How we really measure the CMB

Ritoban Basu Thakur

Astro 448
Did I plug in the Isocurvature
amplifier ? fluctuations and
the slope of
slow-roll…

Did I include that TLS noise is

⇣
minus sign ? subdominant
q0 @µ
= 1

compared to
µ
A

SQUID …

⇣
q0 @µ
µ= 1
A
 Outline

The what-to and how-to of measuring ?

How to get the light in ?

How to measure the light ?

Some detectors in detail

The light
The light
CMB is ~2.7 K Black Body,
with maximal signal over
foregrounds ~100 GHz

Tiny intensity fluctuations in

frequency and location
contain a plethora of
interesting physics
The light: colors

Large area collector

+ collimator

Broad-band 2.7 K BB
spectrum measured
across three decades
 The light: colors

Not prisms …

Fourier Transform Spectrometry

at the heart measurements.

Z L
˜
I( 1
)=4 {I( ) 0.5 I( = 0)} cos(2⇡⌫ / )
L

Fourier transform of Interferogram

The light: colors

COBE FIRAS
has the best
spectral
measurements,
constraining
dI/I to .10-5

TB[K]
 https://lambda.gsfc.nasa.gov/product/cobe/

The light: colors

COBE Far Infrared Absolute Spectrophotometer (FIRAS)

Fits to the cosmic spectral distortion

parameters give 95% confidence limits of
|μ/kT| < 3.3 x 10-4 and |y| < 2.5 x 10-5 -
The light: colors

Lot of interesting physics in CMB Spectral Distortions

Probing energy injection from DM interactions / decays

Recombination lines etc. Several papers by Jens Chulba et al.
 The light: colors

Proposed PIXIE satellite FTS can do

much better than FIRAS can open up
interesting avenues. Ask Steve Meyer!
https://arxiv.org/pdf/1105.2044.pdf
The light: patterns
`min = ⇡/(N ✓res ) Map-size

`max = ⇡/✓res Beam-size

Relation between time/frequency

and ell, given a scan speed
Small(ish) area collector
+ Scanning
f [Hz] vx [rad/s] vx [deg/s]
= =
` ⇡ 180
straints are approaching that level. However, there is no such limit on the B-modes,
The light: patterns
s (and a potential detection) with CMB-S4 will rely on measurements of C`BB – most Very roughly, sensitivity to some
4
egree-scale “recombination” feature in the primordial B-mode spectrum.
ΔT scales with # detectors
Angular scale
4 10 ° 1° 0.1 ° Space based experiments
10 −1 Stage−I − ≈ 100 detectors
10
Temperature

Approximate raw experimental sensitivity (µK)

Stage−II − ≈ 1,000 detectors
Stage−III − ≈ 10,000 detectors
W Stage−IV − ≈ 100,000 detectors
M
AP
2
10
( + 1) C / 2π µK 2

