Fig. 1. The IUE Ground Track.
Frederick Gordon, Goddard Space Flight Center, Greenbelt, MD
Michael C. W. Sandford, Science and Engineering Research Council, London, U.K.
The receiving site for NASA's ground system was, until April 1986, located at the Goddard Space Flight Center (GSFC), Greenbelt, Maryland. Subsequently it has been removed to NASA's facility at Wallops Island, Virginia (WPS). NASA's spacecraft and science operations facilities have always been located at GSFC. Commands and received data are now transmitted between GSFC and WPS by a commercial communications satellite link.
NASA has three principal areas of responsibility in operations: first, overall responsibility for monitoring and maintaining the health of the spacecraft; second, providing a backup system for the purpose of spacecraft safety during the shift controlled by ESA's ground station; and third, operating IUE for two 8 hour shifts each day, designated US1 and US2. The spacecraft-related tasks are carried out on a 24 hour per day basis by the GSFC IUE Operations Control Center (IUEOCC). The scientific operations are carried out by the IUE Science Operations Center (IUESOC), also at GSFC, in conjunction with the IUEOCC. The IUESOC comprises a Telescope Operations Control Center (TOCC) from which astronomical observations are controlled during the US1 and US2 operational shifts. An Image Processing Center (IPC), located near GSFC, carries out the standard processing of the IUE data.
All the elements of ESA's ground segment are located at Villafranca del Castillo near Madrid (VILSPA). ESA provides spacecraft control and science operations for one shift, designated VILSPA, and in a second shift the standard data processing is carried out. Because the ESA station has limited backup facilities, e.g. only one computer capable of controlling IUE, GSFC maintains readiness to take control and keep the spacecraft safe in the event of a failure in the VILSPA station.
The operational concept of sharing, internationally, the responsibility for operations was made possible by the choice of a geosynchronous orbit for IUE. Constraints on the orbital parameters arose since continuous viewing of IUE is required from the NASA receiving/transmission site and for at least one shift per day IUE must be in view from VILSPA. To insure reliable communications, IUE must be at least 10° above the local horizon at VILSPA for more than about 10 hours per day, which is the time required for an 8 hour shift, a shift handover of about 30 minutes, and a monthly adjustment by 2 hours in the shift starting time. The latter arises since the orbit is fixed in sidereal time but for the convenience of the operations staff the shift start remains at a fixed Universal Time during each month.
An elliptical geosynchronous orbit was chosen with a period of 23 hours 56 minutes, eccentricity 0.2, inclined at 28° to the equatorial plane and with the ground track initially centered over the Atlantic at about 70° west longitude. With this positioning the VILSPA viewing time usually averages 12 hours. As the orbital plane precesses westward due to natural perturbations the viewing time falls and when the 10 hour minimum is reached an orbit adjustment must be made. A maximum of 15 hours may be achieved at its most easterly position. Partly in consequence of these orbit adjustments, and partly as a result of other natural perturbations the ground track of the orbit has gradually evolved during life of the mission as shown in Figure 1. For example, as of late June 1994, the eccentricity has dropped to 0.13, while the inclination has risen to 35°.
Fig. 1. The IUE Ground Track.
An early project objective was that the IUE system would be an international research facility, available to a wide community of astronomers, and be organized much in the way that many ground based observatories are operated. The operations management plan and the ground system were specifically oriented towards achieving this. IUE has been operational since its launch, 26 January 1978, for over 16 years at the time of writing. It continues to provide excellent facilities to the scientific community. It can be unequivocally stated that all scientific objectives of the mission have been met or exceeded. The mission's success was made possible not only by the outstanding performance of the spacecraft and its scientific instrumentation, but also by the excellent cooperation of the technical and scientific staff of the participating agencies. These achievements have been widely acknowledged in the general scientific community as being unique in space exploration.
