Notes on Keck AO for PSF Recovery Project


General AO documentation can be found here Main Keck AO public Webpage

AO Sign Conventions and Geometry

Keck DM to WFS geometry: Square subaperture Hartmann sensor lenslets. Actuators are located at the corners of each subaperture. This geometry is sometimes referred to as a "actuator centered Fried Geometry" see pictures below. Detail on numbering and orientation can be found here KAON 368 . The actuator spacing is 7 mm at the DM corresponding to 0.200 mm at the lenslet array, 562.5 mm at the telescope primary mirror and 74.8 mm on the virtual pupil image formed by the telescope pupil secondary. The secondary obscuration (including mirror support and baffling) is circular with a diameter of 2.65 m on the center of the primary mirror.

The following IDL code and associated text files sub_ap_map.txt and act_map.txt are useful for plotting and displaying AO data queried from the telemetry system (TRS) with the IDL tv command.


AO Algorithms

Centroid, servo loop and other realtime control algorithms are documented in KAON 517 . The reconstruction algorithm is documented here KAON 356 and the actual IDL code is here

Microgate Documentation

The update to the Keck real time reconstructor is described in the overview by Johansson (SPIE 2008). More details can be found in the Microgate Detailed Design Document and the Microgate As Built Document.

Wavefront Sensor Plate Scale

The both Keck I and II wavefront sensors have a SciMeasure camera that features an e2V CCD-39. (While the Microgate electronics will also support a Lincoln Labs CCID-56 that camera was never purchased.) The CCD-39 has 80x80 pixels. It can be used in both the 80x80 mode or a binned 2x2 to a format of 40x40. The wavefront sensor was equipped with 3 lenslet arrays all with the same pupil sampling (~20x20 see sign conventions above) but with different focal lengths resulting in the following plate scales (These are preliminary numbers and are still to be confirmed). To reduce the effects of charge diffusion the CCD is most often operated in the binned mode (40x40).

Lenslet Name CCD Format Plate Scale ("/pixel)
2.4 40x40 2.74
2.4 80x80 1.38
1.0 40x40 2.28
1.0 80x80 1.14
0.6 40x40 1.46
0.6 80x80 0.72

Wavefront Sensor CCD, Read Noise, Gain(e-/ADU),

These measurements are for the spare CCD in the lab, but they are comparable to the K2 AO CCD on the summit. The read noise is a function of the pixel rate, which depends on the camera program selected. Note that the CCD program is recorded in the FITS header of NIRC2 images (WSSMPRG)

Program Mean Illum. (counts) Conv. Gain (e-/DN) Meas. Read Noise (e-) SciMeasure Specs (e-)
0 7102 0.510 3.49 3.5
1 7649 0.486 3.90 4.0
2 6963 0.487 6.91 7.4
3 4935 0.509 8.28 8.7
4 6482 0.495 3.39 3.4
5 6008 0.493 4.12 4.0
6 4739 0.493 6.75 7.3
7 5929 0.541 8.67 8.9

This table shows how the pixel rate varies with camera program: Note that the CCD program is recorded in the FITS header of NIRC2 images (WSSMPRG)

Program Binning Pixel Integration Increment Min Frame Max Frame
    Rate (kHz) (micro sec) Rate (Hz) Rate (Hz)
0 2 80 12.5 1.22 149.04
1 2 250 4.0 3.78 438.42
2 2 1500 0.68 22.16 1763.11
3 2 2500 0.4 37.55 2405.58
4 1 80 12.5 1.19 41.36
5 1 250 4.0 3.70 127.41
6 1 1500 0.68 1.12 667.57
7 1 2500 0.40 36.91 1053.85

AO Control Loop Transfer Function

The AO system servo coefficients (i.e. gain, integrator leak, etc. ) are recorded in the standard TRS save files, see

The basic transfer function for the AO system is calculated from this IDL code the theory can be found in "Control techniques" by Madec Chapter 6 in "AO for Astronomy" by Roddier. The code has an option to use a state space model of the CCD integration in that is based on work by Douge Looze. The tip tilt mirror dynamics are contained in the

The delay that accounts for computer processing by the real time system is given as follows:

Summary of email from Marcos van Dam:

1. The measured values are on pp 87-88 of the Microgate NGWFC DDR document and depend on the CCD program. ; the delays were estimated by looking at the noise power spectra of noise

; t0 is the sum of all the common delays ; t1=150e-6 t2=700e-6 t4=6e-6 t5=65e-6 t7=10e-6

CASE wssmprg OF ; (See wssmprg is the wavefront sensor program )

  • 0: t0=8000e-6
  • 1: t0=2600e-6
  • 2: t0=850e-6
  • 3: t0=700e-6
  • 4: t0=28000e-6
  • 5: t0=9500e-6
  • 6: t0=2000e-6
  • 7: t0=1350e-6


PRINT, 't0', t0

  • t8=40e-6
  • t9=1e-6
  • t10=2e-6 * DTdelay = t0+t9+t10+200e-6 ; TT compute delay

; the 200us term makes the rejection transfer functions fit the measured ones.

