Detectors for AO Wavefront Sensing Mark Downing, G. Finger, D. Baade, N.

Download Report

Transcript Detectors for AO Wavefront Sensing Mark Downing, G. Finger, D. Baade, N.

Detectors for AO Wavefront Sensing
Mark Downing, G. Finger, D. Baade, N. Hubin, J. Kolb, O. Iwert
Instrumentation Division ESO
Examine AO WFS Detector roadmap
CCD50
CCD39
Advanced
Developments /
Tests at ESO
Future Developments
GMT, E-ELT, TMT
CCD60
Past Detectors
CCD220
pnCCD
MIT/LL
MPI/HLL
CCID-26/128
CCID-35
14/10/2009
DfA 2009: AO WFS Detectors
1
Adaptive Optics (AO)
- removing the twinkle of the stars
Deformable mirror
compensates the distorted
wavefront, achieving
diffraction-limited resolution
1
4
Wavefronts from astronomical objects
are distorted by the Earth’s atmosphere,
reducing the spatial resolution of large
telescopes to that of a 10 cm telescope
OFF
3
Control System computes
commands for the
deformable mirror(s)
14/10/2009
2
Wavefront Sensor
measures deviation of
wavefront from a flat
(undistorted) wave
DfA 2009: AO WFS Detectors
ON
2
Challenge =
Speed vs Noise
Biggest challenge for AO WFS detectors is the trade
between:
low noise (RON and dark current/count), and
fast frame rates
– becomes more difficult as format sizes increase
– and finally power dissipation limits feasibility
It is instructive to review how this trade has been achieved
in the past.
14/10/2009
DfA 2009: AO WFS Detectors
3
Add more outputs
Achieves lower read noise at fast frame rates by reading through multiple outputs.
Read noise of output
amplifier
Modelled
Read Noise
16
14
CCD231
CCD230
NES electrons RMS
12
10
5-10e-
8
6
4
Manufacturer
2
Device
CCD50
Pixel
Format
Frame
size
(pixels)
Rate
1.E+05
100 k Frequency (Hz) 11.E+06
M
24μm
128x128
1000
fps
Amplifier
Pixel Rate
pix/sec
CCD39
24μm
80x80
CCID-26/64
21μm
CCID-26/12
21μm
0
1.E+04
10 k
Outputs
RON
( rms)
QE
1.E+07
10 M
e2v Technologies
16
5ee2v
~ 90%
1000 fps
4
10e-
~ 90%
64x64
600 fps
4
6-7e-
~ 90%
128x128
1000 fps
16
5e-
~ 90%
E2v
MIT/LL
14/10/2009
DfA 2009: AO WFS Detectors
4
Improve the amplifier design
Achieves lower read noise by designing better amplifiers.
NES electrons RMS
• Type of amplifier (JFET or MOSFET, “p” or “n”).
Modelled
Read Noise
•RON
Geometry (WxL) of
amplifier.
16
• Oxide thickness.
CCD231
CCD230 and parasitic).
• 14 Reduce capacitances
(floating diffusion
12
→ The higher the conversion gain the lower is the noise.
10
8
Improve
amplifier
6
4
•
2
Barry Burke (MIT/LL) CCID-56 160x160 21 µm pixel 20 outputs:
MIT/LL
–
•
0
1.E+04
New pJFET amplifier
1.E+05design
100
k Frequency (Hz) 11.E+06
10 k
M
– Reporting ~ 2.5 e-Amplifier
at 1 Mpix/sec
(800fps)
Pixel Rate pix/sec
1.E+07
10 M
e2v Technologies
256x256 by 21 µm pixel, 32/64 outputs in development
e2v
Refer back to talk by Vyshnavi Suntharalingam,
“Advanced Imager Technology Development at MIT Lincoln Laboratory”
14/10/2009
DfA 2009: AO WFS Detectors
5
Add EMCCD gain in the serial register
Achieves lower read noise by adding electron multiplication register before the amplifier.
If1
If2
If3
Electron
E2v L3Vision:
Multiplication
<< 1 e- RON at output amplifier speeds of 16 Mpix/sec
Register
•
CCD60 128x128 24 µm pixel.
 1,000 fps with 1 output.
• CCD220 240x240 24 µm pixel.
Rf1 Rf2 Rf3
 1,200 fps withRf2HV
8 outputs.
14/10/2009
DfA 2009: AO WFS Detectors
6
Customize the architecture
Achieves lower read noise by minimizing the number of pixels read out
by custom designing the architecture to the application.
Polar Co-ordinate CCD - talk about later
Curvature CCD, CCID-35 – R. Dorn (ESO), J. Beletic, and B. Burke (MIT/LL).
– 8x10 subapertues,
– RON < 1.2e- at 4 kfps and QE > 80%,
– Successfully used in upgrade to FlyEyes at CFHT.
