Photon Counting with a 512 x 512 Pixel Array at 1 kHz Frame Rate - A New Wavefront Sensor for Adaptive Optics

1
Photon Countingwith a 512 x 512 Pixel Arrayat 1
kHz Frame Rate-A New Wavefront Sensorfor
Adaptive Optics

• Bettina Mikulec, Allan Clark
• University of Geneva
• John Vallerga, Jason McPhate, Anton Tremsin,
  Oswald Siegmund
• Space Science Laboratory, University of
  California

2
Introduction into Adaptive Optics

• Turbulence in the earths atmosphere makes stars
  twinkle
• More importantly, turbulence spreads out the star
  light making it a blob rather than a point
• Temperature fluctuations in the air cause changes
  in the index of refraction
• light rays are no longer parallel when they reach
  telescope and can therefore not anymore be
  focused to a point

Even the largest ground-based astronomical
telescopes have no better resolution than an 8"
telescope!
adapted from AO lectures of Claire Max,
Astro289C, UC Santa Cruz
3
Adaptive Optics
proposal for a new WFS - Optical Medipix tube
4
Adaptive Optics
Example for the enormous improvements using
AO (Lick Observatory).
5
The new generation adaptive optics on 8-10 m
telescopes
Summit of Mauna Kea volcano in Hawaii
Subaru
2 Kecks
Gemini North
And at other places MMT, VLT, LBT, Gemini South
6
Next Generation of Large Telescopes (proposed)
TMT

• 30 m diameter
• California Extremely Large Telescope (CELT) -
• Thirty Meter Telescope (TMT)
• 50 m diameter
• EURO50 on La Palma
• 100 m diameter
• European Southern Observatorys OverWhelmingly
  Large Telescope (OWL)
• All propose AO systems with gt 5000 actuators

Palomar Hale 5m telescope
7
Adaptive Optics

• Determine the distortions with the help of a
  natural or laser guide star and a lenslet array
  (one of the methods). Deviations of the spot
  positions from a perfect grid is a measure for
  the shape of the incoming wave-front.

Shack-Hartmann wavefront sensor
8
Wavefront Sensor Requirements

• High QE for dimmer guide stars (80 optical QE)
• Gate the detector in 2-4 ?s range for operation
  with laser guide stars
• Many pixels in the order of 512 x 512 future
  large telescopes will have about 5000 actuators
  (controlled via 70 x 70 centroid measurements)
• 1000 photons per spot to get a 3 centroid rms
  error with respect to the stellar image size.
• 1 kHz frame rate (light integration, readout,
  calculations, send out 5000 signals and ready for
  new frame) faster than the timescale of the
  atmospheric turbulences
• Very low readout noise (lt 3e-)

Large pixel array, high frame rate and no readout
noise not simultaneously achievable with CCDs!
9
Proposal for a New Wavefront Sensor

• High-QE GaAs photo-cathode
• Matched pair of microchannel plates (MCP) with 10
  ?m pore diameter in chevron configuration
• Medipix2 counting CMOS pixel chip
• Integrate photon events on pixel,
• noiseless chip readout

10
The Medipix2 Photon Counting Chip

• 0.25 ?m CMOS technology (33M transistors/chip)
• square pixel size of 55 µm
• 256 x 256 pixels
• sensitive to positive or negative input charge
  (free choice of different detector materials)
• pixel-by-pixel detector leakage current
  compensation
• window in energy
• discriminators designed to be linear over a large
  range
• 14-bit counter per pixel
• count rate 1 MHz/pixel (0.33 GHz/mm2)
• 3-side buttable
• serial or parallel I/O (min. readout time of full
  matrix 266 µs)

11
Measurement Setup

• A Medipix2 photon counting chip
• A matched pair of MCPs
• Photonis MCPs with 33 mm diameter
• 10 ?m hole diameters, L/D 40/1
• low resistivity (22 MOhms per plate)
• gain was varied between 20k and 200k (1430 - 1680
  V)
• Vacuum tank pumped down to 10-6 torr
• Hermetic feed-throughs (50-pin connector for
  Medipix signals)
• A standard UV Hg pen-ray lamp with collimator
  (10 counts/s -500M counts/s)

12
Feasibility Tests

• Event size function of MCP gain, rear field,
  MCP-Medipix distance and Medipix threshold

