The CCDs will be operated under the multi-pinned-phase (MPP) bias scheme. This greatly reduces the effects of ionising radiation, at the expense of a lower full well capacity. The temperature of the CCD will be stabilized at about -100C, again to reduce radiation induced effects, as well as thermal noise. We envisage the use of passive cooling, with small heaters on the CCDs to regulate their temperature. A mosaic of four butted frame transfer CCDs will cover the focal plane. The 2048 1024 devices can be read from two separate amplifiers to improve readout times, and to provide redundancy should one chain fail. To further reduce readout times the pixels can be `binned' 2 2, and intelligent reading strategies can be devised to spend more time reading the stellar images, and less on the background. With a conservative read time of 2s per super-pixel, there is no problem to maintain a cadence of one frame a second. Frames are co-added to provide integration times of about 10s, before photometric reduction takes place.
Laboratory tests (Buffington et al. 1990) have demonstrated that CCD detectors have the requisite physical stability (in a well controlled thermal environment) to achieve the desired photometric precision, whilst simulations (Appourchaux et al. 1993b) have demonstrated that the noise due to reading the CCD, to data reduction, and to spacecraft jitter are below the photon noise. We should emphasize that we are performing differential photometry in a frequency domain defined by the scientific objectives, and which is well separated from the low frequency noise regime where instrumental drifts and ageing effects have a considerable influence.