Photometric zeropoints are around 21, 22.8, 22.9, 22.85, and
22.1 for UBVRI, where zeropoint is magnitude that gives 1 DN/s. Exposure
times to reach a given S/N depend on the sky background; because of the
relative poor image quality (see below), typical stellar photometry uses
apertures of radius 2-5 arcsec, so there is significant sky contribution.
We can do all-sky photometric calibration observations when conditions
permit, and have done this both in UBVRI and ugriz. Generally, all sky
photometric rms residuals on high quality nights are in the 0.01-0.02 mag
range, although we have not yet been able to achieve this level of
accuracy for U band observations.
Pointing is adequate, but not great; small scale
structure on the drive surfaces is likely responsible. Typical pointing
models give rms pointing performance of about 20 arcsec; for whatever
reason, actual performance using the pointing models is somewhat worse.
Given a 16 arcminute field of view, this does not cause trouble for putting
objects in the FOV. For robotic operations, the standard mode is to
reset the coordinates on a nearby bright SAO star for every object, so
generally, objects observed in this mode end up quite near the center
of the chip.
Because of the relatively poor pointing, tracking
is relatively poor, so guiding is essentially always needed to preserve
image quality for even moderate length exposures, and to keep the object
in the same position over a series of exposures. The telescope has an
offset guider with a 1024x1024 Finger Lakes CCD system, in a guider which
can move radially in the field. The guider FOV is large enough so that
guide stars can almost always be found at the default radial position.
Unfortunately image quality is relatively poor. We have
seen images as good as almost 1 arcsecond, but generally images are in
the 1.5-2.5 arcsec FWHM range. There are likely multiple causes for the
poor image quality, and we hope to be able to make improvements. Problems
probably include thermal inertia of the telescope/mirrors (image quality
almost always improves as the night goes on), dome seeing, and poor
alignment of the telescope (in particular, the current tertiary is very
difficult to adjust).
Flat fields are taken periodically of the twilight sky; no
internal flat field screen is available. The flats appear to be relatively
stable in time, so it appears to be possible to use flats constructed at
significantly different dates than the observation dates.
The E2V detector has fringing when observing at longer wavelengths. It is
possible to subtract out the fringing pattern to a large extent with
fringe frames, although one needs to account for the changing proportion
of the fringe component depending on the amount of moonlight.
There are nonuniformities in shutter timing across the field that are
apparent for exposure times less than a second or so. It should be possible
to correct for these with shutter shading calibration.