See poster Kevin Ho, “Flyeyes: Upgrade of CFHT’s AO System Using an MIT-LL CCID 35 Sensor”
Array design
Sub-aperture design
Image Area – 20x20
18µm pixels
Storage Area
#2
14/10/2009
Buffer
Storage Area
#1
Serial
register
MIT/LL
DfA 2009: AO WFS Detectors
7
Add gain in the pixel => APD
Achieves lower read noise by Electron Multiplication Gain in the pixel.
•
•
Build detector from array of single APDs or better an APD array.
Downside is silicon APDs have statistical variation of gain that results in an
excess noise factor of ~ 2-3
Overcome by operating the APD in high gain “Geiger” mode to discriminate and
count single photon events and thus essentially offer zero read noise.
•
e.g. PoliMI SPADA (Single Photon Avalanche Diode Arrays)




p
80 APDs
QE ~ 40%
dark count rates < 3000 counts/sec/pixel
Photon counter → zero read noise at 20 kfps
n

e
h
p+ substrate
SPADA
14/10/2009
DfA 2009: AO WFS Detectors
8
Detectors for AO wavefront sensing
Mark Downing, G. Finger, D. Baade, N. Hubin, J. Kolb, O. Iwert
ODT / Instrumentation Division ESO
FUNDING: OPTICON FP6-WFS
SH 40 x 40 sub-ap.
6x6 pixels/sub-ap.
240x240 pixels
VLT Instruments SPHERE,
AOF – MUSE and HAWK-I
Future Developments
GMT, E-ELT, TMT
Detectors in Advanced
Development/
Test at ESO
Past Detectors
E2v
CCD220
14/10/2009
MPI/HLL
pnCCD
DfA 2009: AO WFS Detectors
9
e2v CCD220
FP6
Store slanted
to allow room
for multiple
outputs.
OP 4
Gain
Registers
OP 3
OP 2
Gain
Registers
Image
Area
Metal Buttressed
2Φ 10 Mhz Clocks
for fast image to store
transfer rates.
Gain
Registers
Image
Area
Store
Area
Store
Area
240x120
24□µm
240x120
24□µm
8 L3Vision Gain
Registers/Outputs
Each 15Mpix./s.
OP 8
OP 7
Gain
Registers
OP 1
OP 6
OP 5
e2v CCD220:
 Split frame transfer CCD
 240x240 24 µm pixels
 8 L3Vision EMCCD outputs
 << 1 e- RON at 1,200 fps
Next talk Philippe Feautrier
“OCam and CCD220 - World's Fastest and Most Sensitive Astronomical Camera”
14/10/2009
DfA 2009: AO WFS Detectors
10
CCD220 Status
Several Test Cameras in operation
Devices in house that meet specs.
→ built by LAM, LAOG, OHP

ESO’s NGC WFS Camera Head is at
advanced stage of prototype
Reported in Javier REYES poster
“ESO AO Wavefront Sensor Camera”
14/10/2009
DfA 2009: AO WFS Detectors
11
MPI/HLL pnCCD
(Robert Hartmann, Sebastian Ihle, Heike Soltau, Lothar Strueder)
CAMEX 1
CAMEX 2
•
Max Planck Institut /Halbleiterlabor
•
pnCCD 256x256 pixels 51um pitch
•
450um thick fully depleted
 excellent red response & no fringing
 300V backside bias for good PSF
•
Target: RON < 3e- at 1000 fps
•
Split frame transfer
•
Fast readout → Column Parallel CCD
 one output amplifier per column
•
Total of 528 amplifiers but
 CAMEX (mux 132 to 1) for easy I/F
CAMEX 3
CAMEX 4
 Only 8 analog output nodes
MPI/HLL
14/10/2009
DfA 2009: AO WFS Detectors
12
pnCCD: Testing at ESO funded by OPTICON FP6
→ Excellent QE, PSF, and low read noise
Spot scanning used to measure PSF
FP6
QE pnCCD
% of charge in central line of pnCCD versus scan position.
QE %
Charge in Central Line (%)
100.0
90.0
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
51µm
Pixel
100
80
60
40
λ=400nm Vss=200V
20
λ=700nm Vss=250V
0
300
400
500
600
700
800
900
1000
1100
-50 -40 -30 -20 -10
Wavelength [nm]
0
10
20
30
40
Position (µm)
QE:
 Excellent QE into the “Red” → good for Natural Guide Star applications.
 450 µm thick silicon is able to collect the deep penetrating red photons.
PSF:
 Measured < 0.45 pixel over 400-900nm (exceeds specs of < 0.8 pixel).
 Pixels could be halved in size and still meet requirements.
RON:
 < 2.5 e-rms.
