In traditional photographic astrometry, the wider field of view and longer exposure time ensures that there will be several reference stars on every plate. Unfortunately, this is not true in CCD astrometry. Even in the more densely-populated regions of the ecliptic (Scorpius, Sagittarius), very few stars appeared in the CCD frames of Saturn and its satellites. Those that were visible were too faint and too sparse to provide astrometric reference positions.
We therefore turned to other methods to determine the scale and orientiation of the CCD with respect to the celestial reference system. During the 1991 campaign, we tried two techniques: star trails and double stars.
Star trails were obtained by turning off the telescope's tracking drive and allowing a 3rd or 4th magnitude star to drift across the field of view for a specified length of time. However, close examination of the CCD frames showed that the trail was not a uniform straight line but exhibited random motions in declination. It was clear that the telescope was effectively "floating" in its mounting. As a result, the star trails could not be used to establish either a scale or an East-West direction.
We made numerous exposures of the double star 61 Cygni, which had a separation of approximately 60 arc-seconds in 1991. Unfortunately, both components of this star are rather bright, and it was necessary to use very short exposures (0.1 second) in order to avoid saturation. This exposure time was too short for atmospheric "seeing" to produce a circularly symmetric images of each star. Most of the stellar images were irregular in shape and it was therefore impossible to place any confidence in the pixel coordinates of the "centres" of the images determined by IRAF.
In 1993, we took advantage of the larger field of view of the new CCD device in order to obtain images of the globular cluster M92. The brightest stars in this cluster were catalogued by Barnard and Schlesinger at the beginning of the century, but we have produced our own astrometric catalogue based upon photographic plates taken on the JKT. We made one exposure of M92 on each night. Comparison of the positions of the stars in the CCD frame with the astrometric coordinates of the same stars in our catalogue yields the precise scale of the CCD and its orientation with respect to the true north celestial pole of date.
In 1994, we supplemented our set of M92 calibration images with observations of M15, another globular cluster at greater Right Ascension. This cluster was chosen because it has recently been studied by Le Campion et al. (1991) who published an astrometric catalogue. We intend to use M15 as our calibration target in future years because M92 will become less easy to observe as the opposition date of Saturn falls later in each successive year.