−2
10
0
10 es
od
m
E-
Pl
an
ck

-2 BK14
10 −3
10
r=0.05 odes POLARBEAR
B-m s SPTPol
GW ode
m
-4 g B- CM
10 r=0.001 n sin B−
Le S4
−4
10
10 100 1000 2000 2005 2010 2015 2020
Multipole moment Year
Figure 2. Plot illustrating the evolution of the raw sensitivity of CMB experiments
the total
retical predictions for the temperature (black), E-mode number
(red), of bolometers.
and tensor Ground-based CMB experiments are classified into Sta
B-mode (blue)
experiments
imordial B-mode spectra are shown for two representative having
values O(1000)
of the detectors, Stage III experiments having O(10,000) detectors
tensor-to-scalar
Noise-Equivalent-Temperature (NET) of a detector is measured /estimated.
experiment
nd r = 0.05. The contribution to tensor B-modes from (such
scattering as CMB-S4) having
at recombination peaksO(100,000) detectors. Figure from Snowmass CF5 N
document.
m reionization at ` < 10. Also shown are expected values for the contribution to B-modes
ly lensed E-modes Temperature-power
(green). Current measurements of of desired
the B-mode CMB
spectrum feature is noted (Px)
are shown
Array (light orange), POLARBEAR (orange),1.2.1
and SPTPol
Raw (dark orange). The
sensitivity lensing
considerations and detector count
e B-mode spectrum can be partially removed by measuringNET/N
the E and dets < Ppotential)
(lensing x
exploiting the non-Gaussian statistics of the lensing, while the foreground contamination
by a multi-frequency experimental strategy (forThe
details see Section
sensitivity 2.3). measurements has increased enormously since Penzias and W
of CMB
1965, following a Moore’s Law like scaling, doubling every roughly 2.3 years. Fig. 2 sho
In detail …
 so that improvements (and a potential detection) with CMB-S4 will rely on measurem
likely targeting the degree-scale “recombination” feature in the primordial B-mode spe
The light: patterns Suppose we want to discover E-modes at l ~103
Angular scale
4 10 ° 1° 0.1 °
10
Temperature

2
10

( + 1) C / 2π µK 2
Beam has to small enough 10
0
d es
o
Need to resolve our mode E-
m

⇥ ⇤ 3 -2 BK14
✓res rad < ⇡/(` = 10 ) 10
r=0.05 m o des POLARBEAR
B- s SPTPol
G W de
o
-4 g B-m
10 sin
Sky has to be big enough r=0.001 Le
n

Need to capture a wavelength

⇥ ⇤ 3
10
Multipole moment
100 1000

✓sky rad > ⇡/(` = 10 )

Figure 6. Theoretical predictions for the temperature (black), E-mode (red), and tens
power spectra. Primordial B-mode spectra are shown for two representative values of th
ratio: r = 0.0012and r = 0.05. The contribution to tensor B-modes from scattering at rec
Here sky area mapped is ~θ Skyfrom reionization at ` < 10. Also shown are expected values for the contrib
at ` ⇠ 80 and
from gravitationally lensed E-modes (green). Current measurements of the B-mode spe
for BICEP2/Keck Array (light orange), POLARBEAR (orange), and SPTPol (dark oran
contribution to the B-mode spectrum can be partially removed by measuring the E and (
modes, as well as exploiting the non-Gaussian statistics of the lensing, while the foregrou
can be mitigated by a multi-frequency experimental strategy (for details see Section 2.3).

*for a real experiment this will be > 100 Hz

 so that improvements (and a potential detection) with CMB-S4 will rely on measurem
likely targeting the degree-scale “recombination” feature in the primordial B-mode spe
The light: patterns Suppose we want to discover E-modes at l ~103
Angular scale
4 10 ° 1° 0.1 °
10
Temperature

2
10
The experiment needs sensitivity

( + 1) C / 2π µK 2
Dl >1 μK2 or Cl > 2π (nK rad)2 10
0
d es
o
m
E-

-2 BK14
10
r=0.05 o des POLARBEAR
m
B- s SPTPol
G W de
o
-4 g B-m
10 r=0.001 n sin
Le
Suppose that we are scanning
10 100 1000
at 1 deg/sec = π/180 rad/s Multipole moment

Figure 6. Theoretical predictions for the temperature (black), E-mode (red), and tens
-> f (l =103)= 3
10 /180 = 5.55 Hz
power spectra. Primordial B-mode spectra are shown for two representative values of th
ratio: r = 0.001 and r = 0.05. The contribution to tensor B-modes from scattering at rec
at ` ⇠ 80 and from reionization at ` < 10. Also shown are expected values for the contrib
-> Sampling rate >> 11.11 Hz* from gravitationally lensed E-modes (green). Current measurements of the B-mode spe
For every 0.18 sec of scanning time
for BICEP2/Keck Array (light orange), POLARBEAR (orange), and SPTPol (dark oran
and sampling time >> 0.18 sec contribution to the B-mode spectrum can be partially removed by measuring the E and (
we collect one more l =10 mode
3
modes, as well as exploiting the non-Gaussian statistics of the lensing, while the foregrou
can be mitigated by a multi-frequency experimental strategy (for details see Section 2.3).
The light: patterns Suppose we want to discover E-modes at l ~103

