Figure 2.1: Examples of single detector jumps
A. Heras
9 June 1997
This note summarizes the study made on single detector signal jumps which was triggered by D. Boxhoorn's note ``Partial analysis of observation 11205401''. The signal jumps considered here occur in only one detector at a time and can be generally described as a sudden increase or decrease of the signal, which remains constant or has a long recovering time (>> 10 s) after the jump. These jumps are normally seen at low signal levels.
Seventy-nine observations from calibration revolutions performed after PV phase up to revolution 468 have been analysed. They have been divided into the following measurements considered separately: dark currents; diffuse low calibrator; and grating low calibrator (no scan). The reset time in the first two measurements is 2 seconds and in the last one 1 second. Band 4 data have not been analyzed because of the highly perturbed signal due to glitches.
An automatic procedure to detect jumps higher than the average noise level has been run through the observations. Some examples are shown in Figure 2.1.
Figure 2.1: Examples of single detector jumps
Table 2.1 displays the results obtained from the analysis of bands 1 and 2. Jumps in band 3 are of different nature from those in bands 1 and 2 and are therefore not included
| Band 1 | Band 2 | Total | |
| Total jumps | 70 | 34 | 104 |
| Average jump rate (per hour) | 3.43 | 1.66 | 5.10 |
| Jumps during dark | 23 | 18 | 41 |
| Jumps during diff. low | 19 | 11 | 30 |
| Jumps during grat. low | 28 | 5 | 33 |
The jumps in bands 1 and 2 are similar to each other, but different from jumps in band 3. Band 3 jumps generally appear as alternating up and down jumps between two constant signal levels (so-called ``pop-corn'' noise). The frequency of these jumps is higher in bad behaving detectors, i.e. detectors for which the dark currents in the observing period are significantly higher than in the beginning of the routine phase (e.g. 30 or 36).
The following points will refer only to bands 1 and 2.
The jumps can be either positive or negative.
For a few exceptions (e.g. detector 2), the signal jumps observed
for a particular detector are rather variable from case to case.
The jumps are greater during the grating calibrator low measurement
(reset time interval 1 second) than in the other two measurements (reset time
interval 2 seconds).
In half of band 2 jumps, the recovery time of the dark current to
before-jump levels is shorter than the duration of the measurement (5 minutes).
If the cross-talk correction is not applied, no trace of the jump is
observed in the other band detectors and the jump is seen only in one detector
at a time. In OLP processing this cross-talk correction is always applied.
The analysis of the data at ERD level has been done following the suggestions in DRB's notes. As an example, Figure 4.1 displays the difference in bits between sample 24 and 9 in each ramp as a function of the value of sample 24. The two clouds of points for the sample 24 with value 1950 correspond to the first second of the 2-second reset time interval, and the cloud around the abscissa value 2000 to the second second. The points in the lower cloud around x=1950 represent the ramps before the jump and in the upper cloud after the jump. The fact that in the first second of the reset time interval the distinction between ramps before and after the jumps can be seen much more clearly than in the second second, indicates as a possible cause for the jumps the residual effect on the ramps of the after reset pulse.
Figure 4.1: ERD Characterisation of jumps
The association of signal jumps and glitches has been studied with the following results:
For bands 1 and 2, 18 out of 41 and 13 out of 30 jumps (almost 50%) are associated with the detection of a glitch in the dark current and diffuse calibrator low measurements. This proportion is reduced to 4 out of 33 for the grating calibrator low measurement.
There is practically no association of band 3 jumps with glitches.
There is no distinction between jumps associated with glitches and those which are not. Likewise, glitches associated with jumps are not different from no-jump associated glitches.
No relation has been found between the height of the glitch and the magnitude of the signal jump.
The signal jumps observed for each detector vary significantly between different cases, and may be negative or positive.
Jumps in band 3 are of different nature from jumps in bands 1 and 2.
The frequency of jumps in bands 1 and 2 is of approximately 5 jumps per hour.
The signal jump is more clearly reflected on the first second of the reset time interval than in the second one, indicating a relation between the signal jump and a residual after reset pulse effect. This will be further investigated when working on a possible correction for these jumps.
For bands 1 and 2 and for the measurements with reset time interval 2 seconds, half of the jumps are associated with the detection of glitches. However, no correlation has been found between the characteristics of these jumps and and those of the glitches.
Users are recommended to examine the SPD from their observations to look for
such jumps. If they are present the user can either throw the affected
portions of the data away or try to adjust the baseline of the affected
portion to the pre- and post-jump baseline.