introducing mako n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Introducing: MAKO PowerPoint Presentation
Download Presentation
Introducing: MAKO

Loading in 2 Seconds...

play fullscreen
1 / 27

Introducing: MAKO - PowerPoint PPT Presentation


  • 132 Views
  • Uploaded on

Introducing: MAKO. MAKO is a 350 micron KID based camera designed to be a drop-in replacement to SHARC-II Goal: Demonstrate a scalable 350 micron pathfinder instrument using KIDs for large FOV ground-based telescopes Array should be: Photon-noise limited Simple to fabricate

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Introducing: MAKO' - vesna


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
slide2

MAKO is a 350 micron KID based camera designed to be a drop-in replacement to SHARC-II

  • Goal: Demonstrate a scalable 350 micron pathfinder instrument using KIDs for large FOV ground-based telescopes
    • Array should be:
    • Photon-noise limited
    • Simple to fabricate
    • Low per-pixel cost
    • Highly multiplexed
slide3

Can we be photon-noise limited?

J Zmuidzinas, “Superconducting Microresonators: Physics and Applications,”

Annual Review of Condensed Matter Physics, 3 (March 2012).

Noise Source

Photon

Amplifier

GR

TLS

Knobs:Pa,b ->STLS(T,C,Pa)

slide4

Can we be photon-noise limited? Increase Pa

Lk(I) = Lk (0) [1 + I2/I*2 + …]

slide5

Noroozian et al.

Submitted for publication.

ILLUMINATE WITH 5 pW OF

215 mm RADIATION

slide8

Can we be photon-noise limited? Increase Pa

Deep in bifurcation regime:

NEP ~ 2 x 10-16

about 3x above photon noise

However, requires

complicated readout

and doesn’t address

cost and multiplexing issues…

slide9

Can we be photon-noise limited?

Noise Source

Photon

Amplifier

GR

TLS

Knobs:Pa,b ->STLS(T,C,Pa)

  • a 1/sinh(hw/2kT) ~ 2kT/hw
  • MAKO: decrease w using Lk of TiNand big C,
  • keep Pa below bifurcation
slide10

Can we be photon-noise limited?

We think so.

See Chris McKenney’s presentation

slide11

Goal: Demonstrate a scalable 350 micron pathfinder instrument using KIDs for large FOV ground-based telescopes

    • Array should be:
    • Photon-noise limited …. (see Chris’ talk)
    • Simple to fabricate
      • Direct absorption LEKID (see Chris’ talk)
      • Maximum 3 layers lithography
    • Low per-pixel cost
    • Highly multiplexed
slide12

Why focus on cost and simplicity?

http://ccatobservatory.org/

Steve Padin:

“CCAT optical design for 1° FOV”

fl/2 at 350 mm:

6 Million w/ corrector plate 1° FOV

1 Million w/o corrector plate ~ 30’ FOV

Soft goal: ~ $1 per pixel including everything.

In particular, electronics cost dominates. DENSE MULTIPLEXING REQUIRED

slide13

Multiplexing density limited by per-resonator

bandwidth.

Constant power astronomy

Minimum per-pixel bandwidth: Df ~200 Hz

set by telescope scan speed

Min f0 if we can achieve a Q = 105 : 20 MHz

slide15

MAKO

1 Octave

Nmux 1k – 3k

Q few x 105

per octave

mako low frequency advantage
MAKO: low frequency advantage
  • Improved NEPfreq (see Chris’ talk)
  • More octaves per fixed electronic bandwidth. ie 500 MHz ADC spans ½ octave in 1-2 GHz compared with more than 3 octaves below 500 MHz
    • Start designing array with the lowest frequencies first.
    • Low end set by BW needs and fabrication considerations.
    • High Lk material needed to get low frequencies (TiNetc)
  • Eliminate mixing stage and the need for IQ ADC/DAC pairs
build system
Build system
  • Cryostat
  • Readout
cryostat

