close

Вход

Забыли?

вход по аккаунту

?

Cosmic microwave background anisotropy measurements at the thirty arcminute scale

код для вставкиСкачать
INFORM ATION 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.
The quality of this reproduction is dependent upon the quality of the
copy submitted. 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.
UNIVERSITY OF CALIFORNIA
Santa Barbara
Cosmic Microwave Background Anisotropy Measurements at the Thirty Arcminute
Scale
A Dissertation submitted in partial satisfaction
of the requirements for the degree of
Doctor of Philosophy
in
Physics
by
Mark Alexander Lim
Committee in Charge:
Professor Philip M. Lubin, Chairman
Professor Robert Antonucci
Professor James Hartle
December 13, 1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9718599
UMI Microform 9718599
Copyright 1997, by UMI Company. Ail 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.
The dissertation of Mark Alexander Lim is approved:
Date
Date
Chair
Date
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
December 13, 1996
© Copyright by
Mark Alexander Lim
1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dedication
To Waiman and Shen
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgments
All the work that was accomplished would not have been possible without
all the help from others. Tim Koch taught me something about electronics and
worked hand-in-hand with me on some the most detailed ACME observations ever.
Mike Seiffert helped me understand some the nuts and bolts of ACME and its point­
ing system. Josh Gundersen explained and debated the arcana of CMB analysis.
John Staren has been a long-suffering comrade-in-the-field from Texas to Antarctica.
Jeff Childers’ daily enthusiasm has been inspirational. Mark Devlin was a great
help during MAX5 and contributed drive and sympathy. In particular some deserve
special mention. Phil Lubin provided an environment for a student to get into an
experiment and take charge. Stacy Tanaka was a great collaborator and co-worker,
who made MAX5 far smoother than I could have hoped. Finally, Peter Meinhold
truly demonstrated to me what it takes to be scientist with integrity, passion and
compassion.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V ita
Bom December 5, 1966, Oakland, California
June 15. 1989 S.B. Physics-Massachusetts Institute of Technology
October. 1989-August. 1990 Research Assistant, Astrophysics Department,
Nagoya University, Japan
June. 1991 - June. 1996 Research Assistant, Physics Department, Univer­
sity of California at Santa Barbara
September 1990-June. 1991 Teaching Assistant, Physics Department, Uni­
versity of California at Santa Barbara
Selected Publications
1. Degree Scale Anisogropy in the Cosmic Microwave Background:SP94 Re­
sults, J. Gundersen, M. Lim, J. Staren, C. A. Wuensche, N. Figueiredo, T. Gaier, T.
Koch, P. Meinhold, M. Seiffert, G. Cook, A. Segale, P. Lubin Astrophysical Journal
Letters,443:L57-L60,1995
2. Near-Infrared Spectrometer on the Infrared Telescope in Space. Manabu
Noda, Toshio Matsumoto, Shuji Matsuura, Kunio Noguchi, Masahiro Tanaka,Mark
Lim, Hiroshi Murakami. Astrophysical Journal Letters,428,363,132-F5(1995)
3. Measurements o f Anisotropy in the Cosmic Microwave Background Ra­
diation at 0.5 Degree Angular Scales Near the Star Gamma Ursae Minoris M.J.
Devlin, A.C. Clapp, J. Gundersen, C.A. Hagmann, V. Hristov, A. Lange, M. Lim, P.
Lubin, P. Mauskopf, P. Meinhold, P.L. Richards, G. Smoot, S. Tanaka, P. Timbie,
C. Wuensche Astrophysical Journal Letters.430,Ll,165-Bl (1994)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
Cosmic Microwave Background Anisotropy Measurements at the Thirty Arcminute
Scale
by
Mark Alexander Lim
Anisotropy of the Cosmic Microwave Background (CMB) Radiation is one
of the few direct probes of the early universe. Through measurements of the angular
power spectrum of temperature fluctuations we can constrain models of early struc­
ture formation. For example, one can gain useful limits on such cosmological parame­
ters as fi0, Oft, and H0. One can also determine the overall mix of non-luminous mat­
ter; that is whether it is baryonic or non-baryonic, relativistic or not. The fifth flight
of the Microwave Anisotropy eXperiment (MAX) obtained new data in two new re­
gions (HR5127 and Phi Herculis) and one previously observed (Mu Pegasi). Near the
stars HR5127 and Phi Herculis we measured A T/Tanb — ( Cf ar )* =
* 10-5
and 2.2^08 x 10-5 respectively, (error bars reflect 95% confidence interval) Data
taken near Mu Pegasi yielded an upper limit, AT/Tent, < 1.3 x 10-5 (95% confidence
upper limit). These results are consistent with previous observations by MAX. In
particular, the 1996 Mu Pegasi results are consistent w ith the 1993 results.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Frontispiece.
Some of the participants in the MAX5 campaign. In the front row, (1 to r) Josh
Gundersen, Phil Lubin, Newton Figueredo, and John Staren. Standing, (I to r)
Andrew Lange, Shaul Hanany, Peter Meinhold, Stacy Tanaka, Mark Lim, and Mark
Devlin. In the background is Tiny Tim, the main NSBF launch vehicle, holding
ACME just before launch.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C o n ten ts
List o f F ig u re s
xi
List o f T ables
xii
1
S u m m a ry
2
S cientific B ac k g ro u n d
2
2.1 Standard M odel.............................................................................................
2
2.1.1 Hubble E x p a n s io n ...........................................................................
3
2.1.2 Big Bang N ucleosynthesis............................................................... 4
2.1.3 Cosmic Microwave Background R a d ia tio n ...................................
5
2.2 CMBR A n iso tro p y .......................................................................................
6
2.3 Results to D a t e ................................................ " ........................................... 9
2.3.1 Large Scales..........................................................
10
2.3.2 Medium S c a le s ................................................................................. 11
2.3.3 Small Scales....................................................................................... 12
3
E x p e rim e n ta l H a rd w a re for M A X
14
3.1 A C M E ............................................................................................................ 14
3.2 D e te c to r......................................................................................................... 17
3.3 C a lib ra tio n ................................................................................................... 18
4
M A X 5 C am p aig n
24
4.1 Beam P a t t e r n s .................................................................................................24
4.2 Flight P la n n in g .................................................................................................27
4.3 S ynopsis.............................................................................................................28
5
D a ta A nalysis
34
5.1 D ata R eduction.................................................................................................34
5.2 Time Stream A n a ly s is .................................................................................... 35
1
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
x
CONTENTS
5.3
5.4
5.5
5.6
5.7
Systematic E rrors......................................................................................... 36
Foregrounds.................................................................................................. 37
5.4.1 Single Source M odels........................................................................ 37
5.4.2 Point Sources..................................................................................... 38
5.4.3 Interstellar Dust in MuPegasi ....................................................... 39
5.4.4 3.5 cm -1 Amplitude in MuP e g a s i ................................................ 39
Bayesian Methods ...................................................................................... 40
Accounting for F oregrounds.......................................................................... 41
5.6.1 M arginalization.................................................................................. 42
5.6.2 Direct S u b tra c tio n ........................................................................... 43
R esults............................................................................................................ 44
5.7.1 MAX5 Mu Pegasi ........................................................................... 44
5.7.2 MAX5 H R 5127................................................................................. 44
5.7.3 MAX5 Phi Herculis ............................................................................ 45
5.7.4 MAX3 Mu Pegasi R ea n a ly sis......................................................... 45
6
C onclusion
7
F u tu re D ire c tio n s
7.1 Large Scale Surveys
52
55
................................................................................... 55
B ib liography
58
A A C E D ew ar
65
A .l R equirem ents............................................................................................... 65
A.2 D esig n ............................................................................................................ 66
A.2.1 M ech an ical........................................................................................ 66
A.2.2 T h e rm a l.............................................................................................. 66
B
LD B D ew ar T h e rm a l C alcu latio n s
69
B .l Radiative T ra n s fe r...................................................................................... 69
B.2 Thermal A ccounting................................................................................... 71
C
IR A S Low D u st R egions
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
L ist o f F igu res
2.1
CMB Experimental R e s u lts .......................................................................
3.1
3.2
3.3
3.4
3.5
3.6
Servo Gain Map ......................................................................................... IT
ACME g o n d o la ............................................................................................ 18
C a lib ra to r..........................................................................................................20
ARU d r i f t ......................................................................................................... 21
HR5127 S c a n ................................................................................................... 22
ACME Flight R e a d y ....................................................................................... 23
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
Radar R a n g e ................................................................................................... 25
Main L obe..........................................................................................................26
Far Side L o b e ................................................................................................ 27
Flight P l a n .................................................................................................. 29
Sky M a p ......................................................................................................... 30
Jupiter S can.................................................................................................. 31
Jupiter Scan in E le v a tio n .......................................................................... 32
Scan Sim ulation................................................................................................ 33
5.1
5.2
5.3
5.4
5.5
5.6
Foreground S pectra.......................................................................................... 46
HR5127 S p e c tr u m ...................................................................................... 47
Phi Herculis S p e c tru m ................................................................................ 48
HR5127 D a t a ................................................................................................ 49
Phi Herculis Data ...................................................................................... 50
Mu Pegasi D a ta ............................................................................................. 51
A. 1 Schematic of LDB D e w a r ..........................................................................
A.2 Heat Flow D iagram ......................................................................................
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
67
68
L ist o f T ables
5.1
5.2
Dust Spectral F i t s .......................................................................................... 39
Two Component F i t s ....................................................................................... 44
6.1
MAX R e s u lts ....................................................................................................54
B. 1
Dewar Heat Load Spreadsheet........................................................................71
C .l
IRAS Low Dust Constrast R eg io n s.............................................................. 75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p te r 1
S u m m a ry
This dissertation is organized as follows: In chapter 1, we begin with a brief
description of the Standard Model in cosmology. In particular, we discuss the CMB
radiation and causes of anisotropy at 30 ' scales. This chapter includes a reiteration
of results to date at all angular scales. Chapter 2 describes the hardware used in
the MAX experiment. The details have been given in previous dissertations, so we
simply outline the function of each component. Details particular to MAX 5 are
given Chapter 3. An important section in this chapter is the description of the
target region selection process. Data reduction and analysis is discussed in C hapter
4. There were three data sets taken during MAX5 and they each required slightly
different tacks to understand. The Mu Pegasi region was particularly challenging.
We conclude in chapter 5 with a summary of the MAX results over the past seven
years. Since MAX 5 represents the last of the low-sky coverage, one dimensional
scan flights for ACME, it is appropriate to mention what the community thinks are
needed in new experiments. We describe where efforts are heading to fulfill those
needs.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p ter 2
S cien tific B ack grou n d
2.1
Standard M odel
Modern, cosmology has adopted a standard model as a framework for more
exact theories. The universe we see today is believed to have come from a hot, dense
initial state that expanded and cooled. This model assumes general isotropy and
homogeneity with perturbations provided by some mechanism in the earliest epochs.
There are two common explanations for these mechanisms. In the inflationary sce­
nario, there is an initial stage of expansion which occured at an exponential rate [33]
and smoothed a quantum mechanically generated spectrum of perturbations in the
m atter and radiation fields into a overall geometrically flat region. The isocurvature
scenario takes a geometrically homogeneous universe and seeds it with perturbations
from say, phase transitions in the very early universe Whichever their origin, grav­
itational instability enhanced the intrinsic perturbations into the largest structures
we see today, and gas-dynamical evolution created objects less than or equal to the
size of clusters of galaxies [6]. The distribution and evolution of the density fluctu­
ations is influenced by its composition. Some of the mass of the universe may be
in relativistic or non-relativistic particles or even in a cosmologi ;al constant. It is
possible that most of the mass of the universe is non-baryonic and interacts solely
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2. SC IEN TIFIC BACKGROUND
3
through gravity [4]. This picture of a cooling, expanding universe w ith structure
amplified gravitationally is commonly called the standard modeL
The standard model is supported by three areas of observation: the present
mix of light elements (N < 8 ) relative to H, the overall expansion of the universe,
and the presence of a pervasive bath of radiation, the Cosmic Microwave Background
(CMB) Radiation. It should be noted that the standard model may require revision
[2] and that the canonical model alone does not match our observations of the uni­
verse. [13, 2]
2.1.1
H ubble E xp an sion
In 1929 Edwin Hubble observed that distant galaxies have a redshift pro­
portional to their distance. Since the redshift infers a recession velocity, this implied
there was an earlier hot, dense stage. Isotropic expansion with a rate proportional
to distance is a natural consequence of a homogeneous, isotropic universe with kine­
matics governed by general relativity [38j. The constant of proportionality or the
Hubble constant is H 0 and has units km s_ 1Mpc_ l. There is presently much un­
certainty in t
H0 = lOOh km s- 1M pc-1 , where h ~ 0.5 —1.0. Measurements are done by observ­
ing a standard candle, such as la supemovae, globular clusters or Cepheid stars.
Cepheids are the candle of choice for calibrating secondary distance scale measures,
such as the Tully-Fisher relation and the Dn —£ relation. These secondary methods
can reach out to distances (> 100 Mpc) where peculiar velocities are small compared
to the Hubble flow. Presently, type la supemovae and Cepheids give inconsistent
values for the Hubble constant, h = 0.55 ± 0.07 and h = 0.88 ± 0.24, respectively
[41, 53].
he value [34, 53, 29, 66 ], and the Hubble constant is usually written,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H A P TE R 2. SCIENTIFIC BACKGROUND
2.1.2
4
B ig B a n g N u cleosyn thesis
When the universe was at a temperature T ^ 100 MeV, weakly interacting
particles were in thermal equilibrium through reactions such as n <— ►p + e~ + i/e
and ve + n *— ►p + e~. The nucleon number density ratio is
n„
, A me2,
, 1.5 x 1010f f ,
— = « p ( ~ — ) = exp(
------- )
rip
k^T
T
,
(2. 1)
where kb is Boltzmann’s constant, Am is the mass difference between th e nucleons.
