, F. Michel-Gabriel, S. Bann, and J. P. Huignard, ‘Two-dimensional optical
architecture for time-delay beam forming in a phased array antenna,” Optics Letters,
Vol. 16, No. 4, pp. 255-257, 1991.
8. X. S. Yao and L. Maleki, “A novel 2-D programmable photonic time-delay device for
milimeter-wave signal processing applications,” IEEE Photonics Technology Letters,
Vol. 6, No. 6, pp. 1463-1465, 1994.
9. J. Fu, M. Schamschula and H. J. Caulfield, “Modular solid optic time delay system,”
Optics Communication, Vol. 121, pp. 8-12, November 15, 1995.
10. N. A. Riza, “Optical multiple beamforming systems for wireless communication
antennas,” in Wireless Communication, J. J. Pan, Ed., SPIE Proc. 2556, pp. 139150, 1995.
11. N. A. Riza, “Advances in three-dimensional reversible photonic modules for phased
array control,” in Photonics and Radio Frequency, B. M. Hendrickson, Ed., SPIE
Proc. 2844, pp. 274-283, 1996.
12.N. Madamopoulos and N. A. Riza, “Adaptable delay balanced loss switched photonic
time delay modules for antenna arrays,” in Optical Technology fo r Microwave
Applications, A. P. Goutzoulis, Ed., SPIE Proc. 3160, No. 8, pp. 62-68, 1997.
13.N. Madamopoulos and N. A. Riza, “Adaptable-delay balanced-loss binary photonic
delay line architectures using polarization switching,” Optics Communications, Vol.
152, pp. 135-143, 1998
14. N. A. Riza and N. Madamopoulos, “Characterization of a ferroelectric liquid crystalbased time delay unit for phased array antennas,” IEEE/OSA Journal o f Lightwave
Technology, Vol. 15, No. 7, 1997.
305
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15.N. Madamopoulos and N. A. Riza, “Switched photonic delay line for phased array
antenna control using externally modulated microwave fiber-optic link,” in Optical
Technology fo r Microwave Applications VIII, A. P. Goutzoulis, Ed., SPIE Proc.
3160, pp. 45-54, August 1997.
16.N. A. Riza and N. Madamopoulos, “Phased Array Antenna Maximum Compression
Reversible Photonic Beamformer using Ternary Designs and Multiple Wavelengths,”
Applied Optics, Vol. 36, No. 5, pp. 983-996, 1997.
Chapter 8
1. E. Ackerman, C. Cox, and N. A. Riza, Editors, Selected Papers on Analog Fiber-Optic
Links, MS 149 of SPIE Milestone Series (SPIE Press, Bellingham, Wash., 1998).
2. N. A. Riza and N. Madamopoulos, “Phased array radar control using ferroelectric
liquid crystal devices,” in LEOS’96 Conf. Proc.: 9th Annual Meeting (IEEE-Lasers and
Electro-Optics Society, Boston, Mass., 1996), Vol. 2, WG2, pp. 52-53.
3. N. A. Riza and N. Madamopoulos, “Characterization of a ferroelectric liquid crystal
based time delay unit for phased array antenna applications,” IEEEJOSA Journal o f
Lightwave Technology, Vol. 15, pp. 1088-1094, 1997.
4. N. Madamopoulos and N. A. Riza, “Switched three dimensional photonic delay line
using directly modulated semiconductor lasers for microwave radar processing,” in
Radar Processing, Technology and Applications, W. J. Miceli, Ed., SPIE Proc. 2754,
pp. 266-275, 1996.
306
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5. N. Madamopoulos and N. A. Riza, “Directly modulated semiconductor laser fed
photonic delay line using ferroelectric liquid crystals,” Applied Optics, Vol. 37, No. 8,
pp. 1407-1416, 1998.
6. N. A. Riza and N. Madamopoulos, “Photonic time delay beamforming architectures
using polarization switching arrays,” in Advances in Optical Information Processing
VII, D. R. Pape, Ed., SPIE Proc. 2754, pp. 186-197, 1996.
