LDEF IDE SENSOR STATUS
INTRODUCTION
While relatively rugged and reliable, the MOS detectors of IDE can be damaged under some circumstances. The most likely form of damage results in a "shorted" sensor in which the thin dielectric layer has been short-circuited. In order to determine microparticle impact fluxes, it is necessary to know the effective area of each detector array. For this reason, a scheme was developed to try and detect the number of shorted sensors as a function of time during the mission. Sensor status data was made part of the "housekeeping" record recorded every 2.5 hours during the first 346 days of the flight.
IDE DETECTOR STATUS CIRCUITRY
The circuit for a single detector is represented in Figure 1. The applied bias was 40 volts for the 0.4 micron dielectric thickness and 60 volts for the 1.0 micron dielectric thickness sensors. Normally, the device acts as a capacitor, with only a very small value of leakage current flowing from Vbias to Vdd. The Impact/Discharge measuring output is capacitively coupled so it only registers a signal when the MOS detector is discharged (usually induced by an impact). The status output normally lies at (approximately) Vdd. If a sensor becomes short-circuited due to some failure, the status output takes on a value close to 0 volts (set by the degree of short circuit current flowing and the 470K/4.7M voltage divider. If the status output dropped significantly below Vdd towards 0 volts, this fact was registered as part of the IDE status data recorded at each data dump.
Figure 1 Impact and Status detecting circuitry for IDE MOS impact detectors.
DETECTOR STATUS OBSERVATIONS
Analysis of the detector status data indicate that, in addition to detectors becoming permanently shorted, several of the panels also had periods during which a number of detectors displayed transient indications of being shorted. In some cases, every sensor on a panel displayed this indication. There is reason to believe that this transient behavior was related to the position of LDEF in its orbit relative to sub-satellite latitude and/or the sub-satellite solar direction. The SOUTH 0.4 panel displayed a high level of transient detector behavior as well as a relatively high particle hit rate. For this reason, the following discussion will concentrate on the SOUTH 0.4 panel.
Figure 2 Impact activity and status data for the South 0.4 micron (high sensitivity) dielectric thickness sensors during the first 719 orbits.
In Figure 2 the upper trace indicates particle hit rate and the lower trace indicates the number of detectors indicated to be inactive (shorted) for the first 721 orbits of the mission. Full scale for the lower trace is reached when all the detectors are indicated to be inactive. The entire data taking portion of the mission is displayed in a similar fashion in Figure 3. The transient behavior of the detector status is evident in both figures. In addition, it appears that the there is no immediately obvious correlation between hit rate and status behavior. It is possible to see that, in addition to the transient behavior, some detectors do become essentially continuously inactive, with the number of such detectors increasing as the mission progresses. Note, for example, the thickening of the "baseline" of the lower trace of Figure 2 for the last 1/4 of that trace. A similar increase in baseline thickness is evident in Figure 3, though it is somewhat obscured by the continuing transient activity.
Figure 3 Impact activity and status data for the South 0.4 micron (high sensitivity) dielectric thickness sensors during the entire 346 day data taking period.
CORRELATION OF ORBITAL POSITION AND TRANSIENT BEHAVIOR
It was quickly apparent that the transient status behavior was not randomly distributed, but was related in some fashion to the terrestrial latitude directly beneath LDEF at the instant of the data dump. Extensive analysis suggests that the important causes of transient behavior are the subLDEF latitude and solar time. The occurrence of transient status behavior is plotted in Figure 4 as a function of terrestrial latitude and LDEF geocentric ecliptic longitude minus solar ecliptic longitude (that is, the difference in the direction to LDEF and the direction to the sun, as seen from the center of the earth). It is clear that the transient behavior is highly concentrated near the equator at about 60 degrees; equivalent to about "3:30 PM". No good model has yet been developed for this behavior.
Figure 4 The incidence of detector status transient behavior as a function of terrestrial latitude and direction to the sun (L-Lsun=0 at subLDEF "noon").
The Significance (or lack therof) of the Transient Status Behavior
In order to interpret observed particle impact rates in terms of particle fluxes, it is necessary to know the effective area of the detector surface. Hence, it is necessary to know the number of active detectors at each time during the mission. This can be determined easily in the case of permanently shorted detectors; the transient behavior described above, however, introduces a problem. The detector status data provide only a "snap shot" of the instantaneous detector status at the time of a data dump. There is no indication of how long before (or after) the data dump the detector remained inactive. Since the data dumps were relatively infrequent (occurring approximately every 2 1/2 hours) it is not possible to investigate this possibility directly. If the cause of the transient behavior was limited to a portion of the orbit it is possible that the detectors were inactive for only a small fraction of the time. Therefore, a statistical approach to analyzing the hit rate data and the detector status data was developed.
Figure 5 Illustration of the locations of the four data analysis segments, relative to a data dump time. The duration of one LDEF orbital period is also indicated.
If the transient detector inactivity significantly reduced the effective area of the SOUTH 0.4 panel, then such a reduced effective area should show up as a reduced counting rate. Therefore, there should be a correlation between the number of detectors indicated inactive during a data dump and the counting rate during the period represented by that dump. Since the length of the period of inactivity is unknown, it is possible that reduced counting rates will shown up to a larger degree for the time period adjacent to a data dump than for more distant time periods.
With these possibilities in mind, the hit data were sorted based on the number of inactive detectors recorded in the data dump nearest in time to the time of impact. Further, the period adjacent to each data dump was divided into four "time segments" as illustrated in Figure 5. and the data were grouped by segment. These counts were then converted into hit rates (in units of hits per data dump period) by dividing by the number of data dumps with the appropriate number of inactive detectors.
Figure 6 Histogram of the number of occurances of "N" sensors reported "inactive".
In Figure 6, the number of data dumps (out of 3417 in the mission) are plotted as a function of the number of inactive detectors indicated during the dump. The resulting hit rate (hits per dump) is plotted in Figure 7. It is immediately obvious that the hit rate is almost independent of the number of detectors reported inactive. It is especially important to note that the hit rate at the far right of this plot (when the status data indicated that no detectors were functioning at the time of the dump) is virtually the same as the hit rate at the far left (when all detectors were indicated functioning). The average impact rates for the four time segments A, B, C, and D are given in Table 1. These numbers confirm that the transient detector behavior did not appear to effect the detection of particle impacts.
Figure 7 Impact rate on the South 0.4 micron detectors as of the number of sensors reported "inactive" in a data dump.
Table 1:
Segment A B C D
average rate 16 21 19 17
standard dev 2.3 3.5 3.3 2.6
Table 1 Average impact rate (hits/data dump) for the two time segments before a data dump (A and B), and the two time segments after a data dump (C and D).
This document produced by
John Oliver; oliver@astro.ufl.edu