Oral History Transcript — Dr. Steven Murray & Dr. Gerry Austin
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
See the catalog record for this interview and search for other interviews in our collection
Steven Murray & Gerry Austin; April 9, 1998
ABSTRACT: Uhuru X-ray detector; Cross-grid charge detector; HEAO-2 mission (1978); development of imaging detectors; microchannel plate development; change from building prototypes to flight-ready instruments; demonstration of how the ROSAT detector is put together; AXAF detector design.
DeVorkin:For identification for the video, this is April 9th, and we're in Room A101. Right now we're looking at the Uhuru [phonetic] detector and some of its collimator elements in front, and Jerry Austin on the right with Steve Murray on the left. Jerry is going to walk us through what we're looking at here in the Uhuru detector.
Okay. This is a proportional counter. It's a device that is used to detect X-rays. It consists basically of a hollow tube with a single small-diameter, probably two thousandths of an inch diameter, wire down the middle. The way this device works is that an X-ray, which goes through this thin window on the front, which is a thin beryllium foil, enters the gas, interacts with the gas. There's a high-voltage bias applied to the wire. Interaction occurs in an avalanche resulting in electrical pulse at the output, which is proportional to the energy of the X-ray that went into the detector to begin with. So we use this to both detect the existence of an X-ray and to make some estimate of the energy of the X-ray. The Uhuru payload consisted of two banks of these detectors. There were actually six of them next to each other on one side and six on the other side, looking in opposite directions. But as you can see from looking at this window, X-rays could come in from any direction. If it was just looking out at the sky like this, it wouldn't be too useful because it would be looking at the whole sky. So it's necessary to restrict its field of view. In this era, which is the middle sixties, when we began to develop this — it ultimately flew in 1970 — this work all predated the use of any sort of imaging optics on orbit.
So it was necessary to build something which we call a collimator in front of this to define a certain field of view on the sky. That was done by using these thin-walled aluminum tubes. We wanted the wall thick enough to stop X-rays from going through in directions they shouldn't, but not so thick as to increase the weight, weight being always a critical parameter in the design of these payloads. These are just individual thin-walled aluminum tubes. They are square tubes on this side, approximately a half-inch square. The tubes on the other side were also a half-inch high but were only fifty thousandths of an inch, one tenth of this width. I don't happen to have any of those anymore. But the resulting collection of these when they're all glued together — and it was a considerable task to develop the fixturing to be able to bond them together so that they weren't doing this in a stack-up — resulted in something that, in this case, looks like a square box. The other one looks more like a radiator, actually. A lot of people said it looked like a Model T radiator. The tubes themselves were an interesting technical challenge. Wall-thickness-to-diameter ratio of a tube like this is very high. Very large diameter, very thin wall. Represented a state-of-the-art drawing at the time it was done. It's probably still technically challenging. In fact, the aluminum was so thin that it was necessary for sample tubes in process to actually draw them on a copper tube inside, to make sort of a sandwich. Then once the aluminum-copper sandwich was drawn to size, then the copper was etched out to leave these tubes.
DeVorkin:Can you put that — again, we can see sort of a copper — [Tape recorder turned off.]
Austin:Here's an example of one of the tubes in an intermediate stage of the manufacture that shows the aluminum drawn on top of a copper tube to give it enough strength and to support the relatively softer aluminum, which would tear. The sandwich was drawn in an ultrasonically excited dye to help the extrusion process, and then, once it was drawn, it was then cut to length and the copper pieces inside etched out to make these lighter weight final tubes, which were then bonded together. There was a block of these tubes made that was this wide and as high as a sixth of these and mounted in a frame, and the detectors were all mounted on the back. And then the entire payload rotated about the vertical axis. It was a quarter of a revolution per minute, I think.
Murray:Twelve minutes per revolution, I thought it was.
Austin:Okay. And thus the collimator would scan across the sky, and we could then tell by knowing where the spacecraft was pointing at any instant in time exactly which portion of the sky the X-rays came from, and we would make scans in one direction and then cross in another direction and so on to more precisely pinpoint the location of the X-ray sources that were found. But you know more about that than I do.