7. N. A. Riza, “25-Channel nematic liquid crystal optical time-delay unit characterization,”
IEEE Photonics Technology Letters, Vol. 7, pp. 1285-1287, 1995.
8. Displaytech Shutters User’s manual, Version 1.1, February, 1994, Displaytech, Inc.,
Boulder, CO, USA.
9. N. A. Riza, and N. Madamopoulos, “Characterization of a ferroelectric liquid crystal
based time delay unit for phased array antenna applications,” IEEE/OSA Journal o f
Lightwave Technology, Vol. 15, No. 7, pp. 1088-1094, 1997.
10.Selfoc Product Guide, Nippon Sheet Glass Co. (NSG) America, Somerset, New
Jersey, 1995.
11. J. Kim and N. A. Riza, “Fiber array optical coupling design issues for photonic
beamformers,” Advances in Optical Information Processing VII, D. R. Pape, Ed.,
SPIE Proc. 2754, pp. 271-282, 1996.
12.QLINK1-XXX series, Lasertron 1996/97 Product Guide, Lasertron, Inc., Bedford,
Massachusetts, USA.
13.G. K. Gopalakrishnan, R. P. Moeller, M. M. Howerton, W. K. Bums, K. J.
Williams, and R. D. Esman, “A low-loss downconverting analog fiber-optic link,”
IEEE Transactions on Microwave Theory and Techniques, Vol. 43, pp. 2318-2323,
1995.
307
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 9
1. E. Ackerman, C. Cox, and N. A. Riza, Editors, Selected Papers on Analog Fiber-Optic
Links,. MS 149 of SPIE Milestone Series (SPIE Press, Bellingham, Wash., 1998).
2. C. H. Cox, “Gain and noise figure in analogue fibre-optic links,” IEE Proceedings-J,
Vol. 139, No. 4, pp. 238-242, 1992.
3. N. Madamopoulos and N. A. Riza, “Switched photonic delay line for phased array
antenna control using externally modulated microwave fiber-optic link,” in Optical
Technology fo r Microwave Applications, A. P. Goutzoulis, Ed., SPIE Proc. 3160, pp.
45-54, 1997.
4. N. Madamopoulos and N. A. Riza, “Adaptable-delay balanced-loss binary photonic
delay line architectures using polarization switching,” Optics Communications, Vol.
152, pp. 135-143, 1998.
5. N. Madamopoulos and N. A. Riza, “Adaptable-delay balanced-loss binary photonic
delay line architectures using polarization switching,” Optics Communications, Vol.
152, pp. 135-143, 1998.
6. N. A. Riza and N. Madamopoulos, “Characterization of a FLC-based time delay unit
for phased array antennas,” IEEE/OSA Journal o f Lightwave Technology, Vol. 15,
No. 7, pp. 1088-1094, 1997.
7. J. Kim and N. A. Riza, “Fiber array optical coupling design issues for photonic
beamformers,” Advances in Optical Information Processing VII, D. R. Pape, Ed.,
SPIE Proc. 2754, pp. 271-282 1996.
8. OZ-Optics Fiber Collimator/Focusers, OZ-Optics, Carp, Ontario, Canada, 1995.
9. A. M. Yurek, S. W. Merritt, and G. Drake, “Determining the cascade parameters of
externally modulated links,” Microwave Journal, Vol. 38, No. 8, pp. 80-86, 1995.
308
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 10
1. N. Madamopoulos and N. A. Riza, “Switched photonic delay line for phased array
antenna control using externally modulated microwave fiber-optic link,” in Optical
Technology fo r Microwave Applications VIII, A. Goutzoulis, Ed., SPIE Proc. 3160,
pp. 45-54, 1997.
2. E. Ackerman, C. Cox, and N. A. Riza, Editors, Selected Papers on Analog Fiber-Optic
Links, MS 149 of SPIE Milestone Series (SPIE Press, Bellingham, Wash., 1998).