The experiment needs sensitivity Cl > 2π (nK rad)2

Suppose that one detector has noise RMS given by w1 [μK/√Hz],

simplified white noise power spectral density

Suppose we have Ndets and we are scanning for Ts ( >> 0.18 s) s

We know that the noise variance will scale with 1/Ndets

Longer scan duration (Ts) -> more modes captured, i.e. higher SNR

2 2
w1 ✓sky ⇥ ⇤2
 2⇡ nK-rad
Ndets Ts

This is the (Knox) sensitivity limit for our case

The light: patterns

detector output [nV, Hz, μA…]

Receiver Amplifier Detector time [s]

FFT
The noise RMS w1 [μK/√Hz], is the white
noise level value, or the median value of
the noise power spectral density,
referenced to a temperature scale.
 The light: patterns
4 H. McCarrick, D. Flanigan, G. Jones, B. R. Johnson, P. A. R.
5 K. N. Abazajian, P. Adshead, Z. Ahmed, S. W. Allen, D. Alonso,
MKID
Example of noise power spectra
FIG. 4. Thefor
spectral density of
two different the fractional
detectors, in frequency time
series data at module temperatures 65 mK and 114 mK, as
units of ydet vs. of
well as probe tone powers
frequency
114 and 111 dBm at the
feedline. To ease comparison between the di↵erent test con-
ditions the noise floor set by the low noise amplifier (Samp )
was modeled and subtracted from the data. The data have
been binned in logarithmic frequency intervals for clarity. The
dark portion of each curve shows the data used in the fits.
The best fit models are shown by the gray lines. The black
6 H. G. Leduc, B. Bumble, P. K. Day, B. H. Eom, J. Gao,
dashed lines show the red noise components of each fit. The
blue dotted line shows the white noise level in the detector
band (Sw ) for the 114 mK data. The spur at 1.4 Hz is due to
the pulse tube cooler. The steeper noise below 1 Hz is largely
correlated between detectors and is suspected to be caused by
module temperature fluctuations and microphonics.
7 A. E. Lowitz, A. D. Brown, V. Mikula, T. R. Stevenson, P. T.
TES
nating these LEKIDs and studying their noise properties
8 S. W. Deiker, W. Doriese, G. C. Hilton, K. D. Irwin, W. H.
in more detail. An added advantage of low-Tc films is
that photons with frequencies much greater than ⌫c will
4
H. G. LeDuc, M. Limon, P. Mauskopf, A. Miller, C. Tucker, and
J. Zmuidzinas, “Development of dual-polarization lekids for cmb
The noise RMS w [μK/√Hz], is the
observations,” (2016). 1
Noise Equivalent Temperature (NET)
Ade, D. C. Araujo, K. Bradford, R. Cantor, G. Che,
P. Day, et al., “Horn-coupled, commercially-fabricated aluminum
lumped-element kinetic inductance detectors for millimeter wave-
lengths,” Rev. Sci. Instrum. 85, 123117 (2014).
Measurement / detector units are
K. S. Arnold, C. Baccigalupi, J. G. Bartlett, N. Battaglia, B. A.
usually Voltage, frequency etc. as a
Benson, C. A. Bischo↵, J. Borrill, V. Buza, E. Calabrese, R. Cald-
well, J. E. Carlstrom, C. L. Chang, T. M. Crawford, F.-Y. Cyr-
Racine, F. De Bernardis, T. de Haan, S. di Serego Alighieri,
function of time (or freq. by FFT). Let’s
J. Dunkley, C. Dvorkin, J. Errard, G. Fabbian, S. Feeney,
S. Ferraro, J. P. Filippini, R. Flauger, G. M. Fuller, V. Glusce-
call these units y
vic, D. Green, D. Grin, E. Grohs, J. det.
W. Henning, J. C. Hill,
R. Hlozek, G. Holder, W. Holzapfel, W. Hu, K. M. Hu↵en-
berger, R. Keskitalo, L. Knox, A. Kosowsky, J. Kovac, E. D.
Kovetz, C.-L. Kuo, A. Kusaka, M. Le Jeune, A. T. Lee, M. Lil-
ley, M. Loverde, M. S. Madhavacheril, A. Mantz, D. J. E. Marsh,
J. McMahon, P. D. Meerburg, J. Meyers, A. D. Miller, J. B.
Munoz, H. N. Nguyen, M. D. Niemack, M. Peloso, J. Peloton,
L. Pogosian, C. Pryke, M. Raveri, C. L. Reichardt, G. Rocha,
A. Rotti, E. Schaan, M. M. Schmittfull, D. Scott, N. Sehgal,
1. Convert this to NEP or noise
S. Shandera, B. D. Sherwin, T. L. Smith, L. Sorbo, G. D. Stark-
man, K. T. Story, A. van Engelen, J. D. Vieira, S. Watson,
equivalent power.
N. Whitehorn, and W. L. Kimmy Wu, “CMB-S4 Science Book,
First Edition,” ArXiv e-prints (2016), arXiv:1610.02743.
S. Golwala, B. A. Mazin, S. McHugh, A. Merrill, D. C. Moore,
O. Noroozian, A. D. Turner, and J. Zmuidzinas, “Titanium
2. Convert to NET by using a
nitride films for ultrasensitive microresonator detectors,” Ap-
plied Physics Letters 97, 102509 (2010), arXiv:1003.5584 [cond-
Black-Body-Jacobian
mat.supr-con].
Timbie, and E. J. Wollack, “Design, fabrication, and testing of a
tin/ti/tin trilayer kid array for 3mm cmb observations,” Journal
of Low Temperature Physics 184, 627–633 (2016).
Rippard, J. N. Ullom, L. R. Vale, S. T. Ruggiero, A. Williams,
and B. A. Young, “Superconducting transition edge sensor using
dilute AlMn alloys,” Applied Physics Letters 85, 2137 (2004).
The light: patterns Black-Body intensity I has dimensions of W/Sr/m2/Hz