Cooldown by Cryomech

PT410 pulse tube

Cryostat
  • Vibration isolation for pulse tube,
  • compliant straps at 50 and 4-K
  • Aluminum vacuum jacket (two section)
  • Warm magnetic shield:
  • 2-layer Amumetal~100 attenuation
  • Aluminum 50-K cold plate + radiation shield
  • Copper 4-K cold plate, aluminum rad shield
  • G-10 CR supports
  • Aluminum 1-K stage, CFRP hexapod
  • support
  • Copper UC stage

12”

D= 21”

50 K

4 K

34”

1 K

D= 18”

cryostat cold head
Cryostat – Cold head

3He head

1μW @ < 220 mK

3He buffer head

30μW @ < 350 mK

1 K head

250μW @ < 1 K

Cycle time ~3 hours

Hold time 24 – 36 hours

cryostat filter s tack
Cryostat – Filter Stack
  • 300 K: 1 mm HDPE window, 2.5” diameter
  • 50 K: 2 mm quartz blocking filter, LDPE AR coat
  • 4 K: 2 mm quartz block filter, clear + black LDPE
  • AR coat. (Porex scattering filter?)
  • 1 K: QMC 300 μmlow pass
  • FP: QMC 350 μm band pass (10% bandwidth)
            • 32 mm diameter ~ 1256 fl/2 pixels
cryostat optics
Cryostat - Optics

Tertiary focus

Cold Aperture Stop

Filters

Cryostat

bottom

Dewar Entrance Plane

Relay Optics

Ellipsoid

  • Focal plane fed at f/# = 4.48
  • fl/2 = .78 mm
  • Entrance plane – aperture stop 2”
  • Aperture stop – focal plane 5.6”
readout
Readout

Server

price

$2k

C++ (CUDA) cuFFT

CPU

or Disk

Nvidia

m2090

~$2.8k

Pentek 2x ADC, 2x DAC

500 MSPS $15k

1st stage: $3500

WeinrebSiGeCryo Amps

2st stage: $500

Miteq.001-500 MHz

Full instrument.

C/C++ programming.

250 MHz bandwidth

$30k for 2 lines ($15k/line)

slide24

Need appropriate reconstruction/

anti-aliasing filters here. 500 MHz BW.

PLLs used to

upconvert SRS rubidium

10 MHz source.

Separate A/D, D/A & FPGA

PLLs

D/A:

2 Bytes/S

A/D:

1.5 Bytes/S

Use minimum FPGA

to keep $/pixel down.

x8 PCIe Gen2 supports

4 GB/s transfer rates.

Data rate is only 1.5 GB/s

readout gpu
Readout - GPU

Can the GPU keep up?

FFT Costs:

5Nln(N)

FLoating-pointOPerations (FLOP)

for N = 2^25, 3 GFLOP-per-fft

Rate:

(1e9 Samples-per-second/2^25 samples-per-fft)*3 GFLOP-per-fft

= 89 GFLOPS

or 60% of benchmarked rate.

(m2090 has 25% greater max capacity, thus ~47% of GPU potential)

readout adc performance
Readout – ADC Performance

Texas Instruments ADS5463 500 MSPS, 12-bit A/D

Does the ADC have enough dynamic range?

Worst SNR ~ 65 dBFS

PSD = (N/BW)/S = 1/[SNR*BW] = 1/[SNR*(fs/2)]

Noise-to-carrier PSD (dB):

-65-10log(250e6) = -148 dBc/Hz

Compare with kTn/Pread where Tn is ~ 3.5 K

and Pread is max before bifurcation, 4 K TiN:

10*log(1.38e-23*3.5/129e-15) = -94 dBc/Hz …. 54 dB cushion

thank you
Thank you!

Loren Swenson

Chris McKenney

Tony Mroczkowski

Hien Nguyen

Peter Day

Matt Hollister

OmidNoroozian

  • Erik Shirokoff
  • Steve Hailey-Dunsheath
  • Rick Leduc
  • Matt Bradford
  • Attila Kovacs
  • Byeong Ho Eom
  • Jonas Zmuidzinas