On-orbit performance of the Hubble SpaceTelescope fine guidance sensors
David J. Eaton, Richard A. Whittlesey, Bruce W. Allen, Roy Stoll,Linda Abramowicz-Reed and Marcel Margulies
The observed and measured on-orbit performance of various aspects of the fine guidance sensors ispresented and discussed in the light of the original requirements and predictions. The fine guidancesensors are shown to meet or exceed the original requirements concerning dynamic pointing errors,photometric repeatability, and moving-target tracking capability. Calibration accuracy has been suffi-cient for observations to date, and fine-lock acquisitions are approaching a 100% success rate.Improvements to the fine-guidance-sensor tolerance of telescope spherical aberration, the South Atlanticanomaly, and solar-panel vibrations have been made, and further improvements are expected.
Key words: Hubble Space Telescope, fine guidance sensors, on-orbit performance.
1. IntroductionThe three fine guidance sensors of the Hubble SpaceTelescope (HST) have two principal functions: (1)providing an absolute pointing reference to the tele-scope-pointing control system and (2) serving as anastrometry instrument. Some of the most impor-tant fine-guidance-sensor design requirements in-clude the following:
(1) A star-brightness range of 9-14.5 visual mag-nitude (Mv) and 10-17 Mv for astrometry.
(2) A fine-lock dynamic pointing error of 2.8 marc-sec (object space) rms over a 5-HZ bandwidth at 13mV.
(3) A calibration accuracy of 3 marcsec.(4) Moving-target tracking velocities of 0.21
arcsec/s and accelerations up to 0.1 arcsec/s 2 withoutthe rate feedforward.
(5) In astrometry the photometric measurementrepeatability of 1% standard deviation in < 60-s inte-gration time.
(6) Digital control electronics with a versatileuplinkable parameter set.
D. J. Eaton, R. A. Whittlesey, B. W. Allen, R. Stoll, and L.Abramowicz-Reed are with Hughes Danbury Optical Systems, 100Wooster Heights Road, Danbury, Connecticut 06810. M. Margu-lies is with Perkin-Elmer Corporation, 761 Main Avenue, Norwalk,Connecticut 06859.
This paper presents and discusses fine-guidance-sensor performance with respect to these require-ments. In Section 2 we describe the design of thefine guidance sensor, while Section 3 presents theon-orbit performance. A brief summary follows inSection 4.
2. Design of the Fine Guidance SensorsThe above requirements placed stringent demands onthe fine-guidance-sensor subsystems, specifically theoptical, servo, mechanical, and thermal control sub-systems. The design of the fine guidance sensor isdescribed in Refs. 1 and 2 and summarized here.
A. Optical/Mechanical DesignIn the HST there are four radial bay modules, threecontaining a fine guidance sensor (Figs. 1 and 2).Light from the outer portion of the HST field of view,10-14 arcmin from the optical axis, strikes the fine-guidance-sensor pickoff mirror and is collimated anddirected through apertures in a pair of star-selectorservos. The beam then passes through beam split-ters and Koester prisms before reaching four photo-multipler tubes (PMT's). The angular positions ofthe servos determine which 5 arcsec x 5 arcsecportion (instantaneous field of view) of the fine-guidance-sensor field of view enters the field stops ofthe PMT's.
One key to a successful fine-guidance-sensor designwas producing a thermally stable structure thatdistorts minimally from temperature variationscaused by changes in spacecraft orientation withrespect to the sun or by day-night transitions. This
Fig. 1. Optical schematic of the fine guidance sensor.
requirement necessitated the use of a thermallystable optical bench, exotic athermalized flexures,and complex mirror mounts that provided kinematicsupports.
To ensure that the optical bench is a thermallystable assembly, we made it of graphite-epoxy lami-nates that have a very low coefficient of thermalexpansion, 0.1 ppm/0 C. An oven consisting of graph-ite-epoxy plates covered with strip heaters was builtto surround the bench and hold the bench tempera-ture variation to ±0.1 C. Other components, suchas electronics, servo motors, and encoders, wereisolated from the bench, and their temperatures wereactively controlled. Finally the oven and the entirefine-guidance-sensor assembly were enclosed in insu-lating blankets.
OPTICAL CONTROLSYSTEM OVEN
Fig. 2. Fine-guidance-system radial bay assembly diagram. GR,graphite.