Three-axis stabilization of IUE was initially accomplished by an Inertial Reference Assembly (IRA) composed of 6 gas bearing single-degree-of-freedom gyroscopes operated with pulse rebalance electronics. The ACS was designed to hold a 1 arc second diameter star image within a 3 arc second entrance aperture to permit an integrating exposure of at least 1 hour by the spectrograph camera. Now, under the two-gyro/Fine Sun Sensor (FSS) control system, the FSS is also used as a sensor for spacecraft control. The ACS uses the outputs of the attitude control sensors (gyros, FES, and/or FSS) as inputs to the control program in the On Board Computer (OBC). The OBC then controls the spacecraft attitude and slews by changing the rotation speed of the pitch, yaw, and roll momentum wheels.
The primary power for IUE is derived from the two solar cell arrays which are shown in the general layout, Figure 2. The power output is a function of the angle between the Sun and the satellite pointing direction (+x axis). The supplement of this angle is designated Beta. In normal operations the satellite is rolled about the x-axis to keep the Sun close to the x-z plane. Thus the single angle Beta is useful to define the approximate orientation of the satellite with respect to the Sun, and will be so used here.
Fig. 2. IUE Configuration.
Central to the spacecraft control are the command system and the data multiplexer unit (DMU). The Command Decoder receives the commands, checks for errors, and then routes valid commands or OBC data block loads to the correct subsystem. The DMU samples the performance data from the spacecraft subsystems and generates two telemetry streams: one is dedicated to the OBC and its control of the spacecraft; the other is the ground telemetry stream, which includes both the science data and the spacecraft `housekeeping' data. The OBC also uses the ground telemetry stream.
The final key spacecraft element is the communications system, which warrants description in some detail. Two wavebands, S-band and VHF, are used for communication between IUE and the ground. The normal data telemetry link from IUE to the ground receiving site is by S-band (2249.8 MHz). This telemetry can be transmitted at several different bit rates: 40, 20, 10, 5, 2.5 and 1.25 kilobits per second (kbs). There is a convolved, half rate, data mode used to increase accuracy. (The 40 and 2.5 kbs telemetry rates are not used operationally because there is either a timing conflict in the generation by the DMU of the separate ground and telemetry data streams, or the OBC has a timing conflict in reading the two data streams.) There are two telemetry streams, one special for the OBC and the other a ground telemetry stream that is also read and used by the OBC. The dedicated OBC telemetry rate is normally held at 20 kbs for smooth spacecraft control. The telemetry rates are selected by the operations controller. Both spacecraft housekeeping data and science data come down in this S-band telemetry. There are four S-band power amplifier/antenna combinations distributed around the spacecraft, only one of which can be switched on at a time. Two are at the bottom on the sun- and the anti-sun-side respectively, and two are on the satellite upper body in similar locations. As a result, no matter what the attitude orientation relative to a ground station, there is at least one S-band antenna that can be used. There are two VHF transmitters (for redundancy) that operate at 136.86 MHz for the transmission of satellite data. A VHF transmitter is used during range and range-rate operations to determine IUE's location and also under conditions when S-band is not available, such as: spacecraft to ground S-band data link problems; during eclipses because it uses less power than the S-band; and during spacecraft emergency operations. The maximum data rate of the VHF system is 5 kbs. There are two VHF command receivers (148.98 MHz) and decoders for redundancy on board IUE. The command bit rate is 0.8 kbs. All operations and control are achieved through this command system.
At GSFC the command of IUE is initiated either in the TOCC, or in the IUEOCC. VILSPA also divides control between a main control room and the observatory room. Operations of an observatory-type satellite are very complex and are achieved for IUE by calling up a set of pre-coded and extensively tested procedures, each carrying out a particular sequence of operations, such as camera preparation, exposure and reading. The procedures are run on the control centers' computers and can check the status of the satellite via the telemetry before selecting the appropriate command and issuing it at the appropriate time. A medium level and user-tolerant language known as Control Center Interactive Language (CCIL) was developed which is used by the operations staff to develop new procedures to meet the requirements of the guest observers (GOs).
