This document is obsolete as of March 03, 2011
For further info please contact: P200 Support Last modified: May 25, 2013; by Kevin Rykoski
Most of this document was written by Eric Bloemhof
PALAO OVERVIEW: (A) INTRODUCTION
This OVERVIEW gives a brief outline of PALAO, the adaptive optics system for the Palomar 200" telescope, including some of the theory of operation and information needed for advance planning of an observing run. For a step-by-step guide to operation at the telescope, go back to the PALAO homepage and click on the PALAO COOKBOOK button. It is very important to understand that this document is a simple overview. The AO system is operated by the Night Assistant, and the observer runs the science camera Pharo. SO IF YOU ARE A NOVICE PALOMAR AO USER, CONCENTRATE ON LEARNING PHARO FIRST!!!
Below is a highly schematic drawing of PALAO's optical path that includes most of the chief components; the control loop is indicated as well.
...conceptual schematic of PALAO...
The beam expands beyond the f/15.7 Cassegrain focus until it is collimated by an off-axis paraboloid (OAP-1) and directed onto the fast-steering mirror (FSM), which provides tip-tilt correction with a closed-loop bandwidth of about 30 Hz. Still collimated, the beam then hits the 341-element deformable mirror (DM), which corrects higher-order atmospheric effects at a closed-loop rate of roughly ~50 Hz.
[NOTE: The FSM and DM are driven by meaurements from the wavefront sensor (WFS) at open-loop rates as fast as 2000 Hz; closed-loop speed has to be at least a factor of 10 slower, by general control-loop theory].
From the DM, a second off-axis paraboloid (OAP-2) refocuses the beam onto the near-infrared science camera, PHARO. En route, the visible portion of the beam is picked off by an infrared-transmitting dichroic and sent to the wavefront-sensor (WFS) camera, which can measure the phase perturbations across the telescope pupil of light from a reference guide star close to the science object of interest. The operation of the WFS, which is of the Shack-Hartmann type, is described in an appendix (see menu list at left).
The dichroic is also called star-selection mirror 1 (SSM1); it is (slowly) steerable, as is another flat mirror called SSM2 that relays the beam to the WFS camera. The two SSM's are used in tandem to automatically select any desired guide star and send its light into the WFS camera without changing the science field of view. The two SSMs are shown in the following raytrace diagram of PALAO:
...optical layout of main bench of PALAO...
These optics are bolted onto the f/15.7 Cass focus with the plane of the diagram (plane of the optical bench) perpendicular to the optical axis of the telescope. The beam from the telescope is injected by a 45-degree flat (FM1).
A useful Glossary, which includes some theory of operation of the instrument, is given in Appendix C.
It is useful for the observer to understand how the SSM mirrors operate, but once again remember that the AO system is the responsibility of the night assistant. If there are minor aspects of AO that you do not understand it is not critical to your observing. Ask the night assistant or the support astronomer if you have questions.
From the second diagram of section (B), you will notice that moving the SSMs has no direct effect on the position of the infrared image on the PHARO chip. It will, however, clearly move the optical image in the field of view of the wavefront sensor. More than this, the SSMs are carefully set up to permit (almost) independent relative pupil alignment between the wavefront sensor and the deformable mirror, in addition to the optical/WFS image motion just mentioned.
As part of the afternoon AO setup, the SSMs are moved until the image of an internal white-light source is brought onto the field stop, and then they are moved according to a different algorithm until the pupil alignment as seen on this white-light source is tweaked up. Opening up to the sky will give a good lock with almost no further adjustment (just a fresh WFS sky, and a fine DM registration, to be precise). The night assistant moves the telescope to bring the star onto the field stop. Turning on the tip/tilt and DM locks will then pull the star image into the same "hot spot" or "sweet spot" that the internal white-light source had in the afternoon setup.
One may think of the "sweet spot" as being the place on the PHARO chip where the image ends up when tip/tilt and DM loops are locked. (It shifts only a little with DM lock; it shifts a little if a fresh pupil alignment is done).