B. Fine-Guidance-Sensor Control and Modes of Operation
The fine guidance electronics (FGE), an electronicsunit that includes two microprocessors, controls thefine guidance sensor. The FGE receives commands,accepts pulse trains from the four PMT's, and re-ceives data from the servo encoders. FGE algo-rithms define a set of complex operating modes.Among the FGE commands are those that set thevalues of 38 uplink parameters, which are used toadapt the system to specific mission requirements orspecial operating conditions. The FGE also outputsdata, such as PMT counts, servo encoder positions,and status bits.
The fine-guidance-sensor modes relating to acquisi-tion, pointing, and tracking are search, coarse track,and fine lock. In search the servos define a spiralscan of a region of the fine-guidance-sensor field ofview until the target star image enters the PMT fieldstops. When the summed output (counts) of thefour PMT's exceeds a commanded threshold, the fineguidance sensor autonomously terminates the searchmode and enters coarse track. In coarse track theFGE commands the servos to move the images of thetarget star in circular paths (called nutation circles)about the PMT field stops. The coarse-track algo-rithm monitors the total output of the four PMT'sand adjusts the centers of the nutation circles untilthe circles are centered on the field stops.
Approximately 70% of all observations are donewith two fine guidance sensors operating in thecoarse-track mode. The remaining observations re-quire the lower jitter provided by fine lock. Also, forus to achieve the positional accuracy required byastrometry observations, fine lock is necessary.While the search and coarse-track modes use the sumof the outputs from all four PMT's as feedback, finelock uses the PMT's as part of a unique Koester prism
A. Star Brightness RangeBecause of the spherical aberration in the telescope, itis necessary to operate the fine guidance sensors withthe two-thirds aperture stop in place. However,with the aperture stop, the number of photons reach-ing the PMT's for a given star is reduced, and thelimiting star magnitude is consequently 1 Mv brighterthan originally expected. Figure 3, in which theerror signal is plotted as a function of the truepointing error, illustrates the effects of sphericalaberration and the aperture stop on the error signalcomputed by the FGE in the fine-lock mode. Thiserror characteristic is frequently termed the s-curve.The figure shows that the aperture stop can recovermuch of the s-curve modulation that was lost tospherical aberration.
The fine guidance sensors must also be restrictedagainst operating in the South Atlantic anomaly(SAA), where a charged-particle flux increases thebackground noise. In the coarse-track mode thegain of the computed error is also reduced. Thisgain reduction can reduce the stability margins of theHST pointing control system to the point wherelow-frequency oscillations of the entire spacecraft areobserved. In the fine-lock mode the SAA also re-duces the gain of the error; this gain reductiondegrades the ability of the fine guidance senor totrack the pointingjitter. Furthermore, if the vehicleis oriented so that one PMT is shielded from the SAA
IDEAL INTERFEROMETER S-CURVE
NORMALIZEDINTERFEROMETER
OUTPUT
A I I I I
ZA~~~ I
X (MILLIARC-SECONDS)
S-CURVE MEASURED ON-ORBITWITHOUT APERTURE STOP
X (MILLIARC-SECONDS)
S-CURVE MEASURED ON-ORBITWITH APERATURE STOP
X (MILLIARC-SECONDS)
Fig. 3. Use of the fine-guidance-sensor 2/3 aperture stop torecover the s-curve modulation lost to spherical aberration.
flux, a large bias is imparted to the fine-lock errorsignal. This bias has led to a false lock duringfine-lock acquisitions (transitions from the coarse-track mode to the closed-loop tracking in fine lock).Algorithms for future fine-guidance-sensor designshave been demonstrated to mitigate the effect of theSAA.
To acquire a star in fine lock, the fine-guidance-sensor line of sight, as determined by the servopositions, is offset from the center of the coarse-tracknutation circle, the center being the best estimate ofthe star position in coarse track. The servos arethen stepped along a straight path toward the esti-mated star position (Fig. 4). When the servo posi-tions are within 0.040 arcsec of the interferometernull along an axis, the error signal increases above apreset threshold, and the FGE reduces the speed ofthe walk toward the star. For each axis, if the errorsignal exceeds the threshold for three consecutivesamples, that axis is locked; that is, closed-loopoperation of the servos is initiated with the interfero-metric error signal as a position feedback. Whenboth axes are locked, the fine-lock acquisition iscomplete, and the servo loop acts to maintain thepointing error at null.
Based on data taken over the period from day 358to day 364, 1990, the percentage of successful fine-lock acquisitions was 99% in the guidance mode witha limiting magnitude of 13.5. In the astrometrymode the success rate was 100%, 93%, and 93% forfine guidance sensors 1, 2, and 3, respectively. Allfine guidance sensors have locked onto stars of 14.9magnitude in astrometric observations.
