The following changes/fixes are needed to allow ground aspect processing when Chandra PCAD is configured to use IRU2 as the primary IRU:

• Update CALDB IRU characteristics file:
  
  Change values of config2_chan column from [4 5 6 7] to [5 4 7 6]. This is needed to create the correct mapping between physical gyro axes (spec'd in the corresponding values of gyro_sc_align) and gyro count telemetry MSIDs AOGYRCT<1-4>. An example modified file is:
  
  /proj/sot/ska/analysis/IRU_swap/iruD1999-08-12charN0001.fits
  
  
• Update standard aspect processing PED file (asp_l1_std.ped):
  
  Set bias_source=1 in param list for asp_forward_kalman This has the effect of initializing the gyro bias from PCAD engineering telemetry.
  
  
• Investigate and correct the 2-second transient at beginning of aspect solution when bias_source=1 is used.

Testing Setup

Test results and data are in the directory below, and any files or paths are specified relative to this directory:

/proj/sot/ska/analysis/IRU_swap/obs3921

In order to test the changes, I used flight data from obsid 3921, which was observed while IRU2 was powered and counts were being reported in PCAD subformat telemetry (MSIDS ASPBGYC0A, ASPBGYC1A, ASPBGYC2A, and ASPBGYC3A). I took those raw count values which are reported once each two seconds and interpolated them to the higher sampling (each 0.256 sec) of the primary gyro count MSIDS (AOGYRCT[1-4]).

The IRU2 data were then written into the corresponding AOGYRCT[n] columns of the pcad*_3_* engineering telemetry for time span of the observation. In addition, the OBC estimate of gyro bias AOGBIAS[n] columns were adjusted accordingly, and the gyro config flag (AOGYRSEL) was changed to indicate configuration 2 ('CFG2'). The new files were created with iru2to1.pl and are in iru2/pcad0.

Finally, I created a local version of the CALDB IRU characteristics file iruD1999-08-12charN0001.fits in which the config2_chan values were set to [5 4 7 6]. This reflects the correct mapping of physical gyro channels and gyro count telemetry MSIDs AOGYRCT[1-4]. This IRU characteristics file was also used with standard IRU1 data reprocessing to verify backward compatibility.

Testing Results

When I ran the current standard aspect pipeline (Release 6.13.2) on this dataset, the asp_forward_kalman tool failed. The cause is the higher gyro bias level in IRU2, approximately 5 times that of IRU1 in magnitude. Initially the ground kalman filter assumes a zero bias level, which for IRU2 results in rejection of all star data because the kalman propagator gives a very poor prediction of the spacecraft attitude. Because initial star data get rejected, the bias estimate never gets updated, and all subsequent star data are similarly rejected.

The fix is to set bias_source=1 in the parameter list for the asp_forward_kalman tool. This causes the pipeline to initialize the gyro bias from the OBC estimate in engineering telemetry. (This had been disabled in an attempt to make the pipeline independent of OBC telemetry, but at this point the benefits of this strategy are minimal.)

After setting bias_source=1 for asp_forward_kalman (via a customized PED file asp_l1_iruswap.ped), processing ran normally. The final aspect solution matches that from IRU1 data (processed with the standard pipe) very closely, with one exception noted below. The estimates of aspect quality from the V&V tool are identical to within the noise.

One clear inconsistency is an initial transient spike in the aspect solution during the first 2 seconds. The spike has an amplitude of ~1 arcsec and appears only in the data processed with the asp_l1_iruswap.ped file. IRU1 data were reprocessed with asp_l1_iruswap and show the same transient. The sign of the transient is different in IRU1 vs. IRU2 data, although the magnitude is comparable. No such transient is found at the end of the processing interval.

I do not understand the cause of the initial transient, and it should be investigated by Roger Hain.

Other issues relevant to IRU characteristics CALDB file

• The new IRU characteristics CALDB file uses pre-launch values of the gyro alignments. As expected, this has no noticable effect on ground aspect during science pointings. Nevertheless, after the swap to IRU2 is complete and the operation and OFLS/FOT calibration values are verified, we will update the IRU characteristics with more accurate alignment / scale factor values.
  
  
• The IRU characteristics file has a column named config_curr which is nominally defined to give the current PCAD gyro configuration (0 or 1). To my knowledge this value is not used in any processing tools. Can we somehow declare the value to be obsolete?
  
  
• With the exception of the point above, the new file is valid for all dates since launch, so the file name should be the same except for N0002. The current config_chan2 values are/have been incorrect at all times in the mission.

V and V Aspect Quality

    IRU1:
    STATISTICS:
      slot         id    mag   n_pts  med_dy  med_dz    dr68    dr95
         0   ACIS-S-1   7.07     732  -0.010   0.041   0.009   0.015
         1   ACIS-S-4   7.09     731   0.016  -0.019   0.007   0.012
         2   ACIS-S-5   7.13     732  -0.032  -0.009   0.007   0.011
         3  816721800   9.56    1464  -0.032  -0.059   0.111   0.175
         4  816725616   9.40    1464   0.058  -0.340   0.184   0.294
         5  892732184   8.20    1464  -0.363  -0.175   0.082   0.128
         6  892734656   9.35    1463  -0.219   0.105   0.089   0.141
         7  892755368   9.83    1460   0.569   0.484   0.116   0.188

    IRU2:
    STATISTICS:
      slot         id    mag   n_pts  med_dy  med_dz    dr68    dr95
         0   ACIS-S-1   7.07     732  -0.010   0.041   0.009   0.015
         1   ACIS-S-4   7.09     731   0.016  -0.019   0.007   0.012
         2   ACIS-S-5   7.13     732  -0.032  -0.009   0.007   0.011
         3  816721800   9.56    1464  -0.031  -0.061   0.113   0.176
         4  816725616   9.40    1464   0.057  -0.340   0.184   0.293
         5  892732184   8.20    1464  -0.363  -0.175   0.082   0.130
         6  892734656   9.35    1463  -0.219   0.105   0.088   0.143
         7  892755368   9.83    1460   0.568   0.484   0.117   0.187