1. Conversion to NEP or noise equivalent power:

Measure ydet for varying incident BB power and obtain the responsivity,
as a function of frequency (or FFT of time streams)

ydet Spectral density of

R= ! Pdet (f ) = R(f )ydet (f )
P this power is the NEP

2. Conversion to NET by using a Black-Body-Jacobian

Z
P (T ) = d⌫ dA d⌦ (W (⌫) · I(⌫, T ))
@P
! J=
@T
w1 = NET = NEPJ 1
The light: patterns NEP and photon noise limit

Noise Equivalent Power: input signal power that produces SNR = 1

at the output of a detector, given data-signaling rate / modulation
frequency, and effective noise bandwidth.

Thus it is the minimum detectable power per √bandwidth.

The response of a detector can vary with frequency: NEP(f)

 The light: patterns NEP and photon noise limit

SuperSpec:
For a detector with negligible intrinsic (thermal) andareadout
mm-wave on-
(laboratory) noise, photon counting determinesfor the high-redshift
measurement limit as

Shirokoff

p
hnrms i = n + n2

h⌫ ⌧ kB T h⌫ kB T
Bose bunching Poisson

Discussion about Noise Equivalent Power and its use for photon noise calculation. Samuel Leclercq. 2007-03-02.
 The light: patterns NEP and photon noise limit

For ground based experiments

NEP (CMB + hot optics + hot Sky)
limit is ~50 aW*/√Hz

SPTpol, single detector <102 aW,

therefore with >103 detectors
CMB can be measured.

