A noiseless 512 x 512 wave-front detector for adaptive optics with

Download Report

Transcript A noiseless 512 x 512 wave-front detector for adaptive optics with

A noiseless 512 x 512 wave-front detector for
adaptive optics with kHz frame rates
University of Geneva (Allan G. Clark, Bettina Mikulec) - SSL Berkeley - Medipix Collaboration
Motivation
Detector Concept
• Adaptive Optics are indispensable for new generation
ground-based telescopes!
1. A photon releases a photo-electron from the high-QE
photocathode (GaAs).
• Detectors for wave-front sensors (WFS) require large pixel
arrays (512 x 512), noise <3e- per pixel, high quantum
efficiency (QE) and kHz frame rates*
2. The electron enters one of the tiny pores of a microchannel-plate (MCP) detector where it gets amplified (gain
variable between several thousands to millions).
• Current CCD have high QE, but must trade noise
performance and array size for speed
3. The exiting charge cloud gets detected by the Medipix2
pixel circuits. If detected charge > threshold --> pixel
counter gets incremented.
4. Noiseless chip readout after programmable shutter
time.
Photocathode
Photon
e-
Q = 104e-
Pij = Pij + 1
Point-like objects get blurred
by turbulences in the atmosphere.
WFS measure the atmospheric turbulences, send this information
to deformable mirrors that compensate online for the distortions.
Window
Medipix2
MCP



Shack-Hartman correction method
using an equally spaced lenslet array.
AO reveals a previously undetected star in the Orion cluster. The
surface plot shows the dramatic increase in intensity and sharpness.
* Angel, R. et al. ‘A Roadmap for the Development of Astronomical Adaptive Optics’,
July 6, 2000; http://www.noao.edu/dir/ao/
Medipix2* Photon Counting Pixel
Readout ASIC
• 256 x 256 pixels, 55 m square
Measurement Results
• Detector concept works!
• Flood fields show MCP fixed pattern noise that divides out
take 2 independent uniform
illuminations
(flood fields)
• window discriminator, 14-bit counter per pixel
• 3-side buttable (512 x 512 arrays), serial (LVDS) or
parallel readout (32-bit CMOS bus; 266 s @ 100 MHz)
histogram
• ~500 transistors/pixel; 0.25 m CMOS technology
Flood field (500 Mcps).
Ratio = flood1 / flood2.
• Spatial resolution consistent with theory (Nyqvist sampling
of 55 m pixels)
increase
shutter time
Test pattern; 100 s exposure.
The spots correspond to individual
photon events.
Group 3-2 visible
~9 lp/mm.
Test pattern; 1 s exposure.
• Parameters can be tuned to optimize spot size
Rear Field = 1600V
Spot Area vs Rear Field
20
Gain 25k
35
Gain 50k
18
Mean Spot Area (pixel)
40
Gain 100k
30
Spot Area (pixel)
•Developed within the framework of the Medipix Collaboration;
http://medipix.web.cern.ch/MEDIPIX/
Histogram of ratio is consistent
with counting statistics.
Gain 200k
25
Gain 400k
20
15
10
5
0
16
G=20k, Area
14
G=50k, Area
12
G=100k, Area
10
G=200k, Area
8
6
4
2
0
200
400
600
800
1000
1200
1400
Rear Field (V)
1600
0
0
5
10
15
20
25
30
35
40
Lower Threshold (ke -)
Spot area versus rear field.
Spot area versus Medipix2 low threshold.