3. A. S. Daryoush, E. Ackerman, N. R. Samant, S. Wanuga, D. Kasemset, “Interfaces
for high-speed fiber-optic links: analysis and experiment,” IEEE Transactions on
Microwave Theory and Techniques, Vol. 39, No. 21, pp. 2031-2044, 1991.
4. C. Cox IE, G. E. Betts, and L. M. Johnson, “An analytic and experimental
comparison of direct and external modulation in analog fiber-optic links,” IEEE
Transactions on Microwave Theory and Techniques, Vol. 38, No. 5, pp. 501-509,
1990.
5. C. Cox HI, E. Ackerman, R. Helkey, G. E. Betts, ‘Techniques and performance of
intensity-modulation direct-detection analog optical links,” IEEE Transactions on
Microwave Theory and Techniques, Vol. 45, No. 8, pp. 1375-1383, 1997.
6. G. E. Betts, C. H. Cox, and K. G. Ray, “20 GHz Optical analog link using an external
modulator,” IEEE Photonics Technology Letters, Vol. 2, No. 12, pp. 923-925, 1990.
7. E. Ackerman, S. Wanuga, D. Kasemset, A. S. Daryoush, N. R. Samant, “Maximum
dynamic range operation of microwave external modulation fiber-optic link,” IEEE
Transactions on Microwave Theory and Techniques, Vol. 41, No. 8, pp. 1299-1306,
1993.
309
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8. K. J. Williams, L. T. Nichols, and R. D. Esman, “Photodetector nonlinearity
limitations on a high-dynamic range 3 GHz fiber-optic link,” IEEE/OSA Journal o f
Lightwave Technology, Vol. 16, No. 2, pp. 192-199, 1998.
9. A. M. Yurek, S. W. Merritt, and G. Drake, “Determining the cascade parameters of
externally modulated links,” Microwave Journal, Vol. 38, No. 8, pp. 80-86, 1995.
10.N. Madamopoulos and N. A. Riza, “Adaptable-delay balanced-loss binary photonic
delay line architectures using polarization switching,” Optics Communications, Vol.
152, pp. 135-143, 1998.
11.L. Xu, R. Taylor, and S. R. Forrest, ‘True time-delay phased-array antenna feed
system based on optical heterodyne techniques,” IEEE Photonics Technology Letters,
Vol. 8, No. 1, pp. 160-162, 1996.
12. W. W. Ng, A. A. Walston, G. L. Tangonan, J. J. Lee, I. L. Newberg, and N.
Bernstein, “The first demonstration of an optically steered microwave phased array
antenna using true-time delay,” IEEE/OSA Journal o f Lightwave Technology, Vol. 9,
No. 9, pp. 1124-1131, 1991.
13. L. Cardone, “Ultra-wideband microwave beamforming technique,” Microwave
Journal, Vol. 28, No. 4, pp. 121-131, 1985.
14. A. P. Goutzoulis, D. K. Davies, and J. M. Zomp, “Hybrid electronic fiber optic
wavelength-multiplexed system for true time-delay steering of phased array antennas,”
Optical Engineering, Vol. 31, No. 11, pp. 2312-2322, 1992.
15. H. Zmuda, R. A. Soref, P. Payson, S. Johns, and E. N. Toughlian, “Photonic
beamformer for phased array antennas using a fiber grating prism,” IEEE Photonics
Technology Letters, Vol. 9, No. 2, pp. 241-243, 1997.
16.D. Dolfi, F. Michel-Gabriel, S. Bann, and J. P. Huignard, ‘Two-dimensional optical
architecture for time-delay beam forming in a phased-array antenna,” Optics Letters,
Vol. 16, No. 4, pp. 255-257, 1991.