*1 aW = 10-18 W
Journal of Low Temperature Physics June 2012, Volume 167, Issue 5, pp 865–871
 PTER 5. DETECTOR PERFORMANCE 104

The light: patterns

amplifier noise. In converting to NEP, these current sources are divided by the current
NEP and photon noise limit
onsivity, which reduces to vb at high loop gain. Additional frequency dependent terms
account for the bolometer response have been omitted, as these terms do not become
ficant until well outside of both the SPT data bandwidth and the digital anti-aliasing
— Here, Psat is the power measured
detector w/ at turn-around (L = 1) estimated from da
applied prior to re-sampling to a 100 Hz data rate.
optical
dark detector
t of the
p array. For all devices, loading
Tbath = 0, 1mK and Tc = 550 ± 30 mK. G = @P/@
ξ =275
aW/ Hz, are the expected contribution from thermal carrier noise, NEPcar est , th
dout system and bolometer Johnson noise, NEPcar est , and the total which include
nd a Bose term for a correction factor of ⇠ = 0 1, NEPtot est . Measured noise
quency band from 3 5 Hz during a single 10 minute observation in good weath
e common mode has been removed from each wedge at every time step to remo
heric low frequency noise.

p
re 5.6 Measured NEP for individual detectors in the season 3 focal plane. Data from a
e 10 minute stationary observation in August 2009 is shown in gray. Predictions based
power P
model parameters areopt will be hn
in black. On the left, a dark bolometer from wedge SA12,2with
i= n+n rms
= 15 pW, and a plot of the predicted NEP with no optical loading. On the right, an
✓
al device from wedge X16, with Psat = 24 pW, Popt = 10.5 pW, and ⌧opt = 25 ms. The
ed line includes predicted NEP for all non-photon contributions. Lower and upper solid2
◆ ✓ 2
◆
are the total noise for Bose factor ⇠ = 0 and
⇠P
2 ⇠ = 1, respectively. In all cases, theopt f
NEP = 2h⌫ P + 0 curves.
onse of the digital anti-aliasing filter has been applied to predicted opt · 1+
⌫ (2⇡⌧opt )2
he measured bands, measured optical loading, and resulting predicted optical NEP for
SPT season 3 focal plane is listed in table 5.2. Device and bias parameters2and dark
+4 k T GNE
contributions from both carrier noise and readout noise, as well Btotalc expected
as the
are listed in table 5.3. The measured NEP during a 10 band-center
minute noise stare inband-width
good
2
(L + 1)
her in November 2009 is also listed. In general, the agreement between predicted and
+ 4k T P
ured NEP is reasonably close, especially considering that there is likely to be residual
Bose-bunching- B mode
amination from changing atmosphere as well as additional low frequencyc noise collection
bias from the 4
efficiency Shirokoff
Let the light in
 Let the light in

WMAP

Penzias
& Wilson
 Let the light in
Reciprocity theorem:

Receive and transmit properties of an antenna are identical.

Radiation pattern in transmit mode = pattern in the receive mode.
 Let the light in

~1% of TV noise is the CMB

But we can do better using horns …

 Let the light in

Shape the lobes, so as to focus

the CMB from the sky and not
pick up terrestrial junk

Horsing around, but more

seriously….
 Let the light in

A well defined
corrugated horn
antennae can
couple the
entire power in
the main lobe
 Let the light in

Planck

SPTPol
 Let the light in

A horn has one well defined length-scale (LH), and thus the
response or gain will drop for all ν < c/LH

But ideally we want to pick up “equally” at large bandwidths and

then chop it into bands that we care for

r
te
La
 Let the light in

Log periodic / fractal shape is “coherent” across decades

The broad-band signal is segmented by lumped-element on-

chip filters (RLC) and passed to detectors for measurement

SPT3G, PB2 etc.

