Introduction and Summary

As a consequence of a transponder failure, the GO data for RXTE observations performed since Sept 1st 1999 may include obsID directory names marked with a G, C, or D. A G-designation indicates on-source data that may suffer a gap in science, housekeeping or attitude information of more than 10% of the total predicted good time. C- and D-designations indicate a slew to or from a source during which attitude information is not available, and thus the exact limits of the slew are undefined. Note that only a very small fraction of the data taken with RXTE currently suffers from such problems.

The tools in our existing RXTE software package will process these data transparently. The definition of Good Time Intervals (GTI, included as an extension to each FITS file) has been amended to only include intervals where attitude information is available. Under certain circumstances, data outside these GTIs may be useful for analysis.

Typical examples of data loss include the following: (i) satellite attitude/position data may be missing, but the Std2 or user-selected EDS mode data may still be usable for timing analysis. (In this case, you may wish to get the most out of their data by disregarding the given good time intervals, generating your own GTIs with less stringent criteria, and reanalyzing portions of your data, accepting the possibility that these data may have a less-reliable background subtraction); (ii) Std2 data may be missing but the user-selected EDS-mode data may contribute to your science goals; (iii) some HEXTE data may be unavailable but the PCA data may be complete (or vice versa).

This recipe describes the categories of data loss, the changes we have made to the directory naming scheme to indicate that data loss may have occurred, the compensatory changes to the GTI definition, and the possible effects of all this on your RXTE data analysis. Familiarity with the RXTE directory structure and 'conventional' RXTE data analysis is a prerequisite for understanding the text that follows.

In addition, RXTE data obtained during the interval 2000 September 6-13 may include obsID directory names marked with a F, T, or U. These designations indicate data that include errors in attitude determination. (Note the important difference between missing attitude data, as above, and attitude data that is present but incorrect.) An Addendum to this recipe describes the problem and the additional data categories in more detail.

The Transponder Failure

On Sept 1st 1999, one of the two transponders on board RXTE suffered a failure. As a result, only one of the two high gain antennas (HGA) carried by RXTE can currently be used. Operational changes in the SOF have succeeded in minimizing the effect of this transponder failure on data capture. After a two-week adjustment period, the incidence of data gaps has fallen dramatically, and many days may currently elapse without a data loss of any kind. However, if you intend to analyze data obtained after Sept 1st, 1999, you should be aware of the information contained in this recipe.

Categories of Data Gap

Because of the way data are stored in RXTE's virtual recorders, a data gap will generally affect certain types of RXTE data, leaving other types intact. So far we've seen three main classes of data gap (some of which can occur concurrently):

• ATTITUDE/HK/HEXTE ARCHIVE DATA MISSING; STANDARD2 AND OTHER SCIENCE DATA INTACT.
  • Specifically, the following ApIDs are missing for all or part of the observation: 0-47, 82-85, 88-233, 251-254. These cover the attitude control system (ACS) data (including pointing position, orientation, and satellite position), the PCA and HEXTE housekeeping data, and HEXTE Archive data (the HEXTE Science Mode data will still be present). Lacking these data, you won't be able to determine pointing position, satellite orientation or positional information information and your ability to create a PCA background will be limited.

• PCA AND/OR HEXTE USER-SELECTED DATA MODES MISSING; OTHER DATA INTACT
  • In this case, the user-selected PCA/EDS modes may be missing for all or part of the observation, including GoodXenon, binned or event modes, or the user-selected HEXTE Science Mode data. PCA Standard modes, HEXTE Archive modes, and attitude/HK data will still be present. Such a gap affects the following ApIDs: [55,59,63,79] and/or [80,86].

• PCA STD2 AND STD1 DATA MISSING; OTHER DATA INTACT

  • ApIDs 70 and 74 are missing in this case, though most science objectives may be achievable with the user-selected data modes.

The GTIs supplied in the 2nd extension of your RXTE data files will include ONLY those times when attitude information is present (ApID 15), and RXTE is not in SAA or Earth occult. These criteria are then folded in along with the availability of the particular data you're analyzing - for example, the GTIs supplied with the Good Xenon data will specify those times when Good Xenon data are available AND the attitude criteria are met.

