The telescope after nominal mechanical optical alignment has poor pointing performance. However, the pointing can be fairly well modelled using standard terms for an alt-az telescope implemented, e.g., by the TPOINT software, which we use. When making a pointing model using 100-200 stars around the sky, we typically achieve rms pointing performance of 20 arcsec.
Pointing accuracy appears to be limited by apparent small changes in scale (encoder steps/degree for alt and az, motor steps/degree) with time. The changes, at least in azimuth, do not appear to be correlated with the temperature.
In addition, we appear to very occasionally have major shifts in some component that require us to do an entirely new pointing model. This is not well understood, but it appears to happen very infrequently.
Tracking performace directly comes from this pointing performance because we track using encoder information in the same way that we point. The tracking performance varies with location in the sky. Typically, we estimate that moderately good tracking, i.e. image quality can be maintained, for roughly 3-5 minutes.
We have implemented guiding software to allow for longer integrations. With the current system, we perform relatively slow guiding; we can update roughly once every 1-2 seconds fastest. Our usual mode of operation is to operate significantly slower, usually computing an offset every 1-2 seconds but averaging five of these before sending a correction to the telescope.
The performance of the guiding still depends on the quality of the pointing model because we only have position measurements of the guide star in two coordinates while the telescope is a three axis system. We currently assume that the rotator position is correct, and guide in alt-az alone. This appears to be sufficient so long as the pointing model is good; when it is not good, we get noticable rotation around the position of the guide camera when we are guiding. Better guiding can also be obtained if the observer is careful to rezero the coordinate pointing immediately before guiding (i.e. setting the pointing to be correct to first order).
Our current guiding performance has yet to be fully quantified.
We are continually working on image quality. In its original condition, the telescope never never achieved image quality much better than 1.7 arcsec, with a strong wavelength dependence such that the best images were achieved at the longest optical wavelengths, and significantly worse performace was seen at shorter wavelengths.
Based on analysis of the images, we decided that the optics were likely the cause, and we sent all three mirrors to be refigured/repolished in June 2002 to Rayleigh Optical Corporation in Baltimore. They found that the primary mirror indeed did not have a very good figure. The mirrors were returned in October 2003.
Since then the image quality is better, although still leaves some room for improvment. Our best images are now around 1.25 arcsec FWHM, with a moderate wavelength dependence in the same sense as before. However, we don't usually seem to be able to obtain this for longer exposures; currently, we are investigating focus drifts, collimation drifts, and seeing effects.
We have attempted to collimate the telescope several times. We have twice done mechanical collimation by
From this initial mechanical alignment, we have refined the collimation using secondary tilts to minimize the apparent coma in the images. The secondary can be tilted using motor control, but not decentered.
Comments from before mirror repolishing
Out-of-focus images show no strong signs of coma or astigmatism. However, they do show very prominent, roughly azimuthally-symmetric, zonal errors. At the current time, we suspect that the poor image quality arises from one or more misfigured optics. The primary support system is also a possible culprit, but it is not clear how it would produce azimuthally symetric features, as the primary support structure consists of a set of three ``tripod'' cup mounts.
There is also some clear evidence of spherical aberation from out-of-focus image pairs. The various pairs shown here were taken on 12 March 1998 at 4 different back focal distances; the top pair was taken at the ``nominal'' CCD position (with the old guider), the next pair in 6 cm, the next pair in 3 cm, and the final pair out 1.5cm from the ``nominal'' CCD position. Note with the new guider, we are closest to the bottom position.
The image quality appears to be rather strongly dependent on wavelength, despite all reflective optics (apart from the dewar window). Best image quality as a function of wavelength is given by the following table:
The primary mirror was removed from the telescope in February 2000 to be realuminized; the tertiary had to be removed at the same time, but was not realuminzed, and the secondary was not removed at all. Upon reinsertion of the primary, we redid the collimation being more careful about the positioning of the secondary mirror. However, after several attempts at collimation, it appears clear that the image quality degraded since before the primary was removed. This is currently a large puzzle. Focus data from October 2001 give:
Coarse focus run data obtained on 15 March 2000 is shown here. Note that the vertical streaking seen in stars near focus is from the CCD which was being operated with the shutter continually open.
Careful measurements have not yet been made on these images as compared with previous images (not clear if sufficient archival data exists), but by eye, it seems that the spherical might be a bit worse now than before; of course, this would be rather hard to understand.
Errors in primary-secondary spacing will lead to the introduction of spherical
aberration. From Schroeder, we have
Note that the shift in the focal plane is given by .