This is probably due to the fid light beam (which is collimated by various optical elements in the Fiducial Transfer System) moving from one surface of the Retro-Reflector Collimator (a corner cube) to another when the SIM translates in Z. Because the surfaces of the RRC are not precisely 90 degrees to one another, a small angular shift in the fid light beam occurs. T
This ECR is to create and populate 4 new columns in the CALALIGN ARD which will correct for this effect. Based on the SIM-Z offset, a linear correction term is applied to the effective LSI positions of the fid lights.
simz_offset = fidprops.lsi0_stt[2] + fidprops.stt0_stf[2]; = LSI0STT3 + STT0STF3 (in the header keywords)The current offset data for fid 'i' can be characterized by a linear function:
dy[i] = P0_y[i] + P1_y[i] * SIM_Z_offset dz[i] = P0_z[i] + P1_z[i] * SIM_Z_offsetHere dy and dz are changes (millimeters) which are applied to the effective physical position of fid lights.
y[i] = y[i] + dy[i] z[i] = z[i] + dz[i]In order to allow for future calibrations which are more accurate, we are allowing for an offset described by a 4th order polynomial. In addition, a cut-off is introduced to prevent an unexpected extrapolations of the offset equation.
dy[i] = Sum(j=0,4) P_y[j,i] * SIM_Z_offset^j dz[i] = Sum(j=0,4) P_z[j,i] * SIM_Z_offset^j dy[i] = dy_max[i] if (dy[i] > dy_max[i]) dy[i] = dy_min[i] if (dy[i] < dy_min[i]) (and similarly for z)These new coefficients are added as new columns (5 to 8 in the list below) in the CALALIGN1 HDU of the CALALIGN files. Specifically:
ColNo Name Unit Type Range 1 FID_NUM Int4 - Fid number (1-14) 2 FID_SI String[9] Fid SI (ACIS-S, HRC-I etc.) 3 FID_NUM_SI Int4 - Fid number wrt SI (1-6 or 1-4) 4 FID_POS_LSI[3] mm Real8(3) -Inf:+Inf Fid position in LSI --> 5 fid_y_corr[5] Real8(5) -Inf:+Inf Polynomial coefficents for LSI correction --> 6 fid_z_corr[5] Real8(5) -Inf:+Inf Polynomial coefficents for LSI correction --> 7 fid_y_lim[2] Real8(2) -Inf:+Inf Offset limits for LSI correction --> 8 fid_z_lim[2] Real8(2) -Inf:+Inf Offset limits for LSI correctionThe new values of the polynomial coefficients and min/max values to be used are all 0.0, except for ACIS-I fids 2 and 4 and HRC-I-1, which have the values:
ACIS-I-2: P_y = [ 0.010059 -0.002576 0.0 0.0 0.0 ] dy_min = -0.05 dy_max = 0.05 ACIS-I-2: P_z = [-0.025355 0.006493 0.0 0.0 0.0 ] dz_min = -0.10 dz_max = 0.10 ACIS-I-4: P_y = [ 0.010825 -0.002772 0.0 0.0 0.0 ] dy_min = -0.05 dy_max = 0.05 ACIS-I-4: P_z = [-0.023910 0.006123 0.0 0.0 0.0 ] dz_min = -0.10 dz_max = 0.10 HRC-I-1: P_y = [ 0.011159 -0.002825 0.0 0.0 0.0 ] dy_min = -0.05 dy_max = 0.15 HRC-I-1: P_z = [-0.020947 0.005303 0.0 0.0 0.0 ] dz_min = -0.25 dz_max = 0.05These have been calculated with the program calc_fid_dist.pro, which uses the database of fid positions STATS.rdb. The CALALIGN files were generated with fix_calalign_files.pl, which is a Perl script that runs fparkey.
/home/aldcroft/Aspect/flt/fid_sim_z
Obsid Detector SIM_Z (arcsec) ------ ----------- ------------------ 1843 ACIS-I -173.3 1719 ACIS-I -17.9 1112 ACIS-I 57.5 1302 ACIS-I 78.1 (nominal) 531 ACIS-I 137.9 910 ACIS-I 228.1 1841 ACIS-I 426.4 1133 HRC-I -711.8 558 HRC-I -232.0 87 HRC-I -74.2 559 HRC-I 7.9 (nominal)
The improvement in the fid positions used in processing can be seen by
comparing the mean fid residuals DY and DZ calculated
by the Aspect V&V program. These values are the residuals when fitting
the constellation of predicted fid positions to the measured positions.
Plots for ACIS-I and HRC-I are shown below, with blue representing the
Release CALALIGN residuals, and red corresponding to residuals using the
new files.
The HRC-I data show that the corrected fid positions are not as accurate as for ACIS-I. This is because the HRC-I data are not well fit by a line, but there are not enough data points to fit a higher order polynomial. In any case, at large SIM-Z the HRC-I residuals are dramatically reduced.
In addition to checking the fid residuals, the shift in absolute aspect was checked. The aspect solutions (Release and new) for Obsids 1843, 1719, 1302, and 1133 were compared. The show that for the nominal SIM-Z value, no shift in absolute aspect is found. At off-nominal values, shifts are present which improve the absolute aspect accuracy.
More details are available in the analysis directory in
NOTES.verify.