The overall ground system is designed in such a way as to resemble functionally the operations of a modern ground-based telescope. A Resident Astronomer (RA) provides the necessary support to the GO. The Experiment Display System (EDS), which consists of an interactive control keyboard and display terminal, is operated by a Telescope Operator (TO), usually working in the TOCC. These personnel, the RA and the TO, possess the required knowledge of spacecraft maneuvering, target acquisition, and S/I operations needed to advise the GO how to carry out critical operations in an efficient way. They also actually carry out the operations, since the GO is not permitted to use the control functions of the EDS console. The EDS provides the observer with all the information needed to plan maneuvers, identify the target, and verify the quality of the spectral image and carry out a `quick-look' analysis of it.
Once a GO is granted observing time with IUE, usually in the form of a number of 8 hour shifts, all targets specified in his or her proposal will be checked against Sun, Earth, Moon, spacecraft power, and thermal constraints in order to investigate target availability throughout the year. This information, combined with any time-dependent requirements specified by the GO, is used by the two ground stations to construct the schedules that assign GOs to specific shifts. In due time the GOs are contacted to carry out their observations. On arrival for his or her observing shifts the GO makes final preparations with the aid of an observatory RA, who checks the overall plan for the shift, in order to confirm the feasibility of the proposed observations, prepare any special procedures required, and establish whether suitable finder charts are available for star identification, etc.
If the GO already has some observing experience with IUE and the RA approves, the GO may carry out the observations in either a `remote' or `service' observing mode, in which case he or she does not need to be present at the ground station for the observations. In both cases, the TOCC staff keep in contact with the GO either by phone or over the computer networks as needed. Remote observers receive the raw data over the network and can display it using software similar to that used by the TOs to carry out their quick-look analysis, although they do not have the observatory's software to carry out the data reduction. (They may well have their own software to carry out the reduction themselves, however.) Service observers are not interested in looking at the raw data, and so do not need to have any special software.
Routine spacecraft operations include: maintaining communications with the spacecraft, principally by switching S-band antennas and sometimes adjusting the data rate; monitoring all spacecraft housekeeping data against given safety or operational limits for each system, carrying out corrective action when necessary, e.g. slewing to a power positive attitude to prevent excessive battery discharge; planning and execution of maneuvers ensuring pointing constraints will not be violated; correction of gyro drift using information from the FES; dumping of angular momentum by means of the hydrazine jets when the speed of a momentum wheel is too high; and recording processed telemetry data and its analysis for spacecraft trends. In addition, many of the problems described in Section 5, and the testing of new operating procedures give rise to new routine operational requirements. Finally, the operations team must always be prepared, in the event of an emergency, to take immediate action to insure spacecraft safety pending a more detailed analysis by the relevant experts.
The commands to generate these operations are transmitted from the ground by running procedures in the ground station's computer with appropriate parameters. The options available are described in more detail below.
As the target is read out the data are transmitted immediately to the ground and, shortly after completion of the whole read, a copy of the resulting image is transmitted from the control center computer to the EDS in the TOCC. There it can be examined by the observer using simple image processing facilities. The most important aspect of such examination is usually an assessment of the level of the exposure. Due to the restricted dynamic range of the SEC vidicon and the wide range of intensities present in many spectra, a repeat observation with a different exposure time may sometimes be necessary. Within 15 minutes of the termination of an exposure the observer can make a decision on the subsequent program based on a quick evaluation of the image. This time is not always wasted, since an exposure in another mode may have already been initiated on the same target, or, when it is clear that a repeat is unlikely, a new target slewed to and another observation started. In camera operations, as in planning of slewing, an efficient observer can optimize the scientific return from an 8 hour shift.
Each ground station has to keep up with the continuing daily flow of raw data images without incurring an ever-increasing backlog. The goal is to give the GO a data tape within 24 hours of the observations, prior to departure for his or her home institution. For GSFC IUESIPS runs on a VAX workstation which receives the image data via an archive tape from the commanding ground computer. At VILSPA there is only one computer for both operations and data reduction, so the latter is carried out during the US shifts following the VILSPA observing shift.
