A common goal of observing with Mosaic is to
derive photometric magnitudes for many objects. Due to the wide field and
multi-CCD nature of Mosaic, though, there are several issues that may affect
the accuracy of one's results, and in particular, the hope of achieving 1%
photometry.
CCD-to-CCD sensitivity differences: While flat-fields remove the gross sensitivity differences across the CCDs, each CCD exhibits its own spectral response. Relative to the B-band, U sensitivity varies by up to 10% and V by 5% from CCD to CCD. Consequently, the color terms for one CCD are not applicable to data taken with another CCD. At a minimum, therefore, one must know the color terms for each CCD. But there is another quirk: if data are taken in "dithered" mode, then some stars may have been exposed across several different CCDs, each with different spectral properties. Photometry from the stacked image will be sensitive to the combination of color terms.
Variations in spectral characteristics on a single CCD: even across a single CCD, there are variations in sensitivity approaching 4% (U relative to B). This problem, though, is not new; the old TI and Tektronix CCDs also exhibit variations almost as large.
Shutter corrections: The large shutter for Mosaic is not perfectly accurate. That is, a command for a 1 second exposure will not open the shutter across the entire array for exactly 1.000 seconds. As of November 2003, a 1 second exposure held the shutter open for 0.97 seconds at the top of the array and 0.96 seconds at the bottom. This effect can be measured and calibrated out, but it can also be avoided with longer exposures since the shutter error is a constant offset in time.