ECR: Update CALALIGN align matrices (2002-Apr)


This ECR is to update the values of the CALALIGN columns aca_misalign and fts_misalign in order to improve absolute astrometric accuracy. The new alignment values are based on re-analysis of a carefully selected sample of observations spanning the entire mission. Past alignment updates were done via piecewise linear fits to data for each detector. The current update, which is valid from 1999-10-07 to 2002-12-01, uses a global fit of a single decaying exponential (per axis) to offsets for all 4 detectors at once. The fit is very good, indicating a common underlying mechanism for the alignment shift. The best candidate is a change in the FTS periscope alignment, although the magnitude of the shift is larger than expected.

Determining alignment drift

The first step in determining the new alignment values was to reprocess the aspect for large set of observations spanning the duration of the mission. The new aspect was applied to the event file with reproject_events, and new sources positions derived with celldetect. Reprocessing removes any variations that still exist in the archive, e.g. fid positions changes. All observations were reprocessed with the same CALALIGN file from 1999-08-18. In this way the relative change since launch is clearly seen.

Only observations with a bright point source identified with a Tycho optical source or an ICRS (internation celestial reference system?) radio source were taken. In addition, to minimize residuals due to fid drift at off-nominal SIM-Z, observations were chosen with small or zero SIM-Z offsets. A total of 70 observations were selected and reprocessed.

The results of reprocessing are given in the plots below, which as usual show the offset of observed source position from the known celestial location. The offset is relative to Chandra body coordinates, with DY in red and DZ in blue. The solid line is a fit to the data with a exponential decay curve. The fit equation and best-fit parameters are:

  dy = exp_off_y[i_det] + exp_amp_y * exp(exp_coeff * day_of_mission)
  dz = exp_off_z[i_det] + exp_amp_z * exp(exp_coeff * day_of_mission)

                ACIS-S    ACIS-I    HRC-S     HRC-I
  exp_off_y = [-2.81803, -3.2013, -2.62313, -1.95124 ]
  exp_off_z = [ 1.91393,  1.8397,  1.83264,  1.44095 ]

  exp_coeff =  -1.0 / 357.741
  exp_amp_y = 2.75
  exp_amp_z = -1.19

As you can see, the fit is quite good except for the very early ACIS observations, which were excluded from the fitting. This discrepancy is probably because of the rapid outgassing of the ACIS mount points. As a result, the new alignment values are only applied to CALALIGN files starting from 1999-10-07.

The raw data used for the above plot is available here. The best fit values were determined using Sherpa and this script. The original analysis directory is


Creation of new CALALIGN files

New CALALIGN files were created by applying the fitted exponential offsets for DY and DZ to the baseline alignment matrices in the 1999-08-18 CALALIGN file. The IDL script was used to do the processing. The files with dates between 1999-10-07 and 2001-06-01 (inclusive) were updated, and new files (at monthly intervals) up through 2002-12-01 were created. Note that this involves extrapolating the decaying exponential, which if the current trend continues will keep future aspect offsets to a minimum.


A total of 67 of obsids were reprocessed through Aspect L1 using the Release build of the aspect pipeline. In each case, up to the first 15000 seconds of the observation was reprocessed using both the new CALALIGN files as installed by the CALDB manager into the test_CP area.

The resulting offsets are shown in the plot below. (There is no difference between the lightly shaded and darkly shaded plot symbols).

The deviations are small, although there appears to be a very slight systematic sinusoidal residual in the ACIS-S data. The raw data are here. The original analysis directory is


Aspect Information main page

Comments or questions: Aspect Help

Last modified:12/27/13