It is the right of every astronomer to move sources around on the infrared chip, and the SSMs allow this while loops are locked. (In fact, you'd only want to do this closed-loop). This operation is called "closed-loop dithering", and is performed transparently with controls and macros on the PHARO GUI that specify magnitude and direction in which the PHARO display window-frame will move. See the cookbook for more info. The effect will be to move the infrared source on PHARO, which of course involves moving the telescope (see that PALAO raytrace diagram above), but the SSMs are moved in a synchronized and compensating way to hold the guide star steady on the WFS and maintain adaptive lock. To pull this off, the telescope move has to be fairly small (see the rules below).
Each step should be no more than 5 arcseconds.
Wait 5 seconds to let things settle down; watch the wavefront-sensor display to see that lock is maintained.
If the total distance dithered is more than 10 or 15 arcseconds, depending on the quality of PSF you want, you should tweak up the pupil alignment when done. This is done by the night assistant with a DM registration (fine) command.
Somewhere, as you move the guide-star 5 or 10 or so arcseconds beyond the edge of the PHARO field of view (in 40 arcsecond mode), one of the two SSMs will be the first to hit its hardware limit, the other will bravely carry on, and the two will go out of synch. The symptom is no light to the WFS, the cure is typing ao restore DEFAULT in the command line (TAO) window.
Be warned that small dithers, eg manoeuvering under slit or coronagraphic spot, are subject to hysteresis, stiction, etc..
If you attempt a large (e.g. 60 arcsec) telescope move while locked, it will be like breaking rule 4...the SSMs will go out of synch, and you will get zero light through to the WFS .
For given settings of the AO system, there are two basic things that can be refreshed to provide a better adaptive lock (and hence corrected PSF):
WFS sky frame: this only takes a couple of minutes and is easy to do. The night assistant will normally do this after every slew. It is especially important to do thia when close to the moon or when the moon has risen or set.
Pupil adjustment: takes a little longer (5 minutes) and is typically done for each new object.
Also, you can never get a PSF any better than the PSF calibration done at the start of the run and embodied in the "centroid offset" file (see the Glossary in Appendix C); this best-possible calibration-limited PSF may be displayed at any time simply by imaging the internal white-light source. Ask the night assistant or support astromer, if you want look at the PSF/Strehl of the white light source.
A) THE MAIN TACTICAL DECISION at your disposal is the speed at which to run the WFS (wavefront sensor) camera. The experience of the night assistant is your most reliable indicator. Here are some points to ponder:
If the source is bright enough (certainly V=10 and brighter), just run the system at top speed. As of 2004 the top speed is 2000 Hz, unfortunately there is some debate that the DM actuators may not be able to keep up with this speed especially in cold temperature conditions. The top speed for now is probably 1500 Hz.
If the seeing is good, particularly if the seeing on the WFS display shows no clear directional flow (like a wind blowing by), but just sits in a stationary, "percolative" pattern, you may be able to lock on fainter guide-stars with negligible loss of quality of the lock by turning down the WFS camera speed. In outstanding seeing we have locked at 100 Hz, effectively increasing the integration time and sensitivity of the WFS CAM by a factor of 5, and extending the guide star limit by almost 2 mags down to about V=13.5.
An indication of lock quality that you might expect is provided by PALAO's real-time plot of subaperture flux. This value should be greater than 200 counts, but we have "pushed" the flux below 100 with debateable results.
An indication of lock quality that you might be getting is provided by PALAO's real-time plot of DM residuals (a great lock has as low as 0.04; 0.20 is getting poor)...dmresiduals can be wrong, however...
BUT!!! the bottom-line indication of better/worse performance is the PSF seen by the observer watching the PHARO screen, although this is more difficult and more subjective to judge.
B) OTHER PALAO ADJUSTMENTS:
Loop Gains, both FSM (t/t) and DM...some recent investigations have revised our "best" choices somewhat, but it is not believed that large performance gains can be efficiently realized by observers adjusting these values. The night assistant has a list of these gains for different WFS speeds derived empirically during engineering runs.
Reconstructors and other files...again, different loop-control files can be considered, but experimenting with these is not recommended.
There exists three choices for actuator offsets on the DM. These offset files correspond to good, bad, and moderate seeing. Normally the observer will take a ~60 second uncorrected image of the guide star on Pharo to get the uncorrected seeing. The night assistant will then try locking with two of the files, and the observer can check which choice gives the optimal PSF/Strehl.
