Adaptive Optics
AO Team
Outline
• Solar AO – What is different?
• High order AO development – a prototype for
ATST AO
• ATST AO requirements
• Design Concepts
– wavefront sensor
– DM
– WFS optics
Solar AO
•Small r0 (visible&day-time seeing)
•Near-ground turbulence
•High temporal frequencies
•Extended object
•Object evolves in time (sec –min)
• Photons are plentiful (broad-band)
Wavefront Sensor Noise
Night time AO:
•S/N limited by # of photons collected and detector noise (<1-3e-)
•Limiting magnitude
•For faint objects: laser guide stars required
Solar AO:
•S/N limited by image contrast (Michau et al 1992) - granulation 1.5 –2 %
contrast for d ~10cm and high frequency content in object (Poyneer 2003)
•Larger FOV to track on large scale structure: Yes but, average over many
isoplanatic patches > only turbulence near telescope is corrected
•Flat Field Problems are deadly!! Partially filled apertures are problematic!
•Background: Photon noise dominates - Detector noise is not an issue.
CCDs with large wells are preferred.
Progress in steps
• Low- Order AO: 24 subapertures @ 1.2-1.5 kHz
• High-Order AO: 76 subsparture @ 2.5 kHz
• Next: ATST- AO: order 1000 subapertures,
>2kHz
The NSO low-order AO system
24 subapertures
Correlating SHWFS
Dyson
IF
Collimator/Camera
lens
Video ,
AO
corrected
Wavefront
Sensor
WFS camera
DM 97
Disk Center
Intensity &
Magentogram:
6302 A
Exp: 18 sec
FeI 5576A line:
h~200km
Intensity Map &
Velocity Map
Dark:
downflow
Bright:
upflow
First direct
measurements of
flows in magnetic
flux tubes
Exposure: 30sec
Large variations in Strehl on short time scales
•Lack of consistent time sequences
•Interpretation of spectral, polarimetric data becomes
difficult  High order AO
HO-AO – 76 subapertures
high Strehl for median r0
maintains reasonably high Strehl as seeing fluctuates
High order AO WFS geometry
Pupil image & lenslet
d=7.5 cm subaperture –
pushing it for granulation
200 Pixels
20 Pixels
200 Pixels
20 Pixels
Subaperture
images
2-d x-correlations
Camera arrangement
Parallel processing using DSPs
Ch0
4 DSP Cluster
Ch1
4 DSP Cluster
Ch2
4 DSP Cluster
40 DSP Block
Diagram
Ch3
Camera
Ch4
NSO
PhotoBit
Ch5
200x200
Ch6
Camera
Link
to
Link
Port
4 DSP Cluster
4 DSP Cluster
Link
Port
to
RS422
Deformable
Mirror
4 DSP Cluster
Tip/Tilt
Mirror
4 DSP Cluster
33MHz
per
channel
Ch7
4 DSP Cluster
Ch8
4 DSP Cluster
2500
fps
Ch9
4 DSP Cluster
Monitor
See K. Richards
for details
Control
Host Computer
Keyboard
Intelligent 2.5kfps CMOS AO
camera
Poster by
K.
Richards
DSP WFS&Reconstructor
Mostly off-theshelf parts
Performance
• Detailed performance characterization in progress:
Strehl > 0.7
• Update rate: 2500 Hz
• Servo delay:
– 400 μsec readout + 250 μsec processing = 650 μsec
– Bandwidth:
~130 Hz
(0dB cross-over error attenuation)
WFS
DLSP
UBF
First light Dec. 2002
High order AO:
Digitized real-time video
Seeing: mediocre&highly
variable
High Order AO + UBF:
FeI 5434 wing intensity
FeI 5434 bisector velocity (dark = downflow)
Summary
• The high order solar AO operational DST
• Closed-loop bandwidth: 130 Hz
• Diffraction limited imaging over long periods of
time
• High Strehl ratios
• First Scientific results – MHD confirm
fundamental model predictions
• Successful stepping stone towards ATST AO!
Requirements:
see SRD
• The ATST shall provide diffraction-limited
observations (at the detector plane) with high Strehl
(S > 0.6 (goal S>0.7) during good seeing conditions
(r0(500nm) > 15cm); S> 0.3 during median seeing
(r0(500nm) = 10cm) ) at visible and infrared
wavelength.
• The wavefront sensor must be able to lock on
granulation and other solar structure, such as pores
and umbral and penumbral structure.
• Time sequences of consistent image quality are
required for achieving many of the science goals.
• Robust operations.
