3.5. Signal Processing and Storage

  The CCD is read out 15 times per second. A sampling of the CCD's average minimum output is fed back to a circuit that adjusts the electronic zero reference level to insure that changes in the CCD's dark current do not allow the background level to go too close to a zero (or even negative) signal amplitude or go to a high enough level to consume a good portion of our dynamic range. The output signal from the CCD is then amplified and digitized to an accuracy of 8 bits at a rate of 1.67 megapixels s^-1. The (adjusted) background level was typically 50 DN (Data Numbers), except when a bright spectrum signal was present, thus causing the dark level servo to over-respond and drive the true zero to 10 DN (this was largest correction that we encountered; it was for the brightest star observed, $\beta$ Cen).

If a photoevent lands exactly in the middle of a CCD pixel so that little charge is shared with adjacent pixels, the signal is about 45 DN above the background level. Thus, in a single frame, about half of the pixels will be saturated at 255 DN in regions where the accumulated flux is as high as about 5 photoevents pixel^-1.

Successive frames are added into the contents of an accumulating memory that has a capacity of 16 bits in each pixel address location. With some exceptions, images made during the first ORFEUS-SPAS flight discussed in §6 were accumulations of 511 frames, representing an exposure integration time of 34 seconds. If desired, shorter accumulations can be made, with options of 1, 63, 127 and 255 frames.

Images can be stored within any of 128 exposure locations in volatile memory. At appropriate times, the images are transferred to permanent storage on the spacecraft's tape recorder. Because of the limited bandwidth of the tape recorder, the transfer takes 49 seconds. Camera exposures and data transfers must be done in sequence, and thus the transfer time represents a real loss in observing time.