^
« 1 at T = IQ12K . The nucleons are in thermal equilibrium. Deuterium is
created through n + p <— ►D + 7 but because the binding energy is 2.225 MeV,
it does not accumulate. When the universe had cooled to a tem perature of « 1
MeV, the nuclei of the light elements were able to form. When the nucleons fall
out of thermal equilibrium,
He4, the He4 mass fraction
density ratio above,
— 0.20. Assuming all the neutrons are bound into
4nn
o
Given the nucleon number
~ 0.30. Walker et al. (1991) find Xfjg* = 0.24. This
rough agreement with a simple estimate of the He4 mass fraction is an argument in
favor of the standard model. Observations of other elements, 3He, D, and 7Li at the
present epoch are all consistent with theoretical predictions for a narrow range of
possibilities [39] for the one relevant parameter, the photon-to-baryon ratio
is equivelant to the baryon density parameter,
This
since we know the temperature
of the CMB. The density of baryons is constrained to be 0.015 <
77
77.
< 0.026 if
= 4 —7x 10~10. In particular, because D is not created in present day astrophysical
processes, the observed abundance must be a lower limit to the primordial quantity.
Based on observations of D abundances, fifth2 < 0.037. This means that one cannot
have n o = flft = 1 and that some of the dark matter must be non-baryonic if the
universe is fiat.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 2. SCIENTIFIC BACKGROUND
2.1.3
5
C osm ic M icrow ave Background R ad iation
The existence of an almost (10 ppm) isotropic bath of black body radiation
is good evidence of the existence of an earlier hot, dense stage. The photons and
baryons were tightly coupled via Thompson scattering and able to come to thermal
equilibrium. Measuring the present value of the tem perature of the CMB radiation
has seen considerable effort in the past [48, 60, 35). COBE/FIRAS has determined
Tcmb to be 2.726 ± 0.010 K from 5000 to 500 pm [21]. Deviations from this temper­
ature appear as a departure from the Planck spectrum. Present efforts are directed
at finding the far infrared background from primordial galaxies or the low-frequency
10G H z) distortions from early energy sources. As the universe cooled, it reached
a point where the electrons can begin to combine with the protons. For historical
reasons, this is called "recombination”. The degree of ionization is described by the
fractional ionization X = rn 2b-. The universe is said to be recombined when X = 0.10.
Kolb and Turner (1990) give the Saha equation, the evolution of the fractional ion­
ization in equilibrium, as
-1- y / ' = K t j { — )3/2 exp(B /k T )
A
nig
(2.2)
where K is a constant of order unity, me is the mass of the election, B is the binding
energy of the hydrogen atom, 13.6 eV and k is Boltzmann’s constant. Large values of
fib shift the moment of recombination to earlier times by changing nb, but the epoch
typically occurs at z = 1300. To be precise, one must account for the interaction
between the hydrogen excited states and the plasma [5]. Since the CMB is 2.726 K
now, this means that the universe recombined at a tem perature of Tree — 1300 x
2.726K
= 0.31 eV. T hat this is much less than B is illustrated by the smallness
of t) and T /m e. Another way of putting it is that with the large number of photons
available per baryon, the photon temperature must drop well below 13.6 eV before
there are few ionizing photons.
Eventually, the plasma becomes rarefied such that the photon mean free
path is about the size of the Hubble length and the photons can then free stream to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
CH APTER 2. SCIENTIFIC BACKGROUND
the observer. T hat is
ii o
ilo
1
<u>
where T, the interaction rate is approximately ne times the Thompson scattering
cross section, o r . From the evolution of ne in Equation 2.2 we can determine the
time of decoupling.
This epoch is called decoupling, which refers to the fact th a t the photons
and baryons can no longer be treated as a single fluid. Most of the features of the
CMB radiation have been frozen in at this time, although now microphysical effects
can evolve the power spectrum as we will see below. Decoupling occurs at a z of
~ 1100 or when the universe had a temperature of ~ 3000#.
2.2
C M B R A nisotropy
Anisotropy experiments measure temperature variations in the direction x*
on the sky,
which can be expanded in spherical harmonics on the sky,
J-cmb
=
E E
1=2 m = - I
(2-4)
where the Y[m are the spherical harmonic functions, aim are random variables whose
distribution function is given by the particular cosmological theory and F/m is a
filter that depends on the particulars of the experiment. The two-point correlation
function is then written in terms of the spherical harmonics,
(A I W T W j = £
cmb
Iml'm'
(2 . 5 )
Due to rotational invariance [18], (a;maz*,m,) = CiSuiSmm’. Ci is the power spectrum
th at is defined by theory. In order to make comparisons between experiments it is
convenient to use a theory independent power spectrum, Ci oc1/1(1 + 1)or ’’flat band
power (FBP) spectrum”. The name comes from the fact that the FBP spectrum is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
CH APTER 2. SCIENTIFIC BACKGROUND
constant in the equal power per logarithmic frequency interval plot, 2k/(fjrxyj vs. Z,
often used in the literature.
For one dimensional scanning experiments, it is a good approximation that
Fim be independent of m. The two-point correlation function becomes
,A T {x i)A T ( x j) .
( ------------
and
»21 + 1
. ,9 _ .
- ) = J 2 ~ 7 Z ~ CllF «l
i
,
• Xf)
v—*2Z-F i
= 2 2 - — c iw i(0v )
i
4
(2-6)
where Pi are the Legendre polynomials, fly = cos(x* • Xj), and we have identified
the window function Wi = |F/| 2P/.
T he window function is determined either by numerically integrating a beam
pattern over a chop cycle, or constructing an analytical form based on fits to beam
patterns. There has been much work in this area [59, 64] and we adopt the analytical
form of the window functionfor MAX,
WiiQij) = JV2 exp(-i*(Z + l)o-2) SZ
4) cos( ( Z - 2r ) f l y ) (2.7)
where N2 = 1.11 for MAX5, a is 0.4248 times the Gaussian beam full width at half
maximum (FWHM) and fly is now the angle between centers of the ith and jth pixeL
We have defined
G2(a, <t>) = J?((Z - 2 r)a )i02((Z - 2r)0 )
(2.8)
where J\ and jo are the regular and spherical bessel functions respectively, <t>is the
pixel size, and a is half of the peak-to-peak chop angle. iV2 is the normalization of
the window function. The normalization depends on how one chops the beam and
calibrates the instrument. For MAX,
^ = ( ^ ) i F i ( i , 2 ; - 72)
where j =
(2.9)
and 1F 1 is the confluent hypergeometric function.
Despite the elaboration required to describe temperature anisotropy, all
the physics lies inthe C/’s. In general, as one progresses from larger to smaller
angular scales,the important processes take effect from earlier to later times. At an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 2. SCIENTIFIC BACKGROUND
8
angular scale ~ 1° there is an important boundary that corresponds to the size of the
horizon at decoupling. At this scale and less, dynamical effects become important
and produce both the highest fluctuations and most of the structure in the power
spectrum.
At the largest scales, the anisotropies are primarily from intrinsic fluctua­
tions in the curvature or the entropy of the universe. Differences in the gravitational
potential or the Sachs-Wolfe effect [52] will red or blue shift the CMBR photons
as they travel to the observer. In the first case, these are called adiabatic fluctua­
tions since the entropy is a constant. The distribution of these adiabatic fluctuations
should be Gaussian since they were formed from quantum fluctuations in the in­
flationary era. In the absence of a theory, it is assumed th a t there is no preferred
scale and that the spectrum should be a scale-invariant power law, or the HarrisonZel’dovich spectrum. The second case is called the isocurvature fluctuation. Here,
Spphotons — —Spmatter so the curvature of the metric is constant. Variations in the
entropy cause density perturbations to grow when they cross into the horizon. Phase
transitions in the early universe can cause isocurvature fluctuations, for example,
cosmic strings and monopoles.
At medium to small scales, a multitude of dynamical effects come into
play. The picture is that of a fluid with several components, baryons, photons,
neutrinos, plus some form of dark m atter, with different couplings through different
forces between the populations. Hu and Sugiyama (1995) simplify this picture by
noting that just before the era of decoupling, the photons and baryons are tightly
coupled through Thompson scattering, whereas the other particles interact primarily
(or solely) through gravity, and so the former can be treated as a single fluid.
Perturbations to the density create perturbations in the potential which
cause the fluid to oscillate If the period of the oscillation is faster than the time
it takes sound to cross the perturbation,-the perturbation will grow. Furthermore,
the surface of last scattering in the oscillation will Doppler shift the radiation during
decoupling.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 2. SCIENTIFIC BACKGROUND
9
After decoupling, photons must travel through a varying potential to reach
the observer. Because causal effects evolve the fluid and hence the potential, the
Sachs-Wolfe effect must be integrated along the line of sight. The exact effect depends
on the mix of material components and the type of intrinsic fluctuations.
Additionally, there are two geometric effects th at affect the power spectrum.
One is the overall geometry of the universe. As the subtended angle of an object is
proportional to Qa, if the universe is open, features in the power spectrum will be
shifted to higher I. The second is the projection of high frequency oscillations on
the celestial sphere, which shifts them to lower frequencies. Thus power is shifted to
lower I for all models.
At the finest angular scales, (f > 500) the peaks are greatly reduced by
photon diffusion or Silk damping [57]. As the fluid progresses toward decoupling,
the photons leak out of the baryon overdensities. Smoothing of CMB features comes
from the photon diffusion and to some extent the drag on the baryons. The baryon
component does not diffuse since a baryon’s mean free path is a factor of tj smaller
than that of a photon.
2.3
R esu lts to D a te
It is conventional to classify experiments by the angular scales they are
optimized to m easure Mainly this is because they measure different physics and
also because they tend to use similar procedures and equipment for a given scale.
Large scale experiments cover a large portion of the sky and have low resolution.
Medium scale experiments were usually designed to probe an angular scale ~ 1° and
have small sky coverage. Small scale experiments are probe angular scales < 1°.
Results are often given in the amplitude seen in the window function for a
FBP spectrum also called the flat band power estimate (FBPE). This is considered
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
C H APTER 2. SCIENTIFIC BACKGROUND
preferable since it is model independent. We define the FBPE,
AT_ 2
=
Tomb
C[ 1(1+1) =
K
2tt
'
{dT/Tcmb)~
.
.
E S 2(l + 5 ) ^ / W + D)
where the rms temperature variation,
(2 . 1D
the average window function,
w* = J - f > z(0 = o)
Mpix l=o
(212)
where Npix is the number of pixels and
/=
LE& m .
(2.13)
In the past (> 5 years), results were reported with respect to a Gaussian
Autocorrelation Function (GACF), where
(A
r (x j)> = Coexp(_ ^ . )
TZmb
(214)
2<t>Z
where <t> is the angle between x* and Xi, 4>c is the correlation angle (an arbitrary
parameter), and C qis the amplitude of the fluctuations.
This had the disadvantage
of requiring an extra parameter and also of not being applicable to non-Gaussian
models. Results in the text will be for a flat band power spectrum unless noted
otherwise.
2.3.1
Large Scales
CMB anisotropy finer than the dipole was first detected with the DMR
on COBE after the first year of operation [58]. This was verified by FIRS soon
thereafter, through correlation with the FIRS 170 GHz band [31]. The DMR is a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 2. SCIENTIFIC BACKGROUND
11
differential radiometer with angular resolution of 7° and three channels at 31.5, 53
and 90 GHz. FIRS is a single horn bolometer-based experiment referenced to an
interned load. The magnitude of the quadrupole, Q, is measured to be 15.3*0.8/^.
COBE can probe the angular power spectrum down to I = 40 and reports results
consistent with a scale-invariant Harrison-Zerdovich spectrum.
2.3.2
M edium S cales
Cold Dark M atter theories predict most power to be at an angular scale of
% 30' or I « 200. Thus, a lot of the experimental effort of the past few years has con­
centrated in that area. Ground-based microwave telescopes have been enormously
successful; for example, the UCSB South Pole expeditions[30, 54, 22], recent Python
results [51] and the Saskatoon experiment [46]. Except for Python, which is a 4 pixel
90 GHz bolometric photometer, these are HEMT based direct amplification receivers
operating at 30-40 GHz with meter class primary mirrors. They all have comparable
beam sizes, 0.75 —1.0°, with the exact FWHM being frequency dependent. They all
single diffence chop, and add on a double difference to account for atmospheric gradi­
ents, with the exception of the UCSB South Pole experiment which removes gradients
in software. The latest UCSB South Pole result is A T/Tcmb = 1.3 3 ^ 7 3 x 10-5
(68% c.l.). Ruhl et ad. (1995) report A T /T c o , = 2.1 ^ 5 x 10-5 for Python’s 1993
season. The latest out of Saskatoon is perhaps the most interesting. By synthesizing
beam sizes, they probe the angular power spectrum from I = 60 to I = 404. They
see a rise from I = 60 to I = 300 with amplitv.de of 80/xK and a drop at I = 350.
However, the error bars are too wide to definitively say th a t the first Doppler peak
has been seen. See Figure 2.1.
The higher frequencies afforded by bolometers enables us to probe down
to 30' w ith meter class telescopes. There are two successful balloon borne exper­
iments which are comparable in design. MAX has flown five times and observed
varied regions of sky [1, 24, 23, 25, 11, 14, 61, 43]. Tanaka et al. (1996) report
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 2. SCIENTIFIC BACKGROUND
A T/Tanb =
12
l-2 ^ol3 x I ®-5 (68% c.1.) in the region near HR5127. An in depth
discussion of MAX results can be found in the Conclusion. MSAM is also a multi­
channel balloon-bome bolometer. While it can double difference, the scan and fre­
quency coverage are less optimal than MAX. Recently, Cheng et aL (1996) confirm
their first measurement with A T/Tanb = 1.291 q;7 x 10-5 (90% c.l.).
2.3.3
Sm all Scales
While the angular power spectrum is expected to drop at high resolution
this region of the spectrum will ultimately be the most important since the loca­
tion and height of the higher order doppler peaks break the model degeneracy in
the parameters of the first peak.
