The strong source correction (see Section 5.7) was made with
Saturn quarter-second
ramps. Applying this correction directly to observations of strong
sources, where the ramp lengths are one half-second, does not remove the
non-linearities as the data are de-biased more than corresponding
quarter-second
ramps. This is seen in Figure 5.13 where a non-linear
half-second ramp is shown by the + symbols. The dotted vertical
lines mark out the first quarter-second of the ramp and
whole half-second ramp. It can be seen that the ramp
curves over more as time increases. This de-biasing effect is
even more apparent when one considers the
value of the first quarter-second and the whole
half-second ramp (indicated by the slope of the long and short
dashes respectively). This shows that
there is a difference in calibration for quarter and half-second
ramps, due to the de-biasing.
Hence half-second data which suffer from `saggyness' should have
their ramps reprocessed as if they are quarter-second ramps.
This is done by discarding the 22
samples from the end of a ramp, which make up the extra quarter-second.
Also, as in the normal processing, a further sample from the end is
discarded because of the end of the format
point and the usual amount is discarded from the start. This makes the
ramps equivalent to the data used in real quarter-second
ramps, by using exactly the same part of ramp. If any of the remaining
points are unusable, such as being at the voltage limit of the ADC
(see Figure 5.14), they are also discarded.
[The illuminators however are processed as normal half-second ramps,
as described in Section 4.3.6. This is done as their
calibration is based on half-second length ramps and also their
ramps should not be affected by non-linearity.]
In doing this the non-linearity
suffered will be to the same degree as those ramps of the correction
source, Saturn, and the correction can be applied successfully to the
photocurrents found from these `quarter-second ramps'.
![]() |
For LW5, with quarter-second ramps, the strong source correction is not needed as these ramps are linear. However, when strong sources are observed with half-second integrations the ramps may become non-linear. Hence for this detector the quarter-second and normal processed data (Figure 5.15 shows an example of these data) are compared with the corrected scaled LW4 data. The data which better match those of LW4 are used.
![]() |