The probability of successful fine-lock acquisitionhas been improving as a result of fine tuning theappropriate fine-lock uplink parameters (also knownas K factors. Figures 4 and 5 define the uplinkparameters that are relevant to fine-lock acquisition.
B. Pointing PerformanceThe dynamic pointing error refers to pointing errorsthat vary over an observation period and degrade thepoint spread function of the image. We evaluatedindirectly the on-orbit fine-guidance-sensor dynamic
KB, KD, K10 -DETERMINE STARTING POINT, STEP SIZE, ANDSCAN DIRECTION DURING ACQUISITION (FIGURE 4
Fig. 5. Uplink parameters relevant to fine-lock acquisition.
pointing errors in fine lock by measuring the timevariation in the pointing of fine guidance sensors usedfor guidance during quiet periods, i.e., periods unaf-fected by terminator/solar array-induced spacecraftjitter. This time variation in pointing, which wascalculated from records of PMT counts and starselector servo encoders, was compared with the origi-nal HST pointing requirement of 0.007 arcsec rms.This is a conservative comparison, because the mea-sured variation represents the total of HST andfine-guidance pointing errors.
Two examples of short-term HST dynamic-pointing-error measurements are cited here.3 4 In Ref. 3 thepointing error was found to be near or below 0.007arcsec rms and not strongly dependent on star magni-tude. In a second study4 the variation in fine-guidance-sensor pointing over 20 1-min quiet Au-thor queried intervals was measured. Differentguide-star pairs, ranging in brightness from 9 to 12Mv, were used in each interval. The HST dynamicpointing error was found to be 0.003-0.004 arcsecrms.
The above studies show that the short-term point-ing error of the HST is below the 0.007-arcsecrequirement during quiescent periods with the fineguidance sensors in fine lock. This result indicatesthat the fine-guidance-sensor contribution to theshort-term dynamic pointing error meets require-ments.
For long-duration observations, of up to 24 h,temperature-induced deformations of the internalfine-guidance-sensor components contribute to thedynamic pointing error. As discussed below a directverification of the long-term pointing performancehas not been possible; however, it has been evaluatedindirectly by measuring the temperatures of thosefine-guidance-sensor components that affect pointingstability. To demonstrate that the pointing stabilityachieved on orbit meets requirements, we comparedthe temperature changes during 24 h following a
worst-case slew of the vehicle, defined in Fig. 6,compared with ground (thermal/vacuum) test resultsthat were analyzed and shown to be acceptable. Thedata are presented in Table 1, which shows that allon-orbit temperature changes were equal to or lessthan the ground test temperature changes. In Table1 OV refers to orbital verification, the period of theinitial HST operation and evaluation following launch,while T/V refers to the thermal/vacuum groundtests.
The T/V ground testing and its analysis, describedin Ref. 2, are summarized as follows. Before theywere installed into the HST, the fine guidance sensorswere tested in a vacuum chamber that was equippedwith a star simulator, heater shrouds, and liquidnitrogen-cooled shrouds. The shrouds were used tosimulate the thermal environment that would be seenby the fine guidance sensors on orbit. With the fineguidance sensors in fine lock on the star simulator,the error outputs of the Koester prism interferome-ters and the motion of the star selector servos mea-sured pointing changes in response to variations inthe thermal environment. Internal Fine-Guidance-Sensor temperatures were also measured.
The T/V temperature changes shown in Table 1were found to correspond to acceptable pointingchanges. Therefore, since the worst on-orbit changein the thermal environment resulted in internaltemperature changes smaller than those in the accept-able ground test, we conclude that the long-term(24-h) pointing stability of the fine guidance sensorsmeets requirements.
V3 AXIS
k SUNVI AXIS
BEFORE SLEW
SUN
AFTER SLEW
Fig. 6. Definition of the IST slew preceding the on-orbit temper-ature changes given in Table 1.
HV/pad shelf 1.39 0.18 1.61 0.19 1.72 0.55Interferometer 0.17 0.18 0.39 0.19 0.39 0.0Pickoff arm 0.61 0.37 0.39 0.19 0.39 0.18Bench below image dissector electronics 0.61 0.18 0.50 0.0 0.50 0.19Bench near motor encoder A 0.39 0.0 0.39 0.0 0.39 0.37Motor encoder housing 0.0 0.0 0.0 0.0 0.0 0.0PMT 0.39 0.0 0.22 0.19 0.22 0.0Keel near D Fitting 0.17 0.0 0.39 0.0 0.39 0.18Graphite epoxy oven near the optical control
system 0.39 0.0 0.22 0.18 0.22 0.0
a(Day) 169: (hour) 11:51 to 170:11:51 (1985).b1 2 7 :1 7 :4 8 to 128:17:48 (1990).c199:18:10 to 200:18:10 (1985).d228 :10:35 to 229:10:35 (1985).
