The three SSC cameras of the ASM have independently experienced turnoffs because each triggered the high-count-rate cutoff (4000 c/s) which turns off the high voltage on the affected camera.
SSC1 and SSC2, the two azimuthally-viewing cameras, initially reached this limit (on Jan 6 UT) because the sun was in the field of view and yielded a much higher rate than we had anticipated from our SAS-3 experience with similar windows and gases. This higher rate has been seen on several other occasions where the sun was in the field but toward the edge of the 12 deg - 110 deg (FW) field of view. The reason for the higher rate is probably the substantially lower discriminator setting in the SSC's (1.0 keV) vs. the 2.0 keV setting for SAS-3 combined with the presence of an active region on the sun. The SSC window is 50 micron beryllium which is very opaque below ~1.0 keV. The mask shadows of the sun in the cameras are clean and normal (as from a point source). The difficulty with this situation is that the turnoff of the HV leaves the counter off until the next command pass, causing the loss of much data. We have implemented commanding to turn off the high voltage for the periods that the sun is in the field of view of each SSC.
On Jan. 6 UT, SSC3 also turned itself off due to its high-rate cutoff. Prior to the cutoff, the counter showed an increase of counts on each anode. The shadow of the window support and central support were seen but shadows from the mask were not present. This appears to be particle precipitation, but not at a particularly high rate (3 x normal). Then breakdown commenced on anode #4, indicated by high counting rates, apparently from near one end of the wire. Similar breakdown then began in anode #5, and then the shutoff occurred. During the following days, SSC3 was turned on another 5 times, and in each case it turned itself off within minutes, i.e. the breakdown recommenced on anode #4 and possibly also on anode #5.
On Jan 12 UT, SSC2 began a breakdown on anode 5. The event occurred at a very localized position on the anode about 1/3 of the way from one end. The rate increased until the cutoff occurred. On the next day, the HV was turned on, and there was no breakdown; all anodes performed normally except #5 which showed very few pulses. After several hours, breakdown again occurred on #5 causing another turnoff. There was no apparent precipitating particle flux at the first occurrence, but there was a substantial flux from precipitating particles when the second breakdown occurred.
We have seen HV breakdown in the laboratory and we had instituted changes to preclude this problem (larger antico anodes, better internal cleaning procedures) and believed that we had solved this problem. Both the SSC2 and SSC3 detectors included these latest fixes.
It appears that this problem may be aggravated by high background rates, but we cannot exclude the cause being a particle shaken loose during launch, or some other generic problem with the detectors. We note that this detector is utilized somewhat differently than in our previous space experience in that it views a very large solid angle of the sky and in that the counter body runs at -1800 V. The collimator housing is nearby (1 cm) at ground potential.
We are currently analyzing all pertinent data from these and laboratory events and are investigating ways to avoid regions or times of high background events. In the meantime, at least for a few days, the detectors will remain off.
When the detectors are turned on we have the option of squelching the digital processing of the signals from any given anode. This would suppress the automatic high-count-rate cutoff of the HV and allow the other anodes to function (possibly with pickup noise). However, the breakdown would continue because we run with a hot counter body (-1800 V). Our experience is that persistent breakdown erodes the carbon from the quartz anode and works its way along the anode. We hope that, when the carbon is entirely removed, the breakdown will cease and the counter will become operative with 6/8 or 7/8 of its aperture. If both detectors could thus be recovered, the ASM would be restored to 90% of its intended capability. SSC1 performed well for the several days that it was turned on.
The other instrument we (MIT) delivered to XTE, the high performance computer (Experiment Data System, EDS), which processes both PCA and ASM data on board the spacecraft, has been performing well. Also, the rotation motor/gear train for the ASM continues to work well. It is currently carrying out normal rotations and dwells. We are using this to verify our rotation planning algorithms.
We summarize herein the status of the ASM including the circumstances of the unexpected turnoffs and the results of our data analyses. We also describe our plans for recovery (turn on) of the detectors. Finally we discuss the expected capabilities of the instrument under several different scenarios.
The essential results are:
1. In the first few days of the mission two of the three proportional counters (in SSC2 and SSC3) developed high-voltage (HV) breakdown. All three counters were then turned off pending further analyses. The rotation mechanism of the ASM works well and has continued to operate.
2. Analyses of flight data failed to yield a single initiating cause of the first breakdown events in the two counters, but a high particle flux was implicated in the reinitiation of the breakdown in SSC2.
3. Two modifications to ASM operations are being put into place.
(i) The HV will be turned off when the S/C is passing through high magnetic latitudes where the background rates tend to be high.
(ii) A flexible on-board automatic high-voltage shutoff will be implemented via software. The EDS will sample ASM time-series rates and send a flag to the S/C when commanded counting-rate limits are exceeded. The S/C will send the HV off command to the ASM. This will protect the ASM from unexpected high particle fluxes and from the early stages of breakdown.
4. Tests on a spare counter now underway at MIT should show how such a breakdown can evolve, giving guidance for turn-on procedures.