310
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17. R. F. Mathis, W. L. Floyd, and S. A. Pappert, “High performance fiber optic delay
line,” The Seventh Annual DARPA Symposium on Photonic Systems fo r Antenna
Applications (PSAA-7), The Naval Postgraduate School, Monterey, California, pp. 914, 14-16 January 1997.
18. Optical Variable Attenuator Module, OVA-610, Product Specifications, SANTEC
Corporation, Aichi, Japan, 1998.
19. N. A. Riza, “Advances in three dimensional reversible photonic modules for phased
array control,” in Photonics and Radio Frequency, B. M. Hendrickson, Ed., Proc.
SPIE 2844, pp. 274-283, 1996.
20. S. E. Broomfield, M. A. A. Neil, E. G. S. Paige, and G. G. Yang, “Programmable
binary phase-only optical device based on ferroelectric liquid crystal SLM,” Electronic
Letters, Vol. 28, No. 1, pp. 26-27, 1992.
21. M. O. Freeman, T. A. Brown, and D. M. Walba, “Quantized complex ferroelectric
liquid crystal spatial light modulators,” Applied Optics, Vol. 31, No. 20, pp. 39173929, 1992.
22. J. Kim and N. A. Riza, “Fiber array optical coupling design issues for photonic
beamformers,” Advances in Optical Information Processing VII, D. R. Pape, ed.,
SPIE Proc. 2754, 271-282, 1996.
23. N. A. Riza and S. Yuan, “Demonstration of a liquid-crystal adaptive alignment tweeker
for high-speed infrared band fiber-fed free-space systems,” Optical Engineering, Vol.
37, No. 6, pp. 1876-1880, 1998.
24. V. T. Tondiglia, L. V. Natarajan, R. L. Sutherland, T. J. Bunning, and W. W.
Adams, “Volume holographic image storage and electro-optical readout in a polymerdispersed liquid-crystal film,” Optics Letters, Vol. 20, No. 11, pp. 1325-1327, 1995.
311
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25. N. A. Riza and S. E. Saddow, “N-bit optically controlled microwave signal attenuator
using the photoconductive effect,” in Optical Technology fo r Microwave Applications
VII, A. P. Goutzoulis, Ed., SPIE Proc. 2560, pp. 9-18, 1995.
26. N. A. Riza and S. E. Saddow, “Optically controlled photoconductive N-bit switched
microwave signal attenuator,” IEEE Microwave & Guided Wave Letters, Vol. 5, No.
12, pp. 448-450, 1995.
2 7 .Designer’s Guide to External Modulation, UTP, Uniphase Telecommunication
Products, Electro-Optic Products Division, Bloomfield, Connecticut, USA.
28. B. H. Kolner and D. W. Dolfi, “Intermodulation distortion and compression is an
integrated electro-optic modulator,” Applied Optics, Vol. 26, No. 17, pp. 3676-3680,
1987.
29. H. Goldberg, “Some notes on noise figure,” Proceedings o f the I.R .E., Vol. 36, pp.
1205-1214, 1948.
30. C. Cox, E. Ackerman, G. Betts, “Relationship between gain and noise figure of an
optical analog link,” IEEE MTT-S Symposium Digest, pp. 1551-1554, San Francisco,
CA, 1996.
31.K . Williams, R. Esman, and M. Dagenais, “Nonlinearities in p-i-n microwave
photodetectors,” IEEE Photonic Technology Letters, Vol. 14, No. 1, pp. 94-96, 1996.
32. M. L. Farwell, W. S. C. Chang, and D. R. Huber, “Increased linear dynamic range by
low biasing the Mach-Zehnder modulator,” IEEE Photonic Technology Letters, Vol. 5,
No. 7, pp. 779-782, 1993.
33. R. D. Esman, and K. J. Williams, “Wideband efficiency improvement of fiber optic
systems by carrier subtraction,” IEEE Photonic Technology Letters, Vol. 7, No. 2, pp.
218-220, 1995.
312
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34. G. Drake, and B. Merritt, “High-dynamic range applications of integrated optic
modulators,” in Optical Technology fo r Microwave Applications VII, A.