It is often desirable to check the PALAO PSF (point-spread function) on a nearby star chosen to have similar properties to the AO guide star. The quality of the adaptive lock will depend on the brightness of the calibrator over the sensitive range of the WFS (wavefront sensor), which is roughly V to R in our case. So a PSF calibrator may be chosen to match the guide star at V or at both V and R to some tolerances...the calibrator must of course generally be close by on the sky, as well, to sample comparable atmospherics.
The goal is ultimately to achieve comparable "subaperture flux" [detected photon counts per subaperture of the WFS] when actually observing with PALAO; this quantity is plotted in real-time on the PALAO displays, and should be recorded in your logbook if you want to know how comparable guidestar and PSF calibrator turned out to be.
A quick and easy-to-use star finder that works at V only is:
ESO HST Guide Star Catalogue
More advanced catalogs that permit multi-band searching may be found at:
Imaging: 25 or 40 arcsec field, with 0.025 or 0.040 arcsec pixels
Pharo subarrays are available, and are set with the Pharo GUI in the detector pull down. If you are interested in subarray, ask the support astronomer to show you how to do this.
Spectroscopy: 0.1, 0.2, or 0.5 arcsec slit (spanning any one of J, H, K at up to R = 1500)
Coronagraphy: 0.41, 0.91 focal-plane spot; "std", "med", "big"-cross pupil-plane masks
A description of coronagraphy is contained in the following paper...it also offers observing recommendations to help you choose masks (see Table 1 on page 6):
Oppenheimer et al., (Munich SPIE, March 2000...Proc. SPIE vol. 4007, p. 899)...
PostScript format PDF format
Jason Marshall and Mitchell Troy at JPL have put together an excellent package of IDL software for the basic reduction of FITS images obtained with PALAO/PHARO. The package, called "AOred", can perform sky subtraction, flat fielding, quadrant equalization, extra background flattening, bad pixel correction, cosmic ray rejection, time normalization, and combination of images.
To read more about AOred, or to download it for your own use, check the web page:
AOred Home Page
NOTE: you will not be able to sensibly inspect PHARO-written FITS files with any old display tool (SAOIMG, SAOTNG, etc)...the images are stored as a stack of 4 quadrants, each 512x512. You'll need AOred or some software of your own to put the quadrants together into a single image...
BUT: included with AOred is an interactive display tool for IDL called "ATV" that is like SAOTNG, but capable of dealing with the separate handling of image quadrants.
AO-related Web pages:
There are a number of instructive tutorial pages posted on the Web by some of the other groups and observatories doing adaptive optics. A partial list follows...
Optics Associates Inc.
European Southern Observatory
Mt. Wilson Observatory
The wavefront sensor (WFS) is a Shack-Hartmann device, which senses the phase of the seeing-corrupted wavefront from the guide star across the (reimaged) telescope pupil. It does this by passing the collimated beam through a 16x16 array of small lenses; the projection of each lenslet onto the telescope pupil is often referred to as a "subaperture". The two components of the gradient of the phase (tip/tilt) in the region of each subaperture are found by measuring the deflection of the focal spot from each lenslet.
For a diagram of the Shack-Hartmann WFS, see the tutorial pages listed above.
Obviously, the phase itself can be reconstructed from the resulting 16x16 grid of two-dimensional phase-gradient values, up to an overall constant "piston" term that is unobservable. In practice, the gradient measurement is done by analyzing the data from the 64x64-pixel wavefront sensor camera (from xxx) as if it were from a 16x16 array of quadrant detectors, each having a 2x2-pixel active region surrounded by a guard band 1 pixel wide in each unit cell. The (x,y) position values returned by each quadrant detection is referred to as a "centroid", though it is not intensity-weighted and so is not technically a true centroid.
In a little more detail, the centroid measurement is done with a two-step algorithm that applies an optimal (Kalman) filter to an initial estimate. This approach provides better rejection of noise at large offset values.
Optomechanical layout of PALAO. The beam from the telescope enters from out of the plane at FM1. "SCI CAM" is called "PHARO" in this manual.
ADC = "Atmospheric Dispersion Corrector"...not currently implemented.
Centroid = the (x,y) values returned by an individual subaperture of the WFS acting as a quadrant detector. A non-zero centroid means that subaperture is detecting a local tip or tilt (for x or y, respectively) of the incident wavefront.
