STScI Logo

30.5 Assessing FOS Acquisitions

[Top] [Prev] [Next]


Nearly all FOS observations included an onboard target acquisition that used instrument observing modes to locate the science target in the desired aperture. Under normal circumstances, prior to any FOS acquisition or science observation sequence the FGS performed a guide star acquisition and commenced FINE lock guiding. A guide star re-acquisition was also performed at the start of all subsequent visibilities in the same visit. Occasionally, some aspect of the acquisition or re-acquisition failed. Appendix C provides for more information about guide star acquisition failure. Chapter 35 (for the GHRS) also contains useful discussions of a variety of guide star acquisition and guiding anomalies.

Before June 1993, the FOS aperture door was closed for the remainder of any exposure during which a guiding anomaly occurred. As a safety measure, commencing in June 1993 the FOS shutter door was always closed for the duration of the obset (typically remainder of the visit) if guide star lock was not established at the start of any exposure. After October 1995, a significant number of FOS acquisitions failed in this manner due to anomalous losses-of-lock caused by FGS1.

This section describes each type of FOS acquisition by giving a detailed discussion of the acquisition method (in the "Understanding" subsection), a discussion of pointing accuracy limitations with the method, a description of and references to examples of the mode-specific paper products, tips for understanding and using the paper products to assess the observational data quality, a description of the structure and content of the output data products, analysis information, and, in some cases, very detailed background information for the most inquisitive users.

No FOS acquisition exposure (e.g., an ACQ/BINARY exposure or a single scan-pattern of an ACQ/PEAK) could be interrupted by Earth occultation.
Some additional practical information on target acquisition is given in Chapter 2 of the FOS Instrument Handbook version 6.

30.5.1 Binary Acquisition-ACQ/BIN

Binary search acquisition mode is indicated by a raw data header file (.d0h) OPMODE value of ACQ/BIN. This method was used for targets that are point sources with well known energy distributions. Although ACQ/BIN was designed to select the nth brightest object in a crowded field, this option was never correctly used except in calibration programs. The following discussion will describe how the brightest target was acquired.

Understanding ACQ/BINARY

The binary search algorithm used MIRROR and began by mapping the 4.3 aperture (only the 12 central diodes which cover the 4.3 aperture provide the data) with 3 y-scans which are stored in the raw data file as 3 groups. The first raster (i.e., group 1) was of the central region of the 4.3 aperture, the second (group 2) was of the lower region of the 4.3 aperture, and the third (group 3) was of the upper region of the 4.3 aperture. The binary search algorithm summed in each group the counts from all pixels that are bounded by the user-specified FAINT and BRIGHT limits. The program then compared the count levels in the three scans, and located the target in one of the three scans by finding the total number of peaks in the scan and the number of counts in each individual peak. The brightest peak, representing the target, was then selected. If more than five peaks per raster were found by the binary search algorithm, the target acquisition failed. If there were two peaks (within a diode width) in two adjacent scans, the algorithm summed up the number of counts in the two peaks to determine the brightness of the target. The search program then continued (up to eight more tries) to electronically deflect the image of the target in the y-direction (note the telescope was never moved in this search process) until the target image was on the edge of the diode array, i.e., until the number of counts in the peak was half the maximum number of counts observed in the initial three scans. For each deflection the data were stored as a group.

The offset required to position the target on the edge of the diode array was stored in the .shh file. This quantity was then used to determine the slew that was necessary to place the object in the nominal center of the 4.3 aperture. Note that no further refinement of position was made in the binary search algorithm.

If ACQ/BINARY failed for any reason (except 8-steps-to-success described below), the telescope did not slew to place the target at the nominal center of the 4.3 aperture, and subsequent exposures in the visit were executed at the initial blind pointing position of the telescope.

Pointing Accuracy: ACQ/BINARY

If a binary search was successful, we can only determine the amount of telescope slew that occurred to place the target at the STScI-determined nominal aperture center on the diode array. We cannot locate the exact position of the target in the aperture unless a confirming image was acquired after the acquisition.

Since binary search determined the offset to place the target on the edge of the diode array and used that offset to calculate the offset slew required to place the target at the expected center of the 4.3 acquisition aperture, the accuracy of centering was dependent on both the accuracy of the half-power point algorithm and the 0.1" accuracy with which the location of the center of the 4.3 aperture was known (see FOS ISR 139).