 SPT-3G Receiver Assembly at SiDet
Let the light in
SPT-3G Optics
Layout 0.75 meters
Prime Vacuum Window
Focus
50 K Field Lens

50 K Radiation
Shield

Field Lens 4 K Aperture Lens

Secondary
Mirror 4 K Lyot Stop and
Infrared Filters
2.107 meters

Aperture Lens 4 K Radiation

Shield
Lyot Stop
4 K Collimating
Lens
Collimating
Lens
Focal Plane
Image plane

Various optical elements, lenses, collimators, filters etc. are non-trivial at

mm waves, especially for big (3G-like) instruments. R&D in microwave
material science is happening, and more effort is required
Seeing the light
 Seeing the light

At the heart of all CMB experiments, is quantum excitation

We will discuss:
WMAP receivers, SPT detectors and one CMBS4 technology
 Seeing the light: WMAP

WMAP looked at the difference in signals from two horns

The “radio” signals are amplified by HEMT amplifiers

The amplified power is measured with diodes

https://arxiv.org/pdf/astro-ph/0301164.pdf
 i
Combined responses, ravg (ν), of representative radiometers in each of MAP’s five frequency bands. Note that the frequency ranges of the Ka and Q
eters overlap slightly.
ch phase-matched leg the components resides. for the left and right detectors where sl and sr are th
dent gain of the HEMT amplifiers has been ities of the Seeing two detectors. theIn light: WMAP
general these respon
to emphasize that it is not constant, as is usu- function of the input microwave frequency, but for
t is assumed that all the components between assumed frequency independent and equal so that
ees and 2 warm hybrid tees are perfectly phase- The outputs of the two detectors then become
y terms / kb Tintentional
A involving A ⌫ differential phase #! 2 2 "
s A +B
he legs are presented. The effects of the various Vl = + n21 g21 (t)
e discussed in subsequent sections. 2 2
! 2 2
"
nals ‘A’ and ‘B’ are thermally induced noise A +B 2 2
wo separate loads they are uncorrelated, and + + n 2 g 2 (t)
2
elations $
A=B=0 1. 2. (4) 3. ∓ (A2 − B2 )g1 (t)g2 (t)
AB = 0 (5)
and
AA ∝ kB ∆νTa (6) #! 2 "
s A +B 2
BB ∝ kB ∆νTb (7) 2 2
2 V r = ( + n 1 g 1 (t)
tzmann’s B / k
constant,
b T B ∆ν ⌫is the bandwidth over 2
! 2
2
"
A +B 2
es A and B are measured and Ta and Tb are the 2 2
ature of the loads in Kelvin. The over-bar indi- + + n 2 g 2 (t)
2
age with a period long compared to the period $
Test setup used to measure the radiometer’s bandpasses. The calibrated power meter and synthesizer are configured to provide a constant power at the
eroomsignal,
temperaturebut
30 dB short compared
parameters of this to theandtime scale 2 2
coupler. The coupler, all the fixturing ±(A
waveguide used to route the signal to the cold hybrid tee at the − B )g1 (t)g2 (t)

n any 1.characterized
input were Amplifiers
of (HEMT)
with the VNA, and
the microwave withapplied
corresponding corrections
components. gain g1/2bandpass
to the measured
Given anddata.noise n1/2
ns, the 2. voltages
Phase present offsets at the inputs of the two where equations 4 and 5 have been used together
s are that the intrinsic voltage noise of the HEMT amp
3. ADiodes
+B responds
A−B linearly to input power correlated to the signals from either of the loads, i.
√ and √ (8)
4. Answer 2 =>2 Vr - Vl ni A = ni B = 0.
ence in sign of the B signal reflects the 180 ◦
 Seeing the light: SPT

Transition edge
sensors (TESs)

Resistance change of
a superconductor as
you heat it up a bit.
 Seeing the light: SPT
Voltage-bias the
detector to sit on
transition:

⬌
Small dT -> large dR
large dR -> large dI

Optical-heating
Joule-heating from CMB, Sky etc.
from Voltage-bias
 Seeing the light: SPT
We have to read
thousands of TESs! Ch 1, Ch i, … Ch N
Clever but complicated
We have to read thousands of Seeing the light: SPT
TESs! Clever but complicated

Couple a TES to a LC system.

