Fine Guidance Sensor Instrument Handbook for Cycle 14
 |
 |
 |
 |
 |
|
| Fine Guidance Sensor Instrument Handbook for Cycle 14 | |||||
| ||||||
4.1 Position Mode OverviewChapter 4:
Observing with the FGS
4.1.1 The Position Mode Visit
4.1.2 The Position Mode Exposure
4.2 Planning Position Mode Observations
4.2.1 Target Selection Criteria
4.2.2 Filters
4.2.3 Background
4.2.4 Position Mode Exposure Time Calculations
4.2.5 Exposure Strategies for Special Cases
4.2.6 Sources Against a Bright Background
4.2.7 Crowded Field Sources
4.3 Position Mode Observing Strategies
4.3.1 Summary of Position Mode Error Sources
4.3.2 Drift and Exposure Sequencing
4.3.3 Cross Filter Observations
4.3.4 Moving Target Observation Strategy
4.4 Transfer Mode Overview
4.4.1 The FGS Response to a Binary
4.4.2 The Transfer Mode Exposure
4.5 Planning a Transfer Mode Observation
4.5.1 Target Selection Criteria
4.5.2 Transfer Mode Filter and Color Effects
4.5.3 Signal-to-Noise
4.5.4 Transfer Mode Exposure Time Calculations
4.6 Transfer Mode Observing Strategies
4.6.1 Summary of Transfer Mode Error Sources
4.6.2 Drift Correction
4.6.3 Temporal Variability of the S-Curve
4.6.4 Background and Dark Counts Subtraction
4.6.5 Empirical Roll Angle Determination
4.6.6 Exposure Strategies for Special Cases: Moving Targets
Position and Transfer modes each have distinct calibration requirements and observing strategies. This chapter describes the detailed characteristics of the two observing modes and the observational configurations which maximize science return.
4.1 Position Mode Overview
4.1.1 The Position Mode Visit
A Position mode visit yields measurements of the location of objects in the FGS's total field of view (FOV), and hence their relative angular positions. The objects are observed sequentially according to the sequence of exposure lines in the proposal. The target list of a typical Position mode visit consists of the science object(s) and reference stars used to define the local reference frame. A subset of the targets, referred to as check stars, should be observed several times during the course of the visit to track any spurious motion of the FGS's FOV on the plane of the sky (e.g., thermally induced drift or OTA focus changes). The changes in the positions of the check stars are used to model the drift as a function of time so that its contaminating effect can be eliminated from the astrometry.
An FGS astrometry visit begins when the HST computer - the 486 - commands the Star Selector Servos to place the IFOV at the predicted location of the first star specified in the visit (as per Phase II proposal). Control is transferred to the Fine Guidance Electronics (FGE) microprocessor, which commands the FGS to acquire and track the target (Search, CoarseTrack, and FineLock). Later, at a specific spacecraft clock time, after the exposure time + overhead has expired, the 486 resumes control of the FGS, terminates the FineLock tracking of the object and slews the IFOV to the expected location of the next star in the sequence. This process repeats until the FGS has completed all exposures in the visit. The spacecraft's pointing is held fixed on the sky under the control of the guiding FGSs for the entire visit unless otherwise instructed by the Phase II proposal. The status flags, photometry and instantaneous location of the IFOV is recorded every 25 msec (40 Hz).
Further discussion of how these impacts will affect the observatory and the instruments can be found in a separate Handbook, the HST Two-Gyro Handbook for Cycle 14. See the Two-Gyro Handbook for detailed information. All text in this FGS Handbook assumes three-gyro control.
4.1.2 The Position Mode Exposure
During a Position mode exposure the object is tracked in FineLock (see Appendix1). After the target is acquired by the Search and CoarseTrack procedures, FineLock begins with a WalkDown, a series of steps of the IFOV toward the CoarseTrack photocenter. At each step, the IFOV is held fixed for a period defined by FESTIME (Fine Error Signal averaging time) while the PMT data are integrated to compute the Fine Error Signal (FES, the instantaneous value of the S-Curve) on each axis. Once the FES on both axes have exceeded a pre-set threshold, FineLock tracking begins. The star selectors are continuously adjusted after every FESTIME to re-position the IFOV in an attempt to zero out the FES during the next integration period. The objective is to present the Koesters prism a wavefront with zero tilt.