The statistical 1-sigma uncertainty for ACQ/BIN is 0.08" for post-COSTAR FOS/BL and 0.12" for post-COSTAR FOS/RD. Pre-COSTAR uncertainties are estimated to be of similar order, but an additional worst-case uncertainty of up to 0.15" due to uncorrected GIM motion must be added in quadrature for all ACQ/BIN acquisitions prior to April 5, 1993.

If an ACQ/BIN spectrum has 11 groups, it may have been considered to have failed. Before June 1, 1993, all such acquisitions were returned to the blind pointing. Commanding was changed on that date to recognize the fact that any additional correction to be applied after the algorithm reached the eleventh step was typically of order 0.02" and of no significance. After this date, these so-called 8-steps-to-success "failures" resulted in the position implied by the last ACQ/BIN iteration being chosen as the pointing. Figure 30.12 illustrates some of the characteristics of ACQ/BINARY.

Figure 30.12: Binary Target Acquisition (pre-COSTAR sizes in parentheses)


Paper Products: ACQ/BINARY

Figure 30.3 provides an illustration of the FOS paper products ACQ/BINARY diagnostic plots. The one-dimensional images (counts vs pixel) for each group are plotted as well as a separate map of Y-bases chosen for each deflection step. The final x- and y-offsets, and their equivalent V2 and V3 offsets required to center the target are given. A compass rose, oriented identically to the spatial coordinates of the individual plots, is also provided. Conversion of offsets to right ascension and declination is described in "Converting X,Y to RA, Dec" on page 33-9.

Output Data Products: ACQ/BINARY

The binary search target acquisition mode is identified by the OPMODE value of ACQ/BIN. The ACQ/BIN data (.d0h file) contains up to 11 groups (depending at which stage the algorithm stopped the search) and each group has images of 64 pixels (12 diodes scanned with NXSTEPS=4 and OVERSCAN=5). The central pixel is defined to be number 32 of the numeric range 1-64. The calfos calibration only corrects the data for paired-pulse effects and converts the raw counts into a count rate. No further calibration procedures are applied.

Analysis: ACQ/BINARY

If there are fewer than four groups in the ACQ/BINARY data files, the binary search target acquisition has failed.

A binary search may have failed due to one of the following reasons:

30.5.2 Peakup or Peakdown Acquisition (ACQ/PEAK)

This acquisition mode was used for bright targets, variable targets, targets whose energy distributions were not well known, and especially for those cases in which very accurate pointing was required. This acquisition mode was used extensively throughout the FOS operational lifetime. The highest pointing accuracies achieved with FOS used ACQ/PEAK and after July 1, 1992, all FOS flatfield, photometric calibration, and external wavelength calibration observations used high-precision multiple-stage ACQ/PEAK acquisitions.

Understanding ACQ/PEAK

An ACQ/PEAK stage consisted of a series of NXSTEPS=1, OVERSCAN=1 exposures made at each point in a user-specified scan pattern. All successful ACQ/PEAK stages ended with a telescope slew to the pointing corresponding to the brightest (for peakup) or faintest (for peakdown) exposure in the scan pattern. No attempts were made to interpolate positions between individual pointings in the scan pattern. Evaluation of the brightness of the signal at each dwell point was performed onboard by simple addition of the counts accumulated in all pixels with no correction for background, dead or noisy diodes, paired pulse, or even edge-pixel exposure effects.

As for ACQ/BIN, the first stage of a peak-up target acquisition was normally done in three integration steps with the 4.3" aperture and either the camera MIRROR or a disperser. As with all FOS acquisitions, the original target coordinates for the first stage were required to be accurate to about 1", so that the object fell within the 4.3" aperture. After this first stage the measured signals from the dwell points were stored in the raw data file as three groups. The dwell point with the maximum number of counts was determined and the telescope was positioned with the center of the aperture corresponding to this dwell point. The second stage, a 2 x 6 step pattern, using the 1.0" aperture, traced the location of the source within the 1.3" x 3.7" (1.4" x 4.3" pre-COSTAR) area where it was found in the first stage and these measures were recorded as groups in a separate data file. This narrowed the area in which the target was located to the surface area of the 1.0" aperture (to the one of the 12 integrations that had the most counts). The third stage of a peak-up sequence was normally a 3 x 3 point scan of the area of the 1.0" aperture, with the 0.5" aperture and a step size of about 0.3". This led to a pointing accuracy of about 0.2". If higher accuracy was needed, the surface area of the 0.5" aperture had to be scanned with the 0.3" aperture, possibly in two steps with decreasing step sizes. A variety of patterns and pattern-sequences were used by observers. Please refer to Table 32.3 for a listing of the FOS team-recommended patterns and nominal worst-case pointing accuracies.