The Cambridge Cosmic Anisotropy Telescope,
a three element microwave interferometer, has observed anisotropy at the level of
A T / T = 2 .0 ^ 4 x 10—5[55]. OVRO is a classic double differenced two-hom Dicke
switch with a MASER amplifier operating at 20 GHz. They report an upper limit
1.7 x 10" 5 [49].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
CHAPTER 2. SCIENTIFIC BACKGROUND
10
A
CAT
O
OVRO
A
II MAX5
0"
10
100
1000
Multipole I
Figure 2.1: The line is a plot of a CDM model with h = 0.7, n = 1.0 which has been
normalteed to COBE/DMR [58]. Various experimental FBPE have been superim­
posed. SP results are plotted with the open squares[30, 54]. MAX results are plotted
with solid squares. Flights are bunched in the same I bracket. MAX4 is leftmost
[11, 14] and MAX5 is rightmost[61, 43]. Mx3MP and Mx3G are MAX3 Mu Pegasi
and Gamma Ursae Minoris results respectively[25, 23]. The MSAM92 results [8] are
plotted with an open circle. The three beam chop is on the right, the two beam chop
on the left. CAT is represented with the upward pointing open triangle[55]. OVRO
is shown with the solid triangle[49]. The latest Saskatoon results (SK96) are plotted
with the stars [46]. The downward pointing open large triangles represent ACME
results averaged at a particular angular scale.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p ter 3
E x p erim en ta l H ardw are for
M AX
MAX is a balloon borne experiment which consists of a pointing platform
with a 1 M telescope provided by UC Santa Barbara and a 4-channel bolometer
detector provided by UC Berkeley. The platform has been described extensively
elsewhere, [45, 10, 44] as has the detector [12], so only an overview for completeness
will be given.
3.1
ACM E
The Advance Cosmic Microwave Explorer (ACME) is a 1 M off-axis Gre­
gorian telescope mounted on a pointing platform capable of 1 arcminute rms error.
Chinguanco et al. (1991) [9] and Meinhold et al. (1993) [44] discuss the instru­
ment in detail. Absolute offsets are typically 6 arcminutes with a drift of 1 degree
per hour. Orientation is measured by the A ttitude Reference Unit (ARU), which
is a military surplus gyroscope system. This pointing information is provided to
the onboard computer which controls the servos. The azimuth is regulated by a
Proportional-Integral-Differential (PID) control loop while the elevation servo uses a
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 3. EXPERIM ENTAL HARD W ARE FOR M A X
15
Proportional-Integral loop. Proper tuning of the loops is necessary for efficient oper­
ation. To place this in context, we note, that a control system’s signal (for example,
a motor’s control voltage)
u(f) = K pe(t) +
+ K i j e(t)dt
(3.1)
where K p, K d, K i are the proportional, differential and integral gains, respectively,
and e(t) is the error signal. The error signal, e(f) = x(t) —x 0(f) is the amount the
system x (t) (e.g. the position), is away from the set point or desired value, x 0(f).
(Note that x 0 is a function of tim e) Essentially, if e(f) is big, the controller must
work harder to restore the system to x0. If the system is rapidly moving toward or
away from x Q, the controller must work harder in anticipation of future deviations.
The effect of integral gain (K i) is to minimize the DC offset in the error signal.
Figure 3.1 is a map of the effects of changing gain values. The cross hairs in the
upper figure represents a well-tuned position for ACME weighing ~ 1300 lbs with no
baffle or roll bars. The arrows are labeled with the effect of changing the gain and
point in the direction of increasing effect. The units are arbitrary 1 but the values
are those used in the servo code directly.
In general, the error signal from a step function set point can be described
as a damped oscillator.
e(t) = exp(—t/r ) sin(wf)
(3.2)
Heavy damping causes the error signal to look like a heavily damped oscillator (r <C
l/ui). Regular damping allows the servo to make several oscillations and gives an
error signal like that shown in the lower diagram of Figure 3.1 ( r « few x l/ui).
The ringing length is 2tc/ u>. The region below the horizontal line in Figure 3.1 is
where l / u < r . Overshoot height is the amplitude of the initial swing of the system
as it adjusts to the set point. Increasing stability makes the system generally more
lThe units are not completely arbitrary. They are normalized to produce volts after using Equa­
tion 3.1 with the calculus done over a time window where the unit of time is an integer number of
interrupt cycles in the servo computer. Typically this is six cycles with the interrupts clocked at
18.3 Hz.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3. EXPERIM ENTAL H ARDW ARE FOR M AX
16
sluggish but lowering it too far will cause ACME to oscillate or even deviate away
from the the set point.
We consider the servo to be well tuned if it moves to the set point accurately
and quickly. By "accurately” we mean that e(f)
the beam size. "Quickly” means
th at the previous condition is true for > 90% of the time in a scan cycle. In tuning
the system, one finds that stability and responsivity are competing requirements
and the optimum values depend on the scan type used. Experience has shown th at
smooth scans have efficiency > 95%, whereas step scans can achieve « 90% with
scan amplitudes of a few degrees.
A good reference for PID loops is Kuo (1991)[40], although the particulars
for ACME are described in Chinguanco (1991) [9].
The ARU is susceptible to drift in its absolute pointing as can be seen
in Figure 3.4. This is removed by the operator by referring a CCD camera image
of a star or star field to the desired image. Actual performance during a seem, is
illustrated in Figure 3.5. The backup system is a magnetometer and angle resolver,
which provide azimuth and elevation information respectively. Both backups assume
that the gondola is oriented vertically when making measurements, hence rocking or
pendulation is not accounted for in the servo system, although it is detectable in the
CCD camera.
ACME has two CCD cameras, although only one is used at a time. One
is equipped with a zoom lens capable of changing the field of view from 2° x 3° to
11° x 16°. It is also light intensified and at maximum zoom can see down to visual
magnitude M v = 8 on the ground. It does about two orders of magnitude b etter
at altitude. The older backup camera has a field of view approximately 2° x 3° and
can see down to Mv = 6 on the ground.
The optics are designed to m in im ize sidelobe contamination by eliminating
obstructions to the main beam. Figure 3.2 illustrates the optical path.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3. EXPERIM ENTAL H ARD W ARE FOR M A X
17
>.
i55
3600
3400
£
Damping
Heavy Damping
3200
s
3000 -
I<s
2600 _
I3"
2600
500
-•-------- 1--------■-------- 1-------- ------1------- 4-------- 1-600
700
800
900
a ----- 1
1000
1.0
Ring Length/Damping
-0.5
0
20
40
60
80
100
T im e (Arbitrary Units)
Figure 3.1: The upper figure is a map of effects of varying proportional (K p) and
differential (K j) gains. The units are arbitrary and set for the ACME servo code.
The lower figure is of a damped oscillator to illustrate the effects as they appear in
the servo error signal. The figure is described in detail in the text.
3.2
D etector
The MAX photometer is a single pixel, bolometer based system with three
refrigeration stages. Passive cooling is provided by LNo and L4He, the latter of
which is pumped down to 1.7 K. A 4He refrigerator cools the second stage. Finally,
an adiabatic demagnetization refrigerator cools the four-band photometer to 85 mK.
Successive dichroic filters split off the bands from highest to lowest frequency. The
four frequency bands are centered at 3.5, 6, 9, and 14 cm -1 with respective fractional
bandwidths 0.5, 0.5, 0.4, and 0.2. Laboratory measurements of the sensitivity are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3. EXPERIM ENTAL H ARD W ARE FOR M A X
18
.o .
Figure 3.2: Sketch of the ACME gondola showing some the relevant parts. A)
Azimuth control system. B) Dewar. C) The chopping secondary. D) Inner Frame
and pivot point. E) 1M primary. The rays are approcimately the 3db surfaces of the
beam.
409, 193, 126, 148 /J.K/V H z (Rayleigh-Jeans). The most complete discussion of the
dewar is found in Clapp (1994) [12).
3.3
Calibration
The MAX photometer includes an external calibrator which is used to mea­
sure the long term drift of the output during flight. It consists of a thin film reflector
and black body load. The geometry is illustrated in Figure 3.3. W ith the reflector
in one of the chopped beams, the signal in volts is V = SfiightR^Tload ~ T3ky), where
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
19
CHAPTER 3. EXPERIM ENTAL H ARD W ARE FOR M A X
Sfught is the responsivity of the detector in volts/Kelvin, R is the membrane reflectiv­
ity Tioad is the black body load temperature and Tsky is the sky temperature. Before
flight we measure the reflectivity in two tests. First, a LN2 black body load is placed
into the optics so that it intercepts both beams. The membrane is inserted into one
beam, so now the signal is V\
= S\R{Tioad —TTK).
Then, the membrane is removed
and a water bath black body load is placed into the beams. The load is divided into
two regions, like a theta function sky, w ith one half at ambient temperature (300K)
and the other at 273K. Therefore the calibration is such th at a chopped beam cen­
tered between sky regions with temperatures T\ and T2 would yield AT = T\ —T2 in
the absence of instrumental noise The signal is Vo = So(300lif —273K). If Si = So
we would be done But because the optical loads are different for the two tests, this
is not true. In practice, it’s not the optical responsivities that are used at all. We
take the ratio of the two equations
Vi = SiRjTiogd - 77K)
Vo
S2{300K - 27ZK)
^’ '
and then assume that the ratio S 1 / S 2 is the same for electrical as for optical loading.
Then one can calculate R. We assume R is the same in flight as in the lab and use
the membrane as a transfer standard from the calibrations on the ground to those
in flight. In order to confirm our calibration we observe a known source in the sky,
usually Jupiter. Other candidates that have been considered are Saturn and Venus.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 3. EXPERIM ENTAL H ARD W ARE FOR M A X
20
A n b ie n t
C a lib r a to r
S eco n d a ry
M irro r
MM
Tl
vwv
AAM
T2
Figure 3.3: Schematic of optical path for the calibration tests. In the LN2 bath,
T l = T2 = 77 K. In the water bath test, the membrane is removed and T l = 300 K
and T2 = 273 K.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3. EXPERIM ENTAL HARDW ARE FOR M AX
f
-
0.10
-
0.11
21
-0-12
q„> -0-13
f
-0-14
=5 -0.15
cc
-0.16
•0.17
0.0
5.0x1
1.0x10*
1.5x10*
2.0x10*
2.5x10*
3.0x10*
2.0x10*
2.5x10*
3.0x10*
2.0x10*
2.5x10*
3.0x10*
Time(sec)
25.0
24.5
o>
<D
9 . 24.0
1> 23.5
£
23.0
I
22.5
<
22.0
0.0
5.0X103
1.0x10*
1.5x10*
Time (Sec)
-0.05
j? -0.06
_©
O
) •0.07
c
<
c
o
■0.08
5 -0.09
in
0.10
■
0.0
5.0X103
1.0x10*
1.5x10*
Time (Sec)
Figure 3.4: The drift of a stationary ARU (SN#1033) in roll, azimuth and elevation.
The ARU was aligned twice and left to sit on a bench for the duration of the test.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3. EXPERIM EN TAL HARDWARE FOR M AX
22
0.40
035
I
030
|
035
I 020
0.15
0.10
-6
-4
-2
0
2
4
6
Azimuth Error (deg)
Figure 3.5: Azimuth error vs. elevation error for the HR5127 scan. The fiducial star
is located at azimuth error = 0 and elevation error = 0.2. The non-zero elevation
offset is due to a software error in the code. The vertical lines at the center are from
operator updates. The vertical spread is 0.1 °, which is less than half the size of the
angular bins used in the analysis. The spread is due to pendulation of the gondola.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3. EXPERIM ENTAL H ARDW ARE FOR M A X
23
Figure 3.6: The ACME gondola ready to fly. The nutating secondary mirror is inside
the snout-like baffle. This baffle is designed to reject reflections to the sky. The main
baffle protects the primary mirror from direct illumination by the earth and other
sources of stray light. It is tapered at the edges to m inim ize diffraction. Inside the
main baffle one can see the non-intesiffed CCD camera. The intensified camera is on
the other side of the primary mirror. The saddle bag on the right contain batteries.
The suitcase strapped on top is the MAX dewar electronics box. The left set of
saddle bags contain ACME electronics and the servo computer.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p ter 4
M A X 5 C am p aign
The MAX5 Campaign, was conducted at the National Scientific Balloon
Facility in Palestine, Texas. We shipped ACME by truck on April 24, 1994, and met
our Berkeley collaborators in Texas. The experiment was ready in a record 4 weeks.
However, due to weather and position in the flight queue, we did not fly until June
19, 1994.
4.1
Beam P atterns
Before flight, we focused the optics and obtained beam patterns of the main
lobe and deep side lobes. In MAX4, the optics were not optimized, having a FWHM
of 0.75 ° in all channels except the 3.5 cm-1, which had a FWHM of 0.55 0 [11, 14].
This time we made a special effort to ensure the optics were correct. We used the
east bay of the new staging building and the nearby water tower as a radar range.
The arrangement is illustrated in Figure 4.1.
We used two sources for beam pattern testing. The first was an Impat
coherent source at 90 GHz. A double or tripler could be attached to increase the
frequency to 180 or 270 GHz. The second was a thermal source borrowed from the
MSAM group. This was simply an oven with variable aperture stops. A lens could
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 4. M A X5 CAMPAIGN
25
S o urce
Hangar
W a t e r Tower
20 deg
ACME
190 m
Figure 4.1: A sketch of the radar range used for beam patterns.
be added to increase the intensity.
We use beam patterns to check the focus of the system. We want the beam
to be symmetric in all frequencies with a width at the target 30 ' FWHM. It should be
well described by a Gaussian function. Additionally, we want to measure the off-axis
response of the whole telescope to put an upper limit on side-lobe contamination.
Main lobes were checked by doing elevation scans. Reflections from the
structure of the tower and surrounding terrain prevented doing passes in azimuth.
Typical main lobe patterns are given in Figure 4.2. We measured the main lobe with
the mirror at its neutral position and canted at the extreme position of its chop. This
did not affect the size of the beam. We used both the Im pat and the thermal source
for main lobe patterns, but chose the focus based on the thermal source. This was
because we could measure all bands simultaneously without worrying about single
mode effects in the 3.5 cm -1 band that the coherent Impat might create. The final
ground based main lobe sizes were 0.42 0 in the 3.5 cm -1 and 0.52 0 in the 6, 9 and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4. M AX5 CAMPAIGN
26
(D
CO
55
(D
GC
E
I
1
■(3
BevationPngle (Deg)
Figure 4.2: Main lobe measured in elevation for the 3.5 cm 1 channel using the
thermal source.
14 cm -1 bands.
In practice, we are only able to measure the far side lobes by starting
with ACME pointed below the source, passing through the main lobe and scanning
higher in elevation until we hit the noise floor. In order to measure deep enough, we
locked-in on the Impat at 90 GHz. Changes in the beam shape were less important
than the sidelobe level Irregularities in the surrounds prevent us from making a
similar measurement in azimuth. Figure 4.3 illustrates the far side lobe pattern.
