CalDB Release Notes
Public Release Date: 07 November 2006
SDP + PUBLIC
includes only the addition of six new ACIS time-dependent gain
correction files (T_GAIN), upgrading T_GAIN Epochs 21 through 26 (i.e.,
Feb 2005 though April 2006). These upgrades are being introduced for
(1) To address a discontinuity in the I0 and I2 gains versus time which
occured at approximately 01 August 2005.
(2) To introduce new time-dependent gain corrections for Epochs 25 (Nov
'05-Jan '06) and 26 (Feb '06-April '06).
SUMMARY OF CHANGES
A. ACIS T_GAIN
updates (Epochs 21-26)
acisD2005-02-01t_gainN0005_revA.fits (EPOCHS 21 and
(EPOCHS 21 and 22 repeated)
(EPOCHS 23 and 24)
(EPOCHS 24 and 25)
(EPOCHS 25 and 26)
(flat file, current in processing) (EPOCH 26 only)
DS 7.6.9 ACIS
L1 pipeline and CIAO 126.96.36.199 tool acis_process_events,
"tgainfile", L1 Event File Keyword "TGAINFIL".
"Create a New
Level=2 Event File"
Time-Dependent ACIS Gain Correction"
Please refer to
the web page "How CalDB 3.2.4 Affects Your Analysis"
See also the
ACIS Time-dependent Gain "why" page:
A. ACIS T_GAIN updates
upgraded T_GAIN correction data for the Epoch 22, 23, and 24,
principally due to the I0 and I2 gain discontinuity seen at
approximately 01 August 2005, and the prevailing high particle
background, which over time has reversed the trend of the
time-dependent gains on other chips (notably ACIS-I3!). Furthermore we
have reduced ECS data and calibrated the T_GAIN corrections for Epochs
25 (ECS data for 01 Feb 2006-30 April 2006) and Epoch 26 (01 May
2006-31 Jul 2006). This results in the production of five new T_GAIN
files, as listed in table 1, along with the input files, EPOCH#
keywords, and Calibration Validity Start Dates (CVSD* keyword set).
Table 1: New T_GAIN files and components, with the files
being replaced. The EPOCH1 and EPOCH2 values given
are for the NEW files.
(UTC date in sec)
(UTC date in sec)
| (new epoch)
| (new epoch)
value is the midpoint between the CVSD value of the given file and the
CVSD of the NEXT file in the series. The EPOCH2 value is the midpoint
(numerical average) of the CVSD values from the next TWO files in the
sequence. For the last file in the table, EPOCH2 is the midpoint time
between 2006-08-01T00:00:00 and 2006-11-01T00:00:00.
previous definition implies that a change in the CVSD of a given file
will affect the EPOCH1 value of the previous T_GAIN file. Similarly,
such a change also affects the EPOCH2 value of the file two epochs in
advance of the given file. The CVSD of the August 2005 file was changed
from 2005-08-01T00:00:00 to 2005-08-02T00:00:00. Hence there is a need
to update EPOCH1 for the 2005-05-01 file, and the EPOCH2 value in the
tagged as version N0002 above have the same data in DELTPHA2 as in
DELTPHA1, and hence the file interpolates over two identical data
values, and so the result is an identical value. Hence such a file is
known as a "flat file", preventing two-point trends from being
extrapolated too far into the future. When EPOCH 27 (August-October
2006) is released, the N0002 file will be upgraded to an N0005 file,
the permanent, interpolating T_GAIN file for Epoch 26.
DISCONTINUITY ON I0 and I2, ~01 Aug 2005:
The ACIS gain @
-120C page, at http://space.mit.edu/~cgrant/gain/ includes the
following illustrations of the raw time-dependence in the gain of the
several ACIS chips. Clearly illustrated are gain discontinuities
located near 01-Aug-2005 which are problematic for the above CalDB
T_GAIN file definitions. In short, the system should not attempt to
interpolate the "corrgain" data across these discontinuities for these
1: Average gain values for the ten ACIS chips while operating at -120C.
These are measured from the ECS line source data, however they only
include the bottom twenty rows of the chips, to avoid any CTI effects.
to be made in the T_GAIN interpolation for I0 and I2 must not at the
same time destroy the accuracy for the other eight chips across 01 Aug
2005. The file affected by this discontinuity is
acisD2005-05-01t_gainN0005.fits, released in CalDB
ACIS Calibration team has decided to use the previous datasets released
in the file acisD2005-02-01t_gainN0005.fits to substitute for the data
in the former file; however both files need to change the EPOCH2 data
to adjust for the new end date of 02 Aug 2005. Hence the two files
designated *N0005_revA.fits are included in this release.
Trends in I3 and S3:
plots have been produced by A. Vikhlinin of the ACIS M&A team:
Fig. 2: I3 T_GAIN
adjustments as derived from ECS data for the three X-ray lines
illustrated. These data are taken MID-CHIP, and so include the
radiation level effects on the electron traps which cause the CTI
S3 T_GAIN adjustments as derived from ECS data for the three X-ray
lines illustrated. These are also near the mid-chip region.
The I0 and I2
discontinuities occur between Epochs 22 (May-July 2005) and 23 (Aug-Oct
2005). However, the net T_GAIN correction for I3 in particular,
including the particle background, as derived from the ECS data, shows
that the correction for I3, while not discontinuous between Epochs
22-23, are also not monotonic in time. Hence there will be T_GAIN
corrections versus Epoch that will reverse prevailing trends in
previous data sets, such as for I3. S3 shows some similar behaviors
over Epochs 15-17.
level changes the CTI gain correction by fully or partially filling the
numerous e- traps between the selected detector row and the readout at
the bottom of the chip.
Calibration Source) data have been used as follows to verify
improvement in the I0 T-GAIN corrections using the CIAO 188.8.131.52
software, currently released. Here is a summary J. DePasquale's results:
~ 5.895 keV
TABLE 2: Testing
results for new T_GAIN files to be released.
(out of spec)
behavior is as <we> would expect for OBSID 59536 with the N0005
and N0005_revA Tgain files. The 2005-05-01 N0005_revA (new) Tgain
file produces a peak which is justbarely out of our spec (but very
close) and the N0005 (old) file produces a peak which is significantly
lower and out of spec. This is consistent with <our>
expectation that the N0005 file interpolates over the gain
discontinuity and the N0005E file does not interpolate over the gain
 The results
for OBSID 59365 and 59243 are within spec.
 We get the
same results for OBSIDs 59365 and 59243 when we used the new gain