Over the many years of the mission, various aspects of the IUESIPS reduction have been changed to take advantage of new algorithms or to calibrate changing properties in the detectors. An unfortunate consequence is that a spectrum taken early in the mission cannot be directly compared with a recent spectrum of the same target, unless both have been specifically processed using the same routines. To remedy this problem, and to make the data more readily available to the astronomical community, the staffs at both GSFC and VILSPA have undertaken to develop a data reduction system, NEWSIPS (NEW Spectral Image Processing System), which uses new algorithms and re-derived calibrations to process the voluminous data taken by IUE. All images will be processed with the same routines to create a `Final Archive' of IUE data, making direct comparisons easier for observations of the same target. Among the many additional benefits of the Final Archive are a reduction in the noise level present in many spectra, an error estimate for each extracted point in the spectrum, and more accurate relative and absolute calibrations. As of mid-1994, the algorithms for high dispersion spectra are nearing completion, while those for low dispersion are complete and are being used to process the archival SWP spectra.
IUE can be, and has been, pointed at the Earth and the Moon, but as far as normal operations are concerned these bodies are considered to form constrained zones as they obscure a part of the celestial sphere. Furthermore they produce a high background in the FES due to scattered light within about 10° of the sunlit limb. Unlike the Sun constraint these are not hard coded into the maneuver generator and so may be overridden.
Fig. 3. Solar Array Output.
Although the batteries can be used to supplement the solar cell power and support operations over a wider range of Beta, this is minimized to prolong battery life and conserve their capacity for the more critical eclipse seasons. An FOD limits the number of times IUE may be used in a power-negative or power-neutral situation in any one year and the duration of any one such session. There are also constraints on the magnitude of current drain and other related operational parameters.
In planning the operations of IUE a very high priority has been given to the safe preservation of the satellite and conservation of resources within the constraints of providing a scientifically effective and productive mission. Even with this approach anomalies can and have occasionally occurred in several systems, but those in three systems, the gyros, the OBC, and recently the FES, have caused the most concern and are described below. Other systems have degraded in an expected (non-anomalous) fashion, e.g. the solar cells which result in Beta constraints as described in Section 4.3. Problems have also occurred in the S/I. Those occurring in the camera system have been potentially the most serious and these are also described below.
Fig. 4. Inertial Reference Assembly.
The first three gyro failures occurred as follows. During the third eclipse season, in the Spring of 1979, three of the then six operating gyros (Gyros 2, 4 and 6) were turned off to conserve power. At the end of the eclipse season Gyro 6 failed to restart, although a number of attempts were made. (Hope has not been permanently abandoned for Gyro 6, as will be seen below.) In the middle of 1981, maneuver accuracy decreased and telemetry analysis indicated that Gyro 1 was suspect. It was taken off-line and a further analysis finally led to the conclusion that the feedback loop was open and the gyro was not recoverable. Gyro 1 was designated a permanent loss in March 1982. In July of the same year IUE began to slowly drift. Gyro 2 telemetry showed a rapid increase in current indicating that the gyro had stalled. At this point Gyro 2 was written off and the control software changed so that the spacecraft would be operating with the remaining Gyros 3, 4, and 5, at which point IUE had no spare gyros left. Seemingly, one more failure would mean the end of scientific observations. The IUE mission managers had anticipated this possibility even before the loss of the third gyro and had asked that the attitude control experts should prepare contingency plans.