The defining parameters of an exposure are the target magnitude, the filter, the FESTIME and the exposure time. These topics are discussed in the following sections.
4.2 Planning Position Mode Observations
4.2.1 Target Selection Criteria
When targets are selected for FGS Position mode observations, several options and requirements should be considered. These options are described below.
Brightness
The bright limit for FGS1r is V = 8.0 without the neutral density filter in place. With the F5ND filter, objects of V = 3.0 or fainter can be observed. The faint limit is V ~ 17.0.
Near Neighbors
FGS target acquisition in Position mode will be unreliable if the target has a neighbor of comparable or greater brightness within a radius of 10 arcseconds. In essence, the FGS's IFOV - a 5" x 5" box - expects to encounter the target star within the search radius. Companions of similar brightness within this search radius may be mistakenly acquired instead of the target. However, for magnitude differences
m > 1, companions within ~ 6" will not affect the acquisition of the brighter target. Note that binary stars with component separations less than about 0.5" can be successfully acquired in Position mode, regardless of the
Target Field
The target field consists of the science target and reference stars. Observations of the reference stars will be used to define the local reference frame for relative astrometry. Since the optical field angle distortions are calibrated most accurately in the central region of the FOV, the pointing of the spacecraft (via POS_TARG commands - Chapter 6 for more details) should be specified to place the target field (as much as possible) in this area.
If the visit also includes Transfer mode observations of an object, the spacecraft pointing should be chosen to place the object at the FOV center, as this is the only location calibrated for Transfer mode. If the target field geometry requires the Transfer mode observations be executed at other locations in the FOV, special calibrations will be needed. Proposers should consult STScI's Help Desk for assistance.
Reference Stars
Ideally, reference stars should have the following characteristics:
- Have magnitudes in the range 8 < V < 15 to avoid the need for an F5ND cross-filter calibration and to minimize exposure times.
- Be geometrically distributed around the target.
- Have at least 10 arcsec distance between one another and from the science target to avoid acquisition of the wrong object.
- Should fall within the FOV for all HST orientations specified in the proposal.
Check Stars
Check stars, which are a subset of the target list, are observed several times over the course of an orbit (visit). Two or more check stars, distributed across the field, provide the information needed to characterize the drift of the FGS's FOV on the sky (which is typically about 4 mas over the course of the visit). Each check star should be observed at least three times. The best check stars are brighter than 14th magnitude to minimize exposure time, and should include the science object for the highest accuracy astrometry.
4.2.2 Filters
Table 4.1 is a listing of the FGS1r filters, their calibration status and applicable brightness restrictions. (Refer back to Figure 2.8 for the filter transmissions as a function of wavelength.)
PUPIL Not Recommended for Position Mode
Occupying the fifth slot on the wheel is the PUPIL. It is not a filter but rather a 2/3 pupil stop. Use of the PUPIL significantly reduces the degrading effect of spherical aberration (which does not necessarily improve Position mode performance) but collaterally alters the field dependence of the distortions. Consequently, the OFAD calibration for the F583W filter cannot be applied to PUPIL observations. In addition, PUPIL observing attenuates the object's apparent brightness by nearly a full magnitude, which sets the faint limiting magnitude at about V=16 while making observations of stars fainter than V = 14.5 excessively time consuming.
FESTIME and Signal-to-Noise
Photon statistics dominates the noise in the measured position of stars fainter than V ~ 13.0. To track fainter objects, the Fine Error Signal must be integrated for longer periods. Table 4.1 lists the default FESTIMES for various target magnitudes. The default FESTIMES, determined from the Phase II target magnitude, are appropriate for most observations, and are set to ensure that photon noise, when converted into the Noise Equivalent Angle (NEA), does not exceed a predefined angular error threshold. The NEA is given by the relation
The NEA is used by the proposal processing tool (APT) to set the default FESTIME time. The parameter C is the total count rate expected from the target summed over all four PMTs, B is the background count rate, and t is the FESTIME. The NEA is plotted as a function of magnitude and FESTIME in Figure 4.1. C as a function of filter and magnitude for FGS1r is given by:
The constant f-factor is a function of the filter and the target's spectral color. Table 4.3 provides the f-factor for each combination of filter and color. The default FES times used by the proposal processing software for Position mode measurements are listed in Table 4.1.