It has not been possible to verify directly that thefine guidance sensors meet the long-term dynamicpointing error requirement because of frequent lossesof fine lock induced by solar-array vibrations as thetelescope passed through day-night transitions. Asecondary disruption to long-term pointing has beenrandom bit changes in the FGE semiconductor mem-ory caused by charged particles in the SAA. Thesesingle-event upsets resulted in improper functionalperformance of the FGE while and after the HSTpassed through the SAA. We nearly eliminated thisproblem by refreshing the uplink parameters at 5-sintervals. Future fine-guidance-sensor designs willeliminate the use of components that are susceptibleto single-event upsets.
C. Calibration AccuracyThe three fine guidance sensors were calibrated formagnification, optical field angle distortion, and rela-tive alignments of the three fine guidance sensors.Interim calibrations comparing fine guidance sensormeasurements with ground-based information re-sulted in distortion, magnification, and alignmentknowledge to the 30-marcsec rms level. While there
TRACKINGVELOCITY
ARCSEC/SECONC1.0
0.8
0.0
0.4
0.2
0.0
0.2
0.4
0.0
0.0
1.0
FINE LOCK FINE GUIDANCE SENSORLINE OF SIGHT RATE VS. TIME
21:26:12.0 21:26:19.5 21:26:27.0
FINEGUIDANCESENSORNO. 3
STARTINGDATE OFPLOTTEDDATA:1990.222
TIME (HOURS: MINUTES: SECONDS)
Fig. 7. Fine-guidance-sensor fine-lock tracking performancethrough a day-night transition. The velocities and accelerationsexceed design requirements.
have been significant changes in the relative align-ments of the sensors,5 the calibrations have substan-tially improved the accuracy with which targets canbe placed in the apertures of science instrumentsincluding any fine guidance sensor used for astrome-try, and the present calibration accuracy has beenentirely sufficient for all observations to date. Therepeatability of the calibrations is being investigated.
After the planned installation of optics correctingthe telescope's spherical aberration, there will be aneed to improve the HST calibration accuracy to 10marcsec of which 5 marcsec are allocated to each fineguidance sensor, so that targets can be placed in thesmall aperture of the high-resolution spectrograph.The improved accuracy may also be needed for thefaint-object spectrograph. Since the long-term stabil-
Fine On-Orbit On-Orbit On-OrbitGuidance Days 180-185 Days 264-265 Day 315 Percent Change fromSensor Filter Ground Test 1990 1990 1991 1990.264-5 to 1991.315
aThe numbers in the table represent the total (sum of four) PMT counts over 0.025 s for a visual magnitude of 13. The factor 0 -04(Mv-13)scales the calibration to a given visual magnitude.
bAperture stop.
ity of the calibration solutions is not sufficient toachieve 5 marcsec, monthly tests are planned toupdate the alignment between fine guidance sensorsand the distortion in fine guidance sensor 3, which isused for astrometry.
Fine-guidance-sensor calibration is discussed inmore detail in Ref. 5.
D. Moving-Target TrackingWhether used for guidance or astrometry, the fineguidance sensors must have the ability to track starsthat change position in the fine-guidance-sensor fieldsof view because of pointing changes of the HST.
When the HST passes through a day-night transi-tion with the fine guidance sensors in fine lock, one ormore of the fine guidance sensors may lose lock.However, the cases in which the fine guidance sensorsmaintained fine lock through a day-night transitionshowed that they have the ability to track stars atvelocities and accelerations exceeding the originalrequirements. For example, Fig. 7 is a plot of fine-guidance-sensor pointing velocities, as determinedfrom a history of star-selector servo angles, for aperiod following a day-night transition. The fineguidance sensor did not lose lock in this period.Note that the maximum tracking velocities and accel-erations were significantly higher than the designrequirements.
Simulations have indicated that tracking perfor-mance can be improved by the adjustment of uplinkparameters. For example, the integral gain parame-ter K14 in the FGE fine-lock control law was shownby simulation to improve the tracking of bright starsin the presence of low-frequency (0.6-Hz) spacecraftoscillations observed during and following day-nighttransitions. However, solar-array jitter sometimesconsists of amplitudes and/or frequencies that can-not be tracked even with increased FGE control lawgains; thus the effectiveness of uplink parameteradjustment is limited.