06 April 2004
single photon events
gain 106, rear field 427 V
gain 50k, rear field 980 V
It works!
13
Flood Fields

• Take image with collimated UV source at 50ke gain
  and 1600 V rear field (5000 counts/pixel).
  Average single spot area 2.4 pixels
• Fixed pattern noise from dead spots on the MCPs
  and MCP multifibres divides out.

take 2 independent uniform illuminations (flood
fields at 500Mcps)
Histogram of ratio consistent with
counting statistics (rms 0.02)
Ratio flood1 / flood2.
14
Resolution

• The Air Force test pattern was used to
  demonstrate the imaging properties of the
  detector, in particular the resolution.

increase shutter time
100 ?s exposure the spots correspond to
individual photon events.
1 s exposure.
Group 3-2 visible (9 lp/mm corresponding to the
Nyquist limit of 55 ?m pixels)
15
Event Centroiding

• Centroiding individual photon events to achieve
  sub-pixel resolution
• Take many very low count rate images with larger
  spot area to avoid overlapping spots. (100-150
  counts/frame 1000 frames)
• Identify unique spots and reject overlapping
  events (counts ? 2), count spots, record their
  size and calculate the centroids.

centroiding
Group 4-2 starts to be resolved (17.95 lp/mm
55.7 ?m corresponding to 28 ?m pixels).
Could be useful for low rate imaging applications!
16
UV Photon Counting Movie
Air Force resolution mask, 100 ms exposures
17
Spot Size vs. Gain

• Pinhole grid mask (pitch 0.5 mm x 0.5 mm) to
  simulate Shack-Hartmann spots

Rear field 1600V, gap 500 ?m, threshold 3
ke- Gain 200 000 Gain 20 000
18
Sub-Pixel Spatial Linearity
19
Average Movement of 700 Spots
1 pixel

• Achieved a 2 ?m rms centroid position error with
  550 events/spot.

20
Electron Detection

• First test results with beta sources
• QE 46 for Ni and 63 for the Tl image
• increasing efficiency with e- energies above 50
  keV consistent with literature.

Gain 60k, rear field 1600 V Medipix
threshold 38 ke-
63Ni, 67 keV max. 300 counts/pixel
204Tl, 764 keV max. 100 counts/pixel
21
Conclusions

• New detector concept proven to work!
• Performed systematic tests varying different
  detector parameters
• No fixed pattern noise yet detectable except MCP
  imperfections
• Resolution at Nyquist limit and below (for
  event-by-event centroiding) demonstrated
• Images presented with both UV and electron
  sources ? detector has a great capacity to be
  used for various wavelengths and particles

22
Ongoing work

• Test new ceramic chip carrier ( tube backend)
  thermal cycling tests with Medipix2 chip mounted
• Tube fabrication at SSL Berkeley and at
  commercial firm with GaAs photo-cathode
• Test prototype parallel readout board designed in
  collaboration with ESRF
• reduce output bandwidth by using an FPGA goal 1
  kHz continuous frame rate with 2x2 chip
  arrangement
• Test prototype tubes at the AO laboratory at
  CFAO, U.C. Santa Cruz
• Final test at a telescope
• SSL received 3-year NOAO grant, 1.5 more years
  to go, some funding for Univ. Geneva from Fonds
  National and F. Schmidheiny
• Consider other applications for such a detector

23
Parallel Readout Board
24
Tube Design
25
Tube Design
26
Backup Slides!
27
The Setup at SSL - Photos
28
Spot Size
Spot area versus Medipix2 low threshold.
Spot area versus rear field.
29
Soft X-Ray Photocathodes
30
EUV and FUV
31
GaN UV Photocathodes, 1000- 4000Å
32
Isoplanatic Angle (?0) Sky Coverage
Bright stars ?0 1 sky coverage
Telescope Primary mirror
33
Laser Guide Stars
Can achievegt70 skycoverage withlaser guide
staradaptive optics!
34
Laser Guide Star Parallax

• Star more of a streak
• Shape changes over pupil
• Can use pulsed laser to limit spatial extent
• Requires gated detector

35
Advantages of Multi-Pixel Sampling of Shack
Hartmann Spots
5 x 5
2 x 2

• Linear response off-null
• Insensitive to input width
• More sensitive to readout noise

36
Deformable Mirrors

• Range from 13 to gt 900 actuators (degrees of
  freedom)

300mm
50 mm
Xinetics
37
Position Error (550 Events/Spot)
rms 2.0 µm
38
Medipix readout of semiconductor arrays
39
X-ray of Fish
( with silicon detector)