DMdelay = t0+t8 ; DM compute delay

What all these terms mean is explained in the Microgate NGWFC DDR document. I would ignore the 200e-6 term, which is probably due to the response of the tip-tilt mirror.

In the case of LGS observation with STRAP (the tip tilt sensor) the DTdelay is very small and is set to zero

Telemetry TRS

The basic method of querying the TRS system using our IDL library is described here Keck Telemetry Webpage the most important thing on that web page is the description of the SQL tables and the field names. An example of using the IDL library is This code queries both the full frame rate table (ffb) and the configuration table (configuration). We have been using variations of this tool to save AO telemetry data in the IDL save file format (*.sav). These files can be read into IDL with the restore command. The IDL help and help,/structure command are useful to see what has been saved. SQL tables names are generally preserved as field names (i.e.offsetcentroids, dmcommand, etc.) of the relevant data structure.

Seeing Tool

Seeing estimator for Keck it uses closed loop DM commands to estimate r0, Some explanation can be found here: Keck Atmogui public web page


NIRC2 is the main imagining camera used with the Keck 2 AO system. It is positioned behind the AO bench on the Left Nasmyth Platform of Keck II telescope. The instrument operates from 1 to 5 Ám, providing three selectable cameras to cover the expected range in image sizes. Two filter wheels with 18 positions each provide a variety of filters and/or grisms, while a focal plane mechanism provides slits and occulting spots for coronography. A dedicated slide carries larger grisms for spectroscopy. Six selectable pupil masks are available to reduce background noise sources; four of these rotate in concert with the telescope pupil and one is specific to spectroscopy.The detector is a 1024x1024 Aladdin-3 InSb array with four-quadrant readout into 32 channels.The main public NIRC2 web page is here: NIRC2 Home Page and much useful information is here NIRC2 Observers Manual

Field of view 10x10 arcsec (narrow camera)
  20x20 arcsec (medium camera)
  40x40 arcsec (wide camera)
Pixel scale 0.009942 arcsec/pixel (+/- 0.00005)
  0.019829 arcsec/pixel
  0.039686 arcsec/pixel
Filters J, H, K, Ks, Kp, Lp, Ms, H2, Fe II, Br Gam, plus others
Pupil mask 1 circular fixed, circumscribes the telescope pupil
  1 circular rotating, inscribes the telescope pupil and masks the spiders
  3 hexagonal, rotating


A number of telescope, AO system, and instrument "keywords" that are useful for observers are added to the NIRC2 FITS files as additional header fields. The ones used by the Flicker PSF estimation code include the following:

FITS Header Definition
OBJECT Object name
UTC Coordinated UT (h:m:s)
DATE-OBS UT date of obs Year-Month-Day
NAXIS1 Number of pixels in axis 1
NAXIS2 Number of pixels in axis 2
COADDS number of coadds
SAMPMODE Sample mode 1=Single; 2=CDS; 3=MCDS
NREADS Number of reads per integration, if SAMPMODE = 3
ITIME Integration time per coadd
FILTER Composite filter names (both wheels)
EFFWAVE effective wavelength of passband (microns)
MINWAVE minimum wavelength of passband (microns)
MAXWAVE maximum wavelength of passband (microns)
CAMNAME camera slider's named position, one of: narrow, medium wide
PMSNAM pupil wheel's named position, one of: INCIRCLE, LARGEHEX, MEDIUMHEX, SMALLHEX, OPEN, ...
PMRANGL Pupil mask drive angular position (deg)
WSFRRT Frame rate for WFS camera (Hz)
WSSMPRG AO WFS SciMeas program (see WFS section above)
ROTMODE Mode of AO system K-mirror one of, position angle, vertical angle, or stationary
DTCLP OPEN/CLOSE Tip Tilt Closed Loop Mirror Positioning
AODTSTAT AO deformable mirror loop state
AODMSTAT AO deformable mirror loop state

NIRC2 Pupil and PSF

The following IDL code can be used to compute theoretical telescope pupils and NIRC2 point spread functions. These code use NIRC2 keywords to determine the correct image scale and pupil mask size. See NIRC2/Telescope pupil code and NIRC2 theoretical PSF code


Spiders, Pupil orientation, Segment wavefront maps, and Segment exchange information

Links, KAONs, and other References

-- ChrisNeyman - 20 Aug 2010

Edit | Attach | Watch | Print version | History: r17 < r16 < r15 < r14 < r13 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r17 - 2011-11-18 - ChrisNeyman
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback