SummaryCXC automatic pipeline aspect processing failed to produce a good aspect solution for obsid 4611. This was revealed in the V&V process as a very large (> 600 arcsec) residual error in the star centroids. The root cause was identified as a single anomalous value in the raw counts for gyro channel 1 at time 2004:167:05:03:34 (Chandra time 203663078.24). This is highly likely to be a telemetry glitch since there is no indication of corresponding anomalous behavior in any other PCAD / gryo telemetry. This was confirmed by discussion and review of telemetry with PCAD FOT engineers. The glitch can be seen in the plot below which shows raw gyro channel 1 counts as a function of time.CXC ProcessingProcessing of this obsid can be carried out in the short term in two ways:
The GlitchThe actual data values near the glitch are:ciao-436: dmlist pcadf203660505N001_15_eng0.fits'[time=203663077.3:203663079][cols time,aogyrct1]' data -------------------------------------------------------------------------------- Data for Table Block AXAF_PCAD -------------------------------------------------------------------------------- ROW TIME AOGYRCT1 1 203663077.4722892940 -12022 2 203663077.7285393178 -11998 3 203663077.9847893119 -11974 4 203663078.2410393357 20817 5 203663078.4972893298 -11927 6 203663078.7535393536 -11903The binary representation of 20817 is 0101 0001 0101 0001. If the leading zero is flipped to a 1 and the value interpreted as a signed integer, one gets a value of -11951. This is precisely what would be expected based on the surrounding values. For conspiracy theorists, the fact that the pattern 0101 0001 gets repeated twice may be important. Aspect pipelineThe aspect pipeline should seamlessly handle short glitches such as occurred. The gyro_process tool needs to be evaluated to understand why the point was not correctly filtered. The effect on the aspect pipeline output is seen below in the plot of aspect solution declination versus time. The gyro_process tool did detect a problem with the data, as can be seen in the final status bits, plotted below as a function of time. The value of 18 indicates that the data were rejected by the sigma clipping algorithm and interpolated. However, the actual bad point was apparently not interpolated or flagged.Aspect Information main page Comments or questions: Aspect Help Last modified:12/27/13 |