P.
Goutzoulis, Ed., Proc. SPIE 2560, pp. 2-8, 1995.
Chapter 11
1. W. Ng, A. A. Waltson, G. L. Tangonan, J. J. Lee, I. L. Newberg, and N. Bernstein,
“The first demonstration of an optically steered microwave antenna using true-timedelay,” IEEE/OSA Journal o f Lightwave Technology, Vol. 9, No. 9, pp. 1124-1131,
1991.
2. W. Ng, R. Loo, V. Jones, J. Lewis, S. Livingston, and J. J. Lee, “Silica-waveguide
optical time-shift network for steering a 96-element L-band conformal array,” in Optical
Technology fo r Microwave applications VII, A. P. Goutzoulis, Ed., Proc. SPIE 2560,
pp. 140-147, 1995.
3. D. Dolfi, P. Joffre, J. Antoine, J.-P.
Huignard, D. Philippet, P. Granger,
“Experimental demonstration of a phased-array antenna optically controlled with phase
and time technology,” Applied Optics, Vol. 35, No. 26, pp. 5293-5300, 1996.
4. M. Y. Frankel, P. J. Matthews, and R. D. Esman, “Wideband array transmitter with
two-dimensional fiber-optic beam steering,” in Revolutionary Developments in Phased
Arrays,
1996 1FJRK International Symposium on Phased Array Systems and
Technology, pp. 425-428, Boston, MA., October 1996.
5. M. Y. Frankel, P. J. Matthews, and R. D. Esman, “Fiber-optic true time steering of an
ultrawide-band receive array,” IEEE Transactions on Microwave Theory and
Techniques, Vol. 45, No. 8, pp. 1522-1526, 1997.
313
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6. A. P. Goutzoulis and J. M. Zomp, “Development and field demonstration of an eightelement receive wavelength-multiplexed true-time-delay steering system,” Applied
Optics, Vol. 36, No. 29, pp. 7315-7326, 1997.
7. Fiber Collimators-Focusers, OZ-Optics, Ltd., Carp, Ontario, Canada.
8. N. A. Riza and S. Yuan, “Demonstration of a liquid crystal adaptive alignment tweeker
for high-speed infrared band fiber-fed free-space systems,” Optical Engineering, Vol.
37, No. 6, pp. 1876-1880, 1998.
9. J. Kim and N. A. Riza, ‘Tiber array optical coupling design issues for photonic
beamformers,” in Advances in Optical Information Processing VII, D. R. Pape, Ed.,
SPIE Proc. 2754, No. 30, pp. 271-282, April 1996.
10. N. Madamopoulos and N. A. Riza, “Switched photonic delay line for phased array
antenna control using externally modulated microwave fiber-optic link,” in Optical
Technology fo r Microwave Applications VIII, A. Goutzoulis, Ed., SPIE Proc. 3160,
pp. 45-54, August 1997.
11. LAH-58 OHARA glass Specification Sheet., OHARA Corporation, Rancho Santa
Margarita, CA, USA.
12.N. Madamopoulos and N. A. Riza, “Adaptable-delay balanced-loss binary photonic
delay line architectures using polarization switching,” Optics Communications, Vol.
152, pp. 135-143, 1998.
13.Jay Stockley, Private Communication, Boulder Nonlinear Systems (BNS), Inc.,
Lafayette, CO.
14. N. A. Clark and S. T. Lagerwall, “Submicrosecond bistable electro-optic switching in
liquid crystals,” Applied Physics Letters, Vol. 36, No. 11, pp. 899-901, June 1, 1990.
15. N. A. Riza, “Liquid crystal-based optical control of phased array antennas,” IEEE/OSA
Journal o f Lightwave Technology, Vol. 10, No. 12, pp. 1974-1984, 1992.