5. The first turn-on attempts will be made on 20 Feb. and will continue during that week.
6. The capability of the ASM will be nearly total if SSC2 and SSC3 can be made operational with only the loss of the already damaged anodes. If only SSC1 survives, a large portion of the original objectives can be attained because SSC1 alone covers 80% of the sky for a given spacecraft orientation.
The Turn-off Events (Solar and HV breakdown)
The HV supplies of the three SSC cameras in the ASM were first turned on on 5 January. In the next few days, each of the three SSCs independently experienced turnoffs because each triggered its hardware high-count-rate cutoff (4000 c/s) which turns off the high voltage of the affected camera. The HV supplies for all three SSCs were turned off on Jan 13 after two different SSCs had experienced serious failures (breakdown). The rotations of the ASM continue normally.
SSC1 and SSC2, the two azimuthally-viewing cameras, initially reached the 4000 c/s limit (on Jan 6 UT) because the sun was in the field of view and yielded a much higher rate than we had anticipated from our SAS-3 experience with similar windows and gases. This higher rate has been seen on several other occasions where the sun was in the field but toward the edge of the 12 deg - 110 deg (FW) field of view. The reason for the higher rate is probably the substantially lower discriminator setting in the SSCs (1.0 keV) vs. the 2.0 keV setting for SAS-3 combined with the presence of an active region on the sun. The SSC window is 50 micron beryllium which is very opaque below ~1.0 keV. The mask shadows of the sun in the cameras are clean and normal (as from a point source). The difficulty with this situation is that the turnoff of the HV leaves the counter off until the next command pass, causing the loss of much data. We have implemented commanding to turn off the high voltage for the dwells when the sun is in the field of view of each SSC and to turn it on again thereafter.
Several hours later on Jan. 6 UT, SSC3 (the polar camera) also turned itself off due to its high-rate cutoff. Prior to the cutoff, the counter showed a modest increase of counts on each anode. The shadow of the window support and central support were seen but shadows from the mask were not present. Then HV breakdown commenced on anode #4, indicated by high counting rates, apparently from near one end of the wire. Similar breakdown then began in anode #5, and then the shutoff occurred. During the following days, SSC3 was turned on another 5 times, and in each case it turned itself off within seconds or minutes, i.e. the breakdown recommenced on anode #4 and possibly also on anode #5.
During this period, SSC1 and SSC2 were operated during non-SAA orbits and when the sun would not be encountered. On Jan 12 UT, SSC2 began to break down on anode 5. The event occurred at a very localized position on the anode about 1/3 of the way from one end. The rate increased for some 14 minutes until the HV was automatically turned off. On the next day, the HV was turned on, and there was no breakdown; all anodes performed normally except #5 which showed very few pulses. After several hours, breakdown again occurred on #5 causing another turnoff. There was no apparent large charged-particle flux at the first occurrence, but there was a substantial flux from energetic charged particles when the second breakdown occurred. The S/C was at the edge of the SAA at that time (but outside the formally-defined SAA region).
The detectors in the SSCs have the following Metorex serial nos.
SSC1 azimuthal view direction SN 05/02
SSC2 azimuthal view direction SN 06
SSC3 polar view direction SN 01
The two counters experiencing breakdown, SN 01 and SN06, both have the latest engineering modifications which include polyimide coatings on the Be windows to forestall leaks and antico anodes of increased diameter to forestall breakdowns. Both counters were tested at Metorex for breakdown at voltages at least 150 V above the operating voltage. This testing was done with the flight gas mixture before final welding (sealing) of the detectors and again after the counter was welded shut. For example, before sealing, the SN 06 antico anode network was tested at 2100 V and each quartz anode was tested at 2000 V. (Its operating voltage is 1850 V.) After sealing and final filling, the antico was tested at 2000 V followed by a test of the entire quartz anode/antico system at 1950 V for an 8 hour period. SN05/02 is an earlier version; it does not contain these two design changes.
After integration with its electronics, each counter underwent extensive environmental testing at MIT to mimic spacecraft environmental testing, because both counters were installed as replacements after S/C testing. Shake, thermal, thermal vacuum, and calibration runs in an x-ray beam were carried out. Each counter was run for many hundreds of hours during and in addition to these tests. We were gratified that our previous catastrophic breakdown problems did not reoccur; i.e., the larger antico anodes appeared to be doing their job. Previous breakdown incidents had often started in the antico system.
The operation of SN 01 and SN 06 during these tests at MIT was fault free except for temporary breakdowns that removed a bit of carbon from one spot on each of two anodes. These occurred during calibration testing where a very high counting rate (2000 c/s) was applied through a narrow slit to a 1-mm portion of the 8 anodes. In SN 01, anode 8 had the primary involvement and anode 7 had some secondary involvement. In SN 06, a similar occurrence affected the same two anodes but in reverse order. Both temporary breakdowns occurred while the most energetic calibrations x rays (Cu K; 8 keV) were being used. After the initial event, the counters were remarkably stable; subsequent scans at the same rates did not restart the breakdown, nor did subsequent thermal vacuum and other tests. We had seen this phenomenon in other counters on previous occasions and never saw breakdown restart.