The unchopped off-axis response in the 3.5 cm -1 band is > 70 dB below the on-axis
response at angles from 15° to 25° in elevation above the boresight.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 4. M AX5 CAMPAIGN
27
10P
iff1
iff6
20
25
30
36
45
50
AncJe(Deg)
Figure 4.3: This is the far side lobe plot for the 3.5 cm 1 band. This is actually
several files taken with different gains and then sewn together.
4.2
Flight P lanning
Except for the Mu Pegasi scan, we selected regions of the sky expected
to be free of foreground contamination at our sensitivity level. This is folded with
the time constraints imposed by sky rotation and the launch and termination tim e
First, patches of low dust contamination were selected on the basis of their ISSA
map rms intensity. The list of regions is given in Appendix C. Within those maps
we selected pointer stars to meet one of two criteria: Was the pointer easily visible
to the intensified camera (Did it have a visual magnitude Mv < 6 ) and did it have
companions from which one could determine orientation? Was the pointer easily
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 4. M A X 5 CAMPAIGN
28
visible in the non-intensified CCD camera (Afv < 4)? The dimmer stars were chosen
for optimal position with respect to the inter-stellar dust (ISD). The brighter stars
were used as backups in case the intesified camera failed and we had to use the weaker,
old camera. We generated scan trajectories for different times and superimposed
them on the ISSA maps along with potentially visible radio point sources. We could
then determine a tim e window for that target. There is also a window defined by the
elevation range of ACME and proximity to the horizon, which limits the field-of-view
to 25°-55°. The logical AND of the windows defines the final window of availability.
Furthermore, the total time at float begins at launch time plus approximately two
hours to either dawn at float (which equals dawn at surface minus «30 min) or when
telemetry range is reached (300 miles). We hoped for eight hours or so at altitude.
We prioritized targets based on the science we would get out of them.
CMBR
integrations were given top priority. The Mu Pegasus revisit was next, followed by
the Coma Cluster, on which we hoped to do a Sunyaev-Zel’dovich effect measurement.
The flight plan we followed is given in Figure 4.4.
4.3
Synopsis
The instrument was launched from the National Scientific Balloon Facility
in Palestine, Texas a t 1.16 UT 1994 June 20. We observed CMBR anisotropies in
three sky regions near the stars HR5127 (a = 13tl37?12, 6 = 24?34'), Phi Herculis
(a = I 6 h8“ 6 , 6 = 44?57'), and Mu Pegasi (a = 22h49?7, 6 = 24?34'). Their positions
on the sky are shown in Figure 4.5.
We calibrated the instrument before and after the observation using a mem­
brane transfer standard [24]. We observed Jupiter from 4.86 UT to 4.95 UT to mea­
sure the beam size and position and to confirm the membrane calibration. Using
the best-fit beam size and the membrane calibration, the derived temperature of
Jupiter agrees with Griffin et aL (1986) to within 10%. We therefore assume a 10%
uncertainty in calibration. Azimuth cut beam patterns are shown in Figure 4.6. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H A P TE R 4. M AX5 CAMPAIGN
29
PhiHac
HR5127
F
os?
0
2
4
6
8
10
12
UT (Hours)
Figure 4.4: The bands show when th e object was available at float. The triangles
indicate when events actually occured.
shape is consistent with a Gaussian beam with sinusoidal chop and sinusoidal lockin.
The beam width is weakly dependendent upon the channel: 0.49, 0.52, 0.53, and
0.54 0 in the 3.5, 6 , 9, 14 cm -1 bands respectively. There is some squint evident as
shown in Figure 4.7.
We observe with a constant velocity scan in azimuth of ±4° relative to the
pointing star. The left-hand lobe of the antenna pattern was coaligned with the field
of view of our CCD cameras and centered on the fiducial star. Gyroscope drift was
taken out every 400 seconds. The relative offset between the center of the chop and
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
CHAPTER 4. M AX5 CAM PAIGN
•30
6 .0.
1. 20 -
Figure 4.5: Positions of the MAX scans in an oblique cylindrical equatorial pro­
jection. The horizontal centerline is approximately the horizon at the surface in
Palestine Texas. The sky rotation is arbitrary. Superimposed for reference is the
galaxy at 408 MHz. The contours are in mK. The abbreviated names are: PH = Phi
Herculis, SH = Sigma Herculis, GUM = Gamma Ursae Minoris, ID = Iota Draconis,
MP = Mu Pegasi
the target star was 0?55 in azimuth.
During the Mu Pegasi scan (7.22 UT to 7.76 UT) the gyro malfunctioned
and moved the scan center with a trajectory tilted I0±1.5 degrees from horizontal
That is to say, Mu Pegasi was still the center of the scan, but the gyro horizon
was tilted with respect to the actual horizon. See Figure 4.8. The orientation of
the gondola was still vertical. We verified the orientation and trajectory with the
positions of stars in the CCD camera field of view. We did not observe in the same
orientation as in MAX3 and we do not expect the morphology to be identical. The
other observations displayed no significant tilt.
The Phi Herculis scan was terminated with a ballast destruct. Because
our altitude was slowly dropping from 115 kfeet to 105 kfeet over the course of the
flight, NSBF ballasted between each scan. In an effort to keep us from dropping
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4. M AX5 CAMPAIGN
31
Jupiter Beam Patterns
Iz
1.0
-
0.5
-
ec
d
•H
H
(A
•o
®
-
1 .0
eE
- 1.0 -
-
N
-2
1 .0
-
0.5
-
I
0
I
-2
1. 0
•I
0
I
1
0
1
-
u
o
2
0.0
-0 .5 -
-1.0 -
- 1.0 -
■2
1
0
Sean A ngle (dog)
1
-2
Sean Angle (dcg)
Figure 4.6: Chopped beam pattern for all four channels. The signal has been nor­
malized to one. Some of the d ata has been cut due to cosmic rays and telemetry
glitches.
past our minimum, they dropped all the remaining ballast by blowing open the
container. Although bobbing severely, we observed 3C345, a radio point source, near
the star Eta Herculis at 9:08 UT. At 9:49 U T we moved to and scanned Saturn
in azimuth and elevation. We did a membrane calibration at 10:05 to compare to
Saturn. We began to observe the region near Delta Casseiopea at 10:23. However,
the ADR became warm at 10:27. In fact, its performance had been degrading since
HR5127. Initial analysis determined that only data up to Phi Herculis was usable.
After the detectors useful life, we performed two systematic tests to understand the
malfunction at Mu Pegasi. We scanned in four quadrants to understand the scan
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4. MAX5 CAM PAIGN
32
1 .0 x 1 0 * - i
3cm-1 jip te r el scan
FWHM=0.541 deg, GaLESianfit
0 .0 -
35.0
35.5
36.0
36.5
37.0
Figure 4.7: Main lobe for the 3.5 cm 1 band. Note the small amount of squint. The
peak is calibrated to the first membrane calibration.
orientation as a function of azimuth. We then did fast rotations with the gondola to
map out the gyro horizon.
The conclusion about the gyro is that its perceived horizon was tilted with
respect to the true horizon. The axis of tilt at the time of the Mu Pegasi scan was
approximately east-west. Since Mu Pegasi was in the east, it had the most tilt. The
HR5127 scan and Phi Herculis scan, being approximately north, had some tilt, but
not to a measurable degree. In addition, we found that the tilt precessed slightly
during the systematic tests.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 4. M AX5 CAMPAIGN
Ujy/Str
30
450
micro*
Figure 4.8: The Mu Pegasi scan is superimposed over a IRAS 100 fim map. The
circles represents the FWHM of the beam at each apex of the chop at the center
of each of the 29 bins. The map is a 12.5° x 12.5° (500 x 500 pixels) gnomonic
projection in RA and DEC centered on Mu Pegasi (1994 precession).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p ter 5
D a ta A n a ly sis
5.1
D ata R eduction
We removed transients due to cosmic rays using an algorithm described
by Alsop et al. (1992). A F F T on a 30 second sliding window was performed and
cosmic-ray hits were identified by the rise in th e low-frequency noise and white noise
level We then removed data within a few detector time constants of the cosmic-ray
hit. This procedure excluded approximately 18-20% of the d ata from each scan. We
demodulated the detector output using the sinusoidal reference from the chopping
secondary to produce antenna temperature differences A T a on the sky. This created
a data set in phase and a data set 90° out of phase with the optical signal. The noise
averaged over the Mu Pegasi observation gives respective CMBR sensitivities of 440,
240, 610, and 5100 y K ^ /s in the 3.5, 6, 9, and 14 cm -1 bands. Similar sensitivities
were obtained for the other scans.
The signal in each band was significantly offset from zero. We attributed
the offset to chopped emissivity differences on the primary mirror and chopped atmo­
spheric emission. For example, in the Mu Pegasi data, the averages of the measured
instrumental offsets in antenna temperature were 0.6, 0.15, 1.4, and 2.8 mK in the
3.5, 6 , 9, and 14 cm -1 bands. The offset drifted in the higher frequency bands with
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA AN ALYSIS
35
amplitudes of 700 fiK and 1000 nK in the 9 and 14 cm -1 bands over a time scale of
3 minutes.
We dealt with the offset drift differently for each data set. All data sets have
the offset removed. HR5127 was a clean, stable scan and so a gradient was removed
with a least squares fit to each pass going from zero to -4° to zero or from zero to
+4° to zero. The effect is to remove only instrument drift and to leave in any sky
signal gradient. In Mu Pegasi, we subtracted the gradient with a linear least squares
fit to each pass going from -4° to +4° or +4° to -4°. Here we intentionally removed
a gradient on the sky for reasons that will become clear later. In Phi Herculis we
removed gradients as in HR5127. Each half scan takes 72 s.
For each observation we calculated the means and standard error of the
antenna temperature differences for 29 pixels separated by 17' on the sky. Figures
5.4 - 5.6 show the antenna temperature differences as a function of scan angle for
each scan. In each scan there was significant structure (x 2 = 38, 86, 86 , 79 for 27
DOF) that was well correlated (R ^ 0.5) in all channels of the in-phase data. The
out of phase components were consistent with Gaussian random noise for all scans.
(x 2 = 22, 32, 18, 22 for 27 DOF).
5.2
T im e Stream A nalysis
We analyzed the data for stability by splitting the data in half, binning
each part, and then fitting the binned data to each other, channel by channel. This
is known colloquially as ”jack-knifing”. For Mu Pegasi, we found that the signal is
stable in the 3.5, 6 , and 9 cm -1 bands, but not so in the 14 cm -1 band, (x2 = 17,
12, 26, 30 for 29 DOF for 3.5, 6 , 9, and 14 cm-1 ) The instability in the 14 cm -1 band
could be caused by sidelobe pickup or atmosphere. If we do not use the gradient
removal scheme above, the jack-knife of the 14 cm -1 band has a reduced x2 of 50/27.
Stacy Tanaka analyzed the time stability of HR5127 and Phi Herculis. HR5127 was
a stable scan, with x2 = 20, 26, and 28, for 29 DOF in the 3.5, 6 , and 9 cm -1 bands
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
C H APTER 5. DATA ANALYSIS
respectively. Phi Herculis is stable only in the 3.5 and 6 cm 1bands, with reduced
X2 = 21/29 and 22/29.
5.3
System atic Errors
CMB observations are bought with possible systematic errors.
A non-
inclusive list is radio-frequency interference (RFI), electrical or mechanical pickup,
tem perature fluctuations of the baffles and optics, and sidelobe contamination from
the balloon, earth or moon. In the MAX4 era, we conducted RFI tests with the
408/416 Mhz, 2W transmitter used for telemetry on ACME. There was no rise in
the noise level at the chop frequency with the window capped (sealed against EM
radiation) or uncapped. Both forms of pickup were limited by integration tests done
before flight. We cap the dewar so that it is looking back at itself (4.2 K black
body), and run the entire experiment. We integrated while scanning and chopping,
using telemetry to command and record data. The integrations were consistent with
no signal after 30 m inutes, limiting pickup and interference to < 6.4, 2.3, 2.2, 3.1
fiK in the 3.5, 6 , 9 and 14 cm -1 bands. The balloon was > 24° above the beam in
elevation during the Mu Pegasi and HR5127 scan. During the Phi Herculis scan the
target star came to within « 15°. This is stil well within the far side lobe region
of the beam pattern, with rejection > 70 db. The moon was not full during the
observations. HR5127 was ~ 56° away from the moon. Mu Pegasi was ~ 137° and
Phi Herculis ~ 66° away. Temperature fluctuations in baffles and the primary were
lim ite d by our measurements of their gradients in flight.
MAX3 systematic tests
show that temperature gradient rejection was ~ 10-3 [15]. The primary mirror, for
example, had a temperature gradient of 250 ± 10 mK during the data taking part of
the flight. This corresponds to a 250 fiK signal. However, the time constant of the
mirror was « 8 hours, so at worst the gradient contributed a slowly varying offset
to the data. One would need to drive scan synchronous tem perature gradients (time
constant 72 seconds) to appear as a sky signal.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
C H APTER 5. DATA AN ALYSIS
5.4
Foregrounds
5.4.1
Single Source M odels
Possible astrophysical sources for the signal in the data are free-free or
synchrotron radiation, interstellar dust (ISD) emission, radio point sources, or CMB
radiation. Figure 5.1 shows the relevent spectra at the frequencies of interest. The
monotonic drop in antenna temperature diffence in HR5127 and Phi Herculis is
indicative of free-free, synchrotron or CMB radiation, but not ISD. It may be possible
to construct the morphology in those two scans from point sources. The d ata from Mu
Pegasi have a more complicated spectrum that, as we will see, is not well described
by a single source.
Spectral Analysis
The multi-frequency nature of the experiment allows us to discriminate
between CMB radiation and other possible signal sources. Because the band passes
were wide, it is convenient to describe spectra in terms of channel/channel ratios.
These are the antenna temperatures of a sub-set of the bands normalized to the
antenna tem perature in a base band, typically the 3.5 cm -1 channel, for a particular
spectrum. Thus,
T.
Tj
35 ~ r*5 ~
J ‘a.5
2gj&a, JfwMW*
1 '
is the antenna temperature ratio for the ath band over the 3.5 cm -1 band. Here,
is the antenna temperature in the ath band, ta is the transmission of the ath band,
Iu is the brightness of the model spectrum,
is the Rayleigh-Jeans blackbody
function and T is the thermodynamic temperature.