No individual peakup or peakdown stage could be interrupted by Earth occultation. For much of the HST lifetime, telescope commanding imposed a maximum duration, including overheads, for any peakup/peakdown stage (or scan sequence) of 3000 seconds.

Pointing Accuracy: ACQ/PEAK

In the limit of perfect photon statistics, a positional accuracy upper limit for a particular coordinate in any ACQ/PEAK stage is one-half the step-size in that coordinate. This remains a good approximation for most commonly used apertures and scan patterns as the FOS team routinely recommended obtaining 10,000 counts in the peak dwell for so-called critical peakups (those that used the 0.3 or smaller aperture with step-sizes of 0.06" or smaller) and 50,000 counts in the most strongly exposed dwell of critical peakdowns. As Table 32.3 indicates, the most precise nominal worst-case pointing accuracy routinely recommended by the FOS team was 0.025" in each coordinate.

Paper Products: ACQ/PEAK

Figure 30.4 presents a typical FOS paper product ACQ/PEAK diagnostic plot for a 5 x 5 scan pattern. Apart from a period between July and November 1994 when anomalous scan parameter values were written in the ACQ/PEAK headers (see below), the pattern is plotted in the correct orientation and in the correct sense of telescope motion in detector coordinates. The offset to move the telescope to the brightest (or faintest in the case of peakdown) tile of the scan pattern is provided in both the detector (x,y) and telescope (V2,V3) frames. A compass rose is given. PA_APER is also given for easy conversion of (x,y) offsets to the (RA, Dec) frame, although we do not calculate the offset in (RA,Dec). See "Converting X,Y to RA, Dec" on page 33-9 for a description of how to convert to the (RA,Dec) frame. No contour plot is provided. No spectra or images from the individual tiles are plotted, but in some cases you may wish to inspect the individual groups that contain these quantities. The standard jitter ball is provided which typically shows the scan pattern in (V2,V3) and indicates how closely the telescope moved to the desired dwell point.

Output Data Products: ACQ/PEAK

The peakup or peakdown mode is identified by the .d0h OPMODE value of ACQ/PEAK. Each stage of an ACQ/PEAK sequence produced its own set of output data files. An ACQ/PEAK scan-pattern with x rows and y columns produced output data files with (x x y) individual groups.

ACQ/PEAK measures with the MIRROR produced a one-line raster of 96 pixels (NXSTEPS=4, OVERSCAN=5 pattern with 20 diodes readout). Note that even the 4.3 aperture, the largest FOS aperture, illuminated only 12 diodes. ACQ/PEAK measures with a disperser produced 512 pixels (NXSTEPS=1, OVERSCAN=1 pattern with all 512 diodes read out by default). The number of diodes sampled in a dispersed light ACQ/PEAK could be altered by the specification of a limited wavelength range to be read out. These images and spectra are not calibrated by the pipeline, but may contain scientifically useful information. We recommend inspection of these data files (see below).

Analysis: ACQ/PEAK

You can determine the centering accuracy of most ACQ/PEAK stages by inspecting the distribution of counts in the tiles of the scan pattern. Notice that for many acquisitions, especially in the pre-COSTAR era, the centering accuracy is not the same in both FOS x and y. This led in some of these cases to compromised flat field and flux calibration.

The individual spectra in an ACQ/PEAK with either a disperser or MIRROR can yield valuable information about the distribution of light or excitation in extended sources, or provide additional spectroscopic information for point sources if the acquisition disperser was not used for any science observations. Care must be exercised, however, in analysis of these off-center observations as many such spectra will not be fully sampled by the diode array (see "Location of Image on Diode Array" on page 32-7).

In order to convert FOS x,y coordinates into sky coordinates, you need to know the position angle of the aperture, which is documented in the header keyword PA_APER (see "Converting X,Y to RA, Dec" on page 33-9) and which is routinely provided in the paper products.