I came onto the Uhuru Project two weeks after the satellite was successfully launched, so I had all the benefit of having everyone build it, and I had all the fun of actually working on the data. And what Jerry said is correct: the satellite is a spin-stabilized satellite. It rotated around the axis of the proportional counters, and one side swept out of the sky with these half-degree-by-half-degree collimators, giving us sort of a crude image of the sky, and the other side swept out the sky with a half-degree-by-five-degree collimator so that we got a sharper view of the sky. But because it was a smaller part of the sky, we had a little less sensitivity on that side, and so it was a compromise between a high sensitivity and a low sensitivity, modest position and accurate position, at least for those days, operation. What we would do is we would pick a pointing direction for the spin axis, and then we'd let Uhuru spin around as it went around its orbit around the Earth, and that meant that it would sweep out what we call a great circle on the sky.
We would find all of the blips in the data for that, and that would tell us that along that direction was an X-ray source. Then we would tilt the whole satellite into a different spin axis and scan another great circle. Where two great circles intersected, if there was an X-ray source, we'd get two blips, and that triangulation gave us then the actual position of the X-ray source in the sky. That was how we went around building what we called the Uhuru Catalog of X-Ray Sources. In 1972, basically, we were able to go from about ten known X-ray sources on the sky to about 150 known X-ray sources on the sky, and then throughout the rest of the life of Uhuru, we added to that until we had a final catalog of about 400 sources, which was the first all-sky survey of X-ray sources that had ever been done. And in addition to being able to locate the X-ray sources, as Jerry said, we were able to measure their X-ray spectrum and we were able to, of course, measure their intensity, but additionally we were also able to tell the arrival time of each photon at the detector, because the proportional counter gives a signal when the X-ray is detected. You can time-tag that signal. That turned out to be very valuable, because we found, much to our surprise, one of the great discoveries of Uhuru were these things called pulsating binary X-ray sources. These are sources that would fluctuate in intensity on a regular period of pulsations, and then those pulsations would come and go in a regular way. By analyzing that date, we were able to determine what we were seeing was a spinning source of X-rays rotating in orbit around a second star. That was the binary pulsating X-ray source discovery.
The first of those was Centaurus X-3, and then it was quickly followed by Hercules X-1, and that opened up a resurgence in a whole field of astrophysics, which was what happens in close binary pairs of stars. I'd say the second great discovery with Uhuru was that we actually measured the extent, the non-point-like behavior of X-ray sources coming from outside our galaxy, the discovery of standard emission from clusters of galaxies. That also was totally unexpected phenomena, and it was the beginning, again, of another resurgence in that area of study. The discovery of the fact that the medium between the galaxies that are in a cluster of galaxies is not empty, but filled with a hot X-ray-emitting gas, has turned out to be something which we followed up in all the other observatories we've ever built, learning more and more about how the mass balance in the universe is really tilted towards hot ionized plasmas rather than material actually in stars that make up the visible portion of what we see in galaxies.
DeVorkin:Referring back, then, to pre-history, designing Uhuru, you designed it as an exploratory instrument then, I take it. So what kind of information did you have that gave you a sense of what energy range you wanted to be able to detect with these detectors? You mentioned that different detectors had different energy thresholds.
No. Two detectors had the same energy threshold. They had different angular responses, different spatial patterns. But the answer to your question is, from the sounding rocket work that had preceded Uhuru and the discovery of a few X-ray sources, we had some indication about what the characteristics were, what energies they were emitting at, for example. The other part of the answer to your question is that what we really wanted to do was build an instrument which could be sensitive to the lowest energy X-rays we could possibly make it sensitive to, but still make it work. The limitation there is the beryllium window on the proportional counter is an entrance window which is a thousandth of an inch of beryllium, and it's low-energy threshold is right around the one-and-a-half to two-kilovolt band. That's as low as the X-ray energy you can get to. Had we been able to build a system with a different kind of a window and with a gas flow system and had more weight available to build this instrument, we would have gone to even lower energies, because the general rule of emission for X-rays is there are more photons at low energy than there are at high energy.
That's sort of a general statement about the spectral shape of things. So, going to lower energies would have given us more sensitivity. On the other hand, competing with that is our own galaxy is, filled with an interstellar medium which absorbs X-rays, particularly at very low energies. So as you go looking further and further out through more and more of this interstellar medium in our own galaxy, there's a natural cutoff in low-energy X-rays because of all this gas between us and things beyond our galaxy. So, in fact, looking at the sky with a detector like Uhuru's in what was classically called the two-to-ten-kilovolt-X-ray band did not really prevent us from seeing most of what we wanted to see. In follow-up missions that occurred in that same decade of the seventies, there were observations made in what's called the low-energy band, say from about 100 EV to 2 kilovolts, observations in this classical two-to-ten kilovolt band, and then further observations in what we would call the high-energy band, above ten kilovolts. But in terms of finding sources and beginning the exploration, this was a very good band to work in. It turns out that we've missed very few kinds of objects by only working in this band.
