WTTM Focus Error Sensor

Operational Issues & Guidelines

As described in the WTTM Users manual, the autofocus sensor is incorporated as part of the tipt/tilt error sensor.  This sensor measures focus by calculating a ratio of flux in the sensor's four quadrants, also used to sense errors in tip/tilt.  Errors in focus are derived from astigmatic aberration produced by a converging beam of light through a tilted glass element and compensating optic designed to produce a specific level of astigmatism;  astigmatism, coma and chromatic aberrations are introduced as the light passes through the beam splitter, 10mm thickness and at a 45deg incidence angle.  Changes in focus can then be detected by  sensing the change in the amount astigmatism at the focus sensor.

The key to this is that the aberrations as delivered by the telescope must be controlled.  Specifically, if astigmatism or coma are significant, but stable, the focus sensing scheme is valid, but if unstable, aberrations will be introduce onto the four quad-cells and the flux ratios will be changed.  In other words, any reference with the focus sensor that one establishes as "best focus"  will be offset by the change in delivered telescope aberrations.  Note: aberrations introduced by the atmosphere that vary at 1Hz or faster, are averaged out by integrating the focus error signal over a time period that the user chooses, generally 30sec.

Chuck Claver argued at a very early state in the lifetime of WTTM that the telescope aberrations were changing in an unacceptable manner, a statement that I stubbornly refused to accept.  The key is that WTTM is monitoring the changing optics on an almost continual basis, whereas my experience with wavefronts extended over a few hours on any given night and at sampling frequencies of order 7-10minutes.  Therefore, WTTM provided the first opportunity to learn of variation of telescope aberrations.  Note also that Bruce Bohannon, in a internal commissioning report issued in 1995, indicated that astigmatism and coma varied, or it's scatter, was of order of 0.2microns.  The report noted that the scatter was the error in measurement, not the actual changing of the telescope optics.  Indeed, we may have been over-optimistic in the optic's stability.

The conclusion then is that the WTTM Focus Error Sensor is only as reliable as the telescope optics.  The following graphic illustrates the issue.  There are obvious offsets in the focus signal as measured by the WTTM Focus Sensor.  The data sets for the Kron-I and Johnson-R filters were obtained as follows:

1. Establish telescope guiding with the WIYN IAS Guide Probe.
2. Establish best focus on the WTTM CCD for the filter in question.
3. Reference this "best focus" to the output of the WIYN IAS AutoFocus sensor as baseline.
4. Adjust the telescope focus by small amounts and record the averaged WTTM focus sensor signal.
5. Between each telescope defocus, return to the CCD best focus and verify with the WIYN IAS Autofocus sensor.
6. Continue gathering datapoints, occasionally verifying CCD bestfocus via a CCD exposure.

As one sees from the graph, in particular the I-band data, there is a sudden shift in the focus sensor signal, indicated by the arrows.  The only apparent reason was a shift of the telescope coma as verified by wavefronts which bracketed the experiment for I.   The change is coma, by about 0.2microns and some smaller levels of astigmatism. (the shift in coma was the only factor that we could identify which changed, but I do not rule out other influences)  On a good note, one can see the potential of the focus sensor in the R-band data where a reasonable linear response.  NOTE:  Best focus in R, or any other filter, does not correspond to a 0signal of the focus sensor.  This is due to the fact that the two focal planes, focus sensor and the R filter shifted focus of the CCD are not coincident.

You can also see in the R-band data that the focus sensor turns over, or saturates quickly.  In otherwords, the effective dynamic range of the current focus sensor is limited and is heavily dependent on the CCD filter thickness.  The current WTTM configuration is optimized for a filter thickness of 5mm +/- 0.1mm.  Varying from the 5mm will only limit the dynamic range on one side of the focus.  As example, the R-Band is limited to only +20um of secondary defocus.

Last, I offer a theoretical focus sensor curve as comparison.  The curve is to communicate the relatively linear nature of the shifts, maybe too perfectly, and the point where the signal "saturates" the sensor, rolls-over.  This theoretical example is symmetric about the origin, reflection perfect conjugation with the CCD and science filter.
 

Operational Guidelines

• Use only well isolated stars, minimun of 10-arcsec seperation.  (FOV for error sensor is typically 2.00 arcsec, pupil 5)
• Use pupil 5, diameter 2.0arcsec.  Pupils 4 is accetable, 1.414 arcsec diameter , but will require stable conditions.  Lower values of pupil diameters will introduce unwanted effects to properly sense focus.
• Varying levels of scattered light, bright moon and cirrus as example, will adversely effect focus error signal calculation.
• The WTTM Focus Sensor has a much more limited range as seen in the data above.  Long time constants to correct telescope focus should not be used, try to stick to 30-45sec to avoid sensing large changes in focus, greater than 10um.
• Monitor the progress of the focus corrections closely and turn off the focus correction if the magnitude of the corrections fluctuate by greater than 10um.
• Plan to use the WIYN IAS Autofocus Sensor, use the WTTM focus sensor only as a backup.  The WIYN IAS Guide field during WTTM deployement is provided in the user manual.

Possible Future Courses of Action

1. Enhance the astigmatism delivered to the four quadrants by upgrading the corrective element in the error sensor.
2. Limit the focus pupil to only pupil 5.
3. Visit the low-level code and incorporate better error checking and errant signal rejection along with some operational enhancements.