Figure 5.2 is a contour plot though the channel/channel plane of the likeli­
hood for the HR5127 data set at the most likely Q. While the most likely spectrum is
not th at of CMB radiation, it does lie within the 95 % c.l. The two other most likely
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
CHAPTER 5. DATA AN A LYSIS
candidates, synchrotron and free-free radiation are excluded at that level. For Phi
Herculis, the perspective changes somewhat. Figure 5.3 is the likelihood in T®5 and
Q. Here the maximum likelihood is close to CMB radiation, but the 95% confidence
level contour encompasses all possibilities. More properly, the likelihood should be
projected along the r ®5 axis, but the point is made; Phi Herculis cannot exclude
Free-Free and Synchrotron radiation as candidates by spectral arguments alone.
Amplitude and Morphology Analysis
Free-free and synchrotron radiation can be excluded by amplitude and mor­
phology arguments. If we extrapolate the Haslam 408 MHz map [19] to our frequen­
cies using A T a
oc i / -2 -1
for free-free emission and A T a
oc
i / -2-7
for synchrotron
radiation and run simulations from our pointing data, we find that in all three scans,
the former produces < 10% of the signal in the 3.5 cm -1 channel and the latter
< 1%. Furthermore, the morphology does not match th at of the data.
5.4.2
Point Sources
An automated point source search at the NASA/IPAC Extragalactic
Database (NED) 1 yielded no candidates within 90' of the scanned regions that
could produce a signal greater than 10 fiK. Catalogs came from several sources and
include all major surveys [32, 47, 50, 3, 37, 65]. If spectral information was available,
it was used in the estimate of the signal, otherwise the spectrum was assumed to be
flat. The possibility exists for contamination from a rising spectrum point source,
although there is no reason to believe any of the candidates have such a spectrum.
1NED is operated by the Jet Propulsion Laboratory, California Institute of Technology, under
contract with the National Aeronautics and Space Administration
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5. DATA A N A LYSIS
Frequency (cm 1)
15
6
9
14
39
CCC(fiK(M Jy/sr)
30
16
22
30
L) a Reduced x 2
±6
26/27
± 5
56/27
±5
43/27
± 6
28.5/27
Table 5.1: Cross correlation coefficients for fit to IRAS 100 fim dust morphology, a)
These are the ratios of the differential antenna temperatures for the best fit spectrum.
5.4.3
Interstellar D u st in M u P egasi
Previous experience in this region leads us to expect ISD to be the main
contributor to our high frequency signal. From our pointing data we ran simulations
of the scan on the IRAS 100 ^m maps. We found the cross-correlation coefficients
(CCC) that minimized the reduced x 2 from 100
fim
to each channel separately and
then normalized them to the 3.5 cm -1 band. The results are shown in Table 5.1.
The best fit morphology and spectrum are superimposed over the data in Figure 5.6.
If we consider the 6, 9, and 14 cm -1 bands only, theSe CCC are consistent with a
single component dust model
I„ <x uPB v{Tdust)
(5.2)
where 0 = 1.3^ 0^ and Tw.,rf = 19*| K. This is consistent with previous results (44, 28]
and the results of a similar model used with COBE/FIRAS [27]. However, the rise in
amplitude in the 3.5 cm -1 band is not well explained by a single or double component
dust model. We therefore do not use this model to account for the foreground, but
the CCC found above.
5.4.4
3.5 cm -1 A m p litu d e in M u P egasi
There are two possible causes for the rise in amplitude in the lowest band.
One is a high frequency leak in the filters. Pre-flight systematic tests with a thick grill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5. DATA AN A LYSIS
40
high pass filter showed th a t high frequency leakage above 20 cm -1 was less than 0 . 8%
of the total band response to a 300K blackbody chopped relative to a 77K blackbody.
Using measured filter transmittances and the amplitude of dust fluctuations in this
sky region (Fischer et aL 1995), we calculated that maximum modeled high frequency
leakage of power from dust fluctuations contributes less than ~ 1% of the expected
inband power from CMB fluctuations and less than 2% of the observed structure
[62]. Additionally, the d ata in the 3.5 cm -1 band should have suffered the same
instability seen in the 14 cm -1 band. Finally, the Phi Herculis scan showed none
of the atmospheric contamination in the 3.5 cm -1 band that was seen in the 9
and 14 cm -1 bands. Another candidate is a correlated low frequency component.
Kogut et aL (1995) report correlation between HEI and ISD at angular scales > 7°,
which provides the physical motivation for investigating such models. However, our
cross-correlation coefficient 3.5 cm-1 /IRAS 100 pm is 30 ± 6 mK(M Jy/sr )_1 which
should be compared to 4.56 ±3.89 m K (M Jy/sr )_1 for DMR 90 GHz/DIRBE 100 pm.
Furthermore, it is not known whether the spatial correlations will hold at 30 ' as at
7 degrees.
We conclude the following about the Mu Pegasi foreground contaminant:
The correlation between the 14 cm -1 band and the other bands indicates a single
foreground morphology. Whatever the nature of the foreground, the relative ampli­
tudes in the bands are given in Table 5.1 column 2. Because of the excellent fit of
the IRAS 100 p m maps to the 14 cm -1 channel, we assume that ISD dominates over
any other possible high frequency contaminant. Furthermore, because there is no
significant CMB signal in the 14 cm -1 band, we exclude it from any further analysis.
5.5
Bayesian M eth ods
We used maximum likelihood methods assuming uniform prior to set limits
of the rms temperature fluctuation in the data, Q = Qrma-PS = (Qrma)0'5 [58, 20].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5. DATA A N A LYSIS
41
The likelihood, L, is given by
£ .
? p ( - g ¥
:3
(M )
f a " det(M )
where Tia is the data vector of all (i = 1 to 29) bins and (a = 1 up to 3) channels, N
= 29 x 3 and Mijab is the full covariance matrix.
In general, M, is given by
M = C th + s 2
(5.4)
where C th, the theoretical correlation matrix, is given by the cosmological theory
and s is the experimental correlation matrix. Explicitly, C th is written
(5-5)
1=2
The experimental correlation m atrix also depends on frequency and bin number.
s2 =
which is just the variance of the data in bin i, band a, if i = j and a = b,
and is the covariance if a £ b or i ^ j . Note that s 2 is block diagonal. In balloon
borne experiments, the elements off the diagonal are of order 3 - 10% of the diagonal
elements. Thus, it’s not quite justifiable to consider the experimental correlation
m atrix diagonal. The primary effect is the reduction of significance in a likelihood
ratio analysis [17) but can also affect the final value of Q. For the case of a flat band
power spectrum, Mijab becomes
Mijab = £
i { i X \ ) \ Q2Wl + S*ibj
(5' 6)
Since the flat-band power spectrum is the preferred, model-independent way of re­
porting anisotropies, all results will be reported with respect to this approximation.
In this case, Ci =
5.6
and we use the normalization, Co = 4nQ 2/5.
A ccounting for Foregrounds
There are many schemes to try to account for foreground contamination
and somehow extract the underlying cosmological information. We consider two
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA AN ALYSIS
42
here, frequency marginalization and ’’direct subtraction” which is a short hand ref­
erence to analyzing the CMB component of a two component fit. The name arises
from superficial similarity to subtracting out a dust model and analyzing the resid­
uals. Direct subtraction is not used since it does not allow one to construct the
correct covariance matrix. Two component fitting procedure only returns the diago­
nal elements (variances) of the covariance matrix. W ithout the off-diagonal elements
(covariances), one ends up underestimating the size of the covariance matrix and
hence one’s error bars. This could lead to a ” detection” when there is none. Another
way of putting it is that neglecting the covariances can drastically reduce the power
of your test (in a m axim um likelihood ratio test) to not detect anisotropy when there
is none. In the limit of large signal to noise, this is less of an issue, but the CMB field
is far from this regime Marginalization allows us to use the covariances calculated
from the data in d e te rm in in g the new off-diagonal elements. It should be noted that
frequency marginalization is not a method for improving signal-to-noise. It is in fact
a way of finding the data set independent of an assumed foreground and properly
accounting for the error bars after the procedure.
5 .6 .1
M a rg in a liz a tio n
We marginalized the data to account for the best fit foreground spectrum
and offset and gradient removal and analyze them for anisotropy [16, 18, 7]. To do so
we solved
za9a — 0 f°r two orthonormal vectors, zT
a ( r = l to 2 ) which composed
the 2 x 3 matrix z. The coefficients ga = T “5 for a particular foreground spectrum,
as given in Equation 5.1. In a similar procedure we found a 27 x 29 matrix R whose
rows were an orthonormal set that were also orthogonal to a constant and gradient
29-vector. While it is possible to weight the constant and gradient 29-vectors with
the variances (and so weight the elements of R ), we use the unweighted R since the
error bars are similar in each bin. In this case both R r R = 1 and z Tz = 1.
W ith these matrices we constructed a data vector and covariance matrix,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5. DATA AN A LYSIS
43
T*™* = z r R r T and M*™* = zr R r M R z, which accounted for both a single fore­
ground spectrum and offset and gradient removal. Here z and R have been ex­
panded to appropriate dimensions so they act on a per frequency and per bin basis
respectively. The likelihood was then calculated using M*™* and T*™*.
5.6.2
D ir e c t S u b tr a c tio n
Here, we attempted to extract the cosmological information by fitting the
Mu Pegasi data to a two component morphological model. A spectrum was assigned
to each component and held fixed in the fitting procedure, while the morphologies
were allowed to float. The best fit components did not depend on the starting point
of the fit. The spectrum of the first component was chosen to be either the model
from equation 5.2 or a two component model from Reach et. aL
/„ = n i A B v f J i ) + T2V^B u{T2)
where n = 0.25 x 1 0 '5,/?i = 2,T i = 17.5K ,
to
(5.7)
= 1.4 x 10~5,/?o = 2, To = 6.7K.
This equation assumes that the two spectral components are spatially correlated.
The spectrum of the second component was chosen to be that of CMB, synchrotron
or free-free radiation. Results of the fits and correlation coefficients from the fits
to the data are presented in Table 5.2. The dust and CMB models gave the best
fits, although all models were moderately good. The free-free model is within the
68% confidence interval of the CMB model and thus not statistically distinguished,
but the synchrotron model lies outside that region but within the 95.4% confidence
interval.
The error bars from the fit are used to construct <r2. The data from the fit
form T. This information is cam then be used in equation 5.3. This method suffers
from the fact that one cannot get the true covariance matrix. Furthermore, with the
available signal to noise one cannot adequately distinguish from the possible mod­
els. Frequency marginalizations gives a more conservative and statistically rigorous
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA AN ALYSIS
2nd Component
CBR
CBRc
Free-free
Synchrotron
44
Reduced x 2 3.5 cm -1 Correlation® ERAS Correlation 15
52.9/50
(X83
053
52.7/50
0.83
0.57
54.9/50
0.83
0.69
56.8/50
0.84
0.77
Table 5.2: Results from two component fitting. First component is hot dust as
described in the text unless otherwise specified, a) correlation coefficient between
the 3.5 cm -1 band and the 2nd component, b) correlation coefficient between the
IRAS 100 micron morphology and the 1st component, c) 1st component is the hot
and cold dust model described in the text.
method of calculating the likelihood and so it is the method of choice for setting
limits on Q in Mu Pegasi.
5.7
5.7.1
R esults
M A X 5 M u Pegasi
Because of the instability in the 14 cm_1channel and the fact that it pos­
sessed no CMB information, we exclude it from the analysis. Using M 1™* and T*™*
in equation (5.3) yielded an upper limit, Q < 23 fiK (95% confidence level) or from
Equation 2.10, A T / T ^ < 1.3 x 10-5 .
5.7.2
M A X 5 H R 5127
We used only the science channels, 3.5, 6 and 9 cm -1 in the analysis. Since
the spectral analysis indicates CMB as the most likely source we interpret the signal
as CMB anisotropy with Q = 19^g4/x K or A T /Tcmb = 1 -llo J x I ®-5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA AN A LYSIS
5.7.3
45
M A X 5 P h i H erculis
While we could not rule out foregrounds definitively or be sure that we
completely accounted for atmosphere, if we interpret the 3.5 and 6 cm -1 data as
CMB anisotropy, we find, Q = 38^4 or AT/Tcmb = 2.21^8 * 10-5 .
5.7.4
M A X 3 M u P eg a si R eanalysis
In order to compare the MAX5 Mu Pegasi results to MAX3 Mu Pegasi,
we reanalyze the older data using marginalization for the ISD spectrum reported in
Meinhold et aL (1993). In this case, /? = 1.4 and TdU3t = 18K . This yielded Q <
28 nK (95% confidence level) when marginalized and Q = 1 3 ( 6 8 % confidence
level) if we analyze the residuals for anisotropy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5. DATA AN ALYSIS
46
io‘
10°
50
60
200
300
400
SOO
600
Frequency (GHz)
Figure 5.1: Various foreground spectra compared to a CMB spectrum. The MAX
passbands are superimposed w ith a linear vertical scale in bold at the bottom. They
are normalized to be the same height. The upper falling curve is the CMB spectrum.
The lower curve is the same but rescaled to 10 ppm. The rising dotted spectrum is for
interstellar dust with /? = 1.3. The falling dashed line is synchrotron radiation w ith
spectral index -2.7. The falling dot-dashed line is free-free radiation with spectral
index -2.1. The upper spectrum is the atmosphere at 30 km.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA AN ALYSIS
47
1.0
0. 8 -
0. 6 -
0.060
0 .4 -
CUB
0 .2 -
0.0
0.0
0.2
0.6
0.4
0.8
1.0
6/3
Figure 5.2: The 95% confidence level is plotted in T® vs. T3 antenna temperature
ratio space for the HR5127 scan. The crosses are the locations of the spectra for CMB,
free-free and synchrotron radiation. We assume an antenna temperature power-law
spectrum with index -2.7 for synchrotron and -2.1 for free-free radiation. Additionally
we plot the location of the most likely spectrum. A CMB spectrum is included
within the 95 % c.L region, but free-free and synchrotron radiation are excluded.
ISD antenna temperature ratios are of order a few and do not appear on this plot.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA ANALYSIS
48
1.0
0. 8 -
0. 6 -
0.4
0.060
0. 2 -
0.0
0
20
40
60
80
100
Qrms-ps
Figure 5.3: We plot the 95 % c.L region in the T® antenna temperature ratio vs. Q
space for Phi Herculis. The lines are the locations of the spectra for CMB, free-free
and synchrotron radiation. We assume an antenna tem perature power-law spectrum
with index -2.7 for synchrotron and -2.1 for free-free radiation. Additionally we plot
the location of the most likely spectrum. We see that both foregrounds lie within
the 95 % c.l. as well as CMB radiation. ISD antenna tem perature ratios are of order
a few and do not appear on this plot.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA ANALYSIS
49
100
o-2 5 -5 0 -7 5 - 100 75-
O)
25-
0-
*.3 . .