More promising than FGE control law adjustmentare modifications to the HST pointing control systemsoftware; these modifications have significantly re-duced the losses of fine lock from day-night transi-tions. For the long term, replacement of the solar
arrays during the future HST repair mission isexpected to reduce or eliminate HST jitter fromterminator crossings.
E. Photometric RepeatabilityOver time intervals that are of the order of a few dayslong, the repeatability of flux measurements from aset of stars (10-13 Mv) has demonstrated a betterthan 1% standard deviation for measurements aver-aged over 30 s (Table 2). For this brightness range,photometric repeatability is therefore better than therequirement of 1% standard deviation with 60-saveraging. The repeatability for stars dimmer than13 Mv, however, has not yet been determined.
The PMT's in each fine guidance sensor have beencalibrated based on observations of well-documentedcalibration stars. Table 3 shows the results of cali-brations done at several times, including beforelaunch. The differences between before-launch andon-orbit calibrations are sufficiently small to be causedby calibration errors in the reference ground testsource. Although the original design requirementsdid not address long-term (several months) photomet-ric repeatability, the calibration changes on orbitindicate that periodic recalibration is desirable formaintaining the calibrations over periods of monthsor years. However, as Table 3 shows, the changesfrom 1990.264-265 to 1991.315 were within 1%;these data suggest that the rate of change of thecalibrations, and thus the need for recalibration, isdecreasing.
4. Summary and ConclusionsIn general the HST fine guidance sensors have beenperforming quite well with fine-lock acquisitions ap-proaching a 100% success rate, dynamic pointingerrors meeting requirements, photometric repeatabil-ity meeting requirements, and moving target-track-ing capability better than the original requirement.Calibration accuracy has been sufficient for observa-tions to date, and further calibration is expected toimprove the accuracy to meet the demands of futureobservations.
Telescope spherical aberration has reduced by 1magnitude the brightness range of stars that can be
acquired by the fine guidance sensors, and vehiclejitter induced by solar-array vibrations occasionallycauses losses of fine lock at day-night transitions.Modifications to the HST pointing control softwarehave reduced the solar-array jitter and the associatedlosses of lock. The future HST repair mission isexpected to reduce further or eliminate the jitter.
The work reported here was sponsored by NASAMarshall Space Flight Center, Huntsville, Ala., undercontract NAS-8-32700 and by NASA Goddard SpaceFlight Center, Greenbelt, Md., under contract NAS-8-38494.
A large number of individuals from Hughes Dan-bury Optical Systems, NASA, Lockheed Missiles andSpace Company, and other organizations contributedto the design, fabrication, calibration, testing, andoperation of the HST fine guidance sensors. Theauthors acknowledge the extensive work of thesecontributors and thank the organizations involvedfor the privilege of presenting this paper. We alsothank Geralyn Fischer for her assistance in preparingthe manuscript.
References1. G. S. Nurre, S. J. Anhouse, and S. N. Gullapalli, "Hubble Space
Telescope fine guidance sensor control system," in Acquisition,Tracking and Pointing III, S. Gowrinathan, ed., Proc. Soc.Photo-Opt. Instrum. Eng. 1111, 327-343 (1989).
2. R. Stoll and L. Williams, "Thermal design and testing of theHubble Space Telescope fine guidance system and wavefrontsensor," in Proceedings of the Fourth Joint Thermophysics andHeat Transfer Conference of the American Institute of Aeronau-tics and Astronautics and American Society of MechanicalEngineers, publ. AIAA-86-1338 (American Institute of Aeronau-tics and Astronautics, Boston, Mass., 1986).
3. C. J. Burrows, J. A. Holtzman, S. M. Faber, P. Y. Bely, H.Hasan, C. R. Lynds, and D. Schroeder, "The imaging perfor-mance of the Hubble Space Telescope," Astrophys. J. 369,L23-L24 (1991).
4. G. Andersen, Jackson and Tull, Inc., Seabrook, Md. 20706(personal communication, 1992).
5. D. J. Eaton and L. Abramowicz-Reed," Acquisition, pointingand tracking performance of the Hubble Space Telescope fineguidance sensors," in Acquisition, Tracking and Pointing VI,M. Masten, ed., Proc. Soc. Photo-Opt. Instrum. Eng. 1697,236-250 (1992).