The FOS paper products algorithm calculates observed counts per dwell tile from the .c5 files, which have had several corrections applied that were not performed by the onboard ACQ/PEAK algorithm (most notably paired-pulse and edge-pixel exposure correction). In extremely rare cases involving very low count levels, the paper product algorithm may predict a different maximum tile than that actually chosen by the onboard processor. This discrepancy has been documented only in situations in which the FOS aperture door was closed due to guide star acquisition failure.

In mid-1994 the scan direction for some ACQ/PEAK exposures was changed at the same time that commanding overheads were made up to 20% more efficient. Between July and November 1994, ACQ/PEAK scan information was not written to the header files. Analysis of ACQ/PEAK patterns in this period must be done manually with the information contained in the logsheet. The simplest approach to determine scan-pattern directions for your manual analysis is to run the FOS paper products for an equivalent ACQ/PEAK performed after November, 1994.

30.5.3 IMAGE Mode Target Acquisitions

Understanding FOS ACQ Imaging

FOS IMAGE mode observations mapped more than one position in the aperture with the diode array.

Spectral Element: MIRROR

FOS imaging mode with spectral element MIRROR was typically used for EARLY acquisition, INTeractive ACQuisition, or to obtain a verification image of the FOS large aperture to check the target position in the aperture.

Camera MIRROR images for interactive acquisition (logsheet entry INT ACQ) were designated by OPMODE=ACQ. Standard camera MIRROR images of the 4.3 aperture which did not require real-time interaction (logsheet entry ACQ) were designated, somewhat confusingly, by OPMODE=IMAGE. Both types of images have GRNDMODE=ACQUISITION and use the same set of pre-defined instrumental setups. The far less commonly used user-defined IMAGE command (logsheet entry IMAGE, spectral element MIRROR) allowed the user to specify image line and raster parameters for spectral element MIRROR and was designated by GRNDMODE=IMAGE, OPMODE=IMAGE. In the following we use the term FOS ACQ Imaging to refer to all of these imaging modes.

In a standard INT ACQ or confirming ACQ image the FOS Digicon was commanded through a pre-defined sequence of x-steps and y-steps which mapped the aperture. The 4.3\xfd aperture was scanned with 64 strips, each of which was the height of the diode array, beginning at the bottom, -y, edge of the aperture. The post-COSTAR distance between the strips was 16 Y-base units or 0.0786\xfd for FOS/BL and 0.0812\xfd for FOS/RD The pre-COSTAR distance was 0.08958\xfd for both detectors. Each scan read out 20 diodes with the central 12 diodes spanning the 4.3\xfd aperture. In the standard mode (NXSTEPS=4 and OVERSCAN=5) the number of pixels in each strip was 96. Thus the image has 96 x 64 pixels where each pixel were equally spaced on the sky (see Figure 30.13).

Figure 30.13: Digicon Faceplate Sampling Pattern in a Target Acquisition Image


Spectral Element: Any Disperser

The rarely used user-defined dispersed-light IMAGE command (logsheet entry IMAGE with any disperser as spectral element) allowed the user to specify image raster parameters with a spectral element. Spectra with user-specified values for NXSTEPS and OVERSCAN at YSTEPS y-positions, symmetrically spaced about the default aperture center Y-base were obtained. So-called dispersed-light interactive target acquisitions were occasionally employed for solar system moving targets, usually comets, with a dispersed-light IMAGE mode observation at several y-positions in the 4.3 aperture. The maximum integrated count level in the entire spectrum determined the y-position offset of the initial blind pointing from nominal and the deviation of a well-known spectrum line from its anticipated x-position allowed determination of the blind pointing mis-centering in the x-direction.

Pointing Accuracy: FOS ACQ Imaging

Mirror Mode Images: Image centroiding accuracies varied, but for well-exposed images centering accuracies were about 0.1\xfd in both x and y.

Dispersed-Light Mode Images: Uncertainties of <0.5\xfd in both coordinates were achieved. This was typically sufficient for centering of comets in the 4.3 aperture.

Paper Products: FOS ACQ Imaging

Paper products for FOS images with spectral element MIRROR present grayscale renderings of the raw aperture image (see Figure 30.2). No image rectification or deconvolution is attempted. A compass rose provides orientation information and plate-scale fiducials are plotted. An exposure level key is also provided. See "Image Mode Spectra (IMAGE)" on page 30-47 for a discussion of dispersed-light image mode paper products.

