ACIS S3 Calibration Validity Checks

C. Grant, M. Raley - 10/20/03

ACIS response products are now time dependent with the release of the time-dependent gain correction for acis_process_events. This removes trends in the mean PHA as a function of time for the I0-I3, S1-S3 CCDs. The remaining CCDs (S0, S4, S5), are primarily used for grating analysis, where this trend is less important, so their time-dependence is not currently calibrated.

The new time-dependent products have a time resolution of three months. Any changes in the pulse height on time scales shorter than three months may not be completely corrected by the released products.

Users may now check for themselves the validity of the CXC ACIS response products for their particular observation by fitting data from the ACIS calibration source which has strong lines between 1.5 and 6.5 keV. If a user has reason to suspect that the gain calibration or energy scale is incorrect for their observation, for example if fits to spectral lines are shifted in energy, we suggest the user down load observations of the onboard calibration source, taken at times near that of the users observation in order to verify the ACIS calibration for their data.

  1. Find calibration source observations close to that of the observation using the list of calibration observations
  2. (updated 2009_03_02). This list is also available in rdb format. Determine the exact date of your observation:
    dmlist acis_obs_evt2.fits header | grep DATE-OBS
    where acis_obs_evt2.fits is the name of your reprocessed evt2 file, will return the date of your observation. In the case of obsid 3520, a value of:
    0012 DATE-OBS	2003-02-01T07:10:08	String		Date and time of observation start
    We have supplied a quick Perl script to convert the date of your observation to Day Of Year: dates.perl

    When run for this observation will produce:

    -> ../ 2003 02 01
    Entered date is: 02/01/2003
    Day of year 2003 is: 32
    Check observation list for: 2003:032
    Note that this task can also be accomplised using the CIAO tool prop_dates. A quick check of the list of available ECS observations yields:
    Time	OBSID	Chips	FP Temp	 Description
    2003:028:2210	60868	i0-i3,s2,s3	-121	faintmode(musicalfeps)  
    2003:031:1411	60865	s0-s5	-121	faintmode(musicalfeps)  
    2003:036:2313	60856	i0-i3,s2,s3	-121	faintmode(musicalfeps)   
    Accordingly obsid 60865 would be the best choice to compare our data against as it is one day difference in time.

  3. Down load these data from the provisional web interface

  4. Analyze the data using the same extraction region and RMF as the target.
  5. Fit Calibration source line energies and widths: Mn K-alpha (5898 eV), Ti K-alpha (4510 eV), and Al K (1487 eV)
  6. For obsid 60865, the fits for the ECS to the strong lines can be seen:

    Al K (1487 eV)
    Ti K-alpha (4510 eV)
    Mn K-alpha (5898 eV)

    A single Gaussian was used in each range for the lines, the normalization and line energies were allowed to float and the line widths were kept to zero. Each fit returned the exact energy for the line. As is evidenced in the plots, the lines fit these data nicely and we can conclude that with the use of TGAIN, the standard ACIS calibration files provide a good fit.

Caveats: the RMF's used in these examples were created from the FEF (version N0004) which is soon to be released by CALDB. If these are not used, errors of up to 1.3% can be experienced in fitted line energies.