-2 5 -5 0 -7 5 - 100 -
0-2 5 -5 0 -7 5 - 100 75-
-2 5 -5 0 -7 5 - 100 -
0
5
10
15
20
25
30
Scan Angle
Figure 5.4: The means and standard errors of the 29 bins are plotted versus scan angle
for the HR5127 scan. The data are in antenna temperature. The line superimposed
over the 14 cm-1 band indicates the size and shape of the chopped beam.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 5. DATA AN A LYSIS
-4
-2
50
0
2
4
Scan Angle (Deg)
Figure 5.5: The means and standard errors of the 29 bins are plotted versus scan
angle for the Phi Herculis scan. The data are in antenna temperature. The line
superimposed over the 14 cm-1 band indicates the size and shape of the chopped
beam.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5. DATA ANALYSIS
51
150
100
CJt
-50
-HD
-150
10050-
0-5 0 -
I
i
-15D
100 50-
0-50 - 100 -
-150
100 -
■U
0-50-
-150
4
■2
0
2
4
S canA rge(D eg)
Figure 5.6: The means and standard errors of the 29 bins are plotted versus scan
angle for the Mu Pegasi scan. The data are in antenna temperature. The dashed
line is for a spectrum with 0 = 1.3 and T^ust = 19 K . The solid line is for a
similar morphology but with the amplitudes chosen to minimize x~• The line off-set
over the 14 cm -1 band indicates the size and shape of the chopped beam.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h ap ter 6
C on clu sion
The MAX5 flight was the final set of one dimensional scan measurements
taken on the ACME gondola. In its history, MAX has measured six separate regions.
Table 6.1 summarizes the observations and results from each flight. The MAXl
and MAX2 flights used a step-scan strategy which have never been analyzed in
a contemporary fashion.
However, the amplitudes will probably drop rs 50% if
converted to flat band power estimates.
The Gamma Ursae Minoris region has been measured three times and the
Mu Pegasi region has been measured twice. Unfortunately, none of the revisits has
been able to exactly reproduce a previous scan orientation. On the other hand, sim­
ilar measurements are obtained from the same regions. While comparing the Mu
Pegasi upper limits is not too useful, the Gamma Ursae Minoris region has consis­
tently come in high, A T/Tanb = 2 .91^4 MAX3 [23] and A T/Tcmb = 2.01q;|, MAX4,
[61]. As for the regions visited only once (Eta Draconis, Sigma Herculis, HR5127
and Phi Herculis), they too paint a similar picture. The amplitudes are similar, but
combined with the GUM data, show a reasonable spread. In fact, Mu Pegasi is no
longer the lowest point in the set. HR5127 has a A T /T c ^ = 1. 1^0 3 x 10-5 which
overlaps considerably with the 95% c.l for Mu Pegasi, MAX5. While it is interesting
that such a low number can be observed in a clean region without the concern that
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C H APTER 6. CONCLUSION
53
foreground accounting is biasing the answer, it doesn’t answer the question as to
whether or not such variations are allowed for Gaussian fluctuations.
The motivation for the Mu Pegasi revisit was to shed some light on the low
signal with the 3.5 cm -1 band. The expectation was th a t the 3 cm -1 band would
have proved to be independent of the interstellar dust and so be a good measure
of the CMB. This did not prove to be so. Because the 3.5 cm -1 band still had
significant correlation with dust, almost all the structure could be accounted for.
Combined with a relatively high detector noise, an upper limit is sufficient for this
data set. However, the MAX3 Mu Pegasi and MAX5 Mu Pegasi sets are consistent
with each other for similar analysis techniques. Furthermore, the marginalized upper
limit is consistent with the results from HR5127 and Phi Herculis also taken that
flight.
Were it simple, a low frequency (30-40 GHz) measurement in the Mu Pegasi
region would be appropriate. While it is unlikely that a HEMT-based 1-D scan CMB
experiment will ever fly, next generation balloon based experiments inherently cover
a large portion of the sky and perhaps will map the Mu Pegasi region in sufficient
resolution to understand the low frequency anisotropy in that region.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 6. CONCLUSION
Flight
Region
MAXI [24] a
NCP
MAX2 [1] a Gamma Ursae Minoris
MAX3 [23] Gamma Ursae Minoris
MAX3 [25]
Mu Pegasi
MAX4 [61] Gamma Ursae Minoris
MAX4 [61]
Sigma Herculis
MAX4 [61]
Iota Draconis
MAX5
HR5127
MAX5
Mu Pegasi
MAX5
Phi Herculis
54
A r T W i o - 5)
< 6.3
4.5:38
< 1.6
2 . 01 &®
1 53
-Q-S
i “r+u
.8
•*••‘-0.5
1 -j+ 0.8
i -i—0.3
< 1.3
o 0 + 1.4
•^'^- 0.8
Table 6. 1: Table of results from the history of the MAX experiment. Results me for
a flat band power spectrum unless otherwise noted, a) Results are for a GACF at
30' scale.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C h a p ter 7
F utu re D ir ec tio n s
7.1
Large Scale Surveys
Since small sky coverage experiments have reached the end of their useful­
ness, in what directions will the community move toward? It’s generally considered
that what is needed is a large scale survey with fine angular resolution and broad
spectral coverage The experiment must cover enough sky to see the expected rise in
the first doppler peak. The desired angular resolution depends on the science that
one wishes to achieve. An angular resolution of 20 arcminutes will adequetely detect
the first doppler peak. Unfortunately the location of the first peak is degenerate in
cosmological parameters and only by measuring down to the second or third doppler
peak, or 5-10 arcminutes, will one be able to achieve 10 % accuracy on cosmology [36].
Broad spectral coverage is required to adequately distinguish CMB from foreground
sources. Since there are three potential diffuse foregrounds of importance, free-free
and synchrotron radiation and ISD emission, one requires at least four channels to
recover the CMB accurately. The combination of knowing both angular power spec­
trum and electro-magnetic spectrum for each sky component makes distinguishing
them easier[63].
To this end, there are two proposed satellite missions. The Microwave
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 7. FUTURE DIRECTIONS
56
Anisotropy Probe (MAP) is proposed by Goddard Space Flight Center (GFSC) and
Princeton University. The Cosmic Background Radiation Anisotropy Satellite and
Satellite for Measurement of Background Anisotropies (COBRAS/SAMBA) is a dual
mission proposed by the European Space Agency (ESA). Either would be capable of
meeting the requirements above.
MAP uses a single differencing chop, 5 channel HEMT receiver which is
spun and precessed about the anti-sunward axis. As the satellite sits at L2 (along
the earth-sun axis) the scan covers the whole sky in one year. The choice of orbit is
to avoid local systematic effects such as the earth and sun’s thermal radiation, the
earth’s electromagnetic environment and man-made RFI. The optics are based on
an off-axis Gregorian telescope with 1.5 m primary which has a highest resolution
of 20 ' at the highest frequency. This allows Z-space coverage down to Z« 600. The
five bands will provide adequate leverage to remove the three strongest foregrounds
and provide good signal-noise for CMB radiation. Calibration will be provided by
the dipole of the CMB.
COBRAS-SAMBA is a much more ambitious project, being the melding
of two MAP-class experiments into one package. Because COBRAS-SAMBA uses
bolometers as well as HEMT receivers, the designers are presented with great techni­
cal difficulties in cryogenics and optics to overcome. However, it will have a resolution
of 7 ' at 217 GHz (the highest useful data frequency) which coupled with an all-sky
scan strategy, will provide Z-space information down to Z « 2000. The frequency
coverage goes from 30 GHz to 857 GHz, with the four lower channels being HEMT
amplified and the upper four bolometrically detected. COBRAS-SAMBA will also
take advantage of a L2 orbit to avoid systematic effects from local interference.
The alternative to a satellite is to construct a balloon-borne package with
an endurance and flight plan that will give adequate sky coverage. There are propos­
als in for Maxima, BOOMERANG, and QMAP. The first two are bolometer-based
photometers with ultimately long-duration balloon (LDB) capability. QMAP is a
HEMT-based LDB experiment. The Santa Barbara initiative is the Background
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 7. FUTURE DIRECTIONS
57
Emission Anisotropy Scanning Telescope (BEAST), a LDB project which will ul­
timately be developed into a very long duration balloon experiment (VLDB), the
Advanced Cosmic Explorer (ACE). Both are cixcumpolar balloon flights and differ
primarily by their flight time (~ 10 and ~ 100 days respectively) and the engineering
details required to meet those flight durations. A 2 m off-axis telescope will provide
the sub-30' beam required for adequate coverage of the angular power spectrum. A
multi-pixel focal plane will enable us to use several HEMT amplifier bands. Initially
this will be 8 pixels with Ka and Q band amplifiers. Ultimately, we will build a 28
pixel, 6 channel array, with coverage from 10 to 180 GHz. The choice of frequencies
is optimal for detecting CMB anisotropies and identifying and rejecting atmospheric
and foreground contamination. Because of the size of the primary mirror, the finest
beam size (4;) will be a data channel, giving /-space coverage down to I ~ 2700. The
long flight time and limited sky-coverage allows us to achieve high per-pixel sensi­
tivity. Finally, development and flight time is very short compared to a satellite (2
vs. 10 years). While acquiring total sky coverage will be a long and difficult process,
these experiments probe the range of /-space that are significant for many theories
of cosmology and will provide good discrimination between them.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
B ib lio g ra p h y
[1] D. C. Alsop, E. S. Cheng, A. C. Clapp, D. A. Cottingham, M. L. Fischer, J. O.
Gundersen, E. Kreysa, A. E. Lange, P. M. Lubin, P. R Meinhold, P. L. Richards,
and G. F. Smoot. A search for anisotropy in the cosmic microwave background
on intermediate angular scales. Astropkys. J., 395:317, 1992.
[2] J. G. Bartlett, A. Blanchard, J. Silk, and M. S. Turner. The case for a hubble
constant of 30 km s -1 mpc-1 . Science, 267:980, 1995.
[3] R L. Becker, R L. White, and A. L. Edwards. A new catalog of 53,522 4.85
ghz sources. Ap. J. Suppl., 75:1-229, 1991.
[4] James Binney and Scott Tremaine. Galactic Dynamics. Princeton University
Press, 41 William St. Princeton, NJ 08540, 1987.
[5] J. R Bond. Theory and observations of the cosmic background radiation. In
Cosmology and Large Scale Structure, Proc. Lew Houches School, session LX,
1993.
[6] J. R Bond. Cosmic structure formation and the background radiation. CIAR
Cosmology Program, CITA, 1994.
[7] J. R Bond, G. Efstanthiou, P. M. Lubin, and P. R Meinhold. Phys. Rev. Lett.,
66:2179, 1991.
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIO G R AP H Y
59
[8] E. S. Cheng, D. A. Cottingham, D. J. Fixsen, C. A. Inman, M. S. Kowitt, S. S.
Meyer, J. L. Puchalla L. A. Page, and R. F. Silverberg. ApJ Lett., 422:L37,
1994.
[9] Alfredo 0. Chingcuanco, Philip M. Lubin, and Peter R. Meinhold. Model refer­
ence adaptive control for the azimuth-pointing system of a balloon-borne stabi­
lized platform. International Journal o f Adaptive Control and Signal Processing,
5:107-120, 1991.
[10] A. Chinguanco. Modelling and Control o f a Balloon Borne Stabilized Platform.
PhD thesis, University of California, Santa Barbara, 1989.
[11] A. C. Clapp, M. J. Devlin, J. 0 . Gundersen, C. A. Hagmann, V. V. Hristov, A. E.
Lange, M. Lim, P. M. Lubin, P. D. Mauskopf, P. R. Meinhold, P. L. Richards,
G. F. Smoot, S. T. Tanaka, P. T. Timbie, and C. A. Wuensche. Measurements
of anisotropy in the cosmic microwave background radiation at degree angular
scales near the stars sigma herculis and iota draconis. Astrophys. J., 433:L57,
1994.
[12] Andre Charles Clapp. Measurements o f the Anisotropy of the Cosmic Microwave
Background Radiation on Degree Angular Scales.
PhD thesis, University of
California, Berkeley, 1994.
[13] M. Davis, G. Efstathiou, C. S. Frenk, and S. D. M. White. The end of cold dark
m atter? Nature, 356:489, 1992.
[14] M. J. Devlin, A. C. Clapp, J. 0 . Gundersen, C. A. Hagmann, V. V. Hristov, A. E.
Lange, M. A. Lim, P. M. Lubin, P. D. Mauskopf, P. R. Meinhold, P. L. Richards,
G. F. Smoot, S. T. Tanaka, P. T. Timbie, and C. A. Wuensche. Measurements
of anisotropy in the cosmic microwave background radiation at 0. o 5 angular
scale near the star gamma. Astrophys. J., 430:L1, 1994.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIO GRAPH Y
60
[15] Mark Joesph Devlin. Measurements of Anisotropy in the Cosmic Microwave
Background on Degree Angular Scales. PhD thesis, University of California,
Berkeley, 1993.
[16] S. Dodelson and A. A. Stebbins. ApJ, 433:440, 1994.
[17] Scott Dodelson, A rthur Kosowsky, and Steven T. Myers. Noise correlations in
cosmic microwave background experiments. ApJ Lett., 440:L37-L39, 1995.
[18] Mark Srednicki Emory Bunn, Martin W hite and Douglas Scott. Are sp 91 and
cobe inconsistent with cold dark matter? ApJ Lett., 429:L1, 1994.
[19] C. G. T. Haslam et al. A&AS, 47:1, 1982.
[20] E. Wright et aL ApJ, 420:1, 1994.
[21] J. C. Mather et aL ApJ, 420:439, 1994.
[22] J. O. Gundersen et aL Degree-scale anisotropy in the cosmic microwave background:sp94 results. A pJ Lett., 443:L57-L60, 1995.
[23] J.O. Gundersen et aL A degree scale anisotropy measurement of the cosmic
microwave background near the star gamma ursae minoris. ApJ Lett., 413:L1L5, 1993.
[24] M. L. Fischer et al. ApJ, 330:242, 1992.