Output Data Products: FOS ACQ Imaging

The calibrated INT ACQ and ACQ data files have 64 groups with each group having 96 pixels. The raw data file (.d0h) is an actual 96 x 64 image (1 group) file. Output files for user-defined images contain header keyword YSTEPS groups of keyword (NCHNLS+(OVERSCAN-1))*NXSTEP pixels. Routine calibration processing produces .c4 files which are no longer two-dimensional images, rather are multi-group files with one group for each raster of the image. The IRAF/STSDAS task rapidlook converts these processed files back to a two-dimensional representation suitable for examination with standard image processing tasks. A discussion of dispersed-light image mode data products is presented in "Image Mode Spectra (IMAGE)" on page 30-47.

Analysis: FOS ACQ Imaging

The substepping in both the x and y directions, and the elongated shape of the diodes, blurred and stretched the image (see Figure 30.14). A point source (which has the size of a PSF) on the photocathode was recorded by the same diode for four consecutive pixels in the x direction and 16 consecutive pixels in the y direction. The tarestore task in STSDAS can be used to deconvolve the image. In this task the image is trimmed and the flux resampled in the pixels. The modeone task also uses the tarestore task and displays the deconvolved image with the correct orientation on the sky (north and east are indicated).

Figure 30.14: Sample FOS Target Acquisition Image


The position of the target in the image can be found by using the task aperlocy in STSDAS. This task computes the location of the image edges, the midpoint of each axis, the centroid of the image, and the total flux in the image. Note that aperlocy can not be used on restored images and that it will produce meaningful results only for images with a single source present. The output units of this task are diodes for the x-axis and Y-base units for the y-axis. This task also takes into account the elongated shape of the diodes in determining the target centroid in the aperture. The nominal center for the 4.3 aperture is at the pixel (48,32). To determine the nominal aperture center in diode and Y-base units, the following equations are used:

where:

The centroid of the target can then be compared with the aperture center to determine the location of the target in the aperture. Note that the aperlocy task can also be used for images of the paired apertures. See the on-line help files in STSDAS for details.

30.5.4 Firmware Target Acquisition

Understanding ACQ/FIRMWARE

Firmware mode was occasionally used for pre-COSTAR planetary satellite observations, although this was an engineering mode. This procedure mapped the camera MIRROR image of the 4.3 aperture in x and y with small, selectable y raster increments. The microprocessor filtered this aperture map in real time and found y-positions of the peaks by fitting triangles through the data.

ACQ/FIRMWARE was less efficient than ACQ/BINARY and failed if more than one object were found. In practice this mode sometimes failed because the brightness limits were not set accurately by the observer. ACQ/FIRMWARE was used several times in the pre-COSTAR period and was included in an SMOV target acquisition testing program, but no ACQ/FIRMWARE acquisitions were performed in post-COSTAR science programs.

Pointing Accuracy: ACQ/FIRMWARE

Pre-COSTAR accuracies for ACQ/FIRMWARE were similar to those for ACQ/BINARY, approximately 0.15-0.20\xfd one sigma. No post-COSTAR evaluation of ACQ/FIRMWARE pointing accuracy was made.

Paper Products: ACQ/FIRMWARE

There are no ACQ/FIRMWARE-specific paper product displays. The standard spacecraft performance, jitter ball, and calibration status pages are produced.

Output Data Products: ACQ/FIRMWARE

The firmware target acquisition mode is identified by the OPMODE value of ACQ/FIRMWARE. The ACQ/FIRMWARE mode mapped the 4.3 aperture with a certain number (m) of Y-steps acquiring data from 20 diodes covering the aperture in the standard ACCUM mode. The target acquisition data therefore has m groups with 96 pixels (NXSTEPS=4 and OVERSCAN=5).

Analysis: ACQ/FIRMWARE

The microprocessor onboard the FOS searched for the peaks in the data array within a specific window and the target was centered at this location. The m groups could be plotted as a function of location and the position of the target in the aperture could be determined in a manner similar to the ACQ mode.


[Top] [Prev] [Next]


Copyright © 1997, Association of Universities for Research in Astronomy. All rights reserved.

Last updated: 01/14/98 14:29:35