Austin:I could really represent it as well as a combination of the technical limitations at the time. The beryllium foil in the windows — they're 1 mill thick — represented the state of the art in being able to roll beryllium into a thin foil. Beryllium foil is made by taking a piece of beryllium and sandwiching it between two pieces of stainless steel and welding it shut to provide an inert atmosphere for the beryllium so that it doesn't oxidize rapidly, and then rolling, hot rolling it, and when it's done, cutting away the edges and removing the stainless steel to leave this thin beryllium foil. And to try to get that foil leak-tight, which this must be at 1 mill was, itself, a challenge to the beryllium companies to make it. Because, as Steve said, we had no on-board gas system. We did not have the weight space or the power to operate such, and these counters were sealed. This payload wound up working for over four years, I think, on orbit, and as far as I know, we never suffered a single detector failure in the whole mission.
Austin:So obviously what we did worked, but it really represented the state of the art in materials in many areas at the time we built it.
Murray:I think another interesting comment about Uhuru is, when we built up the bank of six of these on each side, the total amount of detector area was about 800 square centimeters of detector, and after you take off the obscuration of the collimators and various other effects, what we call the effective area or the useful area of the detector was about 400 square centimeters per side. That's about the amount of area in half a square of eight-and-a-half-by-eleven paper in terms of amount of actual useful area. Since that time, with the exception of just a few missions, the effective area of most X-ray missions is in the same realm. That is, we have improved how we do the detection. We've improved how we do imagery and so forth, but the typical collecting area of an X-ray mission over the interval from 1970, when Uhuru flew, to 1995, say, where we are now, we're '98, where we are now, is 1,000 square centimeters. So in terms of the size of the detectors or the effective area of the instruments, not much has changed. The only mission that significantly was different than that was with Hio I [phonetic], which was designed to have very, very large collecting area, like 1,000 square centimeters, and it suffered from limitations in terms of what it could do in sensitivity because of the way it was designed to scan the sky. It had source confusion problems. So its sensitivity wasn't all that much greater than even what we had with Uhuru. What it could do, because of its larger area, was collect more photons and do better measurements of spectrum and better measurements of timing, but in terms of its overall sensitivity, it wasn't a heck of a lot different. The way we actually improved sensitivity was not by making collecting area; it was by going to imagery, real focusing optics, that made the difference between the source and the background that much higher in contrast than when you just use these collimators.
DeVorkin:Let me ask about the geometry of the front there where the beryllium window is. Basically I'll ask you, is there any meaning to that geometry of grids in front of the beryllium window where some of them are thick and then there's other squares? And did the collimators actually fit into those squares, or it has nothing to do with them?
Austin:Okay. This sort of a strongback here was designed primarily to support the windows, and there's a couple of design considerations here. One, the thicker ribs at these locations represent the end of a piece of foil. They could only make the beryllium foil this large. It is not one continuous sheet. You need a little extra spot here to seal the ends of them in these areas. Out in here, it is a continuous sheet, and you only need a narrow rib required to support the window. The size of this opening was dictated by structural design considerations as to how large that could be without risking the window rupturing due to the fact that there's about one atmosphere of pressure inside — actually a little over — and in space essentially nothing, and that pressure differential could cause the window to pop out. Ideally, you'd like to make them as large as you could. If you could make one big window, it would be better because all that framework represents lost area. So you made it as large as you could, but still safe enough to prevent the windows from exploding. The collimator tubes do not fit into those openings, and it was not possible to design this and the frequency of the ribs and the collimator to be the same, especially with features like this in the detector. So it's just a question of having to take the hit. There's a certain fractional area lost here, a certain fractional area lost here by the windows. You multiply the two together, and you just lose that much.
DeVorkin:How was this design decided upon, what process?
Austin:Well, actually the development of the detectors itself was a competition between three industrial firms: a company called L&D in Long Island, who actually finally won the job and built these detectors; a company called Reuter Stokes [phonetic] in Cleveland; and a third company called Harshaw [phonetic] Chemical, also in Cleveland. And we had a little initial competition amongst the three firms. We gave them a specification, gave them our envelope dimensions, the thickness of the foil we wanted, the pressure and so on, and then said, "Build us a detector." We gave each of them a contract to build a detector, and one company, Harshaw Chemical, dropped out. They were trying to do their sealing of this front frame using some electron-beam welding techniques that didn't work. They dropped out. Reuter Stokes and L&D both delivered detectors. They both worked. They were tested, and basically it came down to financial considerations. L&D was willing to make these for a fixed cost and Reuter Stokes was not, and so we took the opportunity to take advantage of the fixed-price contract and bought these from L&D, and it really, although L&D had made detectors prior to this, represented a real step forward for them as well. They had not built, really, any aerospace hardware of any reliability before, and they had a lot to learn in the quality areas and so on, and we assisted them with, but wound up producing these detectors, and they produced detectors for us that have flown on subsequent missions as well, and have built general very good reliable detectors.