As a consequence, if you extract light curves and spectra using the given GTIs - i.e. using the default APPLY option at the 'gtiorfile' prompt when using the RXTE SAEXTRCT or SEEXTRCT tools, times when ACS or science data are missing will automatically be excluded from your data products. The data in Filter Files made using XTEFILT will be well-behaved in these regions, and you can apply your usual data cleaning criteria.

New Naming Scheme for ObsIDs Affected by Gaps

Source Data

Data gaps are detected during the processing that creates the FITS files. If an on-source observation suffers data gaps in any ApIDs commonly used in data analysis (i.e. attitude data, PCA and HEXTE housekeeping, all PCA Standard and Science modes, and HEXTE Archive and Science modes), totalling more than 10% of the expected good time for the observation, that obsID is marked with a "G"; i.e. the obsID for these data will become 40066-01-01-00G, rather than 40066-01-01-00. ObsIDs with a "G"-designation have some data missing, although this may or may not affect your analysis goals. ObsIDs without a "G"-designation can be assumed to be 90% complete or better.

For long observations, where the data are split into more than one obsID and data gaps total more than 10%, the letters "H" and "I" are reserved for the continuation. Thus, 40066-01-01-00G may be followed by obsIDs 40066-01-01-00H and 40066-01-01-00I. Because of the logistics of scheduling, this will happen rarely.

Slew Data

Usually we mark a slew onto a source with an "A" (e.g. 40066-01-01-00A), and the slew off-source with a "Z" (hence, 40066-01-01-00Z). However, if ACS data are missing, the ground software cannot directly determine the start and end of a slew maneuver.

• If no ACS data are present when a maneuver to a source is expected, based on the predicted timeline, the software adds a letter "C" to the obsID -- e.g. 40066-01-01-00C. "C"-obsIDs thus flag the 'grey zone' where attitude information is missing and it is unclear whether the slew has terminated. A "C"-obsID can replace the "A"-obsID, or occur in between an "A"-slew and the source observation. In some extreme cases, most or all of the available data from an on-source pointing may end up being included in the "C"-obsID if an attitude gap extends that far. (For this reason, we maintain the one-year proprietary period for "C" and "D"-obsIDs, rather than making them public immediately like the "A/Z" obsIDs.) Also, sometimes such a "C"-obsID may be followed by an 'unlettered' obsID which itself contains no gaps, but which represents only a small fraction of the expected data.

• If no ACS data are present when a maneuver away from a source is expected, based on the predicted timeline, the software adds a letter "D" to the obsID (e.g. 40066-01-01-00D). "D"-obsIDs will occur between a source observation and the "Z"-slew (or replace the "Z"-obsID).

The Effects On Your Analysis

Addendum: Attitude Errors, 2000 Sept 6-13

Between 2000 Sept 6 16:00 - Sept 13 15:40 (UT), the RXTE satellite suffered errors in attitude determination. Corrective action was taken and the problem is believed solved. However, RXTE PIs and archival users should note that data obtained during this interval may be subject to attitude errors of up to 1 degree which will not be reflected in the attitude file. Quantities such as the pointing position, Earth limb angle, offset angle etc, are very likely to be incorrect. Effects on data analysis may include an apparent reduction in the flux measured by the PCA and HEXTE instruments of up to a factor of 5, and poor spectral fitting results. Please be EXTREMELY cautious about drawing scientific conclusions from the affected observations.

PIs of affected observations have been informed directly. Note that the data from public TOOs performed during this period have been released to the community, despite the uncertainty in attitude.

A team working at the Flight Dynamics Facility will attempt to reconstruct the attitude of the satellite during this interval. As yet, we do not know how long this may take or how successful it will be.

Affected obsIDs have been tagged to indicate the unreliability of the attitude data. The new data categories are:

• "F" obsids - e.g. 50117-01-01F, for pointed source data;

• "T" obsids - e.g. 50117-01-01T, for the slew onto the source;

• "U" obsids - e.g. 50117-01-01U, for the slew away from the source;

• For long observations where the data are split into more than one obsID, the characters "V" and "W" are reserved for the continuation obsIDs. Thus, if it had extended long enough, obsID 50117-01-01F would have been followed by 50117-01-01V and 50117-01-01W.

'Failed' Observations, and Re-observations

No official policy or procedure is yet in place for automatically scheduling re-observations of 'failed' observations, where the data gaps render the scientific aims of the proposal unattainable. (In many cases, it may be a complex judgment call whether the aims of the observation can be achieved.)