They concluded that a method of attitude control could be worked out that used two active gyros and the Fine Sun Sensor (FSS). A program was planned and teams formed to develop, build and test such a system. The testing was done on simulators, as tests on the spacecraft were considered an unacceptable risk. By Spring 1983 the two-gyro/FSS control system had been carried as far as it could go, short of spacecraft usage. As it turned out, for three years (from July 1982 until August 1985) the satellite operated satisfactorily on the three remaining gyros. Then on 17 August 1985 the critical situation was eventually reached: Gyro 3 failed. The spacecraft's gyro body angles began to drift in a situation when they should have remained fixed and the Gyro 3 current fell from a normal operating level of 65 mA to 2 mA. The duty IUEOCC Operations Director immediately recognized the situation and took the action necessary to place the satellite in the safe, Sunbath, mode. The Sunbath mode uses analog information from the sun sensors to orient the satellite to Beta = 67° which gives the maximum output from the solar array. However the satellite slowly rotates about the yaw axis (z-axis) thus eventually requiring an attitude recovery exercise. After due consideration, it was decided to try to restart Gyro 3, but to no avail. A conference was called of all experts who could contribute to the situation, and it was decided to try to restart Gyro 6, but not so vigorously as to cause further damage in this already critical situation. If this failed, then the trusted but untried two-gyro/FSS system would be brought into play. Gyro 6 did not restart and the new system was now the only hope. It worked! Not smoothly at first, but with careful nurturing, it evolved in a few more weeks into a system that enabled IUE to gather scientific data essentially as well as in the old days of the three-gyro mode. At the time of writing it continues to do so, always with new improvements being evolved. This was a very great and ingenious achievement by the staff involved.
Even this two-gyro system is not seen to be the terminal control system, for considerable progress has been made in the development of a promising one-gyro/FSS system, and it is possible that, in the event of another gyro failure a workable system could be operational within a few months. The one-gyro/FSS system, unfortunately, may not be useable with the recent presence of the FES streak (see Section 5.3). Finally, even a no gyro operation may be possible using the FES and FSS and making only small slews from star to star but science operations would be greatly restricted.
Another class of anomaly in the OBC was concerned with camera control. A VILSPA study of this phenomenon in June 1980 suggested, but not conclusively, that these problems might be associated with passage of IUE through transitional regions of the radiation belts. These OBC-related camera anomalies still occur on occasion, but are usually caught quickly enough to minimize the effect on the science.
One type of anomaly that has occurred twice, each time causing major concern, is the failure of the OBC to carry through a successful orbital adjustment (DELTA-V) to keep IUE on station. The first time that this occurred was on 12 January 1984. Prior to this DELTA-V there had been eight without incident. The magnitude of the DELTA-V is controlled by the duration of the firing of IUE's high thrust jets. The durations are calculated and the values are inserted into the Worker 19 command sequence which controls the operation. In January 1984 it was planned to fire the appropriate jets for 8.2 seconds but the program aborted after 1.64 seconds, leaving IUE in a slow spin. It took over four hours to restabilize the spacecraft and several hours more to determine its attitude. After analyzing the orbital data a new DELTA-V was planned and successfully carried out on 14 February 1984 without doing anything different except cooling the jets by orienting them away from the Sun prior to firing. The next DELTA-V in November 1984 also proceeded without incident but July 1985 brought a repeat of the fault of January 1984. After collective consultation with the experts it was concluded that differences between successful attempts and failure lay in the yaw phase data. The operational plan was modified so that the next try would emulate the successful DELTA-Vs, which had a negative yaw angle start. On 9 August the exercise was repeated and was successful. It is not known for certain whether the problem had been correctly identified since only eight days later the failure of the fourth gyro occurred and Worker 19 had to be extensively modified for use in the two-gyro/FSS mode. The two DELTA-Vs carried out since have both been successful.
The scattered light intensity correlated very well with Beta, in the sense that higher Beta yielded a higher background, and was virtually non-existent at lower Beta. There were also fluctuations in intensity as time went on for any given Beta which were not predictable. Although initially alarming, the scattered light's presence did not have a substantial effect on IUE's observing capabilities. Faint targets were acquired as blind offsets, long wavelength high dispersion and short wavelength high and low dispersion spectra were not contaminated, and long wavelength low dispersion spectra were contaminated only for exposures longer than several hours.