DeVorkin:So your group's role was at the contractor end, in making sure these met your requirements.
Austin:That's right. We established the requirements, the design requirements. We had to work very closely with these people because what we initially wanted in some cases wasn't always feasible. We would have liked to have half-mil beryllium foil, but it wasn't really obtainable, certainly not in a leak-tight quality. So there were trades that were made and a certain evolution of the specification during that prototyping phase, which led to the final specification for the flight articles, which is what was used to build these flight detectors.
DeVorkin:And the co-ax connection at the end?
Austin:That's just a temporary connection for ground testing. If this was off, you'd see a hollow cavity there. You'd see a place here for a fitting. The high voltage required to operate this detector and the other detectors, as well as the high voltage that was used to operate the photo-multiplier tubes in the Star Tracker was a real technical challenge to us. At one point in the program, we had high voltage power supplies, we had a junction box, we had these detectors, we had photo-multiplier tubes, and none of the high voltage potting was working. We had a big boarding committee came up from Goddard Space Center, about fifteen or twenty people, and we spent the day telling them what we were doing trying to resolve all these issues, and we must have done a good job.
DeVorkin:Are these connections —
Austin:These connections on the back were just to mount a preamplifier board. These signals that come out of these detectors are so weak that you can't afford to run them very far away. It's necessary to give them some initial amplification. So there was a preamplifier that went on the back and a little aluminum housing that covered that up and covered off the end of this. The signal output lead came out this little hole here up onto the preamp board, and the high voltage went in this hole in the front, and there was simply a cover over the end. We wound up encapsulating these with an epoxy, although we subsequently changed on a follow-on mission to a silicon rubber, which we found to be better. The epoxy was very unforgiving. If there was a problem with it, you couldn't dig it out. Basically it was throw the unit away. But we switched. We found a suitable silicon rubber, which had the option of being able to remove it and repair it.
DeVorkin:Is there anything else that we should know about this detector as the visitors come into the museum and encounter this detector? What you've given me already, of course, is a good feeling for how it works, what it did during the life of Uhuru, and I think that's very important, but is there something visual that I've forgotten that the visitor will look at and wonder about, do you think? Any way of being able to predict that? I know that's a tough question.
Murray:Someone might ask you how long this could have lasted. I mean, we mentioned that we ran Uhuru for about four years. The detectors could have lasted probably for ten or fifteen years, and it was the satellite power system, actually, that got us on Uhuru ultimately, but the detectors themselves have a very long lifetime. I wouldn't be surprised if we could turn this on again.
Austin:I was going to say it would be fun to take this in the lab and turn it on and see if it worked. (laughter) It probably would. The lifetime of the detector is basically limited by, in this case, the filling gases, argon and methane, and it's limited by decomposition of the methane gas, which will ultimately build up a thin carbon coating on the wire, which destroys the performance of the detector. That's strictly a function of how many X-rays have been detected by the detector. Unfortunately, there's not all that many X-rays on orbit. Generally these detectors see far more X-rays on the ground during calibration and test than they see in their lifetime. Uhuru had a six-month design lifetime requirement. We obviously met and exceeded that by quite a bit.
Murray:Yes. I think, if I remember right, the average counting rate in the detector with the five-degree-by-five-degree collimator was about thirty counts a second. That's how many X-ray we were seeing, which is mostly sky background and detector background and very few coming from the sources to actually go through the collimators. And on the side with the half-degree-by-five-degree collimator, the average counting rate was about five counts a second. So we were not really worried about total count rates. We were starved for photons, which was are even today.
Austin:Always, yes. I would guess probably a detector like this would have a ten-to-the-eighth or ten-to-the-ninth count lifetime, which we never came near.
Murray:Never came close.
DeVorkin:Okay. Then that does it for Uhuru.
Murray:One test was this classic dot test in the imager, so it's just going down from one corner to the other diagonally, and that looks fine. The other test was a classic dot test and nothing real.
