FOS GIMP Overview

POA - FOS: Geomagnetic Image Motion Problem

ST-ECF

GIMP
The Onboard Fix
Related STSDAS Tasks
Zero Point Shifts
Correction to Blue Zero Point Shifts
Problems with the On Board Fix
The New POA_CALFOS Solution

The "GIMP"
"In orbit HST data taken with the red digicon detector of the FOS show an image drift correlated with the Earth's magnetic field. This drift occurs because the mu metal magnetic shield around the red digicon does not shield adequately. The amplitude of the drift for a typical orbit is roughly equal to the spectral resolution of the FOS. For data taken with readouts every minute or two, resampling and shifting can be used to remove most of the drift in the spectral direction. The drift will compromise photometric accuracy at high signal to noise ratios because the edge of the image of the aperture can drift beyond the diode array in the direction perpendicular to the dispersion. The microprocessor in the FOS can be re-programmed to counter the geomagnetic drift by changing the deflection currents on a time scale of one minute or less. In orbit data from the blue digicon show some image drift that does not correlate well with the magnetic field. The blue digicon mu-metal shield works better than the red side shield by a factor of at least 3." From FOS Instrument Science Report 066 Geomagnetic Image Deflection Problem in the Faint Object Spectrograph

See also:
ST-ECF Technical Reports relating to the GIMP.
FOS Instrument Science Report 082: Lab Test Results of the FOS Detector Performance in a Variable External Magnetic Field
Faint Object Spectrograph Instrument Status and Performance Changes: 1990-1993: Keyes
FOS Geomagnetic Image Motion Problem (or Why the Red Digicon is GIMPy and What to do about it)
Note that, some of these documents were drafts or proposals written around the time that the GIM problem was discovered, therefore some of the details may have been superseeded. However, they are presented here as they contain information which never found its way into more recent documents.

The Onboard Fix
The Space Telescope Science Institute and the FOS Investigation Definition Team, set about correcting this Geomagnetically-induced Image Motion Problem (GIMP) through a real-time on-board correction scheme. This correction required modifications to almost all aspects of the HST ground system as well as additional NSSC1 flight software and the use of an existing software 'hook' in the FOS microprocessor firmware. From FOS Instrument Science Report 098

See also:
The GIMP documents amongst the scanned documents repository.
Minutes from FOS FSW meeting 09/05/91
FOS GIMP Correction Requirements 23/05/91
FOS GIMP Correction Flow Proposal
Preliminary PASS Requirements for FOS GIMP Correction: Balzano, 06/06/91
FOS GIMP NSSC-I Flight Software Requirements Review: Glen Foley, Flight Software Systems Branch 22/10/91
FOS GIMP Flight Readiness Review: Adams, Balzano, Chance, Fitch, Foley, Schneider 16/12/92
Plots - Before and After GIMP Correction: Fitch 16/12/92
Initial onboard test reports in HST Status Report 06/01/93 and HST Status Report 07/01/93
Final GIMP Report: Fitch 19/03/93
Note that, some of these documents were drafts or proposals written around the time that the GIM problem was discovered, therefore some of the details may have been superseeded. However, they are presneted here as they contain information which never found its way into more recent documents.

Related STSDAS Tasks:
Diagram showing the GIMP correction in the calfos pipeline
calfos help pages. Source of the gmpfos task in calfos.
To find the GIMP correction applied for any dataset try the STSDAS task gimpcor. But note that this relies upon the header information (infact it is simply a subtask of calfos).

Zero Point Shifts
A significant, time-dependent, systematic shift of up to 7 quarter-stepped pixels (1.75 diodes) in the zero-point of the FOS/BL wavelength scale has been confirmed in FOS Instrument Science Report (ISR) CAL/FOS-149 (Rosa, Kerber, and Keyes). Relative wavelength offsets within individual spectra are not affected by this shift.

FOS/BL wavelength zero-points shifted in a more-or-less monotonic fashion from launch (April 1990) through de-commissioning (February 1997). An approximate 7-pixel (1.75 diode) shift occurred over this time period. At any given epoch the peak-to-peak scatter about the mean trend is +/- 1-pixel (i.e., 10 times the measurement error). For high-dispersion gratings, 1 quarter-stepped FOS pixel corresponds to approximately 60 km/sec.

FOS/RD wavelength zero-points over the same time period present an apparently random distribution with a peak-to-peak range of 7 quarter-stepped pixels. Therefore, changes in FOS/RD wavelength zero-points are NOT confirmed at the present time. Work continues to determine whether improved characterization of the geomagnetic field and the onboard GIM correction can remove any of this measurement scatter and thereby facilitate a possible correction algorithm for FOS/RD.

Please note that no correction algorithm exists for FOS/BL PRISM, due to the non-linearity of its wavelength scale, nor for FOS/RD observations.

Correction to Blue Zero Point Shifts
We provide an IRAF/STSDAS cl script (foswcorr.cl) and descriptive help file (foswcorrhlp.txt) for use as a quick-fix to correct the wavelength files (.c0* files) for any FOS/BL observation. Figures 1 and 2 (JPG - 42 Kbytes) illustrate the magnitude of the observed shifts and the effectiveness of the quick-fix algorithm provided here. Wavelength zero-point offsets caused by filter-grating wheel positioning uncertainties are not removed by this correction algorithm. Currently corrections are not available for FOS/BL PRISM nor for any FOS/RD observation.

Obtain the cl script: foswcorr.cl (ASCII-text - 9 Kbytes).

Obtain the help file for the cl script: foswcorrhlp.txt (ASCII-text - 2 Kbytes).

Obtain the FOS ISR: CAL/FOS-149 (PDF - 53 Kbytes).

It is expected that the correction algorithm contained in the cl script presented here will eventually be incorporated in standard calfos processing. The availability of an updated calfos will be announced both on the FOS WWW page and in the Spectrographs STAN when available. For the present the cl script presented here is the ONLY procedure provided by STScI/ST-ECF for the correction of FOS/BL wavelengths. Please refer to FOS ISR 149 for a thorough discussion of the correction algorithm and other characteristics of the observed zero-point shifts.

Any additional correction algorithms, improvements to the FOS/BL quick-fix procedure, or assessments of the impact of the observed shifts on other instrumental calibrations will be announced in these pages.

Problems With the On Board Fix
The apparently random distribution of FOS/RD wavelength zero-points stems from a number of problems in the implementation of the on-board fix. In order to remove the zero point shift in a given exposure therefore, it is necessary to know the on-board correction applied, remove it, and apply the appropriate correction. However in attempting to construct a new CAL_FOS pipeline that would implement these corrections, we ran into further problems.

Header Keywords regarding the on-board correction applied turn out to be un-reliable.
Pointing information is often absent from WAVE-CAL datasets. Presumably at the time the pointing was thought to be irrelevant for such exposures, however this information is need for GIMP analysis.
The instructions up-linked to HST's on board computer (essentially coeffiecients of a polynomial that specifies the X & Y deflection voltages at any time) were not archived.
The engineering telemetry AEDP files contain the missing information, but not at sufficient resolution.

See also: Rectification of FOS wavelength Scales, Kerber & Rosa in the June Issue of the ST-ECF Newsletter

The New POA_CALFOS solution
The POA team at the ST-ECF now offer a calibration pipeline package which automates the correction to the blue zeropoint shift and applies some further corrections. See POA_CALFOS.