SRD: 99% of flux within 0.”3
Nordlund, Stein Keller
simulations
Scatter Plots: Stokes V
ATST AO PERFORMANCE
Fitting error & Bandwidth error only
Adaptive Optics for the ATST
NIR (1.6 micron)
High Strehls should be fairly easy to achieve!
The HO-AO system just developed would do
reasonably well
AO Performance
•
•
•
•
•
•
The site is the most important factor
The site will ultimately determine the performance
Cost, Complexity scale with (D/r0)2
Subabperture size ~ r0:
Contrast in subaperture images > WFS noise
Isoplanatic angle > FOV for correlation tracking >
WFS noise and average over several isoplanatic
patches
• Bandwidth: fG ~ v/r0 ; σ2 ~ (fG/fs) 5/3
10 cm subaperture
1232 Subapertures
1313 Actuators
Hammerhead vs. Tiger Sharc
• 80 MHz Clock
• 2 - 32bit float MAC per
clock
• 160 MAC per second
• 2 subapertures per DSP
• 300 MHz Clock (500Mhz)
• 8 - 16bit int MAC per
clock
• 2400 MAC per second
• >15 times as fast!
• 20 subapertures per
DSP
64 DSPs – 300MHz
2400 16bit MACs per second
SMART
INTERFACE
CAMERA
800x800
32 ports
40 MHz
Link
Port
to
RS422
Camera
To
DSPs
Deformable
Mirror
Tip/Tilt
Mirror
Sorts
Pixels
Into
Subapertures
2000 fps
Monitor
D/A
Keyboard
Host Computer
Network
Remote Control
Data Collection
Off-load fixed aberrations
SH-WFS Camera
Need:
~ 8002 pixel camera
> 2000 fps
Custom Camera: CCD or CMOS or Hybrid
• CCD: 32+ parallel readouts @ 40 MHz
•Contacting vendors:
•E2V (doable but $$$)
•1kx1K running at 1kHz exist (in contact with
vendors/developers)
• Design Contract with one or more vendors soon
Alternative (maybe not): split optically (e.g., prisms).
Alignment? Stability?
DM
• A number of ~1000 actuator systems are in
operation
• “Off-the-Shelf” item at Xinetics, Inc.
• Baseline design requires 5mm actuator spacing
• New control electronics, 20 channels on 3U
board, < $100/per channel. Availability: end of
2003
• Big Issue: Thermal Control! (Nathan Dalrymple)
– ~900W/m2 (200mm pupil, R=90%)
– Air-cooled or liquid cooled
Optics
• Integrated AO
• Where do(es) the wavefront sensor(s) go?
– Close to instrument(s) preferred
– Right after DM
• Uncommon path issues, air path to Coude lab
• Other Drivers/Issues:
– Interaction with instrumentation, scanning, modulator,
analyzer
– Complexity due to multiple instrument setup
requirement
DM
Reconstruction
• Modal Reconstruction
• Simple Zonal Approach won’t work because of
rotation between WFS and DM
• Or: Rotate WFS
• Methods very much the same as in night time AO
• Issues:
–
–
–
–
Alignment of WFS and DM actuator grid
Pupil wobble
Develop optimized reconstruction algorithms
Continuously update of reconstruction matrix
PSF Estimation
• Needed for quantitative analysis. E.g.
Photometry
• Important in particular for extended objects
• Interpretation of low Strehl observations
• Should be/Will be standard product of AO
system
• Status: under development, collaboration with
Gemini AO folks (J.P. Veran) and CfAO and
ONERA
Estimation of long exposure PSF from wavefront sensor statistics.
Implement as standard feature!
PSF
MTF
Low-order AO
1.5sec exposure
Reconstructed image
MCAO
Long exposure
w/AO at DST
Fair Seeing
High altitude
seeing
Sum of 11 one
sec. exposures
Destretched
before averaged
Long exposure
w/AO at DST
Good seeing
Good high altitude
conditions
Sum of 11
No destretch
MCAO
1
1
1
1
3
3
3
3
1
4
2
4
4
3
5
5
6
7
6
7
2
8
8
• 3 “guide stars”
8
7
8
8
7
2
1
7
7
8
•3 ROIs in FOV (~10x10
arcsec)
8
7
7
6
4
7
6
6
1
1
4
6
6
5
3
6
5
4
4
6
5
•Large subaperture FOV
(60+ arcsec)
2
3
2
5
5
5
2
5
4
3
1
2
4
3
2
2
8
8
•Enough real estate on
device
•Read-out at sufficiently
high frame rates