Austin:Well, I'll give you one with the copper in it, too. How's that. You're the associate director, and you can tell about that and see what it is. (Murray laughs.) You see the difference in the weight there with the copper. These have been etched as well in preparation for bonding.
Murray:Five-to-one or so?
Austin:Oh, probably at least that, yes.
Murray:Ten to one?
DeVorkin:Okay. We're now turning to discuss the origins of the cross-rib charge detector, circa 1972-1973. Steve, would you carry us through this? But first give us just a short preamble of how you came up with this idea and what it's all for.
Sure. The reason for developing a detector like this starts with the idea that we wanted to make detectors that could actually take pictures, be actual imagers or electronic cameras for X-ray astronomy instead of just sweeping across a field with an external collimator. And since we knew how to make optics that would focus the X-rays onto the detector, the next step was making a detector that would actually take advantage of that. What we started out with originally was an idea of using film, photographic film, behind basically a light amplifier that would take the X-ray, turn it into visible light, make it bright enough, and then expose a piece of film. And we actually built a camera like that and flew it on a sounding rocket. But if you wanted to actually do that for a space observatory, you'd have this problem of how to get the film back. And so the next step was to replace the film with an electronic film, and that was the origin of the need to build some kind of detector that would do that. What happened for me was that I was perusing the literature, the technical journals, and I saw an article about building a detector that would hypothetically do imaging by building what's called a delay line.
It was resistive wires made out of nichrome that would be connected to each other with capacitors, and then the amount of time that it takes for the signal to go from one wire to the next wire to the next wire would vary, and if you could measure the time, then you would be able to tell which wire the signal came on. And I looked at that, and I said, Gee, that seems to me a very complicated way to do something. Might there not be an easier way? And I can't tell you when the epiphany occurred exactly, but I thought, Well, instead of using a delay line, why don't we use a charge division approach. And the idea was to make a bunch of wires and connect them with resistors and say, okay, when charge comes onto a wire and it's connected to other wires with resistors, some of the charge will go down one path and some down another path proportional to the amount of resistance between the paths. And that's the birth of this cross-grid charge detector. Having gotten the idea, the next step was how you build one. And what you see here in front of me is the first early attempts at how to do that. The idea was that we needed to have a plane of wires evenly spaced and then a second plane of wires above it, physically above it, and also evenly spaced so that you could actually detect X positions in two dimensions That was the idea of the crossed part of the cross-grid charge detector. So we built this little glass frame. That was pretty easy to do. But then how do you put the wires evenly spaced and up and down differently? What I did is I went to a little company in Waltham, Massachusetts, that had a diamond ruling machine, and they could just cut grooves in glass, and we made little inserts, these little edges here, ledges, that were sent to this company, and they would just groove a bunch of little lines in the glass.
My idea was I would just take some wire and it would fall into these grooves, and I would just wrap it around. And then I would do the same on the other side, and that would make this cross grid. That was easy to say. It took about a year of experimentation on the size of these little grooves in the glass and the depth and getting the spacing uniform and gluing that piece down onto the glass frame until we actually got wires to sit in their grooves, be parallel to each other and separated into two planes. The result of all that is the device that I'm holding in my hands now, which is one of these glass frames with these steps of ruled glass, and you can see the wires. There's a bunch of wires going this direction and a bunch of wires going this direction, and you can actually see the little epoxy glue that I used to hold the wires in place. What actually was done is the wires were wound around. [Off-tape comment.] So what was actually done is that a continuous piece of wire was wound around in these grooves and through the back and around to the next groove and through the back and so on, and then when that winding was all done, then I put this strip of epoxy here and strip of epoxy here and let it set up, and when the epoxy was finally cured, then I took a wire cutter and cut the wires off at this end and unwound the wires from the back so that I had wires sticking out that I could then connect to pieces. Each wire had to be connected out to this printed circuit board, which we had made up, so that on the other side of this board we had the actual resistors that we were going to connect each wire to its neighbor. These little resistors were put on here, and, of course, we had to span them out so there was actually enough physical space. There was only about sixteen or — I guess thirty-two wires in this regime here, thirty-two resistors, and you'll notice how much bigger it is than the actual space of the detector. And then, arbitrarily, I must say, I chose to take every eighth one of these wires and bring it out to an outside connection, and those would go the actual amplifiers that would measure the amount of charge. The idea, basically, is that if we can have electrons coming down and landing in these wires here, so there will be a big circle of charge, and then some of the charge would go out towards this wire and some would go out towards this wire, and some would go out towards this wire. As I said, basically, in the ratio of the amount of resistance between where the cloud of electrons landed on the wires and the amount of resistance to get from where the charge was to the amplifier.
