---

Cautions, Idiosyncrasies, and Interpretation

• General Discussion of Errors: Users of this pointing information should of course be mindful that accurate absolute astrometry depends on many calibrations, and that such a pointing determination is only as accurate as guidestar and target positions, positional knowledge of the focal plane, and internal instrument calibrations. Typical relative errors in guidestar coordinates are 0.3" to 0.4", and alignment uncertainty between the FGSs has been as high as 0.5". A further discussion of pointing calculations and errors is in the Chapter 2 section, Pointing Calculations and Sources of Error. Information on errors contributed within the SI itself can be obtained by the SI teams.

• STARTIME, ENDTIME: The STARTIME of an observation log should not be confused with the exact start of an exposure. The STARTIME and ENDTIME keywords in the .cmh/.jih header contain a small amount of overhead, the length of which is dependent on the instrument and the mode of observation, (though usually not more than a few seconds). The science data headers should be consulted for more exact exposure times. In nominal cases however, the STARTIME can be relied on for occurring after the guidestar acquisition, during a stationary period just before an observation takes place.

• Bad or Missing Telemetry: Between day 19 1996 and day 60 1997, problems with the engineering data tape recorder meant that HST needed TDRSS satellite visibility to transmit engineering data. Typically 20% of every orbit during this period contained no engineering data and thus no pointing information. Dropouts in engineering telemetry produce time gaps if there is a long continuous gap in the data, or INDEF values for an occasional compromised point; missing keyword values in the ascii .jih/.cmh file can also be seen during substantial telemetry dropouts. Look for keywords TLM_PROB and TLM_GAP at the end of the .jih file for information on such gaps. After February 1997, the second servicing mission restored data recorder capability and complete engineering data coverage.

• Telemetry formats: The engineering data are transmitted in a variety of formats, differing in their rates of update and information content. The formats are AN, FN, PN, and HN, with HN being the most common. Please review the section describing the keyword "TLMFORM" for details. Caution: The FN format does not contain spacecraft position or velocity information. As a result, several keywords and table entries will contain default values (blanks in the header and "INDEF" in the tables). In addition, the velocity and differential velocity aberration corrections which use this information will not be applied. The absence of this correction introduces errors in the pointing data and a drift in the jitter data. Sometimes engineering telemetry switches occur during an observation. The impact of switching to and from FN format is discussed individually for each parameter that is affected.

• Telemetry update rates: The variety of engineering parameters sampled in these tables have different rates of update. For example, the observation logs provide the position of a star in an FGS every 0.05, 0.15 or 0.5 seconds depending on the telemetry mode. The orbital data such as vehicle position and velocity are updated every 6 or 60 seconds, again depending on the telemetry mode.

• Short observation idiosyncracies (less than 60 seconds): Because of the range of sampling rates for the various parameters, (50 milliseconds to 1 minute), short observations must be analyzed with caution. Each row in the table reflects the telemetry value last resident in the buffer. Observations less than the update rate will contain the last valid entry before the observation started.

• "INDEF", blank, or -32767 values: Indicates that no relevant information is available: Any item that is not relevant for a particular mode, unavailable in a telemetry format, or missing because of a telemetry transmission problem will be assigned the value "INDEF" in the tables and blank or -32767 in the header record. IRAF recognizes INDEF; IDL and FORTRAN assign the number 1.6e+38 to INDEF.

• Pointing reconstruction techniques: The guiding mode dictates the technique and input data for OMS pointing reconstruction calculations. There are three modes, differing in implementation and pointing accuracy: two FGS mode, single FGS plus gyro mode and gyro mode.

• Two guide star FGS mode: Two guide stars are used for fine lock or coarse track control, one star controlling pitch and yaw and the other star stabilizing the roll. The observation log software uses data from each FGS to produce the guide star position in V2,V3 space, the RA and Dec of the aperture in question, and the relative jitter of the target in V2,V3 near or in the SI aperture. In the absence of anomalies, and if the vehicle is stationary (no slews and no POS TARG offsets), the RA, Dec, and roll reported in the observation log tables should be comparable to the predicted values to within the reconstruction errors, guide star and target position errors and the jitter. Absolute accuracy of the aperture RA and Dec should not be worse than approximately 1 arcsec. Relative accuracy throughout an exposure will be of order 5-30 milliarcsec.

• Gyro mode: The gyros are used to control the vehicle attitude. When in gyro mode, the actual pointing errors, as well as the errors in the reconstruction of the pointing, are much larger. Both relative and absolute pointing data should not be over-interpreted. Gyro drift can range from 1 to 5 milliarcseconds per second and its magnitude depends on the slew activity and the recent history of Fixed Head Star Tracker updates immediately prior to the observation. In addition, inherent errors in the attitude data and in calibrations between the gyros, Fixed head Star Trackers, and the FGSs also add to the absolute uncertainty. Absolute accuracy of the aperture RA and Dec will be of order 2-100 arcsec and relative accuracy will be of order 1-5 milliarcsec/sec.

• Single FGS plus gyro mode: One guide star is used to control pitch and yaw and the roll is controlled by the gyros. The RA and Dec are also calculated from the gyro data. The jitter at the aperture is reconstructed using the single FGS data. The pointing reconstruction errors are the similar to those described for gyro mode but the pointing stability is better since the drift will occur in roll only.

• Parallel observations: Observation logs are produced for both primary and secondary instruments during parallel observations. The pointing of the secondary aperture will experience a small drift as the telescope moves in its orbit because of differential velocity aberration. The HST pointing control system computes a differential velocity aberration correction to keep the primary aperture stationary. This can produce over the course of an orbit, a variation in the positioning of any other point in the HST field of view of up to 70 milliarcseconds across the field of view. However, this is assuming worst case geometry, where the change in the differential velocity aberration over an orbit is at a maximum, and assuming the entire FOV's 30 arcmin diameter. In fact the most that current parallels can be separated by is about 10 arcmin (600 arcsec), the approximate difference between STIS and NICMOS. Here one would expect worst-case 23 milliarcseconds total variation for 1/2 orbit (48 minutes). This motion will be reflected in the parallel's jitter file.

• Internal observations: There are several types of internal observations, including WAVE,VISFLAT, DARK, BIAS, DEFCAL, INTFLATS. A header, CMJ and CMI tables are produced for these observations but the tables are empty. Note that external flats and EARTH calibrations are not considered internals.

• Slewing and jitter statistics: If slewing occurs during an observation (e.g., target acquisition, spatial scans, moving targets), the jitter statistics in the header and tables will be wrong because they reflect the relative motion of the telescope from the first valid point in the observation. For ease of identification of slews, a "SLEW" flag is included in the tables. Also, a special slew keyword is inserted at the end of the header.

  ---