[25] P. R. Meinhold et al. Measurements of the anisotropy of the cosmic background
radiation at 0?5 scale near the star mu pegasi. A pJ Lett., 409:L1-L4, 1993.
[26] S. L. Wheelock et aL IR A S Sky Survey Atlas Explanatory Supplement. JPL
Publication 94-11, JPL, Pasadena, 1994.
[27] W. T. Reach et al. ApJ, 451:188, 1995.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIO G RAPH Y
61
[28] M. L. Fischer, A. Clapp, M. Devlin, J. O. Gundersen, A. E. Lange, P. M. Lubin,
P. R. Meinhold, P. L. Richards, and G. F. Smoot. ApJ, 444:226-230, 1995.
[29] W. L. Freedman, B. F. Madore, J. R. Mould, R. Hill, L. Ferrarese, R. C. Kennicutt, A. Saha, P. B. Stetson, J. A. Graham, H. Ford, J. G. Hoessel, and
G. D. Illingworth J. Huchra, S. M. Hughes. Distance to the virgo cluster galaxy
m 100 from hubble space telescope observations of cepheids. Nature, 371:757762, 1994.
[30] T. Gaier, J. Schuster, J. Gundersen, T. Koch, M. Seiffert, and P. Lubin P. Mein­
hold. ApJ Lett., 410:L1, 1992.
[31] K. Ganga, E. Cheng, S. Meyer, and L. Page. A p J Lett., 410:L57, 1993.
[32] P. C. Gregory and J. J. Condon. The 87gb catalog of radio sources covering 00°
< 6 < +75° at 4.85 ghz. Ap. J. Suppl., 75:1011-1291, 1991.
[33] A. H. Guth. Inflationary universe: A possible solution to the horizon and flatness
problems. Phys. Rev. D, 23:347, 1981.
[34] G. Jacoby, D. Branch, R Ciardullo, R. Davies, W. Harris, M. Pierce, C. Pritchet,
and D. Welch J. Tonry. A critical review of selected techniques for measuring
extragalactic distances. Publications o f the Astronomical Society of the Pacific,
104:599, 1992.
[35] David G. Johnson and David T. Wilkinson. A 1 percent measurement of the
temperature of the cosmic microwave radiation at lambda = 1.2 centimeters.
A pJ Lett., 313:L1, 1987.
[36] G. Jungman, M. Kamionkowski, A. Kosowsky, and D. N. Spergel. Cosmologicalparameter determination with microwave background maps, astro-ph, 9512139,
1995.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIO G RAPH Y
62
[37] A. R. Klemola, B. F. Jones, and R. B. Hanson. Lick northern proper motion
program, i. goals, organization, and methods. A pJ, 94:501, 1987.
[38] Edward W. Kolb and Michael S. Turner. The Early Universe. Addison Wesley,
New York, 1990.
[39] L. M. KrauJfe and P. J. Keraan. 1994.
[40] Benjamin C. Kuo. Automatic Control Systems. Prentice-Hall, New Jersey, 1991.
[41] et al. Laura Ferrarese. The extragalactic distance scale key project, iv. the dis­
covery of cepheids and a new distance to mlOO using the hubble space telescope.
ApJ, 464:568, 1996.
[42] H. Leuenberger. Compilation of radiation shape factors for cylindrical assem­
blies. In American Society o f Mechanical Engineers, 1956.
[43] M. A. Lim, A. C. Clapp, M. J. Devlin, N. Figueiredo, J. O. Gundersen,
S. Hanany, V. V. Hristov, A. E. Lange, P. M. Lubin, P. R. Meinhold, P. L.
Richards, J. W. Staren, G. F. Smoot, and S. T. Tanaka. A p J Lett., page ac­
cepted, 1996.
[44] P. R. Meinhold, A. O. Chingcuanco, J. O. Gundersen, J. A. Schuster, M. D.
Seiffert, P. M. Lubin, D. Morris, and T. Villela. Advanced cosmic microwave
explorer: A millimeter-wave telescope and stabilized platform. ApJ, 406:12-25,
1993.
[45] Peter R. Meinhold. Anisotropy Measurements of the Cosmic Background Radi­
ation at 3 mm Wavelength and an Angular Scale of 30 Arcminutes. PhD thesis,
University of California, Berkeley, 1990.
[46] C. B. Netterfield, M. J. Devlin, N. Jarosik, L. Page, and E. Wollack. 1996.
[47] E. T. Olsen. Accurate positions of selected 4c sources. ApJ, 72:738-746, 1967.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIO G RAPH Y
63
[48] A. A. Penzias and R. W. Wilson. Astrophys. J., 142:419, 1965.
[49] A. C. S. Readhead, C. R. Lawrence, S. T. Myers, W. L. W. Sargent, H. E.
Hardebeck, and A. T. Moffet. ApJ, 346:566-587, 1989.
[50] H. J. A. Rotterging, M. Lacy, G. K. Miley, K. C. Chambers, and R. Saunders.
Samples of ultra-steep spectrum radio sources. Astr. Ap. Suppl., 108:79-141.
1994.
[51] J. E. Ruhl, M. Dragovan, S. R. Platt, J. Kovac, and G. Novak. A pJ Lett..
453:L1, 1995.
[52] R. K. Sachs and A. M. Wolfe. Perturbations of a cosmological model and angular
variations of the microwave background. ApJ, 147:73, 1967.
[53] B. Schaefer. The peak brightness of snl895b in ngc5253 and the hubble constant.
A pJ Lett., 447:L13, 1995.
[54] J. Schuster, T. Gaier, J. Gundersen, P. Meinhold, T. Koch, M. Seiffert, and
P. Lubin C. A. Wuensche. A pJ Lett., 412:L47-50, 1993.
[55] P. F. Scott, R. Saunders, G. Pooley, C. O’Sullivan, A. N. Lasenby, M. Jones,
M. P. Hobson, J. Duffet-Smith, P. J., and J. Baker. ApJ Lett., 461:L1, 1996.
[56] Robert Siegel and John R. Howell Thermal Radiation Heat Transfer. Hemi­
sphere Publishing COmpany, New York, 1980.
[57] Joseph Silk. ApJ, 211:638, 1977.
[58] G. F. Smoot, C. L. Bennet, A. Kogut, E. L. Wright, J. Aymon, N. W. Boggess,
E. S. Cheng, G. De Amici, S. Gulkis, M. G. Hauser, G. Hinshaw, P. D. Jackson,
M. Janssen, E. Kaita, T. Kelsall, P. Keegstra, C. Lineweaver, K. Loewenstein,
P. Lubin, J. Mather, S. S. Meyer, S. H. Moseley, T. Murdock, L. Rokke, R F.
Silverberg, L. Tenorio, R. Weiss, and D. T. Wilkinson. ApJ Lett., 396:L1, 1992.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIO GRAPH Y
64
[59] Mark Srednicki, Martin White, Douglas Scott, and Emory F. Bunn. Phys. Rev.
Lett., 71:3747, 1993.
[60] Suzanne Staggs. An absolute measurement of the cosmic background, radiation
temperature at 1.4 GHz. PhD thesis, Princeton Univ., NJ., 1993.
[61] S. T. Tanaka, A. C. Clapp, M. J. Devlin, N. Figueiredo, J. O. Gundersen,
S. Hanany, V. V. Hristov, A. E. Lange, M. A. Lim, P. M. Lubin, and G. F. Smoot
P. R. Meinhold, P. L. Richards. A pJ Lett., page accepted, 1995.
[62] Stacy Tanaka. Private Communication.
[63] Max Tegmark and George Efstathiou. A method for subtracting foregrounds
from multi-frequency cmb sky maps. MNRAS, page accepted, 1995.
[64] Martin W hite and Mark Srednicki. ApJ, 443:6-10, 1995.
[65] R L. W hite and R. H. Becker. A new catalog of 30,239 1.4 ghg sources. Ap. J.
Suppl., 79:331-467, 1992.
[66] B. C. Whitmore, W. B. Sparks, R. A. Lucas, F. D. Macchetto, and J. A. Biretta.
Hubble space telescope observations of globular clusters in m87 and an estimate
of h 0. ApJ, 454:L73, 1995.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A p p e n d ix A
A C E D ew ar
While BEAST could conceivably be done with expendable cryogens, ACE
needs to depend on refrigeration technology for cooling the radiometer for such a
long time. We envision a multi-stage refrigeration system. At the core is a 20 K
sorption cooler. This is a closed-cycle refrigerator utilizing Joule-Thompson cooling
of the hydrogen gas refrigerant. Pumping is effected by cyclic heating of sorption
beds that store and absorp the the gas. Reject heat removal and radiation shielding
is provided by two mechanical coolers.
A .l
R equirem ents
The ACE dewar has several extreme design constraints that it m ust follow.
It must contain and cool a large amplifier array, consume little power and weigh
very little. Internal dissipation is approximately 1 W. The refrigerators are allocated
~200 W, primarily for the sorption cooler. The sorption cooler must be able to reject
1.3 W to the 65 K stage. The load on the 20 K stage cannot exceed 500 mW. It must
weigh < 100 Kg. The size of the array necessitates insulating a ~ 1 m 2 cylindrical
surface at 300 K from the 20 K amplifiers which will occupy a ~ 9’ diam eter disk.
Descoping for BEAST removes most of the difficulty.
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPEND IX A. AC E DEWAR
A .2
A. 2.1
66
D esign
M echanical
In general, materials and assembly techniques were chosen to duplicate IR
labs dewars. The dewar shell has a complicated internal geometry. Mostly this is to
accomodate the sorption cooler while minimizing 300K surface area. For example,
the 60K shield line must pass between the heat exchanger coils. But the 150K shield
must also go there, otherwise the 150K surface area will be excessive. The length
of the sorption cooler is not fixed as of this writing. However, adjustment can be
made in the length of the neck, or more conveniently, in the heat strap to the cold
plate. The exact diameter of the shell was defined by what was commonly available
aluminum tubing.
The internal frame is made of G-10 columns. This provides adequate struc­
tural strength with low thermal conductivity and low coefficient of thermal expansion.
The radiation shields are 1/32 stainless steel, rolled and welded. They have been aluminized to reduce emissivity. The cold plate and HEMTs are to be attached to the
end of the G-10 truss. Access to the radiometer is provided by detaching the shields
at their connection to the truss. A one inch diameter hole provides mechanical feed
through from the 300K to 20K side. Most of that area is taken up by the sorption
cooler lines and heat clamp. An overview mechanical drawing is shown in Figure
A.I.
A .2.2
Therm al
The thermal design centers around the sorption cooler requirements. The
main source of heat flow is radiation, both off of the walls and through and from
the windows. Conduction has been minimized by th e choice of G-10 to support
and separate the layers and by the use of coaxial cable to transmit the amplified
signed from the 20 K to the warm electronics. Extensive testing has shown that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPEND IX A. ACE DEW AR
67
Figure A.l: A correctly dimensioned sketch of the main dewar components. The
sorption cooler is attached to the flange at the end of the dewar neck and is self
supporting. The G-10 truss provides all internal support including the cold head and
radiometer. The shield design is actually more complicated than shown. Identified
parts: A) Foam window. B) Cold head. C) G-10 truss. D) Sorption cooler. Note
the heat exchanger coils. Full mechanical drawings are available
aluminizing a metal surface can achieve an emissivity of 0.03, just about necessary
for this design. Simple application of muliti-layer insulation (MLI) has not proved
to be vastly superior to aluminizing. A two-cooler design is able to work on paper,
but has a narrow margin of success. Presently, an extra 10-20% heat load on the
20 K stage will be a failure It is possible to reduce the loads on the 20 K and 65 K
stages by adjusting the temperature of the 150 K stage down. However, this comes
at the expense of added power consumption by the mechanical coolers. Should two
mechanical coolers be inadequate, it is conceivable to add a third. Another option
would be to use something else as the core refrigerator. A LHe4 can with a capacity
of a ~ five liters could provide sufficient cooling for one night. Heat flow and rough
numbers are given in Figure A.2.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPEND IX A. ACE DEW AR
68
H e a t F low D ia g r a n
( V a l u e s in V a t t s )
A n b ie n t L a y e r
5 .4 6
65 K L ayer
0 .3 9 6
1.3
0 .3 2 0 2 0 K
C o ld
H ead
0.01 G
1.93
0.22
-
R e fr ig e ra to r
lo a d s
HEMT D i s s i p a t i o n
— - R a d ia tiv e L o ad s
Figure A.2: Direction and magnitudes of the heat flow in the dewar to and from
individual layers is shown. Values are in watts. For many of the assumptions that
went into the calculations, consult Appendix B.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A p p e n d ix B
L D B D ew ar T h erm al
C a lcu la tio n s
B .l
R adiative Transfer
In the early stages of design, the most uncertain issue was the radiative
transfer between the walls of the dewar. From the accounting sheet below, one sees
that radiation is a substantial proportion of the heat flow into the cold stages. Initial
design requirements were even more stringent and an accurate model was needed to
understand whether the design was even possible
Consider two infintesimal elements of area, d A t and dAa, separated by a
distance r. The differential flux from d A x to dAa is
d ^ ia = / I/dA l - r ^ 2 _ £
lr l
(B.l)
where /„ is the intensity of radiation from dAx (generally a function of frequency
and angle). The total flux from d A x is
K =f
IvdA^rdfl
(B.2)
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX B. LDB DEW AR TH ERM AL CALCULATIONS
70
We define the differential view factor from dA j to d A 2
dFxn = - ^
(B.3)
One obtains the full view factor for two finite surfaces F\n by integrating
over the infinitesmal areas. There are numerous solutions obtained numerically and
analytically in the engineering literature [56, 42]. Note th at if I„ is isotropic, A iF\n =
A 2 F 12 . This is the reciprocity relation. For an enclosed space, such as a dewar,
conservation of energy implies that for a surface i in a volume bounded by N surfaces
N
£ ^ - = 1
j =1
(B.4)
For a given surface we can then write the equation of conservation of energy and note
that in most applications, the bolometric flux C = ecrT4, where e is the emissivity. a is
Boltzmann’s constant and T is the thermodynamic temperature. A solution relevent
for dewar design is that for twosimilar surfaces, Ai and A 2, enclosing a volume, with
.
A 1 interior to An. Examplesare concentric infinitecylinders or concentric spheres.
The heat flow from Ai to An is
_ AMT?
912
~
-L
£1
- To4 )
4.