Figure 4.1: Default FESTIME as a Function of V Magnitude for F583W FGS1r: NEA as a Function of Magnitude and FESTIME
Table 4.3: F-factor Transmission Estimator for Combination of Filter and Color
4.2.3 Background
Background noise includes cosmic ray events, particle bombardment during passages through the South Atlantic Anomaly (SAA), and scattered light falling in the 5 x 5 IFOV. Cosmic ray events are suppressed by special circuitry and the FGS is prohibited from operating while transiting regions of heaviest impact from the SAA. Table 4.4 gives the typical dark + background counts for FGS1r in 0.025 seconds. Typically these values appear to be valid for all observations of isolated targets (suggesting that the dark counts dominate the background contribution). If the background counts for a specific observation are needed for the analysis of the observation, such as when the source is embedded in significant nebulosity or in a crowded star field, it can be obtained from the photometry gathered during the slew of the IFOV to (or away from) the target position. These data extracted by the FGS pipeline package CALFGSA from the FITS files that input are cleaned of spikes from "interloping stars" and can be used to estimate the background levels during post-observation data reduction.
Table 4.4 lists the average dark+background counts/25 msec for each of the FGS1r PMTs. These data were serendipitously gathered over a 45 minute interval from a failed science observation (the target was not acquired due to a guide star problem). These data have proved invaluable for the analysis of Transfer mode observation of faint stars (V>15).
Table 4.4: FGS1r: Dark Counts
4.2.4 Position Mode Exposure Time Calculations
The exposure time is the minimum time that an object will be tracked in FineLock. Based the rate at which the measured location (or centroid) of a star converges (from analysis of FGS1r data) Table 4.5 lists the recommended exposure times as a function of target magnitude. We note that:
- Exposures should be as short as possible to allow for more individual observations during the visit, but should be longer than HST's mid-frequency oscillations (~10 seconds).
- Usually, Position mode observations yield an additional 10 to 20 seconds of FineLock data in excess of the exposure time specified in the phase2 proposal (a result of unused overheads). Hence, specifying a 10 second exposure results in 20 to 30 seconds of FineLock data.
4.2.5 Exposure Strategies for Special Cases
Observing Binaries and Extended Sources in Position Mode
Multiple or extended sources in the FGS's IFOV will result in a reduction of the amplitude of the observed interferometric fringes (relative to that of a point source). This occurs because light from multiple sources in the IFOV do not interact coherently (the observed rays originate from different angles on the sky). Therefore, multiple point source fringes will be superimposed upon one another, each scaled by the relative brightness of the source and shifted by its relative angular displacement on the sky. The result is a composite Transfer Function with reduced fringe visibility.
The fringe visibility reduction for the brighter component of a binary system with an angular separation along the X or Y axis greater than about 80 mas (i.e., when the individual S-Curves are fully separate) is given by:
where fa and fb are the intensities of the brighter and fainter components, respectively. A similar expression, but with lb in the numerator, is appropriate for the faint star S-curve (see Figure 4.3 for examples).
For projected angular separations less that 80 mas, the Transfer Function will be a blend of the merged point source S-Curves. The resultant fringe visibility will depend on the relative brightness and the angular separation of the components (i.e., Fr is more difficult to predict).
Even significant loss of fringe visibility does not pre-dispose the object from being successfully observed in Position mode. To be acquired in FineLock, an object's Fine Error Signal (see Appendix A) must exceed a fringe detection threshold (see Figure A.2). The threshold is set on the basis of the target's V magnitude, as entered in the proposal, to accommodate the acquisition of faint targets. (The fainter the target the more effectively the background and dark counts reduce the fringe amplitude, hence lower detection thresholds must be applied.) If the GO were to state the V magnitude of a binary system or extended source to be sufficiently faint, (regardless of its true value), then the observed fringes will exceed the (lower) detection threshold, and the FGS will successfully acquire the object. However, if a false magnitude is specified, one should also manually set the FESTIME (an optional parameter) to the value appropriate to the object's true magnitude. Otherwise, the observation's overheads will be excessively long.