We attributed the effect to overstressing of the counter; 2000 c/s on 1mm corresponds to 200,000 c/s over the entire 100 mm of the counter, well beyond the 4000 c/s cutoff rate of the automatic ASM trigger level. We speculated that a slight protusion on an anode or cathode had initiated the breakdown and that it had been vaporized by the temporary breakdown. Thus these events did not overly concern us, although maybe they should have. Note that these anodes were not involved in the flight breakdowns. The result of the missing carbon at a local region created a slight non-linearity in the response of the anode to the position of an x ray, a calibratable effect.
SN 05/02, the counter that did not break down, was received at MIT in its current form in May 1995 and placed on the ASM at MIT several weeks later. After a month or more, it was found to be leaking slowly as evidenced by slow resolution degradation (due to oxygen contamination from the air) which may be described as a gain decrease of 1% per month. This has been taking place since around delivery as far as one can tell. (Subsequently we began leak tests just before the final gas filling at Metorex and with a residual gas analyzer at MIT after delivery.) In an RGA test with another counter showing a similar degradation rate, we found the xenon leak rate to be such that the gain increase in space would be negligible (1% in about 6 mos.). However we also have evidence that extrapolation from one counter to another is dangerous. As yet we have insufficient statistics in space with the 55Fe calibration source to determine well the current leak rate, but preliminary results in early January showed no substantial gain change.
As noted, SN 05/02 does not have the latest modifications (polyimide window coating and larger antico anodes). In this it differs from SN 01 and SN 06. Also the small amount of leaked oxygen in the gas could be inhibiting breakdown.
The earlier histories of the detectors are as follows:
SN 06 (SSC2)
This was a newly constructed counter delivered to MIT in July 1995 and placed on the ASM at MIT in August. Both Bradt and Levine were present at Metorex during the final stages of assembly and testing and carried out careful inspections and measurements of the counter before it was sealed.
*SN 01 (SSC3)
This counter was the first delivered to MIT in Aug. 1993. It has had a colorful history. It suffered from the initial problems (conductive epoxy resistance and mechanical distortions in vacuum). It underwent modifications at Metorex to correct these problems.
SN 01 was returned to MIT in Dec. 1994 and was placed in the ASM and on the S/C. A leak in the Be window was detected through gain and resolution changes during S/C thermal vacuum. It was again returned to Metorex, the leak was found and repaired with epoxy without opening the counter body.
It was then redelivered (third time) to MIT in May 1995. It failed with breakdown on the 13th day of the thermal-vacuum test, on anode 5 with later involvement of anode 4; a precursor event may have been a 10-s period of increased antico rate in an earlier thermal test. It was then returned to Metorex for inspection and repair. It was inspected there by Bradt and Levine when it was first opened. Two bare spots (no carbon) opposite one another were found on anodes 4 and 5. These two anodes were replaced along with the 3 ground wires between these anodes, although these ground wires showed no defects.
At this time, an evaluation of the status of the counters led to two new design changes which were implemented to forestall leaks and breakdown: polyimide coating and larger antico wires. Also, an alcohol wash of the completed counter assembly was instituted to better ferret out small particles. As noted, the larger antico wires were installed to lower the antico gain because many of our breakdowns had begun with apparent breakdown in the antico system. SN 01, with these changes, was inspected prior to sealing at Metorex in July 1995 by Bradt.
The final and fourth delivery of the refurbished SN 01 to MIT was on July 1995.
SN 05/02 (SSC1)
This detector consists of the outer case and feedthroughs from the original SN 05 and the internal wire frame from the original SN 02. The counter body was new and the frame which holds all the anodes and ground wires had undergone several refurbishments, once to fix the epoxy and mechanical problem and twice to replace parts implicated in breakdowns.
We have not invoked a formal review but have consulted freely with colleagues who have experience in proportional counters. In particular, a productive review of data from the failure events was reviewed at a meeting in GSFC with Code 666 experts on Jan. 19, 1996. Present were Frank Birsa, Charles Glasser, Jean Swank, Peter Serlemitsos, Will Zhang and Richard Rothschild (formerly of Code 666, now at UCSD). We have also had discussions with Jeffrey McClintock of SAO and with Dan McCammon at U. Wisconsin. The many helpful comments have been taken into account. In April 1995, our earlier breakdown and leakage problems occasioned a review at Metorex. Present were Bill Davis and Chuck Glasser of GSFC, William Mayer of MIT, and Metorex engineers and management. Following this visit (and that of Bradt and Levine in June), the design changes mentioned above (antico anodes and polyimide on windows) were instituted. Also Metorex improved its high-grade clean-room facility by locating and sealing a minor leak. Also, more stringent high-voltage breakdown tests of the completed counters were instituted.
