A warm pixel in the ACA CCD is one of the major reasons for a less than acceptable aspect solution as identified in V&V. A warm pixel located in the centroiding box for a guide star will introduce periodic fluctuations into the computed mean position of the guide star. The effect manifests in processing as the centroid is 'pulled' from the true position of the star. Degradation in centroiding results in large residuals for the affected slot. Large enough residuals can introduce aspect-originated degradation in image resolution of x-ray data. Observations that suffer from warm pixel contamination of a guide star slot are readily reprocessed with removal of the offending slot.
Analysis
The differences in the aspect solution between processing versions with unacceptable residuals for a guide star slot and the corresponding reprocessed versions with the offending slot removed from the aspect solution were examined for 22 recent observations. The root sum of squares for the rms in y and z for the offending slot (BAD SLOT RSS-RMS), used as a measure of the degradation suffered in the aspect solution, were compared to the change in the root sum of squares of the rms in y and z for the overall aspect solution (ASPECT RSS-RMS), used as a measure of the differences in the quality of the aspect solutions between the two versions. The relevant equations summarized here:Dy_i, dz_i = star centroiding residual for measurement 'i' after applying aspect solution
BAD SLOT RSS-RMS = sqrt( Sum_i (dy_i^2 + dz_i^2)/N )
dPitch_i, dYaw_i, dRoll_i = relative difference in attitude of aspect solution processed WITH bad star relative to aspect solution processed WITHOUT bad star
ASPECT RSS-RMS = sqrt( Sum_i (dPitch_i^2 + dYaw_i^2)/N )
The BAD SLOT RSS-RMS ranged from 0.1 to 0.8 arcsec. The ASPECT RSS-RMS ran ged from 0.015 to 0.200 arcsec. It was found that a 0.1 arcsec BAD SLOT RSS-RMS corresponds to approximately 0.025 ASPECT RSS-RMS on aver age (See plots below). This change in rms is taken as the image bluriness introduced by aspect.
Setting V&V Limits
Evaluation of the aspect solution in V&V is decomposed into y and z directions. Therefore, a reasonable limit is articulated in terms of y and z, rather than in terms of the BAD SLOT RSS-RMS of y and z. It was found that a limit of 0.15 arcsec rms in either y or z direction of a star slot identified obsids that lead to a change of approximately 0.05 arcsec in ASPECT RSS-RMS (See plot below). More impor tantly, this selection criteria did not allow observations with ASPECT RSS-RMS > 0.05 arcsec to pass as ``Good'', and the observations identi fied as having bad aspect solutions demonstrated improvement of > 0.04 arcsec.
A comparison among different selection criteria was performed with results from limits set to the rms in either y or z position at 0.1 arcsec and 0.2 arcsec limit for rms in either y or z (See plots below). Setting the ``good aspec'' limit to 0.2 arcsec would result in passing data that would not meet the 0.05 arcsec criteria. Setting the ``good aspect'' limit to 0.1 arcsec would result in the unnecessary reprocessing of data without significant improvement in aspect.
Summary
The improvement of aspect solution after reprocessing with the removal of a ``bad'' guide star slot was investigated. Different selection criteria for reprocessing were investigated, and it was found that a limit of 0.15 arcsec rms in either y or z for an individual star slot maximized the efficiency of selecting data to be reprocessed under the 0.05 arcsec rms for the overall aspect solution.John Bright
Last modified: 03/17/03