The (RDF) ...anc.fits file contains the aspect table which lists the pointing position (RA, DEC, roll) at one second intervals. The fractional time assigned to each set of aspect solutions depends on the clock reset which occurs each January. Thus, for a year (more or less) the fractional time will be a constant. The RDF ...anc.fits files contain the correct fractional times, and we have listed them in the table below. NB: The error occurred during the correction of the photon positions by SASS; thus the photon positions in the RDF *_bas.fits file (and quantities derived from these photon positions) are affected, but the aspect information contained in the ASPECT extension of the RDF *_anc.fits file is correct.
| Reset of S/C clock | GOOD frac. | BAD frac | Difference (s) |
| Aspect t (s) | Aspect t (s) | ||
| 90 Jun 01 (Launch) | |||
| 0.15625 | 0.00122 | 0.155 | |
| 91 Jan 25.386331 | |||
| 0.6875 | 0.00537 | 0.682 | |
| 92 Feb 11.353305 | |||
| 0.140625 | 0.00140 | 0.139 | |
| 93 Jan 18.705978 | |||
| 0.9375 | 0.00732 | 0.930 | |
| 94 Jan 19.631352 | |||
| 0.625 | 0.00488 | 0.620 | |
| 95 Jan 18.169322 | |||
| 0.6875 | 0.00537 | 0.682 | |
| 96 Jan 28.489871 | |||
| 0.609375 | 0.00476 | 0.605 | |
| 97 Jan 16.069990 | |||
| zero | 0 | 0 | |
| 98 Jan 19.445738 | |||
| 0.046875 | 0.00037 | 0.046 | |
| END OF MISSION |
For observations between 16 Jan 1997 and 19 Jan 1998, the fractional time was zero so the bug has no effect. Although rounding errors might introduce a very small percentage of misplaced events, a test run of an observation from this time period showed no apparent changes in the image. Similarly, we would expect the effect to be small for data taken after the 19 Jan 1998 clock reset. The test run on Cen A cited in this document was observed on 1994 Aug with a fractional aspect time of 0.625s.
Even for observations taken when the bug has a large effect, there are some instances when the correction makes very little difference. If the aspect solution is not changing, the effects of the bug are mitigated or absent. The normal wobble excursion is about 3', and thus the expected velocity of the spacecraft is of order 1''/sec (180''/201s, the denominator being half of the 402s wobble period). Since the timing error is always less than one second, we would expect the resulting positional error to be less than one arcsec. However, it is often found to be much larger. We suspect that the ideal wobbling situation is seldom realized in practice; rather star tracker errors and/or actual spacecraft 'jitter' can produce an aspect solution that is often changing more rapidly than 1''/sec. A casual inspection of the aspect table (from the asc.fits file) verifies these suppositions.