314
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 12
1. N. A. Riza and N. Madamopoulos, "Single micro-optical bench fiber-connectorized
delay module using bulk polarization optics," The Eight Annual DARPA Symposium
on Photonic Systems fo r Antenna Applications (PSAA-8), The Naval Postgraduate
School, Monterey, California, Postdeadline paper, 13-15 January 1998,
2. N. A. Riza and N. Madamopoulos, “Reversible fiber-optic switched delay module
using GRIN lens fiber-optic collimators and ferroelectric liquid crystals,” in LEOS ‘98
Coneference Proc.: 11th Annual Meeting (IHJEE-Lasers and Electro-Optics Society,
Orlando, FL„ 1998), ThFF4.
3. JDS Fitel Fiber Optic Catalog, JDS Fitel, Ontario, Canada.
4. UTP Catalog o f Integrated Optical Circuits, UTP, Bloomfield, Connecticut, USA.
5. N. A. Riza and N. Madamopoulos, "Characterization of a FLC-based time delay unit
for phased array antennas," IEEE/OSA Journal o f Lightwave Technology, Vol. 15,
No. 7, pp. 1088-1094, 1997.
6. N. A. Riza, and N. Madamopoulos, "High signal-to-noise ratio birefringence
compensated optical delay line based on a noise reduction scheme," Optics Letters, Vol.
20, No. 22, pp. 2351-2353, 1995.
7. Fiber Bench Coupling Systems, Optics for Research, Caldwell, New Jersey, USA.
8. N. A. Riza and N. Madamopoulos, “All-fiber connectorized fiber-optic delay module
using 3-D polarization optics,”in LEOS’97 Conference Proc.: 10th Annual Meeting,
(IEEE-Lasers and Electro-Optics Society, San Francisco, CA, 1997), Vol. 2, ThW2,
pp. 472-473, 1997.
9. OZ-Optics Fiber Collimator/Focusers, OZ-Optics, Carp, Ontario, Canada, 1995,
10. Garland Best, Personnal Communication, OZ-Optics, Carp, Ontario, Canada.
315
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11.K. Brendin, and B. Tamm, Personal Communications, 3M Specialty Optical Fibers,
West Haven, CT, USA.
12. M. J. Daneman, N. C. Tien, O. Solgaard, A.P. Pisano, K. Y. Lau, R. S. Muller
"Linear microvibromotor
for
positioning
optical
components,"
Journal
of
Microelectromechanical Systems, Vol. 5, No. 3, pp. 159-165, 1996.
Chapter 13:
1. N. A. Riza and N. Madamopoulos, “Photonic delay line using electrically switched
gratings in polymer dispersed liquid crystals,” Optical Engineering, Vol. 37, No. 11,
pp.
1998.
2. L. Domash, C. Gozewski, A. Nelson and J. Schwartz, “Programmable beamlet
generator, dynamic lens, and optical memory using electrically switched holographic
devices,” in Photonics fo r Processors, Neural Networks, and Memories, J. L. Homer;
B. Javidi; S. T. Kowel; W. J. Miceli; Eds., SPIE Proc. 2026c, pp. 642-652, 1993.
3. R. T. Ingwalla and M. Troll, “Mechanism of hologram formation in DMP-128
photopolymer,” Optical Engineering, Vol. 28, No. 6, pp. 586-591, 1989.
4. D. Whitney and R. T. Ingwall, “Fabrication and properties of composite holograms
recorded in DMP-128 photopolymer,” in Photopolymer Device Physics, Chemistry,
and Applications, R. A. Lessard; Ed., SPIE Proc. 1213, pp. 18-26, 1990.
5.
K. Takizawa, H. Kikuchi, H. Fujikake, Y. Namikawa, and K. Tada, “Polymerdispersed liquid crystal light valves for projection display,” Optical Engineering, Vol.
32, No. 8, pp. 1781-1791, 1993.
316
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6. V. P. Tondiglia, L. V. Natarajan, R. L. Sutherland, T. J. Bunning, and W. W.