14/10/2009
DfA 2009: AO WFS Detectors
13
50
Development to add APD output → AApnCCD
(Avalanche Amplified pnCCD)
Part Funded by ESO and OPTICON FP6
FP6
Paper Lothar Strüder MPE/HLL
“Single Photon Counting in the Optical”
Add avalanche stage
before output amplifier
14/10/2009
MPI/HLL
DfA 2009: AO WFS Detectors
14
Detectors for AO wavefront sensing
Mark Downing, G. Finger, D. Baade, N. Hubin, J. Kolb, O. Iwert
ODT / Instrumentation Division ESO
AO WFS roadmap
Future Developments
GMT, E-ELT, TMT
Detector in Advanced
Development/
Test at ESO
• Detectors required?
• Top Level Requirements
• Possible technologies
Past Detectors
14/10/2009
DfA 2009: AO WFS Detectors
15
AO Detector needs for ELTs
Low
Order
AO
Pyramid
TipTilt
Sensors
NGS
Ground
Layer
AO
Shack
Hartmann
Quad -Cell
LGS
MultiConjugate
AO
Other
WFS…
Guiding
14/10/2009
Laser
Tomography
AO
LGS
Ground
Layer
AO
MultiObject
AO
Large visible fast low-noise detector for
Shack-Hartmann based AO WFS
Existing visible high
performance detector
(i.e. CCD220)
Extreme
AO
NGS
Single
Conjugate
AO
3kHz ultra lownoise detector
IR
WFS
IR
TipTilt
sensors
DfA 2009: AO WFS Detectors
IR detectors
development
16
Large Visible AO WFS Detector needed
to sample the spot elongation
Sodium T ~ 10km
layer
H ~ 80km
Sodium Laser Guide Stars (589 nm)
• AO systems operate at ~1 kHz frame rate
• Bright “guide stars” are required
• Only 1% of the sky is accessible with natural
guide stars
• Sodium layer at 80-90 km altitude can be
stimulated to produce artificial guide stars
anywhere on the sky
• Pulsed laser can be used to range gate to limit
laser spot elongation
Predicted spot
elongation pattern
LLT
Pupil
plane
Detector
plane
Distance from
launch site
14/10/2009
DfA 2009: AO WFS Detectors
17
Large Visible AO WFS Detector
Top Level Requirements
(developed from very detailed simulations)
Parameter
Specification
Array Format
1680x1680 pixels
Pixel Size
24-50 µm
460-950nm (NGS)
589nm (LGS)
700 fps
< 3 e- rms
> 80 %
< 0.5 e-/s/pixel
< 4000e-/pixel
< 1.5% .
Wavelength
Frame Rate
RON
QE
Dark Current
Storage Capacity
Cosmetics
Comment
84 x 84 sub-apert. each with 20x20
pixels to sample the spot elongation
Large
Fast
Low noise
High
Low
Expect few photons
Good; very few bad sub-apertures
• Ease of use/compact size:
 integral Peltier
 digital I/F preferred
14/10/2009
DfA 2009: AO WFS Detectors
18
Large Visible AO WFS Detector
Detector Plan
(Multi-Phase Development)
2007
Design
Study
Design
Study
Technology
Validation
Technology
Demonstrator
Scaled-down
Demonstrator
Development
Testing/
Acceptance
2008
2009
2010
2011
2012
2014
2015
Several Design Studies
Retire
 Investigated
many different
technologies
Retire
Pixel Risks
 Most promising –Architecture/
CMOS Imager, APD array
TDorthogonal EMCCD
Process Risks
and
SDD
Full size device
meeting all specs.
Several Technology
Demonstrators
Full Scale
Engineering
exercise
 All CMOS Imagers
- most Demonstrator
likely to succeed
 retire pixel risk by demonstration noise x
Authorize
speed with good imaging capability (noTesting
image lag)
Production
Scaled Down Demonstrator
architectural risks by fab. ~ ¼ imager
 Highly likely CMOS
 Usable device on Telescope
Production
Phase  Retire
14/10/2009
2013
DfA 2009: AO WFS Detectors
Production
30 Science
Devices
19
With recent improvements CMOS now rival CCDs
1.
Pinned Photo Diode → low dark current (10 pA/cm2)
supply
voltage
 0.5 e-/pix/frame with modest cooling (-10 DegC)
2.
reset
High conversion gains (200 µV/e-) → low RON of < 2eBuried channel MOSFETs → reduces/eliminates RTS signal noise
4.
Build from thicker high resistivity silicon and ‘substrate biasing’
Read
transfer
gate
- by reducing sense node capacitance < 0.8 fF
3.
1
2
4
3
FD
PPD
column
output
 low crosstalk and good red response
5.
Multi-sample the pixel at different conversion gains (µV/e-)
 improves dynamic range and linearity
PLUS the long offered advantages of
4.
Fast frame rates → highly parallel readout: ultimate of amplifier per pixel.
