32.3 Location of Image on Diode Array

• At each position, the absolute photometric accuracy and the shape of the spectrum in the 1.0 and larger apertures.
• Target acquisition accuracies in the ACQ/BINARY and, in rare circumstances, ACQ/PEAK modes.

The amount of scatter in the mean Y-base location increased after on-board GIM correction started in April, 1993 as frequent DEPERM commanding (clearing of the ambient magnetic field in the detector) was turned off for all but ACQ/PEAK exposures. The uncertainties in the Y-bases affected both ACQ/BINARY pointing accuracy and FOS photometric accuracy. The size of these uncertainties required that, lest positional and photometric accuracy be compromised, all ACQ/BINARY acquisitions be followed with an ACQ/PEAK stage for those cases in which science was performed in an aperture smaller than 1.0. The photometric quality of the data, especially for FOS/RD and the 1.0 and 4.3 apertures, could be compromised due to the fluctuations in the location of the spectrum (see the discussion in "Image Centering (Image Location) Factors" on page 32-22).

Furthermore, the shapes of the spectra on the photocathode were not linear, but had a curvature of +/- 20 Y-base units (the so-called "s-curve" of the disperser). Figures 32.3 and 32.4 show the scale of the s-curvature with formal one measurement uncertainties for each FOS/BL and FOS/RD grating, respectively. Therefore, the photometric effect of Y-base uncertainty was not a simple matter of losing light in a uniform fashion, rather due to the s-curvature of the spectra, the effect was also wavelength (and disperser) dependent.

No systematic temperature-related Y-base changes were expected in the relatively small (10 degree C.) FOS on-orbit operating range and none were ever observed.

Commencing in mid-1993, FOS Y-bases were updated approximately every six months. The updates were made so as to include the anticipated linearly extrapolated trending over the ensuing six month period from the date of update. With a few notable exceptions (the June 1992 to December 1993 period for all FOS/BL dispersers and the January 1994 to January 1995 period for FOS/BL G130H as seen in Figures 32.1 and 32.2) Y-base updates resulted in worst-case 10 Y-base unit deviations from the final linear trend determined from all of the Y-base measurements. Absolute photometric accuracy and, especially, spectrum shape, could be affected for large aperture observations made with these dispersers in the time periods mentioned above. The dates of all FOS Y-base updates are included in Table 32.3.

FOS Y-base PDB Update Dates

Pre-COSTAR	Post-COSTAR
October 01, 1990	January 13, 1994
October 10, 1990	November 04, 1994
November 11, 1991	October 16, 1995
September 21, 1992	January 02, 1996 (FOS/BL G190H, G270H, G400H only)
August 11, 1993	June 15, 1996
December 17, 1993	December 18, 1996
	February 11, 1997

Summary of Y-base Uncertainties and Impacts

Repeated independent determinations of the same Y-base yield an external +/-25 Y-base unit (pre-COSTAR: ~0.14"; post-COSTAR: ~0.12") full-amplitude scatter for observations taken on the same day. This scatter is attributable to the effects of residual GIM (minimally) and, primarily, FGW non-repeatability uncertainties. The internal measurement error in any individual Y-base value is +/- 5 Y-base units (one ).

The photometric impact of an erroneous Y-base depends on the size of the error, the aperture involved, and the detector-disperser combination employed. On average, for post-COSTAR point sources, if we assume that a Y-base were known with an accuracy of only 20 Y-base units (approximately 1/13 diode height), for the large apertures (1.0") 3-5% of the light in general could be lost and up to 10% at certain wavelengths where the s-curvature is substantial. The light loss is larger, but harder to quantify, for extended objects and pre-COSTAR point sources (see FOS ISR 096).

There is essentially no photometric impact due to Y-base uncertainties for FOS apertures smaller than 1.0 as the displacements required to place the images of these apertures off the diode array are simply much larger than the observed uncertainties.

Y-base errors had no impact on measured positions in IMAGE mode observations made with the MIRROR. The brightness of objects within the field of view could be mis-represented, however.

For large aperture (1.0 and larger) observations, Y-base uncertainty is a prime contributor to photometric, especially spectral shape (color), uncertainties. Refer to Figures 32.3 and 32.4 to determine the spectral regions in which loss of signal is most likely to occur in your spectra. There is little or no photometric impact on apertures smaller than 1.0.

Figure 32.1: FOS/BL Y-bases as a Function of Time

Figure 32.2: FOS/RD Y-bases as a Function of Time

Figure 32.3: Normalized Spectrum Shape on Photocathode ("s-curve") : FOS/BL

The effect of the x-shift in the photoelectron impact point was to effectively shift the spectrum in the dispersion direction as a function of time. This displacement can be seen in data taken before April 5, 1993 by plotting the individual groups of raw data on a single plot (use the STSDAS grspec task) and noting the shift (in x) of the centroids of individual emission or absorption lines. In routine post-observation calibration calfos applies a correction for the GIM x-shift (as long as the OFF_CORR switch is set to "PERFORM") in the creation of the calibrated spectral data (.c0h and .c1h files).³

Pre-onboard GIM Correction

In post-observation processing there is no way to correct for the photometric effects of the shift in y introduced by GIM. The y-shift caused the s-curve of dispersed light to move along the long dimension of the diodes, and as with Y-base error, could produce a wavelength-dependent loss of light off the edge of the array. The resultant time-dependent error in the overall flux level and in the shape of the spectrum was most severe for poorly-centered observations and for the large apertures. The typical size of uncorrected GIM motion was 0.15" or about 25 Y-bases. For well-centered observations with no Y-base error and the 4.3 aperture the error was <5%. You can use the models of light loss as a function of Y-base offset, wavelength, disperser, and detector given in FOS ISR 096 to estimate the effect of GIM motion on your observations.

GIM Summary

• Post-COSTAR: +/-0.02" residual motion (peak-to-peak)
  • x-motion (peak-to-peak): 0.06 diode = 0.25 quarter-stepped pixel = 0.02" = ~15 km/sec.
  • y-motion (peak-to-peak): 4-5 Y-bases.
• Pre-COSTAR: +/-0.15" (peak-to-peak) before correction commenced April 5, 1993
  • x-motion: corrected to nearest pixel (hence, 0.125 diode = 0.5 quarter-stepped pixel = 0.044" = ~30 km/sec.
  • y-motion: up to 25 Y-bases not corrected.

² FOS ISRs 133 and 152.

³ Before May 1991, calfos did not correct for GIM. All data in the HST Archive were reprocessed and given new header keywords after the GIM correction was added to calfos. Therefore, if your FOS data were processed before May 1991, you should retrieve the updated raw data from the HST Archive instead of recalibrating the data on your tape.