Then on 14 September 1992 the scattered light anomaly became even more anomalous. At the end of a maneuver to a target near the Orion Nebula, the staff noticed that there was a bright streak of light in the field of view, comparable to a V = 5 mag star. They were not initially alarmed because they believed this feature was simply the Earth's limb, which they knew was nearby. However, this feature did not go away as they expected it to, and the staff became very concerned that a potentially mission-threatening event had occurred. The feature disappeared without any trace a few hours later while it was being investigated, leaving the staff to wonder if it would be seen again. A month or so later, it did return for a few days, then it was absent, but over time it became a regular feature in the field of view. As it started to make its presence felt, the streak (as it is known and loathed) became more of a problem, filling up to 80% of the field of view and at times saturating the FES detector in parts of that field. The staff concluded that this new problem, like the previous scattered light trouble, is being caused by some of the thermal insulation scattering sunlight, and sometimes Earth light, into the telescope tube. This insulation, while causing the streak, is not physically blocking the tube, since stars can still be seen throughout the field of view; the streak simply adds to the star's intrinsic brightness.
Remarkably enough, science efficiency and data quality have not been substantially affected since the streak's appearance. The staffs at both VILSPA and GSFC quickly developed workarounds for acquiring targets. The same variation with Beta angle that was seen with the scattered light is present with the streak, although there is no Beta at which it disappears altogether. Also, the streak has changed appreciably in the past year, and may be getting less intense as time goes on. As with the previous background, only long wavelength low dispersion spectra show any contamination, but that contamination can now appear in less than one hour. The staffs at both ground stations, as well as several independent observers, are working to find ways of reducing or eliminating this contamination from target spectra. The fact that spectra of the streak are similar to spectra of solar-type stars supports the conclusion that the illumination sources are the Sun and Earth.
The greatest impact has yet to be realized. The one-gyro/FSS system that was previously developed relies criticaly on the FES's ability to identify stars brighter than about V = 9 mag to maintain IUE's stability while on target. The streak intensity at Betas > 75° can wash out all but the brightest stars, which frequently leaves nothing visible in the field of view. This situation is not a problem with the two-gyro/FSS system since IUE can function quite well without the FES in the attitude control loop for the better part of an hour. After that, a short maneuver (on the order of a few arc minutes) is needed to find a star to verify IUE's pointing. But if there is another gyro failure, it is quite likely that, at best, IUE will be confined to observing targets at low Betas, and then only when bright enough stars are in the field of view.
The IUE TV camera tubes are very sensitive to mechanical vibrations particularly at frequencies in the range 500 Hz-10 kHz. This sensitivity arises because the support structure of the SEC target is a 2 cm diameter film of aluminum oxide 50 nm thick, which has high-Q resonances in this frequency range. The target, connected to a very sensitive amplifier, acts as a condenser microphone. During the commissioning of the S/I, severe microphonics frequently occurred. This was tracked down to the panoramic attitude sensor (PAS) which was used to sense the Earth limb in order to determine IUE's attitude during the initial attitude acquisition process. After use, the PAS had been left switched on with its scanning prism rotating but, on discovery, was switched off.
Occasionally important data were lost due to a microphonic `ping' lasting 10 seconds that affected the LWR images by producing a band of interference across the image. It was eventually discovered that this `ping' was induced by the warm up of the heater in the readout gun of the LWR tube. By increasing the delay between turning up the heater voltage and the commencement of the readout scan, it was possible to insure the `ping' occurred before the scan started.
The second problem occurred in the setting up of the digital logic circuits that perform the readout scan of the LWP camera. When the circuits have remained unused for some hours, e.g. during a long exposure, the serially loaded set-up coördinates fail to pass a specific bit in the line scan register. When the scan is initiated the register then counts correctly but from the wrong starting point so that only part of the SEC target is read out. The solution to this, which took quite some time to develop into a reliable form, was to include in the operating procedure the initiation of a dummy scan to insure the logic was unlatched prior to commencing the readout scan.
The third problem has been the intermittent operation of the SWR camera tube arising from a loss of the G1 on voltage, which controls the electron readout beam current. The SWR camera was selected at the outset to be set up and calibrated in orbit for routine use in the short wavelength spectrograph. In the middle of this commissioning period the intermittent operation first occurred. At that time the decision was made to use the SWP camera routinely and this has operated flawlessly to date. At intervals, attempts were made to operate the SWR camera and it was soon observed that successful operation was much more likely when the camera electronics module was cold. As time has gone by successful reads have become less frequent even at low temperatures, and it seems unlikely that the SWR camera would be of any use in the event of the failure of the SWP. One possible cause of the failure is the loss of the clock pulse input to the G1 modulator. Such problems had occurred on the ground due to a poorly mating connector.