5.
Low power → µA instead of mA (CCD) transistor bias currents.
6.
Monolithic integration of support circuitry; biases, sequencer, clocks, ADCs…
 Offers a simple, easy-to-use digital interface.
14/10/2009
DfA 2009: AO WFS Detectors
20
Polar Co-ordinate CCD
Custom orientate each
pixel array to the LGS
trajectory
Customizes CCD architecture to:
 minimize no. of pixels to be read out
 reduce image to store transfer time
 conceived by Jim Beletic, Sean Akins of
KECK; Barry Burke (MIT/LL)
 being developed by Sean Akins, Brian
Aull, Brad Felton (MIT/LL)
-
-
Storage Area
Imaging Area
and Robert Reich)
Image
Clocks
Read out fewer
pixels
14/10/2009
Serial
Register
DfA 2009: AO WFS Detectors
Storage
Clocks
21
Pulsed laser: track spot on CCD
Spot contained in much
smaller number of
pixels and only these
need to be read out
Na Layer
Typical pixel array
 CCD needs to be customized:
Clock annulus
– for each new application, or
– application configured or restricted to use existing CCD; e.g. use
center projected laser and < 60x60 sub-apertures.
 Scalability:
Pulsed
–
Laser
May not be
Clock CCD
with
scalablecharge
to 120x120
the spot
sub-apertures
Laser projection
Typical subaperture
point
 512 amplifiers could be challenging.
and move charge back and forth to
collect photons from several pulses
before reading out.
14/10/2009
DfA 2009: AO WFS Detectors
10 annuli clocked separately
to optimally track spot
22
IR AO WFS Detectors Requirements
•
RON has limited use to low order Tip/Tilt and “Truth” sensors
– Scientific detectors used: HAWAII-XRG (10e-) and the PICNIC (20e-)
–
•
Talk David Hale “Low-noise IR Wavefront Sensing with a Teledyne HxRG”
For E-ELT much better detectors are needed:
– e-APD
– Teledyne Speedster
Specification
Parameter
Array Format
Minimum
Goal
256x256 pixels
Pixel Size
18-40 µm
Wavelength
0.8-2.5 µm
Frame Rate
750 fps
1500 fps
< 5 e- rms
3 e- rms
> 70 %
> 80 %
< 3 e-/s/pixel
< 1 e-/s/pixel
10k
20k
RON
QE
Dark Current
Storage Capacity (e-/pix)
14/10/2009
DfA 2009: AO WFS Detectors
40 µm
23
e-APDs
•
APDs in linear mode have excess noise factor, F > 1
– Typically 2-3 for silicon and 3-5 for III-V materials
•
In HgCdTe: F ~ 1 in linear mode has been demonstrated
– E.g. LETI: gain of 5300 and F ~ 1.05-1.3 at reverse bias of 12.5V.
– impact ionization occurs mostly by single carrier, electrons, rather than the larger,
slower, and less deterministic holes.
•
ESO has funded SELEX to develop a 24 µm 320x256 prototype detector:
– λc = 2.5 µm
– Prototype operational and initial results are encouraging.
Several developments: see session on APDs on Thursday afternoon
Ian BAKER (SELEX),
“HgCdTe Avalanche Photodiode Arrays for Wavefront Sensing and Interferometry Applications”.
Don Hall (IfA and Teledyne),
“Electron-Avalanche and Hole-Avalanche HgCdTe Photodiode Arrays for Astronomy”
Johan Rothman (CEA Leti-Minatec),
“APD Development at CEA Leti-Minatec”
14/10/2009
DfA 2009: AO WFS Detectors
24
Teledyne Speedster
•
Speedster128 (designed 2005)
•
•
•
•
•
•
•
•
128 x 128 pixels
40 µm pixel pitch
Digital input – clocks and biases generated on-chip
Analog output
Two gain settings – high gain for lowest noise
Chip functionality and performance (in low gain) proven
High gain mode (which should be lowest noise) does not work as designed
Speedster256-D (designed 2008)
•
•
•
•
•
fix of Speedster128 design
Improved CTIA pixel
256 x 256 pixels
12 bit analog-to-digital converters on-board
Up to 10 kHz frame rate
Refer back to Jim Beletic’s talk
“Teledyne Imaging Sensors - Producer of detectors with a dynamic range of 1 million....in several dimensions”
27/6/2008
14/10/2009
DfADetectors
2009: AO for
WFS
AODetectors
WFS
25
Conclusion
•
Current detector developments at ESO are on track to meet current
instrument needs.
•
Innovative detector developments are required for the ELTs.
•
ESO is actively involved with Detector Manufacturers to lay the
foundations to meet these needs.
14/10/2009
DfA 2009: AO WFS Detectors
26
THANK YOU
14/10/2009
DfA 2009: AO WFS Detectors
27