Centroid offsets = the stored set of 256 (16x16) (x,y) pairs that the AO system uses as its template of wavefront perfection. In closed-loop operation, the DM will arrange itself so that the guide-star light produces this ideal set of centroids at the WFS (ideal in the sense that they correspond to the best possible image at PHARO, the science camera, not at the WFS itself).
Centroid offset calibration
= The centroid offsets are obtained by a pre-run calibration
process: various amounts of several low-order Zernike components
(currently up to Z10) are injected by trial and error, and the
resulting PSF at PHARO is judged by eye. (This will be automated
someday). Centroid offsets are generally not zero, because they
contain a compensation for "non-common path" aberrations
within PALAO...i.e., the DM configuration that delivers a nice beam
at PHARO is not necessarily centered at the WFS because the optical
path to the WFS has some optical aberrations not experienced along
the science path.
Non-common path errors = differences between the visible-light path to the wavefront sensor, and the infrared optical path to the science camera.
CHECKMAP = a map of values that can be applied to the deformable mirror...it pokes up every other actuator, giving a checkerboard pattern used to check pupil alignment (i.e. alignment of the WFS with the DM).
Dich. = "dichroic"...PALAO uses an infrared-transmitting, visible-reflecting dichroic to separate IR to the science camera from visible guide-star light to the WFS.
DM = "deformable mirror", a 5.5-inch diameter mirror with a thin facesheet of glass whose shape is controlled by 349 actuators that push from behind. In respnose to voltages up to 100 V, each actuator can move a maximum of roughly 4 microns.
FLATMAP = a map of values that can be applied to the deformable mirror...it gives a roughly flat DM.
FM = "Fold Mirror"
FS = "field stop", a small square of metallization (about 4 arcsec) on a transparent substrate in front of the WFS. It acts as a "pick-off" mirror to send guide-star light into the WFS, but you can look around the metallization with the acquisition camera (labelled ACQ on the instrument layout drawing above) to acquire the guide star (this appears on the small monitor).
FSM = "fast steering mirror"...the flat mirror that corrects global tip-tilt in the AO system.
GUI = "graphical user interface", a nice control window with pull-down menus and such, as opposed to command-line control.
IMAGE = image of guide star.
LA = "Lenslet Array"...a 16x16 array of tiny lenses that define the entrance to the WFS. Each lenslet is optically conjugate to a roughly 30-cm-diameter patch of the primariy mirror of the telescope, which is roughly the size of a coherent seeing patch. The beam through each lenslet is then focused into a spot on a different 4x4 subarray of the WFS CCD that acts as a quad cell sensing local tip/tilt.
OAP = "Off Axis Paraboloid"...PALAO uses two, to collimate the telescope beam for the DM and then to refocus it into the science camera and WFS.
PALAO = "Palomar Adaptive Optics"...the name of the adaptive optics instrument.
PHARO = "Palomar High Angular Resolution Observer"...the near-infrared science camera used with PALAO.
Phase Diversity = an algorithm for deriving the pupil illumination from an in-focus and slightly mis-focused image pair (in general, the focal-plane image is the squared modulus of the fourier transform of the pupil-plane illumination, so that a unique inversion to obtain the pupil-plane pattern is not possible without phase info from an OOF image...or you can do curvature/Roddier sensing, with two OOF images, one on each side of focus).
PSF = "Point Spread Function"
PUPIL ...telescope pupil, deformable mirror, and wavefront sensor all have to be aligned...during a run, you control alignment of the last two via a "virtual" pupil motor that is really moving two physical motors, SSM1 and SSM2.
RINGMAP = a map of values that can be applied to the deformable mirror...it pokes up actuators to form a large square pattern. If you are aligned, the square will be centered on the WFS display. If you are misaligned by an entire subaperture of the WFS, then CHECKMAP may look fine but RINGMAP's square pattern will be off-center.
SSM = "science steering mirror"...SSM1 and SSM2 act in concert to allow nearly-independent shifting of pupil without moving image into WFS, and vice-versa.
WFS = "wavefront sensor", in our case a Shack-Hartmann sensor consisting of 16x16 subapertures mapped onto (i.e. optically conjugate to) the primary mirror of the telescope. Each subaperture has a lenslet that focuses light to a spot on a 4x4 subarray of the 64x64 CCD; each subarray acts as a quadrant detector returning local tip-tilt (phase gradient) readings. [From the full set of local t/t samples, the full wavefront can be reconstructed].