CalDB 126.96.36.199 Public Release Notes
Public Release Date: 13 SEP 2019
SDP Installation Date: 2019-09-06T17:31:43 (UTC)
CalDB 188.8.131.52 is an upgrade to the Chandra CalDB, which includes the following items:
For the CIAO 4.11 / CalDB 184.108.40.206 release notes see How CalDB 220.127.116.11 Affects Your Analysis.
II. SUMMARY OF CHANGES
The ACIS calibration team has reported an undetected error in the calibration products for the combined T_GAIN epochs 75+76, and 77+78, covering the bi-epochal periods of Aug 2018 - Jan 2019, and Feb - July 2019. The error occurred due to a single line of code in the recently modified file generator of the calibration interface product files used by the CalDB Manager to generate the ACIS T_GAIN files employed in the CalDB. Only ACIS chips 0, 1, 2, 3 and 6 have been affected. This error produced alternating columns of zero t_gain correction values in the DELTAPHA column of the CIPs, which were then transfered to the CalDB T_GAIN files in their DELTPHA2 and DELTPHA1 columns. The errant files, affected columns, Standard Data Processing (SDP) installation dates, and corresponding corrected files are listed in Table 1 below.
The effect of the T_GAIN errors is a 1%-2% reduction in the PHA, and hence the ENERGY column values for ordinary dithered obervations on those particular chips only.
To determine whether your specific data sets have been affected by these errant files, please check the following conditions. Your data will have been affected only if all of the following apply:
- You have ACIS data including chips ACIS-I0, I1, I2, I3 or S2 (i.e. ACIS-0, 1, 2, 3, or 6 in keyword DETNAM of evt2 files).
- Your data have DATE-OBS later than 2018-05-02T12:00:00.
- Your evt2 files have the header keyword TGAINFIL set to any of the eight filenames in Column 1 "Errant T_GAIN File" of Table 1 below. The text must match of the filenames exactly, and the version number is important!
Notice also that the corrected file for your case will be given in the corresponding row of the final Table 1 column, "Corrected T_GAIN file".
Table 1: Errant ACIS T_GAIN filenames, affected columns, installation dates, and the corresponding corrected files.
|Errant T_GAIN File
|FITS Columns Affected
by the Error
SDP Installation Date
|Corrected T_GAIN File
These errant CalDB files were then released to the public, the first four files on 23 May 2019 with CalDB 4.8.3, and the latter four files on 09 Sep 2019 with CalDB 4.8.4. Hence, any reprocessing by users (e.g. with chandra_repro) with either CalDB would have affected any ACIS observations employing ACIS-0,1,2,3,4, or 6 with a DATE-OBS of 2018-05-02T12:00:00 or any later date.
The error was detected by 05 September, and corrected in processing on 2019-09-06T17:31:43 (UTC). Any data sets processed in SDP after that date are unaffected by the error. (See the DATE keyword in observation data file headers.) Details describing the exact nature of the error are given in the technical details section below.
The CalDB Manager and DS ARCOPS have mutually determined that 212 non-calibration OBS_IDs have been affected in processing by the errors in T_GAIN, including 125 ACIS-I observations, and 88 ACIS-S observations, which had any of ACIS-I0, I1, I2, I3, or S2 chips active. Among these are 18 observations which are on the public archive. All 212 of these OBS_IDs will be reprocessed in Special AP using the correctioned T_GAIN files released in CalDB 18.104.22.168. Users and PIs will be notified in a timely fashion of these errors, with appropriate instructions to make reparations to their data.
ACIS Level 1 pipeline tool acis_process_events.
III. TECHNICAL DETAILS
The error resulted from the change in tiling from 32x32 pixels to 64x64 pixels, which was first implemented in May 2019. The tiling change was approved because of the decreasing count rate from the ACIS External Calibration Source (ECS) with time, to improve the statistics in the ECS analysis. The ACIS team had elected to maintain the same tiling format in the corrgain files as before (32x32), but simply by filling in the extra three 32x32 tiles that would compose each 64x64 tile with the same data. The coding error filled in only one of these three tiles in such a way that every other 32x32 pixel column of the corrgain's DELTAPHA array was left with only 0.000000 values.
The uncorrected corrgain data in Fig. 1a shows the uncorrected data with the vertical stripes of zero T_GAIN correction values. Fig 1b shows how the data should have appeared, from the corrected corrgain file.
In Figures 2a, b, and c below, the error is illustrated using the actual CalDB T_GAIN files released in CalDB 4.8.3 and 4.8.4. The aimpoint position plot in Fig 2.b shows a normal variation between successive epochs for a T_GAIN correction on I3. This is because the nominal aimpoint of (chipx,chipy)=(570,575) happens not to be one of the columns with zeros in it. On the other hand, only 32 x-pixels either to the left or to the right (Figs 2a, 2c) show cases where the points of interest have columns are in all-zero columns, either for only DELTPHA2 (the green curves) or for both DELPHA1 and DELTPHA2 (blue and olive curves, now showing identically 0.0000 corrections on the plots.)
In Figures 3a, b, and c, the corrected CalDB files (each tagged with a "_revA" in their file names) have been used to make the same plots as in Fig 2. The results are obviously much better in Figs 3a and c, as the zeroes have been replaced by the real DELTAPHA data from the ECS analysis.