Objectives and Activities
The objective of our efforts has been to understand as much as possible the failure and to devise methods for turn on and operation that would maximize the possibility of returning the two problem detectors (SSC2 and SSC3) to nearly full capability and to forestall further occurrences. This has entailed an aggressive collection of data from multiple sources and comprehensive analyses. These include correlations of the events on the ASM with view directions, magnetic fields, particle events, earth location, etc. The turn-on strategy we have adopted entails modifications to the flight software so the EDS can command off the ASM HV (via the S/C) and modifications to the ASM Observation Planning system. Also, laboratory tests of a spare counter are also being carried out.
Data Analysis and Conclusions
We have essentially completed our analysis of the data from the SSCs prior to and during the breakdown events. This effort required the development of new tools to extract telemetered data and for the presentation of correlations with other data. Data from all three XTE instruments and the spacecraft were studied. Also data from prelaunch environmental testing was reexamined. Mr. Seppo Nenonen of Metorex, the manufacturer of the counters, provided us with information as needed. Unfortunately no "smoking gun" was found, but we learned a great deal about the breakdown process and about the environment and its interaction with the detector for this particular orbit and detector/collimator configuration.
The several studies yielded the following pieces of information:
Time sequence data for the entire week of operations with special attention to the times of breakdown (expanded plots), including correlations with spacecraft position, HEXTE and PCA data, geographic position, magnetic field direction.
SSC2 and SSC3 first broke down in times of low particle background.
The SSC2 second breakdown occurred during a brief period of intense particle flux; the S/C was at the edge of the SAA at this time. This high background was also seen by the PCA; SSC2 happened to be pointed in the PCA direction.
Periods of high background outside the SAA occurred primarily at specific positions on the earth (high magnetic latitudes). These events are clearly particle events as indicated by coincidences between the two layers of the SSC'.
Some particle events were highly directional, indicative of particles spiraling around magnetic field lines with pitch angles of about 90deg.. As the ASM executed a 4-minute 360-deg. rewind, modest particle fluxes were sometimes seen exactly at 180deg. separations. In general high fluxes were loosely correlated with times when the field of view of an SSC included a large portion of the magnetic equator, e.g., when the long axis of the collimator lies along the magnetic equator).
The first breakdown of SSC2 occurred near one such rewind, but the event began several minutes before the beginning of the rewind. We believe the coincidence with the rewind is accidental.
Position histograms: these show the position of each detected x ray on each of the 8 anodes (per SSC). In fact these plots reflect the ratio of currents detected at the two ends of the resistive wire. Thus they may not represent directly the x-ray position on the quartz anode. These results stem in part from correlations with time-series data.
The first SSC2 breakdown occurred at a highly localized region on anode 5. The antico anodes were not involved. The breakdown began about 10 minutes before the high-rate cutoff turned off the HV. The breakdown worsened with time during this period until substantial carbon was eroded from that location.
After SSC2 was turned on that same evening, the other anodes yielded normal position histograms, but the damaged anode was inoperative. Apparently no other anodes were affected. (SSC2 was turned off after its second breakdown.)
The anticoincidence anodes were not involved in the breakdowns, unlike some breakdowns observed in prelaunch testing.
The exact location of the first SSC3 breakdown is less certain. It occurred on Anode 4, and appears to have started toward the 'A' end of anode; it spread to Anode 5 within a few minutes, again on the 'A' end.
The same two anodes of SSC3 in the same counter body had been involved in breakdown in thermal vacuum in June 1995. However, those breakdowns were definitely near the 'B' end of both anodes; upon disassembly at Metorex in June, bare spots on the two anodes directly opposite each other were found. As noted above, the two anodes and the three ground wires between the two anodes were replaced; no defects could be seen on the ground wires.
The breakdown on SSC3 was preceded by a puzzling rise in counting rate for about 5 minutes in one half ('A' end) of all 8 anodes. These events show the coarse shadow of the Be window strongback but do not show the finer pattern of the mask. Together with their very soft spectrum, this suggests uncollimated x rays coming through the window. Their appearance on only the 'A' end is not characteristic of any celestial x ray source nor of spatially uniform streams of particles. We have speculated that this might be breakdown from the counter (-1800 V) to the collimator housing (0 volts) but we see no noise on the antico circuit which should be strongly capacitatively coupled to any breakdown from the counter body. Neither do we understand how breakdown on the outside of the counter would induce internal breakdown. In short, this phenomenon remains an enigma. No such effect was associated with the breakdown in SSC2.
The initial breakdowns in SSC2 and SSC3 each began with high numbers of counts in the lowest few channels of our 64 channel spectral output. These preceeded the turnoffs by 3-5 minutes.
We have found no explicit cause of the two initial breakdowns. They occurred at a time of benign background rates well within the rates tested before launch. The two breakdowns occurred at different places in the counter; fortuitous events cannot be excluded (e.g., a particle shaken loose during launch).
Breakdown restarted immediately after turnon (SSC3) or when substantial particle fluxes entered the counter (SSC2).
This and the temporary breakdowns seen under high-rate conditions in prelaunch tests (see "Detector histories" above) indicate a vulnerability to high rates. Thus the HV should be turned off during high background regions of the orbit (in addition to the SAA).
The early phases of breakdown have a signature (excess counts in the lowest PHA channels) that can be used to turn off the HV, hopefully before significant damage occurs.
The two conditions that differ significantly from our extensive prelaunch testing are the greater integrated particle fluxes and the interaction of the counter body at high potential (-1800 V) with the particle environment in space. We cannot exclude, for example, a charging-up phenomenon. The more frequent "HV off" commands planned for future operations should help protect against this possibility.
Hardware testing under the proper particle environment would be a major undertaking (e.g. in a vacuum at particle accelerators) as would a program of inducing breakdowns with different counter geometries and gas mixtures. Such testing would perhaps reveal the initial cause of the breakdowns and better design parameters (e.g. more quench gas), but it is unlikely to affect our method of operations with the current orbiting hardware. The counters contain 95% Xe and 5% CO2.
We have had significant prelaunch experience with breakdowns and had done extensive inspection and testing of counters with breakdown problems. We had increased the antico-anode diameters and successfully forestalled breakdowns in these counters over many hundreds of hours of testing in diverse conditions of vacuum, counting rates, and temperatures. Deliberately inducing breakdown in an uncontrolled manner (e.g. by raising high voltage) is likely to preclude meaningful conclusions and to destroy the one spare counter configured the same as the failed flight counters.
Since we have failed to find any smoking gun in the data which could be the basis for a rational test, we are now proceeding with an effort to induce breakdown with high x-ray and gamma-ray counting rates. At this writing, the testing with SN02/05 is showing the anodes to be remarkably stable with whole-counter rates of 1500/s for several days and about 1,000,000 c/s for a half hour, and no sign of breakdown yet!
A objective of this test is to study how breakdowns develop after starting. In previous breakdown events, we generally did not let the breakdown run its full course, lest it begin to involve even more anodes. As more and more of the carbon is eroded away, does breakdown finally stop (if the HVis cycled on and off)? When all the carbon from a given anode is finally removed, does the breakdown stop? Does the vaporized carbon degrade the counter? Do more than two anodes become involved? (PCA-group experience suggest two may be the limit.) If the breakdown continues after all the carbon on an anode is removed, can one operate the counter with the affected channel squelched (i.e. digital signals killed) while breakdown continues? (Limited MIT experience during one such breakdown suggests it can.)
It is not likely that this hardware test will affect the manner in which we choose to turn on the flight SSC's, especially considering the relatively uncontrolled nature of the experiment and the different conditions in space. Accordingly we will use any information gained as guidance in what to expect; of course, unexpected results could be informative. The results should be available before the Feb. 20 turn on, but the turnon schedule should not be delayed for these results.
We plan to turn off the HV of the operational SSCs during all dwells including the sun in the field of view, during passage through regions of high magnetic latitudes (where we found high backgrounds), during passage through the SAA as before, and during rewinds (when directional particles or solar particles can enter the detectors). This, at the least, will protect us from inciting breakdown and will serve to extend the normal life of the detectors. Our modified planning tools for this are nearly in place; at this writing they are complete and initial testing indicates proper operation.
Automatic onboard software trigger for HV off
We are instituting a more sophisticated on-board trigger for automatic turnoff of the high voltages. Currently the electronics of each SSC triggers the HV off when a total counting rate of 4000 ct/s is obtained. The EDS flight software is being modified to trigger on any channel or any combination of channels of the multiple time series data from the ASM. The EDS will send a flag to the S/C which will send the HV command to the ASM within about a minute. (The EDS is not configured to forward such a command directly to the ASM. Initially, we plan to have the system trigger (1) on a high flux of counts in one of the lowest PHA channels (channel 4) as an indicator of the early stages of high voltage breakdown, and (2) on large events (> 10 keV) as an indication of high particle rates.
The purposes of this on automatic on-board cutoff are (1) to limit damage from a breakdown episode, (2) to protect from unexpected high particle fluxes (when the HV is on), and (3) to protect from command-link failures, e.g., when the HV should have been turned off via the DAP but is not.
The implementation of this feature is now in process at both GSFC and MIT. The latter group must implement the change in EDS flight software; the "burst-trigger" software will be applied to the ASM Multiple Time Series data. The engineer who designed the EDS flight-software is available this coming week. The modifications should take about a week, leaving a week or a bit more for testing the changes before the scheduled 20 Feb. turn ons.
Our plan for turn on of the SSCs contains the following elements.
We will turn on the SSCs starting 20 Feb. This date is dictated by the availability of the revised planning tools and the trigger criteria for the automatic onboard HV-off feature. It also gives opportunity for laboratory testing of turn-on methods. The pacing item is the onboard software which should be available and tested by Feb. 20.
The commands for turn on are quite limited. For each SSC, one can (1) turn on HV or turn HV off, (2) 'squelch' the digital electronic signal from any anode, and (3) choose the automatic HV-off trigger criteria (channel and trigger level).
The sequential steps in turning on a detector with previous breakdown should first optimize the likelihood of the breakdown stopping of its own accord (possibly after a HV turn-off and turn-on sequence). One would also attempt to let the breakdown play itself out, but in a controlled manner. The sequential turn-on process for a breaking-down counter could be:
(i) Run the SSC for short periods of 100 s starting every 300 s to see if the breakdown does not reoccur or eventually stops. Continue this for an orbit or a day or more.
(ii) If breakdown ceases, continue on-off cycles, but with longer 'on' times, say 10 minutes, and finally start normal operations.
(iii) If breakdown continues to occur immediately after each turn on, squelch the affected anode for short periods (e.g. , 100s, 600s). This allows the breakdown to proceed.
The exact procedures to be used are being finalized now.
The first day, we will initially turn on each the three SSCs for the low-background portions of three different orbits (one orbit per SSC). SSC2 and SSC3 would be cycled on and off as indicated above and SSC1 would be turned on continuously. All counting-rate safeguards would be in place. Rotation algorithms would proceed normally, and spacecraft pointing operations will not be affected.
The results of the first day will be analyzed. On subsequent days during the week, we will attempt to bring all three SSCs up to normal operations.
Insofar as possible, turnons will be conducted through the DAP to minimize real-time (seat-of-pants) commanding. However, the incoming data will be monitored in near real time and long and frequent command passes are being requested.
It is hoped that by the end of the week of Feb. 19, the turnon process will be complete and that calibrations and 'normal' operations can proceed.
We present here the capabilities of the ASM given various assumptions about the state of the ASM after turn on.
All SSCs fully operational except for damaged anodes
In this case, the ASM would have close to its full design capability. The SSC3 would have 6/8 = 75% of its aperture and SSC2 would have 7/8 = 87% of its aperture. These would cause only a very modest sensitivity degradation, e.g. only 7% for the more important SSC2 (which scans more of the sky). SSC1, being fully operable, would have no degradation, so the average loss in sensitivity would be negligible.
The more frequent HV-off times due to high-background regions will also modestly decrease the average long-term sensitivity, but they will not affect the sensitivity of a given dwell. Each dwell will still last for ~90 s so the scan progress at 6deg. per dwell matches that of the earth's motion. The principal effect of the HV-off times will be in the temporal coverage because, during a given orbit, portions of the sky will not be viewed. Our elementary rotation plan keeps the ASM pointing opposite to the earth; the regions that would be viewed during the SAA are simply skipped. (Rewinds are generally done during off times.) However, as the earth rotates, (1) the skipped regions of the sky rotate around the celestial sphere and (2) the spacecraft will no longer enter the same high-particle regions. Both effects mean that a given source will become visible to the ASM within a few hours.
Under the new operational plan with its conservative assumptions, the total on time was calculated for a given week (Week 2). This week included a fairly large number of S/C slews which should be typical normal operations. It was found that the ASM would be operative in dwell mode 52% of the time, after all turnoffs are taken into account (including spacecraft slews, rewinds, sun in field, and generous estimates of the high-background regions). These extended high-background regions lower the coverage from a previous ~65% (after SAA and S/C slew times are excluded). Of course, even with perfectly operating detectors, much of the high-background data included in the latter percentage would be of limited utility. The net exposure is thus reduced to 52/65 = 80% (or more) of the previous value.
The net effect of all this is that the long-term average sensitivity is reduced by only about 10% while the average time it takes to access the full 85% of the sky encompassed by the SSC FOVs (FWHM) is about 1/(0.52) = 2.0 orbits rather than the 1/0.65 =1.5 orbits previously. (Our advertisements of 80% of the sky in 1.5 h did not fully take into account the HV-off times due to high-background regions.) A more complicated rotation planner could retrieve most of the 85% of the sky in 1.5 hours, but at a loss of sensitivity in each dwell. The present scheme is certainly adquate for the present.
SSC1 and SSC2 are operative but SSC3 fails
This scenario is plausible since SSC2 did operate properly after its first breakdown until it encountered the high background which restarted the breakdown. One could hope it would do so again and that our new planning will turn off the detectors in the high-background regions of the orbit. In this case, the two azimuthal counters would provide almost the entire capability described above, except that 15% of the sky viewed by SSC3 would not be scanned in a particular spacecraft orientation, leaving 70% so scanned by both operating systems. Almost any S/C slew during normal operations would allow SSC1/2 to scan the missed region, typically within a few hours. Crossed lines of position would be obtained leading to arc-min positions of new sources which would allow rapid PCA acquisition.
Both SSC3 and SSC2 fail completely, and SSC1 is completely operational
In this case, one azimuthal camera is operational, and the FOV of the one camera would cover a full 70% of the sky in one rotation rather than the ideal 85%. One would lose the crossed line of position from SSC2, but if the flaring source is previously known, the refined position is not needed (unless source confusion leads to an ambiguous identification). If the source is new (the less probable occurrence), SSC1 would yield a line of position several arc minutes wide and a few degrees long. The PCA could maneuver to this vicinity and another SSC1 scan would provide a crossed line and thus a refined position allowing prompt PCA acquisition. This would probably delay PCA acquisition for an orbit or two.
The overall sensitivity and quality of the light curves would be degraded by the existence of one azimuthal camera rather than two; it would take twice as long to obtain the same statistics on a given source; e.g. one might plot light curves with 3 hour integrations rather than 1.5 h. All in all, if SSC2 and SSC3 both failed, SSC1 would serve quite well the principal objectives of the mission, providing, say, 70% of the science yield of a fully functioning system.
All three SSCs fail
The worst case is that all three SSC's fail because the problem is generic to the counters and/or the space environment. Then the yield from the ASM would be 0%. The rapid response capability of the S/C could still be used to respond to TOO's generated by other missions and ground-based observatories.
We have new evidence from a test counter here at MIT that breakdown such as this will play itself out after it has eroded all the carbon from the affected anode. In the lab case, this left all other 7 anodes working well. Thus there remains hope that all three detectors can be brought back on line, even though we have been unable to find a cause of the initial breakdowns. Al Levine and Ron Remillard are at GSFC for these operations.
It had been planned to turn SSC3 on seven times with no squelch and ten times with the measurement-chain squelch turned on for the two affected anodes. These were to be effected by onboard commanding (via the DAP). The first set of commands was not effected because a previous HV latch from yesterday was still in effect (we had neglected to send the unlatch command). The unlatch command was sent before the second set, but it reset the squelch command which had been sent earlier. SSC3 did turn on but it immediately latched itself as would be expected from breakdown in the unsquelched state.
SSC2 was not turned on. We intend to carry through our turnon procedures, including squelches, for SSC3 before proceeding with SSC2.
In summary, both SSC1 and SSC3 behaved as expected given the commands they received. We learned some new things about command sequences and have revised the new DAP accordingly. The MOC is also implementing a minor modification to the RTS routines to make the safety-turnoff logic more robust.
Beginning at 0h UT 22 Feb (this evening), SSC1 and SSC3 will be commanded to run normally, but with the squelch on for the two affected anodes of SSC3. We expect SSC1 to operate normally. SSC3 could run fairly normally if the breakdown does not cause excess noise in other unsquelched channels. Otherwise SSC3 could turn off. If this happens, we would probably choose to squelch other channels so the SSC3 breakdown can continue until (hopefully) it ceases, having removed all the carbon from the affected anode(s).
Ron and Al are still at GSFC.
For operations on the 24th UT, the MOC had modified the safety turnoff logic of the spacecraft so that, after a safety trigger and (HV turnoff), the next planned HV-ON command in the DAP would be effective. These HV-ON commands occur several times every orbit. This makes unnecessary human intervention to get an SSC running again after a S/C safety trigger shutdown. (The 4000-c/s ASM latchup trigger is still effective.) On the 24th, SSC3 was not run, but SSC1 was run quite successfully (to quote Ron Remillard): "By sundown (the MOC) had modified , tested, and uploaded the revised . . . system, and we are now running a smooth 24-hour DAP routine for SSC1 (can you believe it!). SSC1 and the drive assembly are [at 6AM still] performing very well." Al tells me that it is still chugging along at 4 PM.
We are now moving aggressively to get SSC2 into operation. It will be run for the full 24-hour cycle along with SSC1 beginning 25 Feb. UT. Recall that the previous breakdown in this counter involved only one anode at a well defined location and that it did not exhibit the excess diffuse flux seen in SSC3. We thus hope (1) that the counter will not restart breakdown because we are now avoiding high-background regions, or (2) that a restarted breakdown will burn itself out without involving other anodes (as in our lab test). Recall that SSC2 and SSC1 are the two azimuthal cameras that view about 70% of the sky for a given spacecraft orientation.
On the 23rd, the spacecraft MTS trigger was activated 5 times for SSC3, once for a normal background increase, 3 times due to apparent breakdown after running for 30-40 min, and once due to breakdown pickup in the anticoincidence which had been reenabled. The breakdown events show high rates on additional anodes, but with no apparent damage when next turned on. These events were accompanied by the soft fluxes that show the shadow of the window support structure, similar to the first breakdown event in this counter. There are indications that we could be observing x rays from breakdown outside the counter (as previously suspected), but this is not certain. SSC3 was shut off pending further data analysis. Its prognosis definitely worsened with these events. We plan to proceed soon with operations for only 10-min intervals, with the hope that the breakdown will not commence within this short interval.
Al and Ron (and all of us) acknowledge gratefully the superb support being provided by the SOC and the MOC to these operations.
We are now pressing to complete calibrations for SSC1 and SSC2 so that the ASM can begin to play its proper role.
Unfortunately, after almost 8 days of running, SSC2 began to break down on the anode (#4) that was already damaged. This restart of the breakdown was not unexpected; it is consistent with our lab experience. We disabled (squelched) Anode 4 to keep the counter on the air (and the breakdown proceeding). Later we had to do the same with the antico which was picking up enought noise to trigger the counter off. We were hopeful that the breakdown would propagate along the wire to the end and cease as was the case with the our lab test. However, yesterday afternoon the position histograms on adjacent wires began to show anomalies (peaks at roughly the center of the wire near the central partition). We were, and are still, hopeful that this was noise from the breakdown on anode 4. Eventually six of the other anodes were involved, although they were still intact because a dwell on Sco X-1 showed shadow images on these anodes. At this point we turned off SSC2 by forcing it to latch up. SSC1 was left on. These symptoms have some similarities to the current state of SSC3. As of this morning we do not know if we are dealing with additional breakdowns or pickup from the Anode 4 breakdown. We will make a quick assessment in the next day or two and hopefully will learn enough in that time to determine our next step which may be simply to let the breakdown proceed to its natural end. We might choose to do this with SSC3 before we do it with SSC2.
In the meantime, SSC1 continues to operate well. Also our calibrations are proceeding well. We have preliminary light curves from January and February in our hands.
We then went to work more aggressively on SSC3. We intend to push it fairly hard to force breakdown while leaving SSC2 off until we see how SSC3 plays itself out. For SSC3 we squelched two more anodes to help keep it running, but the counts showed up on other anodes! Tomorrow we will raise the TSM limit and see if it runs without latching. If it does not, we will squelch more anodes.
We are more hopeful today that the problem with the other anodes is electronic pickup and not additional breakdown. When we first turned on SSC3, the 6 good anodes responded perfectly with no apparent damage (albeit with low statistics in a single dwell). The peculiar spikes on these anodes showed up after about 15 minutes. Also, when the other anodes of SSC2 were showing spikes, it viewed Sco X-1 and the images looked normal, and model fits produced low chi-squared, indicating that arcing had not (yet?) eroded carbon from the end anode which changes the position response.
We enter the weekend with SSC3 and SSC1 running in this manner. We should know in a few days how this breakdown plays out on SSC3. SSC2 remains off. SSC1 remains on and is collecting valid data.
So right now, we are still hopeful that we can resurrect both SSC3 and SSC2 with no new damage beyond the known 1 or 2 wires in each.
We are pushing aggressively the analysis of recently acquired data to produce SSC orientation calibrations so that we can begin to produce near real-time intensities with SSC1 at least. It is our hope to have these in preliminary working order within a week or so.
We let SSC3 arc all weekend and Monday while monitoring only one (unsquelched) anode. This morning the breakdown sputtered to a halt (more or less) and we now have SSC3 on line with 6 of the 8 anodes and the antico working normally. Although the 6 anodes appear to be in good condition, we caution that we have only a few hours of running in this state. Evaluation of data from the next 24 hours or so will tell us a lot more. If, as seems likely, this recovery was successful, we will perform the same surgery on SSC2 which has been off for the past week or so. Since we still do not know the inciting cause of the initial breakdowns, our optimism is still somewhat tempered. SSC1 continues to operate well.
We began the burn-in of SSC2 this morning. It is running with the bad anode (#5) and antico squelched, and with all rate-limit safeties in place. The noise from the breakdown yields a total counting rate just below the latch rate of 4000 c/s! If it starts turning itself off, we will suppress other anodes as necessary to keep it on the air.
For the first time since just after launch, all three SSC's are on the air, two of which are acquiring useful data.
Obtaining useful positional response functions and orientation calibrations from the flight data is proving to be more complicated than we had hoped, but progress is good.
The deliberate breakdown of SSC2 continued all weekend. shortly after it started, we squelched anodes 4 and 6 in addition to the already squelched bad anode (#5) and the antico in order to keep it running. It produced noise counts furiously all weekend and Monday, but by this morning it had quieted down. We turned on anodes 4 and 6 so now we have 7 operational anodes on this counter. We will leave the antico off until tomorrow to give time for residual sputtering (if any) to cease. All 7 anodes appear normal based on the modest statistics of single dwells.
Even better is the fact that the 6 anodes of SSC3 continue to work well after some 12 days of operation (including the deliberate breakdown time). Recall that it initially failed on the second day it was turned on. Maybe it is not as vulnerable as it first seemed.
We are now running with 3 SSC's and 21 of 24 anodes (read 'aperture') operating.
Analysis of earth occultation data allows us to get pulse-height spectra with reasonable statistics. For all three SSC's, the gain appears to be consistent within about 1% percent over the two months since our IOC measures. This is reassuring since we know from prelaunch data that there is a slow leak is SSC1 and had hoped that this would lead to a gain increase in orbit of only 1% in 6 months or so. Now we can reasonably hope that this is so.
A new rotation-planner program has been completed by Wei Cui. It minimizes earth blockage of all 3 cameras, avoids sun and high-count-rate regions, and yields dwells of a constant exposure. It should begin operation in a day or two.
Obtaining calibration parameters which will yield reliable intensities are our highest priority now. This will allow us to begin the basic ASM alarm function for TOOs.
Major progress has been made in bringing our production and quick-look analysis systems on line. Calibrations with celestial sources continues and special observations are being carried out for this purpose this weekend.
The ASM team (Bradt, Chakrabarty, Cui, Levine, Morgan, Remillard, Jernigan, Shirey, Smith)