Some binary systems are not reliably observed in Position mode, even with the adjustment to the fringe detection threshold. Objects in this category include those with components exhibiting small magnitude differences (
There is a class of binary stars which cannot be observed in Position mode. In a FineLock acquisition (see Appendix A1), the WalkDown to FineLock is a finite length path (approximately 0.810") beginning at a point which is "backed off" a fixed distance from the object's photocenter. If the fringes of both stars lie outside this path, then neither will be encountered and the FineLock acquisition will fail. The condition for such a failure is the following,
where X is the location of the system's photocenter, ra and rb are the distances from the photocenter to the fringes of the components "a" and "b" respectively, la and lb are the flux from each component, xs is the starting position of the WalkDown, and xl is the length of the WalkDown. If the position of the binary along either the X or Y axis is known to meet this failure requirement, Position mode observations of this system should not be attempted.It is recommended that a proposer contact the STScI Help Desk for assistance with Position mode observations of binary systems.
4.2.6 Sources Against a Bright Background
For sources against bright backgrounds, the fringe visibility function is reduced by I / (I + B) where I is the point source flux and B is the background flux. The proposer should contact the STScI Help Desk for assistance with such observations.
4.2.7 Crowded Field Sources
Crowded fields create two problems for FGS observations:
- A nearby star (< 10 arcsec away) of similar magnitude could be acquired during the search phase.
- The background brightness in the 5 x 5 arcsec aperture may be increased by the presence of numerous faint stars or nebulosity.
The proposer should consult the STScI Help Desk for assistance with such observations.
4.3 Position Mode Observing Strategies
Measurement errors can be minimized by carefully structuring the order of exposures in a visit. This section describes strategies which maximize science return.
4.3.1 Summary of Position Mode Error Sources
The reduction of a Position mode data set requires several corrections and calibrations:
Each of the corrections and calibrations are briefly discussed in Chapter 5, and are thoroughly reviewed in the
HST Data HandbookMany of the corrections specified on the list are determined by analysis of individual observations and removed later in the data reduction processing, (e.g., jitter data is retrieved from the guide star telemetry and removed from the target data). Calibrations, such as the plate scale, the OFAD, lateral color, and cross filter effect are derived from STScI calibration programs. A potentially dominant source of error, the FOV drift (time scale of several minutes), must be measured during the visit, and to that end, the sequence of exposures must be carefully arranged.4.3.2 Drift and Exposure Sequencing
Stars observed more than once per visit ("check stars") are typically seen to drift across the FGS by ~ 2 to 6 mas when two FGSs guide the telescope (or ~ 5 to 20 mas with only one FGS guiding). Because astrometry observations execute sequentially, the errors in the measured angular separations between objects increase as the time between the measurements lengthens. If uncorrected, this drift will overwhelm the astrometry error budget.
Whatever causes the position of an object to "drift" in the astrometer's FOV affects the guiding FGSs as well. The apparent motion of the guide stars are interpreted as "errors" by the pointing control system and are "corrected" by a small vehicle maneuver. The astrometry FGS witnesses the pointing change. Therefore, the check star motion will have a contribution from all three FGSs.
When only one FGS is used for guiding, the telescope is not roll-constrained, and large motions in the astrometric FGS - up to 10 mas - are not uncommon. Nevertheless, this drift can be successfully removed from the astrometry data, provided the proposal specifies an adequate check star sequence. The more check star observations, the more precise the drift correction. Check stars can be reference or science targets. Ideally, both rotation and translation corrections should be applied to the data, implying the use of at least two check stars with at least three measurements each.
The need to observe check stars can be in conflict with other aspects defining an optimal observing strategy, so compromises will be necessary. Overall an optimal Position mode visit is scripted to:
- Define a check star strategy which will be robust against FOV drift.
- Maximize the number of reference stars observed.
- Maximize the number of observations of each object.
For example, a visit could contain 10-35 exposures, provided the overheads are minimized and exposure times are less than 20 seconds. A sample geometry is given in Figure 4.2, where the science target is represented by the central object.
Figure 4.2: A Sample Visit Geometry
For the geometry specified above, the exposures may be sequenced as follows:
1 - 2 - target - 3 - 2 - 4 - target - 3 - 5 - 4 - target --> --> 2 - 1 - target - 3 - 5 - 4 - target - 2 - 1 - target.Additional examples expressed in proposal logsheet syntax are given in Section 6.5.
