Data Reduction and Analysis Therapy Session
#1, 29 October 1999
- Signature: Half of S3 corrupted - looks bright in an image. Can be
dealt with by filtering E<0.3, but with a cost in exposure.
- Cause: bias map corrupted for even numbered columns.
- How often? Mean-time-between failures is about 50 ks. Occurred
in 10 science runs.
- Short term: patch --- when errors occur, turn of half a chip (and
assign FEP0 to a chip where the loss is not critical).
- Directions, Options: turn off a chip; patch
to use 512 rows at specifyiable row offset (difficult); turn of bias
thresholding (e.g., bias=0; but at expense of keeping hot and
- Problem: aspect solution sometimes goes bad during an
observation. Diagnosis: look at zero-order centroid in x or y vs
time. Not clear whether event time error, or aspect solution
error. Not clear if only in manually processed data or also in
automatically processed data. Clealy timing problem in alternating
Gain, Order Sorting Issues
- New "rmfile" is available for order sorting (via tg_resolve_events)
- New gain map and algorithm are being produced for ACIS event
processing. Temperature-dependent files are needed. Currently, DD
has IDL gain correction for serial and parallel transfer, and MW has
used a gain map for position-dependent PHA spectral extraction.
Protoype gain map is:
(explicit or implied; NEW actions get a "*")
- * HLM: check obsid 1450 in automatically processed data for aspect
- * HLM: proved Peter with times and exposure number for obsid 107,
to investigate aspect/timing problems.
- * DPH: request that housekeeping and processing logs be included
with in manual processing products.
- * DPH: provide CXCDS with criteria for handling source-detection
failures in automatic processing: e.g., use RA_NOM, DEC_NOM when no
source is found (short-term, easy fix); use frame-shift-streak and MEG
intersection for zero-order-blocked cases (may not be
worth the effort to do automatically, given frequency of use?)
- * DPH: see about getting a "source size" field on proposal form,
for use in processing for zero-order source region and detection
- * GEA: gain vs time is required in calibration files, and temperature is
needed in headers, along with the file used.
Next meeting: Fri 5 Nov 1999, 1:30p