The fourth major problem has been the high voltage discharge or `flare' that has developed in the LWR camera. A proximity focussed intensifier, whose output is coupled by fiber-optic windows to the TV tube, acts as a UV-to-visible wavelength converter. The high electric field (3.8 kV mm¯¹) is near the maximum that the materials used for the photocathode and the anode can stand without a discharge developing. Any enhancement of the electric field caused, for instance, by a sharp point on the photocathode or contamination by a low workfunction electron emitter can result in an electron (or ion) discharge that gives rise to a point of light in the output image of the converter. The flare intensity increases rapidly as the voltage across the converter is raised, the image is enlarged due to scattering and halation, and further flares with a higher threshold voltage may appear. The phenomenon caused very considerable problems during the development of the IUE cameras. It was only after considerable labor by the project, and the eventually-successful converter manufacturer, ITT, that acceptable converters were made. The development was very successful. The high photocathode efficiency and very low background count rate (in the absence of radiation), which were better than specification, directly contributed to the performance of IUE. A difficult compromise had to be made between tube gain, safety margin from flares and reasonable production yields. The converters were formally rated at a maximum of 6 kV and some testing for flares and background was carried out at that voltage, but for all other testing and operations a voltage of 5 kV was selected.
In 1983 a weak flare was discovered in long exposures on the LWR camera. The center of the discharge was just outside the field of view of the TV camera but well within the working area of the converter, so only a crescent of light was picked up by the TV tube. Subsequently the flare steadily increased in intensity and therefore affected shorter exposures. The size of the light patch has also greatly increased. Fortunately the LWP had been calibrated and was made available to observers since it offered improved sensitivity in some parts of the spectrum. In October 1983 the decision was made to switch to the LWP for all normal observations. At first, use of the LWR was still permitted at 5 kV to complete series of observations of variable objects, but from April 1985 operation has been restricted to 4.5 kV which is at present below the flare threshold. This has reduced the gain by 27%.
On one occasion in 1976, during laboratory acceptance tests on the LWR converter before coupling to its TV tube, an extremely weak pinpoint of light was detected when making a 2 hour exposure at 6 kV but this did not recur on subsequent tests. The position of that flare is exactly coincident with the deduced center of the flare now seen through the TV camera. It is assumed that some aging process has lowered the flare threshold voltage. The extreme difficulty in producing completely flare-free converters forced the project to take the risk of including this one in the flight equipment. With hindsight this gamble has been well justified by five years of trouble-free operation by the LWR. Obviously such aging effects as have been seen in the LWR now raise the question of whether flares may in time appear in other cameras. These were not tested above 6 KV so there is no information as to whether they may have had incipient flare points which could appear at higher voltages.
The scientific efficiency of IUE was not high in the first month or two after launch. The operational procedures and the ground computer software needed some debugging and extensive tuning to reduce unproductive overhead time when IUE was essentially waiting for the next command to be uplinked. As the operations staff gained confidence in the use of the improving software IUE soon achieved an average efficiency (defined as the fraction of time spent collecting astronomical photons) of about 60%. It was expected that the increased constraints on pointing in recent years would tend to reduce efficiency but this has been compensated by increased attention to efficient scheduling. At some point in the future it may be necessary to employ integrated operations so as to maintain efficiency by forming an optimum observing sequence from all the target lists of several GOs in one time interval.
At the time of writing, IUE is still in very good shape, with its scientific performance essentially unimpaired since launch over 16 years ago in January 1978. The targets it had observed by March 1986 included:
The IUE archive has assumed increasing importance (see Giaretta et al. 1987); any image can be obtained after expiration of the 6 months exclusive data rights held by the original observer. Currently over 70,000 spectra per year are requested and the rate of requests now exceeds the rate of acquiring new spectra.