4.3.3 Cross Filter Observations
Targets brighter than V = 8 cannot be observed with the F583W. The F5ND attenuator must be used for such observations. If the visit includes observations of fainter objects with the F583W, a cross-filter correction will be needed for the data reduction. For FGS3, the F583W/F5ND cross-filter effect was found to be astrometrically large (~ 7 mas) and varied with location in the FOV. This effect as been calibrated at the center of FGS1r's FOV by STScI calibration programs. If needed at other locations in the FOV, special calibrations by the proposer might be required.
4.3.4 Moving Target Observation Strategy
In Position mode, the FGS can track a bright target (V
) whose motion is less than ~ 0.1 arcsec per second. However, planning the observation requires extreme precision: the moving target must be accurately located, to within 10 arcsec, in the IFOV at the start of the exposure, implying a very accurate ephemeris. Transfer mode observations of moving targets are discussed in the next section.
4.4 Transfer Mode Overview
4.4.1 The FGS Response to a Binary
If the source is a double star, then its wavefront has two components, each incoherent with respect to the other. Two propagation vectors characterize this wavefront and the angle between them is directly related to the angular separation of the stars on the sky. As the FGS's IFOV scans across the object, each component of the wavefront can be thought of as generating its own interferogram (or "S-Curve"), whose modulation is diminished by the non-interfering "background" contributed by the other component. The resulting relationship between the position of the IFOV to the normalized difference of the PMTs depends on the separation of the stars and their relative brightnesses. The composite interferogram of a multiple system is the linear superposition of the fringes from the individual components, scaled by their relative brightness and shifted with respect to one another by their separation on the sky. Given this, the fringe pattern of a binary system, whose components have an angular separation
(as projected along the X-axis) and fluxes fa and fb, is given by
Figure 4.3 shows the changes in the observed FGS1r interferogram due to binary systems of varied separations and magnitude differences. In Figure 4.3a we display the interferogram of a wide binary pair with component separations and magnitude differences of (200 mas, 1.0) respectively. Figure 4.3b is an example of a system with the same separation, but with a magnitude difference of
Figure 4.3: .Binary S-Curves Generated from FGS1r X-Axis Data
If the angular separation of the stars is greater than the width of the S-Curve, two distinct S-Curves are apparent, but the modulation of each will be diminished relative to that of a single star by an amount depending on the relative flux from each star (see Figure 4.3a and also Figure 3.1). On the other hand, if the angular separation is small, the S-Curves will be superimposed, and the morphology of the resulting blend complicated (as in Figure 4.3d). In either case, the composite S-Curve can be deconvolved using reference S-Curves from point sources, provided the angular separations are not too small and the magnitude difference is not too large. To be more precise, fitting the observed double star S-Curve with two appropriately weighted, linearly superimposed reference S-Curves from single stars leads to the determination of the angular separation, position angle, and magnitude difference of the binary's components. The modulation, morphology, and temporal stability of the point-source calibration S-Curves determine the resolving power of the FGS. For FGS1r, this is about 7 mas for
4.4.2 The Transfer Mode Exposure
Rather than tracking the fringe as in Position mode observation, the IFOV is scanned across the object along a 45 degree path (with respect to the FGS detector axes) in a Transfer mode exposure. Every 25 milliseconds, star selector angles and data from the four PMTs are recorded. From these data, the fringes of the object can be reconstructed. The number of scans and the length of each scan are derived from the Phase II proposal.
For each target, the step size and scan length must be adjusted as necessary to accomplish the goals of the observation
- Step Size: The step-size refers to the angular distance covered by the IFOV along an axis during a 25 millisecond interval. The default step-size is 1.0 mas. Higher resolution is achievable using step sizes as fine as 0.3 mas. However, such a small step size may prohibitively increase the total time to complete a scan, and as a result of least significant bit (LSB) constraints and vehicle jitter, the effective resolution may not be better than 1.0 mas. The goal is to execute as many scans as possible, with the smallest StepSize (0.6 or 1.0 mas), to achieve the highest S/N ratio.