'r
£2
a,
A -2
(B.5)
where Ti and e* are the thermodynamic temperature and emissivity of the respective
surfaces. When Ai
A 2, Equation B.5 tends toward the solution for a closed surface
in a black-body radiation bath at tem perature To,
912 = A X€Xa{T^ - T$)
(B.6 )
If Ai « .4o and ei = en, Equation B.5 tends toward
qi 2 = 2A 1eio-(r 14 - T$)
(B.7)
This is the equation that is the closest and simplest approximation to the actual
radiative transfer and is adequate for design at the 10-20% level The next level of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPEND IX B. LDB DEW AR THERM AL CALCULATIONS
71
analysis is to break the geometry up into N surfaces. The radiative transfer is found
by solving
Aq = C
(B.8 )
Where A is the matrix of F \ 2 and emissivities, q is the vector of heat flow from the
surfaces, and C is the vector of temperature boundary conditions. Explicitly, we
write Aij = 6ij — (1 —ti)F ij, C* = ejcrT^. A shortcoming of this technique is the
great complexity th at some view factors can have. In this case, one may have to use
approximations and Equation B.4 will not be satisfied. The most general technique
would be to numerically integrate the real view factors and solve Equation B.8.
B .2
T herm al A ccounting
This spreadsheet contains the calculations used in the thermal model for
the LDB dewar. As this is an engineering model, there are some approximations.
The radiative loads model each layer as concentric right cylinders, with uniform
emissivity and temperature. Temperature gradient effects on the radiative transfer
are neglibible for Layers 2 and 1, but are about 50 mW at layer 0. Conductive
paths are modeled from the mechanical drawings. Two loads are calculated for each
layer. The left load is for a dewar with waveguide, the right load is for co-axial
cable. The maximum load allowed on a layer is given in brackets. In the case of the
Layers 1 and 2, this is the load at which the sunpower coolers would fail to maintain
the temperature in the modeL For layer 0 this is the limit of the sorbtion cooler.
The largest uncertainty is in the modeling of the window. In particular, it has been
assumed the e -+- a = 1, that is, the reflectivity is zero. For most window materials,
transmission will be ~ 0.8 in the far infrared, but has not yet been measured at the
wavelengths of interest in likely materials, table
Table B.l: Next two pages. Calculation for dewar heat flow and loads. This is a
static model, started by Mike Seiffert, finished by the author and corrected by Peter
Meinhold.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX B. LDB DEWAR THERMAL CALCULATIONS
72
Sorption cooler radiatjye load
2 sunpower/2shield model
1
L a y e r s P aram eters:
l e m ia a iv it y :
[ r a d i u s (in)
L av er 2 P aram eters:
ie m issivity .
Ir a d i u s ( i n ) •
L a y er 1 P aram etera i
re m issivity
ir a d iu s ( in )
0 .03:T em p(K )
6 il ang th( in )
i
300
2 2 . 7 5 ' A r a a ( m A2 ) [
0.69
0 .0 3 'Tem p(K )
5.6 2 5 Jength(in)
150;
1 8 . 5 [ A r e a ( m A2 ) f
0.S4
0.03 !T em p (K )
5 . 2 5 il eng lh( in )
65
l 5 . 5 i A r e a ( m A2 ) '
0 .4 2
L a y e r 0 P a r a m a l a r a^
ie m issivity (
______ i r a d i u s (in)
W indow S iz e s
0 . 0 8 ‘T e m p ( K )
20!
____4 . 7 5 fl e n g t h e n ) ____ _____ 8 [ A _ re a( m ^ 2 )[
* W in d ow e m i s s r v i e s
6 . 0 0 id ia (in)
W av e g u id e
F e e d th rue:
L o a d o n L i y i r 2 (W )
R i d i l l l o j ) ] _____________
____________ | Q 3 2 (W )
I Q21 (W)
m o u n t s L 3 ~ > L 2 fi-10
[num ber
ilength(m )!
l a r e a (m A2)'
i300 K in te g ra l W /m
! 15 0K i n t e g r a l W / m
m o u n t s L2->L1
la y e r 3 w in d o w
[
•am ission (W )
( win dow d i a m e t e r (in)
( w in do w a r e a ( m A2)
:ef fec ti ve t e m p (K)
(w in d ow e m i s s i v i t y
a m b ie n t rad iatio n
l a y e r 2 w in d o w
w i r e s L 3 -> L 2 (32A W Q )
n u m b a r of w i r e s
( le n g th p e r w ir e (m )
;300K i n t e g r a l (W /m )
fl5 0 K integral (W /m )
l A r e a (m A2)
w ir e s L 2-> L1
w a v e g u i d e L 3 - > L 2 (§28 S S )
(number
' l e n g t h (m)
'300 K integral (W /m )
1 15 0 K i n t e g r a l ( W / m )
■Ares ( m A2)
w a v e g u i d e L 2 - > L l (a28)
0.8 5 i
iA m pl ifi e rs
C o-ex
;______________ 7 .7 1 lflQ .0 1 _________________ 7 . 4 4
3.81
4 .0 1 r
-0.20i
I
0.29
4!
- 7.62E-02*
i 8.71 E - 0 5 i
99.5^
361
-
:
------------
!
•
!
i
r
-0.12'
2.97 [
3 .4 9 *
6.00^
1.82E-02!
250
0.85!
.
-
...
•
-
t
___ c o i i ^ __
- 0 .2 3
---------- —
-
_2^75 2[2 .7]
R ad iatio n !
0.19r
iQ 2 1 (W)
0.20 •
[ Q 10 (W )
•0.01
m o u n t s L 2 -> L 1 6*10
0.12:
______ [ n u m b e r
____________ ___
A'
jlen g th (m )i
__
7.62 E -0 2 :
{are a ( m A2)j '
■ 8.71E-0S*
USOK i n t e g r a l W / m
■________ 3 6 _
!65 K i n t e g r a l W / m
«- 4 i
m o u n t a _ L 1 j ^ > L O ^ Q - 1 0 ___ ______ ,__ ____ ___________ :o .o _ i_______
.
!
1
•
__________
0 .3 8 *
^
_
^!1*^5E«<)1;
_________ _________________2 . 6 5
.............. :*• -
0.45~
6 .00
i .82E-02* "
iso ;
0 .8 5 1
. . . .
' c o a x 0 .0 8 5 A g / B e C m 2.37 E -0 1 1
num ber
8*
l e n g t h (m )
0.15'
•H e a t F low ( m W * e m )
450.6726
8'
0 .15
3060!
1000'
5 .6 8 E - 0 6 1
_
.
.
1.82E -02
300
1
i
-0.07
0 .61
—
. _
!
50;
0 .1 1
; 1 1 7 4 7 . 6|^
3373.68 ~
. 3 . 2 4 E *08 i
--
a m b ie n t rad iatio n
8.51
"
,_l .
1.0®
-0.77:
0 .14 =
L o a d o n L i y e r 1J W J ____ r —
la y e r 2 w in d o w
R em ission .
■window d i a m e t e r (in)
___ -w in do w a r e a ( m A2)
__-effective t e m p ( k ) _ ^
____ {window e m i s s i v i t y
_
_
.
.....
_
. _... . _
.
:
i
•
|
*, _____________ ____________________ ;________ ________
T
:
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
APPENDIX B. LDB DEW AR THERM AL CALCULATIONS
a m b ie n t rad iatio n
L3 w i n d o w
0.1 6 1
o.at
'
w i r e s L 2 -> L 1 (I2A W Q )
m u m b a r of w ir es
i l a n g t h p a r w i r a (m)
1 5 0 K in t e g ra l ( W / m )
j « 5 K in t e g r a l (W / m )
' A r e a ( m A2)
w iraa L1->L 0
>0.03
t
0.07
:
86i
0.1:
337 3 .8 8
8 21 .0 8 ’
!
3 .2 4 E -0 8 '
-
.
- *
...
.
•
i
-6 .5 lE -0 3 [
....
0.23'
jnum b ar
I la n g th (m)
j 1 SOK i m e g r a ^ W /m
i6 5 K Jn teg raj W /m
■A r e a ( m A2)
w i v i g g l d 'i L 1->L0 C 23 S S )
73
8
0.15
1000
235’
5.68E-06~
-0.05 i
S o rb tlo n c o o le r reject
.
- -
-
-
•
-
c o a x 0 .0 8 5 A g /B a C u : 0.1 1 5 1 3 2 1
m um bar
8’
.length ( m )
0.15
coax
-3.54E -02 •
1 .3 0 1
0 .4 2 2
L o a d o n L i y » r 0 (W)
HEMT
‘[0 .5 0 0 ]
0 .4 0 9
0.160'
mum bar
idissipation ( W |
la y a r 1 w in d o w
window diam a t a r (in)
. w i n d ow a r e a ( m A2 ) __
refl ect iv e j a m p (K)
■window am ia ai v it y
a m b ia n t rad iatio n
L3 w i n d o w _________
L2 w i n d o w
w i r a a L 1 - > L 0 (32A W Q )
m u m b a r of w i r e s ___
[ l e n g t h p a r w i r a (m)
| 6 5 K int e g r a l ( W / m )
I 20K in t e g r a l ( W / m )
A r e a ( m A2)
-4Ar
m ounta L1 -> L 0
____________ m u m b a r
_ jj en g th _ ( m }_
I6SK in t e g r a l ( W / m ) ~
_|20 K _jn ta gr ml (W/m^)_
7 |A r a a ( m A2)
0 .0 2 .
0.0161
6 .0 0
1 .8 2 E -0 2 ’
65^
o.asT
0.029^_
0.076 i
0 .0 6 8 J
O.OOf'
72,
0 .2 ?
j2 i_ .o ar
91.6:
3.24E -06
0.008
o ia
.
13.35’
1.01 E *04
Q L 1 - > L 0 (W)
0.006:
w a v e g u i d e L 1 - > L 0 ( 8 26 3 5 )
m um bar
jla n gth (m )
;6SK i n t e g ral W /m
|2 0 K i n t e g ral W /m
(m»2)
0.046!
0 .2 0 ‘
23 5T
18.35.6BE-06:
A g/B aC u i
m um bar
^ l e n g t h f m ) ____
H e a t Flow ( m W c m )
3^54E-02^
8;
£.20;
90!
A s s u m e s t h e d e w a r is o p e r a t i n g on t h a g r o u n d in a n a m b ia nj^ an vir on m a n t
W i n d o w l o a d s a r e a d d e d a s s u m i n g ref lec tivity of t h a w in d o w m a t e r i a l is
z e ro a n d th a t absorption * a m is s io n .
sig m a
i 5.76E-08
W /(m A2 * K M )
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
..
A p p e n d ix C
IR A S Low D u st R eg io n s
Table C .l lists ISSA [26] regions that were considered for observation in
MAX4 and MAX5. Each map is a 12.5 0 x 12.5 0 gnomonic projection centered on
the coordinates given in columns 2 and 3. The mean and rms were calculated for the
entire map. Given the existence of point sources and significant structure in some of
them, columns 4 and 5 can only be used as rough guides for the cleanliness of the
regions. Furthermore, it is possible for a clean region to be composed of subsections
of several maps. Therefore, this list not to be considered complete or definitive.
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
APPEND IX C. IR A S LO W D U ST REGIONS
Table C .l:
This is the list of the lowest RMS IRAS re­
gions. T he mean and rms were calculated for each 12.5 ° x
12.5°region.
RMS (MJy/'Sr)
Mean (M Jy/Sr)
-40
0.225
0.246
14h44m
40
0.278
-0.017
I047B4H0
04h00m
-50
0.287
0.171
I353B4H0
13h52m
40
0.310
-0.119
I381B4H0
16h00m
50
0.321
0.016
I356B4H0
16h23m
40
0.326
-0.040
I070B4H0
02h36m
-40
0.329
0.414
I355B4H0
15h36m
40
0.351
0.272
I042B4H0
22h48m
-60
0.373
0.598
I072B4H0
04h20m
-40
0.384
0.248
I376B4H0
llhOOm
50
0.390
0.203
I401B4H0
15hl2m
60
0.390
0.060
I099B4H0
03h04m
-30
0.411
0.381
I380B4H0
15h00m
50
0.420
0.201
I416B4H0
13h52m
70
0.427
0.398
I045B4H0
02h 00m
-50
0.438
0.652
I026B4H0
02h32m
-60
0.445
1.000
I098B4H0
02h l 8m
-30
0.450
0.301
I046B4H0
03h00m
-50
0.462
0.595
I071B4H0
03h28m
-40
0.463
0.239
I400B4H0
13h56m
60
0.485
0.154
I324B4H0
14h34m
30
0.506
0.200
I382B4H0
17h00m
50
0.547
0.470
ISSA Map
RA
I069B4H0
01h44m
I354B4H0
Dec
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
APPEND IX C. IR A S LOW D UST REGIONS
RMS (M Jy/Sr)
Mean (MJy/Sr)
60
0.565
0.064
05h22m
-30
0.597
0.475
I402B4H0
16h28m
60
0.621
0.345
I357B4H0
17h20m
40
0.660
0.947
I100B4H0
03h50m
-30
0.662
0.428
I325B4H0
15h20m
30
0.667
0.860
I101B4H0
04h36m
-30
0.671
0.674
I379B4H0
14h00m
50
0.684
0.040
I417B4H0
15h36m
70
0.685
1.014
I403B4H0
17h44m
60
0.704
1.476
I073B4H0
05hl2m
-40
0.769
0.601
I383B4H0
18h00m
50
0.772
1.521
I023B4H0
22h32m
-70
0.775
1.341
I415B4H0
12h08m
70
0.781
0.377
I293B4H0
16h06m
20
0.809
2.470
I397B4H0
10h08m
60
0.869
0.211
I418B4H0
17h20m
70
0.875
1.954
I326B4H0
16h06m
30
0.876
1.256
I048B4H0
05h00m
-50
0.894
0.589
I027B4H0
03h48m
-60
0.955
1.093
I041B4H0
21h32m
-60
0.979
1.708
I396B4H0
08h52m
60
1.025
1.685
I358B4H0
I 8 h l 2m
40
1.081
2.303
I049B4H0
06h00m
-50
1.121
2.770
I022B4H0
20h48m
-70
1.205
2.653
I428B4H0
18h00m
80
1.230
2.708
ISSA Map
RA
I398B4H0
llh24m
I102B4H0
Dec
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
3 058 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа