[About EXOSAT] [Archive] [Software] [Gallery] [Publications]

REFLECTIONS ON EXOSAT AFTER TWO YEARS IN OPERATION



May 26, 1985 was the second anniversary of EXOSAT's launch and, in the bureaucratic sense, the culmination of the mission with its initially approved and funded operational lifetime of two years.


EXOSAT's origins can be traced to the late 1960's when a mission (code named HELOS) to determine accurately the location of bright X-ray sources, using the lunar occultation technique, was studied by a 'Mission Definition Group' of European scientists. The instrumentation, sealed proportional counters, was to be carried in a small, 150 kg satellite launched by a Delta vehicle into a highly eccentric 'polar' orbit to maximise the area of the sky over which occultations could be performed.


The EXOSAT mission was approved by the ESA Council in 1973 but did not start its phase B until 1977 because of the financial limitations of the ESA scientific programme budget. Also in 1977 it was decided that EXOSAT should be launched on the Ariane vehicle. In the intervening eight years since the HELOS study, the UHURU and Ariel 5 satellites (to name two) had been launched to give the first exciting views of the X-ray sky and NASA's HEAO programme had been restructured to contain a powerful, few arcsecond resolution imaging telescope on the second satellite in that series which was named Einstein after launch.


The Announcement of Opportunity to propose instruments for EXOSAT was issued by the Agency in 1973 with a model payload defined to include large-area proportional counters and crude non- imaging flux collectors for the lower energies. Instrument groups, known as hardware groups in EXOSAT parlance, were selected in 1974 and, following a so-called scientific model phase, the instrument complement were significantly up-graded by the B-phase of the project. It comprised the large area proportional counter array (the ME, medium energy experiment), two imaging telescopes each with transmission gratings, position sensitive proportional counters (PSD) with good energy resolution as colour cameras and channel multiplier arrays (CMA) as high resolution black-and-white cameras and, a newly developed and unique instrument, a single gas scintillation proportional counter (GSPC). It is interesting to note that an array of similar-looking GSPCs formed the major part of the highly successful Japanese satellite, TENMA, launched a few months before EXOSAT in February 1983.


One overall requirement which was maintained throughout the programme was compatibi1ity with the Delta vehicle which constrained mass (though eventually 120 kg of a satellite mass of 500 kg was allocated to the instruments), dimensions (one metre focal length telescopes compared with Einstein's 3.4 m) and, not least, programme cost. Such constraints led to extremely innovative and state- of-the art designs in the areas of the ME detectors' bodies (all beryllium) and collimators (microchannel plate technology), the ultra-lightweight, imaging telescope optics (gold reflecting layers replicated within beryllium carriers), and to the selection of a cold gas (propane) system rather than reaction wheels for attitude control.

While these constrains would limit, a priori, certain performance characteristics, eg. telescope throughput, energy range and resolution, the vehicle compatibility requirement did mean that we could launch in 1983 on a Delta, following the difficult period facing the Ariane programme after the L5 launch failure. As we now know Ariane has performed faultlessly since then - the launcher earmarked at that time for EXOSAT being used for the GIOTTO probe to Halley's Comet on July 2nd this year.


It has been remarked that EXOSAT was "too little, too late".
Given the delay from approval in 1973 to the Phase B start
in 1977, should the programme have been reappraised. then?
Did the 'upgrading' of the instrumentation within the con
straints go far enough and were the constraints reasonable?
Would a satellite, launched in 1978 (earliest time from
mission approval in 1973 should early funding have been
available) centred on occultations and 'lunar offset point
ing' as originally conceived, been enough anyway? Clearly
there is food for thought here, when planning and selecting
future scientific missions.

EXOSAT's development programme both for the spacecraft and the scientific instruments was not without incident and difficulty. Fortunately with the passage of time only a few now come readily to mind and no longer in nightmares, though one remains to haunt us. The major concerns centred on the attitude control system (EXOSAT was ESA's first scientific satellite with a true 3-axis stabilised capability), the timely availability of ME detector collimator elements and thin beryllium windows (where we had to find and 'qualify' an alternative source in the middle of the flight model phase), the long term stability of the gold-onto-epoxy-onto-beryllium X-ray reflecting surfaces and, the haunting one, the PSD's. A full complement of ME detectors was flown but more later of the attitude control and PSDs.


EXOSAT was launched flawlessly by Delta number 169 on 26 May 1983 at 08.18 hours local time (15.18 GMT) from the Western Test Range (Vandenberg), California, at the first attempt, within a few milliseconds of the start of a one-minute long launch window. Such a narrow window was needed to yield the maximum orbital lifetime (limited by celestial mechanics) of just under three years without violating other launch window requirements and constraints. Getting the launch to take place at that moment, rather than a moment twelve minutes earlier, which would have given the statutory two year minimum lifetime, was no easy task but was made possible by the record and experience of the Delta launch team.

Let's back-track now to the beginning of the programme in 1973. At that time it was decided that EXOSAT should be a facility to be used by an 'observing community' on a European-wide basis and its use should not be restricted to the few groups responsible for the hardware. The decision had two important ramifications.


For the first time in the ESA (ESRO) scientific programme it was decided therefore that the instrument procurement would be funded and managed by the Agency rather than nationally and through national groups. (Hipparcos and the ST-FOC are more recent examples of this). However, as noted earlier, hardware groups and instruments were selected through the AD process and responsibility for the instruments shared between the groups and the Agency. In practice the groups were responsible for scientific design, testing and calibration, particularly for the 'frontends', the Agency for the engineering, system aspects, procurement from industry and overall management.


It was further decided at that time that all observing time would be open to competition through the peer review process with no time reserved for or guaranteed to the hardware groups. For a variety of reasons, one being the "quid pro quo" this approach was modified in 1979 by a decision of ESA's Scientific Programme Committee. This granted 'data rights' to the hardware groups for the calibration and performance verification phases with a guarantee of a percentage of observing time in the routine operational phase. Nonetheless, hardware group proposals for this guaranteed time were subject to the peer review process.


Partly as a result of the EXOSAT experience where, perhaps for some, the shared responsibility for procurement was unsatisfactory, and with a view not to load the ESA scientific budget with the costs of instrument procurement, the scheme adopted for the focal plane instruments of ISO, ESA's Infrared Space Observatory to be launched in 1992, calls for PI instruments funded nationally, gives the Pi's commissioning time and a percentage of guaranteed time but makes available to the scientific community the majority of the observing time. A similar scheme is likely to be adopted for ESA's high throughput X-ray spectroscopy mission.


WhiIe EXOSAT was expected to be a facility for use by the astronomical community, the originally approved plans for the mission did not specify how this could be achieved. To be fair of course, the full, final scope of EXOSAT as flown was radically different from that primary occultation mission originally foreseen. Preliminary plans for the ground segment of the observatory were laid in 1978, though within very tight financial limitations, as this was seen as a new requirement, even though by this time IUE was operational. These limitations of course impacted on manpower levels and facilities that could be made available. However, by the time of launch an Observatory team and system had been established at ESOC geared to carry out the scientific: operation, to provide quick-look data for observers, an observation data tape with instrument calibration files to a defined, standard format and a basic automatic scientific analysis (going far beyond quick-look). The basics of an interactive analysis system were also established. The originally foreseen observatory product was little more than a telemetry tape but the "miracle" was achieved with the use and upgrading of HP equipment originally purchased to support the instrument ground test and calibration programme. No VAX clusters here!

In order to review the observing programme proposals and initially to provide input to ESA on the Announcements of Opportunity (AO), the COPS (Committtee for Observation Proposal Selection) was formed and comprised twelve astronomers from assorted disciplines from the community. (Were there robbers?). The first AO was issued in mid-1981, within the ESA member states.


From the overwhelming number of proposals with the available time many-times over subscribed, a selection was made of the observations requiring the full scope of EXOSAT's instrumentation. It was decided not to time-line the observations in any detail due to launch date uncertainty and in case there would be any surprises during the in-orbit commissioning phase. Surprises there were!


Activation of the instruments began some 10 days after launch and initial results showed that all had survived the rigours of that event and were operating apparently nominally in accordance with ground test and calibration. However, the PSD of telescope 2 failed soon after turn-on and by the end of June 1983 the PSD of telescope 1 showed signs reminiscent of those discovered late in the development programme. A fundamental problem with the PSD was discovered -during X-ray beam calibrations of the flight model telescopes in the spring of 1981, at that time within about one year of the planned Ariane launch date. it was found that high energy background events could produce localised sparks (or 'pings' as they became known) in the parallel plate counter geometry, which 'cracked' the methane quench gas, and led to electrode damage, spurious low energy pulsing and eventually continuous breakdown. A solution to the problem was found by modifying electronic component values and by the addition of a small active device known as the 'ping quencher'. Nothing conclusive has been found to explain the in-orbit failures but in the light of the 'ping' saga it is not inconceivable that the PSD's parallel-plate geometry with planar, resistive- disc readout - attractive for reasons of electronic simplicity, but with its demand for very high voltages to achieve the necessary gas gain - possessed little margin of safety to cope with the unforeseen.

Two further failures occurred within the next few months. The grating mechanism of telescope 1 jammed half-in/half-out and eventually was literally dragged out and the CMA of telescope 2 stopped working, started again and finally (?) stopped. Extensive investigations of a spare mechanism on the ground yielded no clues and analysis of CMA 2 data and the implementation of various operational procedures have been to no avail.


The provision of two independent telescopes to maximise throughput was also intended to permit flexibility in observations, eg. a PSD in one telescope together with CMA/grating in the other and to provide a degree of reliability through redundancy or duplication. This concept was undone by the systematic (?) failure of the PSD's and the random, indeed perverse, failure combination of the other two which left us with a working grating and CMA but in different telescopes!


With these failings two important facets of the mission were denied us: broad- band and high resolution spectroscopy in the low energy domain. The results obtained early in the mission did show, tantalisingly, what might have been. However it might be interesting to note that greater observing time was achieved with the EXOSAT gratings in these first few months than with the Objective Grating Spectrometer on Einstein during the whole mission. It is not excluded that further grating observations be undertaken towards the end of the mission, if the grating can be dragged back in. Thankfully for the rest of the operational life, the instruments have operated fully satisfactorily and according to expectation.


On the spacecraft side the major concern has centered on the attitude control system. In the first -months of operations, various anomalies occurred, with the spacecraft switching from star pointing mode to slowly-rotating, sun 'safety mode'. Eventually a working combination of on-board black-box functions was found but not before a considerable mass of propane attitude control gas had been lost. As the mission has progressed, the observing programme timeline has been constructed with increasing emphasis placed on the conservation of this resource - no easy task given the high percentage of observations conducted simultaneously with ground-based observatories and satellites like IUE, IRAS and TENMA.


On January 1st this year, the X-axis gyro malfunctioned and in the following weeks numerous anomalies involving the triggering of safety mode occurred with the resultant loss of a large amount of control gas. Spurious triggering of safety mode has been prevented meanwhile by disabling the hard-wired autonomous safety function and giving the task to the on-board computer.


The on-board computer has proven invaluable for the mission, not only in this unforeseen application, but in its flexibility and application to the various instrument/telemetry operational modes, the vast majority of which have been modified or newly implemented since launch. Again it may be interesting to recall that there was considerable opposition 10 years ago to having such a facility on EXOSAT! Flexibility should not be confused with complexity and the built-in ability to cope with the unexpected or ill-defined is essential in any mission.

The problem with the control gas (propane) is to determine whatremains in the tank since no accurate, direct method isavailable. Currently the results from logging (i.e. estimatingvia telemetry, the usage from thruster activations) and gauging(i.e. measuring rate of temperature rise after switching on theheaters to give a measure of thermal capacity) are converging togive some 4 kg remaining. Providing this is accurate, that thereare no more 'anomalous' events and the current minimum usagestrategy is continued, operations can be expected to last throughto late 1986. However as a precaution the galactic centre regionwill get top priority in the next two months or so. As notedearlier the orbit would decay naturally in April 1986, but byfiring the hydrazine motor (intended to adjust the orbit parameters for occultations) at apogee, the perigee height can beraised. As the hydrazine motor is fired, propane must be used tokeep the satellite pointing the right way. The trick will be toensure that the propane runs out on the last orbit. So althoughnot used for its intended purpose, the hydrazine and the morethan-sufficient-for-two-years propane capacity should extend theuseful mission lifetime of the statutory two years by at least 18months.

While on the subject of useful mission lifetime, it might be recalled that EXOSAT's orbit, primarily chosen for the occultation role, was highly eccentric with a 190.000 km apogee at high northern latitudes. This orbit has allowed uninterrupted observations for 72 hours per orbit. Earth obscuration of the celestial sphere is essentially zero and the detectors do not have to cope with high backgrounds associated with the South Atlantic Anomaly as for earth orbit satellites. On the other hand the particle background in the high orbit is a factor of only 2 or 3 higher than the, low orbit, though solar flare activity can disrupt operations for several hours.


EXOSAT's orbit does allow continuous coverage from a single ground station and permits very efficient operation and control. The satellite design and the orbit together have proved ideal for coordinated measurements and has enabled very quick response to alerts. Many of the most exciting resuls from EXOSAT so far have stemmed from the long, uninterrupted look capability.


EXOSAT's operational efficiency, i.e. useful time on target is very high and would be even higher had the attitude control system been built around reaction wheels rather than the cold gas system to allow high slewing rates. Plans for operation of the Space Telescope (in low orbit) indicate that only some 35% will be spent on target. For future X-ray astronomy missions (like XMM) serious thought should be given to the utilization of a highly eccentric orbit - though it should be more equatorial for orbit lifetime/ stability reasons. The attitude system should of course be capable of high slew rates. The table (later) shows that in two years EXOSAT has spent only 50% of total elapsed time on target, the major contribution to the losses coming from perigee passage (operations only above the radiation belts taken at 70,000 km) and slewing from target to target.

Given the constantly changing on-board situation in the summer and autumn of 1983 it was hardly surprising, at least to those in ESA connected with EXOSAT, that time-lining the observation programme more than a few orbits in advance (forget a few months) was impossible. This view was not shared by some of the user community. Gradually however things improved with time-lines being generated in adequate time, in particular for those EXOSAT observations conducted simultaneously with others - such observations being used as fixed points in the schedule around which non-simultaneous observations were fitted in. It was also impossible to supply data tapes (with calibration data) to observers within the statutory one-month delay and indeed it was not until mid-1984 that the observatory team had caught with the backlog.


The observatory team, who were working flat-out, were certainly not encouraged in the early days by comparisons drawn with other missions and one wondered whether the comparison was drawn for the same relative epoch or whether memories were playing tricks. Having waited about a decade for EXOSAT anyway, waiting for tapes for somewhat longer should have posed no real hardship. What was important was that the observations be done properly with instruments whose calibration was known and understood.


When it was realized that certain of EXOSAT's mission objectives would be compromised by the on-board problems, it was decided in July 1983 not to time- line (i.e. defer) many of the observations selected from those proposed in A01 prior to launch, which it was thought could be affected by the unavailability of certain instruments. The COPS was asked to look again at the deferred proposals and it recommended with very few exceptions that all previously accepted observations should be undertaken. A02 was released earlier than planned on a world-wide basis and indicated to the user community the new situation and emphasized what could and could not be done by EXOSAT AO3 was issued in August 1984 and A04 (the last) will be issued in, August 1985. Since A01 the COPS membership has been changed to bring as broad a range of expertise as possible to bear and expanded to cope with the massive load of proposals that have been submitted in response to each AO.

It might be remarked here that no guidelines were or are established for the a priori allocation of time to small, medium and large observing programmes, or to key projects or to classes of celestial object. The COPS recommended selection of observing proposals from those submitted, naturally trying 'to maintain a reasonable balance between galactic and extragalactic astronomy and the various subsets and of course making sure that the investigations selected are properly matched to EXOSAT's strengths and unique capabilities. It may be interesting to compare this approach with those adopted for IUE, the Space Telescope and even ground-based facilities. Which approach ensures that the best science with expensive facilities is done?

The EXOSAT programme conducted during the first two years and the complete programme approved are shown in the table. It might be noted that no occultation manoevures have yet been performed. One serendipitous occultation observation has been performed to check the system and the hydrazine motor has been fired successfully for calibration purposes.


Object ClassificationApproved
Pointingsl)
Pointings
performed2)
Time
Approved3)
Time
Observed4)
Active Galactic Nuclei5443811043671
Clusters of galaxies 54 36 181118
Deep fields 1 1 1515
Extragalactic (Other) 73679789
High luminosity galactic 3782451175 599
Low luminosity galactic 5243551076561
Miscellaneous 118 78 184121
Occultations 2 0 20
Supernova remnants 11577 274195
Targets of Opportunity 59 59 120120
Calibrations/Operations 107 89 449 424
Performance Verification 21 17 69 53
GRAND TOTAL 1966 1405 4685 2966
1) Down to and including supernova remnants: approved from A01/A02/AO3 responses.

2) Pointings performed as of May 28, 1985 to orbit 196.

3)&4) Units of 104s.

As the mission progressed and observational data were disseminated, requests for help with the analysis began to come in from the community, especially from those members with no previous experience in X-ray astronomy and who perhaps lacked institutional computer and software support. Requests ranged from proposals to change data tape formats to FITs (not done), to distribute auto-analysis software (done on a case-by-case basis) to distribute interactive analysis software (not done) and to provide an interactive analysis capability for external users within the observatory at ESOC (done). This latter was implemented during 1984 by the Observatory team and, following a trial period to iron out the bugs, is now in full use. However it would now appear that the community has got to grips with EXOSAT analysis in that at a recent EXOSAT data analysis workshop at ESOC, the observatory staff outnumbered the external visitors.

The above improvements and indeed the Observatory system as a whole have been implemented within existing resources on a very low budget and shows what can be done by a young, keen Team. However with the hardware development costs of satellites as high as they are, with the flying of ever more complex instrumentation and the ensuing nuances in analysis, just where should ESA draw the line on the services it provides to a user community to ensure the best possible return on the original investment? How much should the 'observer' be expected to have provided through national resources? Is NASA's approach to ST the appropriate one with the Science Institute?


Support is now given to process requests for archival research on those observations conducted a year or more ago. This support is given currently from within available resources, with operations of course having priority. However, this does open up a new window on EXOSAT, and, if IUE archival retrieval and research is any guide, a most important and far reaching one.


What did EXOSAT cost to build and what are the running costs now? When EXOSAT was launched in 1983, the development cost of the spacecraft in industry was some 73 MAU while that of the scientific instruments was about 13 MAU. For those for whom cost per unit mass in orbit is a yardstick, these figures convert to about 200 KAU/kg and 100 KAU/kg for spacecraft and instruments respectively. The total programme expenditure to launch in 1983 including internal costs, satellite testing, launch vehicle procurement, preparations for orbital operations, overheads, etc. came to about 155 MAU. Amortized over a two year orbital lifetime this represented an investment of about 2.5 AU per orbital second. This might explain why the EXOSAT Observatory within given resources, always puts first priority on operations if necessary, at the expense of other non-time-critical functions.

The current yearly cost of EXOSAT is about 5 MAU for 24 hr/day, 7 days-a-week operations, observations, data production, analysis and science.

It would appear that the value of EXOSAT is well recognized by ESA's advisory bodies, the Astrophysics Working Group, the Space Science Advisory Committee and indeed the community as a whole who in the shape of the delegate body, the Scientific Programme Committee, agreed, at their meeting of 27/28 June 1985, that EXOSAT be operated through 1986 to the end of its useful life.


Perhaps this in itself is testament enough to those who, over the course of the project's development, made their contribution and it would be appropriate here to thank on EXOSAT's second birthday:


  • MBB the satellite main contractor and the COSMOS industrial consortium


  • the instrument contractors and suppliers:

    BAe (ME system), LND (ME detectors), Galileo (ME + GSPC collimators), Electrofusion (ME + GSPC windows), Matra (LE focal plane system and LE/ME electronics), SIRA (PSD/CMA detectors), SNIAS (PSD gas system), Laben (LE/ME electronics and GSPC system), AEG (GSPC gas cell) and CIT-Alcatel, ISA and Fichou (X-ray optics);


  • the McDonnell Douglas and NASA launch teams.

  • the satellite project team and the payload team at ESTEC.

While there is still a job to be done, thanks to my colleagues in the observatory team, the ESOC operations group, in SSD, ESTEC and in the hardware groups might be recorded on some future occasion when the job is really finished.


As a final point, the EXOSAT Observatory at ESOC is always open to constructive criticism and we are keen to do the best we can for the scientific community within budgetary resources. If you have suggestions, please let us know - it's not too late!

Brian G. Taylor
Astrophysics Division,
Space Science Department
ESTEC, Noordwijk

[About EXOSAT] [Archive] [Software] [Gallery] [Publications]