- Scan Length: The required scan length is depends on the geometry of the target. A minimum of scan length of 0.3" is needed to fully sample the fringes of a point source along with a sufficient segment of the wings to either side. For binary systems, the scan length should be at least as wide as the component separation plus 0.3".
4.5 Planning a Transfer Mode Observation
4.5.1 Target Selection Criteria
Other than the brightness restrictions specified in Table 4.6 there are several additional considerations when selecting targets for Transfer mode observations.
- Position in the FOV: The morphology of the fringe varies considerably with position in the FOV, as shown for FGS1r in Figure 2.6. As explained in Chapter 2, this field dependence is the manifestation of small misalignment between the star selector optics and the Koesters prism which is greatly magnified by HST's spherical aberration. To obtain the highest quality data with the best resolution, all Transfer mode observations should be obtained at the AMA optimized position at the center of the FOV. Under normal circumstances, STScI will provide reference S-Curve calibrations only at the center of the Field of View.
- Color: The FGS's interferometric response is moderately sensitive to the spectral color of the target. The FGS calibration library contains interferograms from point-source objects with a variety of (B-V) colors. The GO is encouraged to inspect this library to ascertain whether a suitable reference (
(B-V) < 0.3) is available for data analysis. If not, the FGS group should be alerted, and efforts will be made to enhance the library. The color calibrations available in the FGS library are listed in Chapter 5. Up-to-date additions can be found on the FGS web site:
http://www.stsci.edu/hst/fgs/
- Minimum Scan Length: In order for Transfer mode observations to achieve the highest possible signal-to-noise, the length of the scan should be as short as possible (to allow for more scans in the allotted time). The minimum scan length is determined by the expected angular separation of the components of the binary being observed plus the recommended minimum size (for a point source) of 0.3", or,
where the value x is the largest anticipated angular separation of the binary along either axis.- Maximum Scan Length: When considering the maximum scan length, three concerns should be addressed:
- Masking by the field stops becomes relevant when the FOV is moved beyond 2 arcsec from the photocenter. False interferometric features become prevalent.
- Beyond 2.5 arcsec from the photocenter, the intensity of the target's light drops off considerably, resulting in poor signal-to-noise photometry.
- Maximum commandable scan length is 6.7".
- Target Orientation: If the approximate position angle of the non-point source is known, then specifying an orientation for the observation may be advantageous. If possible, (for wide binaries) avoid the situation where the projected angular separation along one of the axes is less than 20 mas. Specific orientations are achieved by rolling the HST to an off-nominal roll attitude or by scheduling an observation at a time when the nominal roll is suitable. It should be noted however that special orientations are considered as special scheduling requirements which affect schedulability.
- Target Field: The Search and CoarseTrack acquisition are vulnerable to acquiring the wrong object if the field is too crowded (with neighbors of comparable brightness within 8 arcseconds of the target). Hence, the problem addressed in Section 4.2.7 for Position mode observations applies equally to Transfer mode observations.
4.5.2 Transfer Mode Filter and Color Effects
Table 4.6 is a summary of the available filters and associated restrictions governing their use.
Table 4.6: FGS1r Transfer Mode Filters to be Calibrated During Cycle 8
The S-Curve morphology and modulation have a wavelength dependence. Experience with FGS3 has shown that the color of the reference star should be within
4.5.3 Signal-to-Noise
In essence, the "true" signal in a Transfer mode observation of a binary system is the degree to which the observed Transfer Function differs from the S-curve of a point source. The signal-to-noise (S/N) required of an observation will depend upon the object being observed; a wide binary whose stars have a small magnitude difference and separation of 200 mas will be much easier to resolve than a pair with a larger magnitude difference and a separation of only 15 mas.
The "noise" in an observation has contributions from both statistical and systematic sources. Photon noise, uncertainty of the background levels, and spacecraft jitter comprise the statistical component. The temporal variability and spectral response of the S-curves dominate the systematic component (these are monitored and/or calibrated by STScI). Provided that at least 15 scans with a 1 mas step size are available, observations of bright stars (V < 13.0) suffer little from photon noise and uncertain background levels, and show only slight degradations from spacecraft jitter (with high S/N photometry, the segments of the data which are degraded by jitter are easily identified and removed from further consideration).