Adams, “Volume holographic image storage and electro-optical readout in a polymerdispersed liquid-crystal film,” Optics Letters, Vol. 20, No. 11, pp. 1325-1327, 1995.
7. S. J. Klosowicz and J. Zmija, “Optics and electro-optics of polymer-dispersed liquid
crystals: physics, technology, and application,” Optical Engineering, Vol. 34, No. 12,
pp. 3440-3450, 1995.
8. R. L. Sutherland, L. V. Natarajan, V. T. Tondiglia, and T. J. Bunning, “Bragg
gratings in an acrylate polymer consisting of periodic polymer - dispersed liquid crystal
planes,” Chemistry o f Materials, Vol. 5, pp. 1533-1538, 1993.
9. L. H. Domash, Y-M. Chen, B. Gomatam, C. Gozewski, R. L. Sutherland, L. V.
Natarajan, V. P. Tondiglia, T. J. Bunning, W. W. Adams, “Switchable-focus lenses
in holographic polymer-dispersed liquid crystal,” in Diffractive and Holographic Optics
Technology III, I. Cindrich, S. H. Lee, Eds., SPIE Proc. 2689, pp. 188-194, 1996.
10.S.-X. Cheng, R.-K Bai, Y.-F. Zou, and C.-Y. Pan, “Electro-optical properties of
polymer dispersed liquid crystal materials,” Journal o f Applied Physics, Vol. 80, No.
4, pp. 1991-1995, 1996.
11. V. Y. Reshetnyak, “Effective-medium theory of polymer dispersed liquid crystal
droplet systems: I. Spherical droplets,” Journal o f Physics D, Applied Physics, Vol.
29, pp. 2459-2465, 1996.
12.H. Kogelnik, “Coupled wave theory for thick hologram gratings,” The Bell System
Technical Journal, Vol. 48, No. 9, pp. 2909-2947, November, 1969.
13. W. Doane, N. A. Vaz, B.-G. Wu, and S. Zumer, “Field controlled light scattering
from nematic microdroplets,” Applied Physics Letters, Vol. 48, No. 4, pp. 269-271,
1986.
317
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14. R. L. Sutherland, V. T. Tondiglia, L. V. Natarajan, T. J. Bunning, and W. W.
Adams, “Electrically switchable volume holographic gratings in polymer-dispersed
liquid crystals,” Applied Physics Letters, Vol. 64, No. 9, pp. 1074-1076, 1994.
15. R. L. Sutherland, L. V. Natarajan, V. T. Tondiglia, T. J. Bunning, and W. W.
Adams, “The physics of photopolymer-liquid crystal composite holographic gratings,”
in Diffractive and Holographic Optics Technology III, I. Cindrich, S. H. Lee, Ed.,
SPIE Proc, Vol. 2689, pp. 158-169, 1996.
16.L. Sutherland, L. V. Natarajan, V. P. Tondiglia, T. J. Bunning, B. L. Epling, and D.
M. Brandelik, “Relation of electro-optical characteristics to materials properties and
morphology in polymer-dispersed liquid crystal holographic gratings,” in Diffractive
and Holographic Device Technologies and Applications IV, I. Cindrich; S. H. Lee;
Ed., SPIE Proceedings, Vol 3010, pp. 142-149, 1997.
17. N. A. Riza and J. Chen, “Ultra-high - 47 dB optical drop rejection multi-wavelength
add-drop filter using spatial filtering and dual bulk acousto-optic tunable filters,” Optics
Letters, Vol. 23, No. 12, pp. 945-947, 1998.
18.T. Todorov, L. Nikolova, K. Stoyanova, and N. Tomova, “Polarization holography
HI. Some applications of polarization holographic recording,” Applied Optics, Vol. 24,
pp. 785-788, 1985.
19.R. Kostuk, M. Kato, and Y. T. Huang, “Polarization properties of substrate-mode
holographic interconnects,” Applied Optics, Vol. 29, pp. 3848-3854, 1990.