So when the charge is sitting right over here, most of the signal comes on this wire. When the charge is sitting, say, between two wires, you get half the signal here and half the signal here, and so on. And by taking the appropriate ratios, you could actually predict — or not predict — measure where the center of the charge was, and that was done in this dimension and then, at the same time, in the opposite dimension. So this was actually the first successful device. You can actually see, if you look real closely in the center of this, you can actually see the oxidation on the wires. That discoloration comes from the electrons that come out of the device that actually converts X-rays to electrons, which we haven't discussed, landing on the wires. The electrons, when they land on the wires, tend to produce some chemical changes occasionally and you get this discoloration, which is really an oxidation effect. So you can see where the signals were actually landing.[Off-tape interruption.] So this was the original first working model. I shouldn't say original. We had lots that didn't work, but this was the first working model. And then what I wanted to do after I built this one was I wanted to understand whether I had optimized any of the design premises, how fine a spacing of wires do I need, what size resistors do I need, how frequently should I make a signal come off from the detector. So I went from the Version One over here to what I'll call Version Two, which looks very similar. You can see already that there are some improvements. For example, the epoxy looks a lot neater on Version Two. It went to a better use of epoxy. The spacing of the wires in Version Two is twice as dense as the spacing on Version One. These are wires which are half the diameter, much finer wires, and there are twice as many for the same linear distance, and that's why this circuit board is actually two circuit boards put together. And now the little resistors are actually connected between the two circuit boards. It's a little more complicated to build. In fact, I got assistance building this one from some of our technicians. The first one I did pretty much my own work, but I couldn't quite get my fingers in there. Then this was used to demonstrate that we got no better performance with twice the density of wires than we did with half the density, which meant that it was much easier to build it. We also found that we didn't need to change the spacing of how often we took out signals. Now, this is only about a quarter of an inch by a quarter of an inch of detector. The next step in developing this unit was to build up a full-sized unit that would be appropriate for the next mission that we were planning to use this on, which was the Einstein Observatory, HEOB, and that was on a mission which we flew in 1978. So this work was going on from the early 1970s through basically 1977, when we finished designing and building the flight model for the Einstein detector.
DeVorkin:You were consciously working toward this imaging X-ray telescope. This was just not blue-sky technical development; you had a mission going on.
Murray:That's right. By 1972, I think it was, we had a start on what was called at that time HEOB, which was the second in a series of high-energy astrophysics observatories. The X-ray telescope for that was conceived and the first optical elements were being worked on. So we knew we were going to have a focusing telescope and we had a complement of instruments that was a series, a suite, of instruments were going to be flown on that observatory, in the focal plane of the observatory, and you'd be able to switch from one instrument to another as you did observing. One of the objectives was to get the highest possible angular resolution out of that telescope, and that's what this detector development was all aimed at.
DeVorkin:Now we go to the third stage?
Murray:The third stage, which represents a very large passage of time, in fact. These developments took about two years, and then the next step was how do you make something that you can actually fly. This is not something that one would want to put in a long-lived observatory in space.
DeVorkin:Just a brief question aside from that. While you were doing this, you were also analyzing Uhuru data?
DeVorkin:How did you make decisions about how your time was apportioned for all of that?
Murray:I worked on detectors during the day, and I worked on Uhuru during the night. (laughter) Basically, work like this that goes on in a laboratory is not something anyone does totally independently. There are technicians and engineers, and lots of people have to support you and machine shops and other locations. So basically you do the hardware development during the day, in all seriousness, whereas data analysis is something that you sit down in front of a computer and work mostly by yourself. So it was really that kind of a split, a day and night kind of a split. And I would say sixteen-hour days were not uncommon. [Off-tape discussion.]
Okay. The original group of scientists and engineers that put together and flew the Uhuru mission split up in the early seventies. The scientific portion of the team left the company that did Uhuru, which was American Science and Engineering, and came here to the observatory. The engineers were still employed at American Science and Engineering. This early development that Steve has just shown us with these glass-based devices was largely done here at SAO without too much involvement by the engineering staff at AS&E. Then in probably the 1974 time frame or so, they came to us and said, "Okay. Here's our concept for a detector that we want to build. Can you take these versions, these laboratory versions here (which we've seen) and produce for us something that we can fly?" And there were additional technical requirements placed on the design at that time. As it was originally envisioned, the HEOB mission would have two types of high-resolution imagers –- HRIs — one of which was to be a microchannel plate-based detector where microchannel plates were used to convert the X-ray photons into electrons that would land on these grids.