Maximizing the S/N for observations of fainter objects requires a measurement of the background level (see Chapter 6) and a larger number of scans to suppress the Possonian noise in the photometry of the co-added product. But with lower S/N photometry in a given scan, corruption from spacecraft jitter becomes more difficult to identify and eliminate. Therefore, the quality of Transfer mode observations of targets fainter than V = 14.5 will become increasingly vulnerable to spacecraft jitter, no matter how many scans are executed.
Systematic "noise" cannot be mitigated by adjusting the observation's parameters (i.e., increasing the number of scans). To help evaluate the reliability of a measurement made in Transfer mode, STScI monitors the temporal stability and spectral response (in B-V) of FGS1r's interferograms. As discussed elsewhere, the FGS1r S-curves appear to be temporally stable to better than 1%, and the Cycle 10 calibration plan calls for observations of single stars of appropriate B-V to support the data reduction needs of the GOs (this calibration will be maintained in Cycle 14). This should minimize the loss of sensitivity due to systematic effects.
4.5.4 Transfer Mode Exposure Time Calculations
The step_size and number of scans determine the number of photometric measurements available for co-addition at any given location along the scan path. Typically, up to 50 scans with 1mas step_size are possible within a 53 minute observing window, (after accounting for overheads and assuming a scan length ~ 1.2 arcsec per axis). The step size and number of scans that should be specified are in part determined by the target's magnitude and angular extent and also by the need to allocate time within the visit to any other objectives, such as Position mode observations of reference stars (to derive a parallax for the binary). The total exposure time for a Transfer mode observation (excluding overheads) is:
where Texp is the total exposure time in seconds, Nscans is the total number of scans, 0.025 is the seconds per step, ScanLength is the length of the scan per axis in arcsec, and StepSize is given in arcsec.
Photon noise is reduced by increasing the number of scans, Nscans, as displayed in Figure 4.4, which demonstrates the benefits of binning and co-adding individual scans. Trade-offs between step size, length, and total duration of an exposure are unavoidable especially when considering visit-level effects such as HST jitter.
Figure 4.4: FGS1r (F583W) S-Curves: Single and Co-Added
Simulations using actual data scaled by target magnitude are needed to relate the Transfer Function signal-to-noise (described in the previous section) to the resolving performance of the instrument. A robust exposure time algorithm is in development. In Table 4.7, we offer some guidelines on the minimum number of scans to use in a visit for various binary parameters. These are derived for a step size = 1.0 mas, so that in a 1 mas bin there would be NSCANS samples per bin. Smaller step sizes facilitate the use of less scans to achieve the same signal-to-noise ratio. (Specificity fewer scans with smaller step sizes can reduce the observational overhead, which can increase the time on target and hence the overall signal-to-noise ratio. However, intermittent vehicle jitter may corrupt the data from some scans to the degree that such data is useless for scientific purposes. These trade off need to be considered when planning the observations).
Table 4.7: Suggested Minimum Number of Scans for Separations < 15 mas
1Note that 60 scans is about the maximum that can be performed in a single HST orbit (assuming a scan length of ~1"). Multi-orbit visits do not necessarily increase the achievable S/N for targets of V>15 since photometric noise makes cross correlation of scans across orbital boundaries questionable. In other words, the data gathered during one orbit is not reliably combined with data from another orbit for faint, close binary systems.
4.6 Transfer Mode Observing Strategies
4.6.1 Summary of Transfer Mode Error Sources
The Transfer mode corrections and calibrations are:
Each of the corrections and calibrations are discussed briefly in Chapter 5 and more thoroughly in the
HST Data Handbook. Those corrections that could result in an enhanced observation strategy are discussed here.4.6.2 Drift Correction
As discussed in the Position mode section, targets observed multiple times per Position mode visit typically drift across the FGS by about 6 to 12 mas when two FGSs guide the telescope. This drift is also apparent in Transfer mode observations, but the cross-correlation of S-Curves prior to binning and co-adding automatically removes the drift. Each single-scan S-Curve is shifted so that the particular features of the S-Curve used for the cross correlation coincides with that of the fiducial S-Curve. The reliability of implicitly removing the drift is only as good as the accuracy of the cross correlation procedure, which, for bright objects (V < 14.5) is accurate to < 1 mas. Analysis is underway to determine the procedure's accuracy for fainter objects.
