38 ACIS OBS_IDs with "Problem Reading TGAINFILE"

NOTICE: The problem below has been fixed completely with the release of CalDB 4.5.0, released on 27 June 2012. Chandra Repro 4 will reprocess all of the OBS_IDs below correctly as it proceeds now. Users who have interest in these OBS_IDs may first upgrade their CalDB to version 4.5.0 immediately, and then proceed with their own reprocessing with the script chandra_repro without concern about the CalDB performance. Their analysis may proceed normally.

There are 38 Observation Ids where manual reprocessing with chandra_repro fails to include the ACIS time-dependent gain (TGAIN) correction. Current archival reprocessing (Repro 4) is correcting for this error, but has not completed reprocessing on all of these Observation IDs. This page is only of interest to people analyzing any of the 38 Obs Ids listed in the ObsId list below. If you are using one of these datasets then below we show you:

• how to check whether you have been affected,
• how to correct the data,
• the ObsId list,
• and explain the problem.

A fix to the CalDB coming in version 4.5.0, anticipated for release in May 2012, will eliminate this issue.

As always, please contact the CXC HelpDesk with any questions or for further clarification.

Have you been affected?

If your data is not one of the 38 Obs Ids listed below then you do not need to worry about this problem.

The TGAINCOR header keyword in the evt2 file records whether the time-dependent gain correction was applied by acis_process_events, so the following check can be made (here we compare the archival data for ObsId 3339 to a version reprocessed with CIAO 4.4 and CALDB 4.4.8):

unix% dmkeypar acisf03339N0002_evt2.fits.gz tgaincor echo+
T
unix% dmkeypar acisf03339_repro_evt2.fits tgaincor echo+
F

The archival data - which was reprocessed as part of Repro 3 - does include the time-dependent gain correction, since the keyword is set to T (the FITS convention for True). The reprocessing via chandra_repro has not included the correction, and so the keyword is set to F (short for False).

So, if the dmkeypar call returns T then the data has had the time-dependent gain correction applied to it and you do not need to read the rest of this document. However, please note that if you intend to re-process the dataset then the following section needs to be followed.

The above check assumes that you did not explicitly turn off the time-dependent correction by setting apply_tgain=no in the acis_process_events call.

How to reprocess the data correctly

Follow the Reprocessing Data to Create a New Level=2 Event File thread, but explicitly set the tgainfile parameter, as shown below, before running acis_process_events. Please contact the CXC HelpDesk with any questions or for further clarification.

Notes:

• if you are analysing Obs Id 2532 then - as discussed below - you also need to set the ctifile parameter;
• the chandra_repro script can not be used to process these observations since it calls punlearn acis_process_events, and so uses the CALDB setting for the tgainfile parameter.

Using ObsId 3339 as an example, we find from the ObsId list that the correct tgainfile for the observation is acisD2002-08-01t_gainN0006.fits, so we set

unix% pset acis_process_events \
    tgainfile=${CALDB}/data/chandra/acis/t_gain/acisD2002-08-01t_gainN0006.fits

before running acis_process_events.

# Obs Id 2532 only
unix% pset acis_process_events \
    ctifile=${CALDB}/data/chandra/acis/cti/acisD2000-01-29ctiN0007.fits

If you do not override the tgainfile parameter setting then you will see the following warning from acis_process_events:

# acis_process_events (CIAO 4.4): WARNING: problem reading tgainfile, tgain adjustment will not be applied. Changing apply_tgain=yes to apply_tgain=no.

Obs Id list

The following table lists those ACIS Obs Ids whose observations span T_GAIN epochal time boundaries. The T_GAIN column gives the tgain file name that should be explictly set for the tgainfile parameter of acis_process_events, as shown in the How to reprocess the data correctly section above.

What happened

The ACIS TGAIN calibration is done in such a way that the calibration files cover 3 month periods ("epochs"), so there is the possibility that observations will take place that span the end of one epoch and the start of the next one. The 38 Obs Ids listed in the table above are all the ACIS observations that have occured during such "epoch breaks".

Unfortunately, since the December 2008 release of CIAO 4.1 and CALDB 4.1, the CALDB query to find such files will fail, resulting in the acis_process_events warning

WARNING: problem reading tgainfile, tgain adjustment will not be applied. Changing apply_tgain=yes to apply_tgain=no.

and the data will be processed without the tgain correction. As noted at the top of the page, this problem has been fixed in CalDB 4.5.0.