The other was called the negative electron affinity detector, that had a very special crystal in the front. It involved making a sealed tube. The output of that tube, which was a visible light image, was then to be converted to an electronic image by what we call the visible light image converter, which was, again, a microchannel plate-based device with one of these cross-grid detectors in it. But that was also a sealed tube. We had a development under way with the Varian [phonetic] Corporation on the West Coast to do that, and that really drove the design for this block, because being situated in a sealed vacuum tube meant that it had to be baked out at around 500 degrees Centigrade or so on in order to be able to seal that tube and give it any useful lifetime. That's a pretty severe requirement, to take one of these laboratory devices and say you've got to make it not only to withstand the dynamic environment of launch, the vibration and so on, but it must also withstand 500 degrees Centigrade temperature bake-out for twenty-four hours. So that's what led us to change from the glass-base material to this ceramic. This white material is a ceramic material. We had to find some wires that would also withstand that, and we had to adjust the tension in the wire, find a wire strong enough so that we could wind it at room temperature and have the wire be tight enough so that when we raised the temperature to 500 degrees Centigrade, the ceramic block would basically not expand very much at all, but the wire would expand greatly. But we did not want it to expand so far that it fell out of the little grooves in the block.
So there was a study that was done to find materials and the right winding tension and so on. We were working very closely with Raytheon in Bedford, Massachusetts, in the development of this block. This is an early non-flight version of the detector. We see here the same thing we saw in the early models, two orthogonal planes of wires closely spaced, slightly separated. The wire in this case we settled on was finally a four-mill-diameter gold alloy wire. We did not use the notion of the inserts in the edges like was done with the prototype models, the diamond-ruled glass pieces. We found that we could, in fact, laser-cut notches in the edge of this block using a machine that was actually built to trim thin-film resistors that Raytheon had, but this block was pretty large, actually, for that machine, but we managed to squeeze it in there and developed a technique to put little laser — little notches in the edge of the block that we used, then, to wind it. And again we wound it just as the early prototypes were done. The wires were wound continuously around, both directions, and then epoxied. And that was another interesting choice of materials. We had originally hoped to be able to use a glass frit [phonetic] material, which was very well suited to high temperatures. We had some interesting experiences when we did it. The first one we did, we wound it carefully with this glass frit material underneath the wires, and we sent it to Raytheon, and they fired it. A glass frit is basically a lead oxide. The first one we did, we did in just ordinary air, and the wires were badly tarnished and oxidized after we did it. We said, well, this is not acceptable at all. So we built a second one. We said, this time we'll do it in a hydrogen-reducing environment so that we will not oxidize the wires. And we wound up reducing the lead oxide in the frit —
Murray:(laughter) Highly conductive.
Austin:— to elemental lead, which is totally conductive and shorted all the wire. That wasn't too effective either. We finally settled — this one was built with one of these early materials. There's this white material we can see in here. It's a sourizing [phonetic] cement. It had a long history of being a cement that was usable in vacuum tubes and so on and could withstand that temperature.
Murray:The bead of white.
Austin:Right here. You see some of that white material, and actually, for redundancy's sake, somebody's even added some out here just to make sure the wires stay in place. There was an extra band put out here.
Murray:An extra piece of ceramic holding it down.
And we built — yes, what these are, these little white strips here are ceramic strips that are put on there to clamp the ends of the wires. These little blocks here are the resistor blocks. You can see the development that went on here in terms of the discrete resistors that were on the circuit boards over here. Look at this and say — now, there's 129 wires in each direction. There's 128 resistors required. And here we only have probably two dozen. So you can imagine trying to squeeze all of these resistors onto this block. It cannot be done with these discrete resistors. So we went to this thin film technology, and it's not possible to see it at this magnification, but there are actually individual resistors there on that ceramic strip. After those strips are bonded in place and these clamping strips are put on there to hold the end of the wires, we came in and we laser-cut the wires out of the middle to separate the continuously wound wire into individual wires bonded to the block. Then it becomes necessary to make the connection now between these resistors and the individual wires, which was done in this unit by these traces on the circuit board, and the resistors were mounted on posts on the other side. Here we used, again, semiconductor industry technology and did what's called ultrasonic bonding.