4.6.3 Temporal Variability of the S-Curve
Measurements of the standard star Upgren69 over the lifetime of FGS3 indicated a 10 - 18% variability of the S-Curve morphology on orbital timescales. The amplitude of these changes have important consequences on the analysis of binary star observations when the separation of the components is less than 30 mas and the magnitude difference exceeds 1.6. These temporal changes also affect analyses of extended source observations. The cause of this relatively high frequency variability in FGS3 has not been determined.
FGS1r appears to be stable at the 1% level over periods of many months to perhaps years (as of July 2002). There appears to have been a slow evolution of the y-axis S-curve however, as shown in Figure 4.5. The changes along the X-axis have been much less. On the assumption that this evolution is due to changes in the alignment of the interferometer with respect to HST' OTA as water vapor outgasses from the instrument's graphite epoxy composites, it is expected that FGS1r will become more stable as time goes on (the rate of outgassing slows with time in orbit). STScI will continue to monitor FGS1r's S-Curves so that this evolution can be calibrated and its effect on science data minimized.
4.6.4 Background and Dark Counts Subtraction
For programs with isolated targets, background is not an issue. For programs with targets embedded in nebulosity, knowledge of the background is required. In order to obtain a background measurement, it is necessary that there be at least two targets in the observing sequence, separated by at least 60. The background will be measured during the slew of the IFOV from one target to the next. If necessary, the proposer should specify a false target at some location in the FOV at least 60 from the science object, and observe it in Position mode for approximately 30 seconds with the same filter as the science target. Care must be taken to avoid observing "background: objects brighter than V = 8.0 with F583W.
Dark counts become important for stars fainter than V=14.5. STScI has calibrated the FGS1r dark counts. Therefore there is no need to acquire such data as part of a science observing program.
Figure 4.5: Evolution of S-curve morphology along the FGS1r Y-axis
4.6.5 Empirical Roll Angle Determination
The science data headers contain the commanded HST roll angle, not a measured angle. The errors that contribute to a difference between the commanded and actual roll include: the relative guide star positional error, the FGS-FGS alignment error, and errors in the predicted ephemeris. The actual roll angle is calculated from the guide star telemetry by the observatory monitoring system, and is reported in the STScI Observation Logs that accompany the science data. More information is available on the following web page:
http://www.stsci.edu/instruments/observatory/The error in the calculated roll angle is estimated to be about 0.04 degrees. If a more accurate determination is needed, the position angle of the observed binary with respect to the local reference frame can be measured via Position mode observations of the target and a reference star (or two reference stars if the target cannot be acquired in Position mode-see Section 4.2.5) should be included in the visit along with the Transfer mode exposures. Please confer with STScI for help designing the visit and calculating the roll from Position mode measurements.
4.6.6 Exposure Strategies for Special Cases: Moving Targets
For Transfer mode observing, a moving target represents a special case. The flight software which enables HST to track a moving target has not been implemented for FGS observations. Nonetheless, the FGS is quite capable of acquiring a moving target provided that the object's angular speed is less than 80 mas/sec. However, during the observation, the scan path is not adjusted to accommodate for the object's motion. The target's fringe will be displaced in each subsequent scan until it moves completely out of the scan path. A method to work around this problem is to specify several observations of the object during the visit. For example:
- The target should be observed in several short exposures rather than in one or two long exposures. The FGS would re-acquire the target with each new Search, CoarseTrack acquisition regardless of the target's motion. The target list should contain enough entries to cover the swath of sky traversed by the moving target, e.g., if the motion takes the object 10 arcsec across the FOV, then at least two sets of target coordinates should be specified.
- The number of individual observations (entries in the exposure logsheet) and number of scans in an observation will be dictated by the object's angular speed.
- Plan the observation when the target is moving its slowest, e.g., at opposition (if possible).
- Take advantage of rolling the telescope to adjust the angle between the target motion vector and relevant FGS reference directions (please contact STScI for assistance).
- Choose the exposure times to be as short as possible.
|
|||||
|
Space Telescope Science Institute http://www.stsci.edu Voice: (410) 338-1082 help@stsci.edu |
