The Accumulate Readout Mode (
ACCUM) generates the simplest basic exposure. In its simplest incarnation,
two array readouts, illustrated in
Figure 8.2, it is analogous to a WFPC2 or ACS readout. Its main difference relative to a
MULTIACCUM exposure with
NSAMP=2 (or two readouts selected) is that the first array read gets subtracted from the second array read by the on-board computer, and only the difference image is sent to the ground.
In other words, the returned image is the difference between the second and the first pass pixel values, and the integration time is defined as the time between the first and second read of the first pixel. The minimum exposure time is ~0.6 sec, and the minimum time between successive exposures is ~8–12 seconds.
ACCUM does not allow pipeline identification of cosmic ray events or to correct for pixel saturation.
In ACCUM mode
multiple initial and final reads can be obtained in place of the single initial and final readouts. In this case, after the detector array is reset, it will be followed by 1–25 (specified by the
NREAD parameter) reads of the initial pixel values which are averaged onboard to define the initial signal level. After the exposure time has elapsed, the final pixel values are again read
NREAD times and averaged onboard. The data downlinked is the difference between the initial and final average signal levels for each pixel. The integration time is defined as the time between the first read of the first pixel in the initial
NREAD passes and the first read of the first pixel in the final
NREAD passes. The use of multiple reads in
ACCUM mode is illustrated in
Figure 8.3 for the case of
NREAD = 4.
The advantage of this method is a reduction in the read noise associated with the initial and final reads. In theory the read noise should be reduced by 1/(n)
1/2 where
n is the number of reads. However, the amplifier glow (see
Chapter 7) adds extra signal and associated photon noise for each read, especially towards the corners of the array. Amplifier glow is an additive noise source large enough that for
NREAD > 9 there is little further gain in noise. In practice, the maximum improvement in effective read noise over a single initial and final read is no larger than a factor 40–50%, due to the added amplifier glow that each read-out adds to the final noise budget. For integrations where source photon noise or dark current noise exceeds the detector read noise, the multiple readouts may not offer much advantage. This option puts a higher burden on the CPU and requires an additional time per readout of 0.6 seconds. This mode does not allow pipeline identification of cosmic ray events or correction for pixel saturation.