20. Q. W. Song, M. C. Lee, and P. J. Talbot, E. Tam, “Polarization sensitivity of
bireftingent photorefiractive holograms and its applications to binary switching,”
Applied Optics, Vol. 31, 6240-6246, 1992.
21. F. Xu, J. Ford, and Y. Fainman, “Polarization-selective computer-generated
holograms: design, fabrication, and applications,” Applied Optics, Vol. 34, pp. 256366, 1995.
318
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22. S. Sato, “Liquid-crystal lens-cells with variable focal length,” Japanese Journal o f
Applied Physics, Vol. 18, No. 9, pp. 1676-1684, 1979.
2 3 .N. A. Riza and M. C. DeJule, “A novel programmable liquid crystal lens device for
adaptive optical interconnect and beamforming applications,” Optical Computing
International Conference, Inst. Phys. Conf. No 139; Part II, pp. 231-234, Edinburgh,
Scotland, August 1994.
24. N. Madamopoulos and N. A. Riza, “Photonic delay lines using polarization selective
holograms for optical signal routing,” in Advances in Optical Information Processing
VIII, D. R. Pape, Ed., SPIE Proc. 3388, No. 16, pp. 120-128, 1998.
25. N. Madamopoulos and N. A. Riza, “Polarization selective hologram-based photonic
delay lines”, Optics Communications, Vol. 37, No. 11, pp._-_, 1998
26. N. A. Riza and N. Madamopoulos, “Characterization of a ferroelectric liquid crystalbased time delay unit for phased array antennas,” IEEE/OSA Journal o f Lightwave
Technology, Vol. 15, pp. 1088-1094, 1997.
27. N. Madamopoulos and N. A. Riza, “Directly modulated semiconductor laser fed
photonic delay line using ferroelectric liquid crystals,” Applied Optics, Vol. 37, No. 8,
pp. 1407-1416, 1998.
28. F. Xu, J. Ford, and Y. Fainman, “Polarization-selective computer-generated
holograms: design, fabrication, and applications,” Applied Optics, Vol. 34, pp. 256366, 1995.
29.Y.-T. Huang, “Polarization-selective volume holograms: general design,” Applied
Optics, Vol. 33, No. 11, pp. 2115-2120, 1994.
30. J. J. Butler, M. S. Malcuit, and T. W. Stone, “Polarization-sensitive holograms
formed with use of DMP-128 photopolymer,” in Optoelectronic Interconnect,
Switching, Processing and Storage, CLEO’97 Proc.,
CWF53, pp. 257-258,
Baltimore, May 1997.
319
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31. Y. Hori, K. Asai, and M. Fukai, “Field controllable liquid-crystal phase grating,” IEEE
Transactions on Electronic Devices, Vol. ED-26, No. 11, pp. 1734-1737, 1979.
32. N. A. Riza and S. Yuan, “Robust packaging of photonic rf modules using ultrathin
adaptive optical interconnect devices,” in Optical Technology fo r Microwave
Applications VIII, Anastasios P. Goutzoulis, Ed., SPIE Proc. 3160, pp. 170-177,
1997.
33. N. A. Riza and S. Yuan, "Demonstration of a liquid crystal adaptive alignment tweeker
for high speed infrared band fiber-fed free-space systems," Optical Engineering, Vol.
37, No. 6, pp. 1876-1880, 1998.
34. N. A. Riza and M. C. DeJule, “Three-terminal adaptive nematic liquid crystal lens
device,” Optics Letters, Vol. 19, No. 14, pp. 1013-1015, 1994.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IM A G E E V A L U A T IO N
TEST TARGET (Q A -3 )
1.0
128
jso
Ui I m
B25
U 9B 1
1 2 .2
t Im III2.0
l.l
1.8
1.25
1.4
1.6
150mm
IIW 1G E . I n c
1653 East Main Street
Rochester. NY 14609 USA
Phone: 716/482-0300
Fax: 716/288-5989
G 1993. Applied Image. Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
13 029 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа