Diapositive 1

1
Neel IRAM KIDs Array

• Cryostat
• Filters
• Cold electronics
• Measurement methods and assumptions for
  Sensitivity calculation
• Antenna-coupled 42 pixels KIDs (first option)
• 32 pixels LEKIDs (second option)
• 196 pixels LEKIDs (TBT)
• Best LEKIDs pixel tested in Cardiff (for 2010 )

2
The Cryostat Optics
Designed for f/1.6 on the detectors (from f/10 at
M8). De-magnifying factor 6
300K ZITEX G110 HDPE (t10mm) window Thermal
Filter (metal-mesh) 150K -- 80K 2 thermal
filters (metal-mesh) 1 low-pass filter
270GHz 5K 1 low-pass filter 240GHz L1 HDPE
lens (t11mm) 1K 1 low-pass filter
210GHz 100mK Reflecting Baffle L2 HDPE lens
(t12mm) Low-pass filter 170GHz High-pass
filter 130GHz (optional) bandpass 130-160GHz
4th Sepember 2009 added new 4K baffle (ECOSORB
inside) ZITEX G110 at 80K
3
The Filters

• 4 thermal (including ZITEX)
• 4 low-pass
• 1 high-pass
• HDPE window and lenses are
• not AR-coated.
• We have 4.5 loss at
• each surface, plus 101112mm
• Bulk absorption. We estimate
• THDPE0.7

Band 130-170GHz Filters transmission 0.5 HDPE
transmission 0.7 So our best estimation of the
in-band transmission is Total Transmission ?
0.35
4
Cold Electronics

• We can now easily (cabled for both down to 4K,
  just has to un-screw the amplifier)
• change between
• IRAM cryogenic amplifier 12 GHz. Tnoise 5K.
  Power dissipated 50 mW
• LNF cryogenic amplifier 39 GHz. Tnoise 5K.
  Power dissipated 4 mW
• Still waiting for the Caltech 112GHz amplifier.
• Both are mounted at 4K. The IN of the amplifier
  is connected to the array with
• 15cm superconducting NbTi 1.6mm semi-rigid COAX
  (thermalised at 1K)
• 16cm Copper 2.2mm semi-rigid COAX (at 100mK)

5
Measurements  protocol 
Everything on XY stage
580 mm
750 mm
CHOPPER 77K / 300K
30x30cm ECOSORB 77K
ECOSORB 77K
12 mm dia. hole in the 2nd ECOSORB 2nd ECOSORB is
to fake M8 in some way (not perfect)
6
Estimation of the power 1/2

• Assuming h?ltltkT (OK)
• So, between 300K and 77K (perfect emissivity) Pe
  0.2 mW/m2 (?max170GHz, ?min130GHz)
• Power emitted over the 2p, both polarisations,
  with a cos(?) distribution.
• For a chopped F12mm surface (S 1.110-4 m2)
  ? Pchopped 22nW
• The Lambertian  beam  is propagating forward
  toward the 4K lenses, d80mm and distance
• D750mm from the chopper. The 4K lens collects
  the part going mostly on the focal plane.
• So the useful fraction of solid angle is
• ?frac (pd2/4) / (2pD2) (1/8)(d/D)2
  1.410-3
• Total transmission (from the chopped source to
  the array)
• Filter lenses TFL 0.35 (0.45 possible)
• Cold pupil cutoff (cutting the external
  ring) TCP 0.8
• Measurement system (from chopper to cryostat)
  TCH 0.5 (n.a. on the sky)

7
Estimation of the power 2/2
So we have P ? ?frac TTOT Pchopped ? 4 pW
Total power, in the two polarisations, hitting
the focal plane and spreading (PSF - gaussian 2-D
distribution) on a number of pixels to be
determined experimentally (aberrations
diffraction). ZEMAX full 3-D simulation
(better calculating the pupil cut and taking
properly into account the Lambertian and
extended nature of the emitter). PZEMAX 5.9pW
(Basically estimating ?frac in 3-D
ray-tracing) Now, the mean beam FWHM is fitted
to be 31mm. It means, roughly, a factor (12/31)2
0.15 Working out the integral of the gaussian
in 2-D we find a factor of 0.098
instead. So, the power hitting the pixel in
this model is 5.9 0.1 0.59 pW (2
polarisations) In case of the measurement of the
SRON array, since we have used an additional
filter 2mm, we estimate 0.4 pW instead of 0.6 pW.
8
The antennas 42 (6x7) pixels array XY scan
PIXELS identifications 37/42 working
pixels. Mean inter-pixel distance 10mm (means
f/1.6 on focal plane, dpixels1.6mm) Mean beam
FWHM 30mm
9
The antennas 42 (6x7) pixels array XY scan
PIXELS identifications 37/42 working
pixels. Mean inter-pixel distance 10mm (means
f/1.6 on detectors plane, dpixels1.6mm) Mean
beam FWHM 29 mm
10
The antennas array FPGA sensitivity
Sensitivity measured with FPGA electronics, LNA
at 4K mounted, Grenoble Up/Down converter. It
seems its still limited by the background.. May
be some excess phase noise ? NEP are for the 2
polarisations !! Should divide by 2 to be fair ..
NEP1Hz ? 210-15 W/Hz0.5 NEP10Hz ? 710-16
W/Hz0.5
11
The antennas array Bonn sensitivity
From Andrey.
12
Antennas MKIDs next steps

• Improve f-distribution (possible before Oct)
• Change to optimum f-distribution on chip
• Use e-beam mask (current is optical)
• Improve coupling
• Current lens array E2.7 instead of 5 (loss 3dB)
• Change extension length (possible before Oct)
• Change lens material (Spring)
• Possible to Niquist reduce oversampling (max gain
  6dB)
• Improve F-noise
• Integrate capacitor at coupler (demonstrated at
  JPL) improves phase noise by 10dB
• Multi frequency pixels ( kid parametric
  optimization)
• Optimise for operation under sky loading

13
LEKIDs 32 pixels focal plane scan
24 working pixels within the 48 MHz band of the
FPGA electronics. FWHM 31/-4 mm
Its OK !
14
LEKIDs 32 pixels unchopped
Image of two hot spots on the focal plane no
chopper
15
LEKIDs 32 (4x8) pixels array (NEP)
16
LEKIDs 32 pixels focal plane scan
PIXELS identifications 28 working pixels, but
only 24 within the FPGA bandwidth Mean
inter-pixel distance 12mm (means f/1.6 on
detectors plane, dpixels 2 mm) Mean beam FWHM
31 mm
17
LEKIDs 196 (14x14) pixels array
CAREFUL still to be tested with the Bonn
electronics (need to change to 16k bins) and in
the same background conditions. Scheduled when
we repair the cryostat.

• S/N 1000 (down to 0.5Hz)
• S/N 200 (at 0.1Hz) but
• (dia. pupil 2cm)
• Assuming 0.5pW power
• NEP 510-16W/Hz0.5
• Ongoing to improve
• 16-32 pixels arrays (FPGA electronics)
• Fabrication on Sapphire
• Change of resonator impedance (phase noise
• and power handling)
• Cryogenic amplifier (amplitude readout)
• Better control of backshort distance
• Superconducting box filter (Cardiff)
• Post-processing KIDs circle calibration
• Test with Bonn electronics to read-out the

CHOPPER
NO CHOPPER
18
LEKIDs Best Pixels Tested

• Border conditions
• - 40nm film (UHV at SRON).
• - Less C fingers
• (helps reducing the phase noise)
• - Tested in low background
• environment (Cardiff).
• (0.1pW per pixel)
• To further reduce the NEP
• - Change the impedance of
• the resonator
• - 2 polarisations design
• fabricate on sapphire
• - further optimise the electrical
• power on the KIDs
• - use an additional AR coating
• - optimise the film thickness

19
Bonn Electronics
Available in Grenoble since end of
July. Bandwidth 400 MHz Max. readout
rate 10 Hz FFT points 8,192 / 16,384 Bins
spacing 48.8 / 24.4 kHz Interfaced with
Acquisition program ? OK
20
Néel FPGA Electronics
Bandwidth 48 MHz Max. readout rate 100
Hz Max. number of channels 27 Can be used
also for fast (MHz) read-out. Interfaced with
Acquisition program ? OK
21
Electronics developments
USA Open Source DAC and ADC boards OK up to
500MSPS. Developing the CASPER Core. Well have
two full copies, with the possibility of using
them in parallel. Same concept as FPGA
readout. Minimal goal 128 channels each board
(256 hopefully) Delivery end 2009 ? LPSC,
Grenoble Making a  copy  of our FPGA, but
with 200MHz bandwidth and able to read at least
64 channels. Prototype delivery November
? Bonn planning 32k bins and 800MHz bandwidth.
22
Conclusions

• Demonstrated NEP is not exceptionally low
• BUT
• Imaging capabilities OK for both antennas and
  LEKIDs
• Expected better performances on the telescope
• (background limiting both Qi and ?qp)
• Need to understand real problems at the 30-m
• Developments are ongoing to reduce the NEP.
• For both LEKIDs and antennas.