[Date Prev][Date Next][Thread Prev][Thread Next][Search] [Main Index] [Thread Index] [HEASARC Archives]

A GIS-3 gain problem

ASCA Guest Observers,

The GIS team has announced a recent GIS3 gain problem which may
affect analysis of GIS3 data taken after August 1995.
Please find the message from the GIS team below, and if you have 
questions, please send an e-mail to ascahelp@athena.gsfc.nasa.gov.




 NOTE:	GIS-S3 gain mis-correction

					by the GIS team (M. Tashiro)

Our calibration with iron lines from the SNR Cas A reveals that the 
absolute gain of GIS-S3 is apparently overestimated by 1 - 2 % since 
around August of 1995. The deviation is gradually increasing, and 
iron-K line center energy exceeds those obtained by GIS-S2 and SISs 
by about 1 % since the second half of 1995. There is no problem with 

This gain drift of GIS-S3 is not real, but likely to originate from 
the current gain correction algorithm. The GIS team investigated the 
phenomenon and found that the deviation is likely to occur in the 
process determining the long-term gain history with pulse-height peak
of the calibration isotope (55Fe) attached to the GIS-S3 sensor. It is 
suspected that the long-term gain decrease of GIS-S3 is slightly more 
pronounced at the fov rim than at the fov center, so that the gain 
correction factor calculated using the calibration isotope (attached at 
the fov rim) tends to be slightly overestimated.

We are working to calibrate the newly discovered effect, using not the 
GIS-S3 calibration source but the intrinsic copper-K line. When it is
established, we will release the function which will describe the long-term 
gain drift due to this effect and the corrected version of gis_temp2gain.fits 
in which the corrected absolute gain will be written.

Guest observers who found the GIS-S3 gain problem can take the following
methods to cope with the problem:
(1) Wait release of the new gis_temp2gain.fits, and run 'ascalin' on 
    GIS-S3 event files using this file. This will be the best way, but 
    not available soon.
(2) In the 'xspec' session, use 'gain' command to see the effect of the 
    gain change by changing the energy scale a posteriori.
    This is the easiest, and the gold-M edge feature at 2.2 keV,
    which originates in the XRT effective area, may be a fiducial mark
    to estimate  the correct gain.  The data of long-term gain drift 
    due to the effect, which will be released
    by the GIS team, will also be a guideline for this 'manual' correction.
    However, please be careful that this method is not precisely 
    correct, since spectral features produced  by the non-linearity  
    of the energy-channel relation (e.g. GIS xenon-L feature at 4.8 keV) 
    cannot be corrected in this manner.
(3) GOF is going to release a new ftool, 'gispi', with which users can 
    redetermine the PI values of the input GIS event file using a user-input
    gain value (which may be fixed, e.g., using the xspec gain command).
    Using the corrected event file, users can re-extract the spectral file
    having a correct energy scale.