Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/ops/swat/safing/96195/doxsey-7-14.txt Дата изменения: Mon Apr 26 17:33:46 1999 Дата индексирования: Sun Mar 2 02:40:25 2014 Кодировка: Поисковые слова: п п п п п п п п п п п п п п п п п п п

From: DOXSEY@avion.stsci.edu
To: LONG@stsci.edu, jay@stsci.edu, sbaum@stsci.edu, mcclure@stsci.edu,
mackenty@stsci.edu, miller@stsci.edu, NORMAN@stsci.edu
Subject: HST entered inertial hold safemode on Saturday, July 13
Date: Sun, 14 Jul 1996 12:27:17 -0500 (EST)


HST Safemode Entry

The HST entered inertial hold safemode at 17:31 UT on Saturday.
The sequence of events appears to have been the following:

At 16:46 one of the four elements of the on-board quaternion
jumped by the equivalent of 7 degrees. The quaternion is the on-board
form of the rotation matrix between spacecraft and celestial coordinates,
and represents the HST's idea of where it is pointing. This discontinuous
jump was consistent with a single bit flip in the words holding the data
(more about this below). As a result of the sudden jump in this component,
loss of guide star lock occurred. The event occurred just before an
earth occultation, so the loss-of-lock recovery was started, but before
it could complete (it would have failed), the commands to put the FGS's
in default for the occultation executed and the recovery was aborted.
The HST then remained in gyro hold during the earth occultation. As
the occultation ended, the commanding for the guide star re-acquisition
kicked off. As the FGS's are brought out from default, a safemode check
on whether the HST is looking at the earth is enabled. Because the
quaternion was in error by seven degrees, this check failed and safemode
was entered. At the time, the HST was, in fact, sufficiently clear of the
earth, so there was no risk to the FGSs.

The safemode entry proceeded normally and all HST and SI
responses since have been as expected. The MOSES staff used
the FHSTs to verify the pointing of the HST and have uplinked
a correction which has reset the quaternion which was affected.

There is considerable uncertainty at this time as to
what happened in a hardware sense. If the upset occurred in
the co-processor memory cell holding the quaternion, then we
would have expected a memory parity error, which did not occur.
It is possible the problem occurred in one of the CPU arithmetic
registers, or along the IO buss between the CPU and memory.
Some problems in these components should also have resulted in
processor errors, but there were none.

The quaternion is used in computations 40 times per second,
even in inertial hold mode. These computations kept processing
right through the event, and have been running without additional
problems for the last 24 hours. All the components involved are
very radiation hardened, so it would take a very unusual particle
event to cause the problem.

So, the event could be a one-time only upset or the
onset of a very intermittent hardware problem somewhere in
the DF-224/Co-processor combination. Operations and software
problems seem to have been ruled out as causes.

The good news is that the project has decided to go
ahead and re-start science observations. The bad news is that
they would like to use one of the two tape recorders to capture
engineering data during ZOEs, to be sure there is data available
should a similar event occur again. The plan is to ensure the
extra engineering data coverage for about the next month. If
the problem is due to an incipient hardware problem it likely
will occur again during that period.

Use of only a single tape recorder for science data will have
a major impact on the science program. The PRESTO staff are working
to rebuild next week's calendar, stripping out enough activities so
that a single science tape recorder will cover the needs. This
replanning will take several days, so we will not re-start
observations until Tuesday, or possibly Wednesday. They will also
have to start replanning the next few weeks to use only a single
recorder. The impact of all this on the cycle 6 schedules will
be unknown for a little while, and will be reported/discussed
in the regular forums (e.g. TTRB, OPM, DOC, DMM, etc).