Amplitude is modulated by R(T),
and each TES gets a resonant fr….
therefore we multiplex
 Seeing the light: SPT
Superconducting Quantum Interference
Device: sensitive magnetometer for
measuring tiny (fT) magnetic fields

Finally a SQUID amplifier

is used to read it all out
 Seeing the light: S4 (potential)

Multiplexing factor is good, but not great: SPT-3G has 15,234

detectors at 68× multiplexing

0.5E6 detectors for CMB S4, thus naively multiplexing factor

should scale up by > 2E3

Of-course, we are likely to split the 0.5E6 detectors across a

few focal planes, and the readout can be further segmented

These can help us by O(10-100), but pushing by another

decade is going to very very challenging.

While 3G operations will be extremely illuminating, we

recognize it is not smooth sailing

Native multiplexing …
 Tc [K] 1.2 9.2
Gao & Mazin (Caltech theses) λL0 [nm] 15.4 33.3
Seeing the light: S4 (potential)
v0 [106 m/s] 1.34 0.28
ξ0 [nm] 1729 39
l [nm] 10000 20
Kinetic Inductance: Superconductors electronics in all generality ∆0 [meV] 0.182 1.395
∆0
kTc 1.76 1.76
Z(!) = R + i(Xc + XL ) =λR + K,
eff (0 i(1/!C + !L)
6 GHz) [nm] 51.4 63.5

−3
x 10
5
Rs
Xs Aluminium sheet
R(⌦) = Re(Z) 4

Z (Ω)
2

1 Tc for Al

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
T (K)

Figure 2.3: Surface resistance Rs and surface reactance Xs of aluminum as a function

High frequency oscillating
T . The materialfields
parameterswill
usedsee
in thea “mass”
calculation for Table
are from the 2.1
Cooper
Im
pairs, leading to a2.2.6.2
phase (Z)
lag, or dependence
Temperature >0
of Z s
Re
 2
S21 = .
2 Seeing
+ (1/j!L the
+ j!C)
light:
Z0 S4 (potential)
This can be rewritten for ! near !0 as
Microwave Kinetic Inductance Detectors (MKIDs)
1
S21 =
1 + jQ(1 !02 /! 2 )
p
using the standard definitions !0 = 1/ LC and Q = !0 Z0 C/2.
For small perturbations from resonance, x = !/!0 << 1, we can write

1
S21 =
1 + 2jQ x

so the power transmission is

1
|S21 | = 2
.
1 + 4Q ->
Photon absorption -> Cooper pair breaking
2 2
x Change in R & L

At resonance, x = 0, and
-> Resonance the transmission
frequency is unity.shift,
shift, phase The power
peak transmission
broadeninggoes do
of 2 (down by 3 dB) when x = ± 12 Q. Thus, the full width at half maximum (FWHM
Ben Mazin and SRON
 Seeing the light: S4 (potential)

Microwave Kinetic Inductance Detectors (MKIDs)

Because each detector is a

mode, we can connect them in
series, i.e. a comb of modes

The amplitude of a mode ~

photon power on that detector

Thus MKIDs are naturally

multiplexed photon detectors

Ben Mazin and SRON For efforts at KICP and further details, see Amy Tang’s talk
Conclusions

We discussed how colors and patterns of the

CMB can be measured

Rest of 448 course has shown the physics

behind these fluctuations

General sensitivity calculations were presented

Types of technologies were outlined

We did not cover: foreground subtraction, map making

algorithms and parameter estimations

Questions ?