The wires, the main grid wires, which are relatively huge — they're four thousandths of an inch in diameter, which is about the diameter of a human hair — were then connected to the resistors with 1-mil-diameter gold wires. There was a bond made basically from the big wire up onto a pad on this resistor network with this wire, and that was done on both edges here, and that was done, again, using equipment that was available at Raytheon. And then every eight wires, we would bring out a tap. The technology at that time was such that we couldn't get all 128 resistors on one pad. There's actually only half of them here and half of them there, which complicated the wiring to be hopping back and forth from side to side. But these early versions required four resistor networks. And that's pretty much the story. We didn't discuss — in this first model you can see here an aluminum block up in the center, which we call the reflector plane, which was biased electrically in a way that it encouraged all of the electrons to be captured in a grid. The grids are 50 percent open. They're 4-mill wires on 8-mill centers, and so some of the electrons are captured by the front grid. Some of the others go through and are captured by the rear grid, and we try to balance it so that we collect roughly equal charge on both grids and then bias the reflector plane to make sure that we collect as much of the charge as possible in the grids and that there aren't some electrons that just go through and aren't detected and therefore wasted. Here we see a different implementation of the reflector plane. Now it is simply a deposited silkscreened gold layer on the block, which is put onto the block prior to winding the wires, rather than being the separate metallic member. And in fact, you can see, maybe, right in here, evidence of the electrical connection that is made to that grid. It's very small. There's actually two wires there for redundancy that come down and go through a channel here and are covered up in there with this sourizing cement.
DeVorkin:You mentioned that you had to convert the X-rays to electrons.
Murray:I'm glad you asked that question. First of all, I should say that from the science point of view of developing technology, this was interesting work. The conversion of a concept to something that could be actually used and flown is at least orders of magnitude more work in many ways. The original thought is interesting and exciting, but to be able to actually see that go from something you could build down in your basement, basically, to something which was of industrial strength was a real feat. And we often don't give credit to the technicians, engineers, and the support people who actually make space science possible. I just want to make a comment here, and maybe you'll keep it or maybe you won't, but that this is what actually is the strength of a science-engineering organization, that we can both come up with concepts and we can turn them into something we can actually use. I spent two years doing this, but the engineers spent three years making this, and I think that's about the right ratio. [Off-tape comment.] I spent two years making these prototypes, but the engineers had to spend considerably more time, three or four years, converting that idea into something that would actually meet all the requirements of being a space-qualified instrument.
Austin:It wasn't too far away from doing this in the basement as well. Actually, these initial blocks were wound — I shared an office with a fellow named Ron Whitty [phonetic], who worked for me, a mechanical engineer, and Ron was the one who was heading up this design activity on these blocks, and we wound them right there in the office with a little winding fixture clamped to the corner of the desk.
Austin:So we've come a ways from it since then.
Murray:I'll also admit that — Jerry reminded me — of these two detectors, only one of them actually survived the program, the negative electron affinity device, which I also worked on, I'm happy to say actually worked, but it had a lifetime of about a millisecond, unfortunately. (laughter) And we never actually completed it for the HEOB Program, and new technologies supplanted the advantages of that technology. So that actually was a great deal of effort from which we learned a lot, but it ultimately did not lead to a flight detector. On the other hand, the simple HRI based on microchannel plate detection has survived and actually has now entered its third generation of use, and we'll get to that, I hope, later on.
Austin:Let me just make the comment, though, that it's interesting. We ultimately did build two visible light image converters at Varian. They were successful. Delivered them, tested them. One of them had a rattle inside of it, but they actually did work. But once the NEAD part of the detector died, the visible light image converter sort of died along with it, and it removed this restriction on the design of this high-temperature bake-out. For that reason, we ultimately were able to make some changes and relax the material choices. The Einstein era, HEOB era, cross-grid charge detectors were made with the sourizing cementing. Subsequent to that, because we eliminated the need for the high bake-out, we were able to switch to epoxy materials, which were more reliable. We occasionally had problems with the sourizing. One of these resistor pads would just simply lift up. The whole grid would just come off, and we always lived in fear that one of your flight detectors, you'd look at it one day and find one of the wire grids not connected anymore. So we changed to better materials, and yet some of the features of the design, the winding tension, the choice of the ceramic and so on, have remained to this day, even though there's not a particular need for them that there was originally. They have persisted throughout the present designs.
Murray:Let me close that chapter off by saying that what superseded the negative electron affinity device and the visible light imaging detector was the advent of commercially available charge couple devices, CCD devices, which made both of those applications no longer relevant for either X-ray astronomy or visible light astronomy. But at the time, we actually did take one of those VLICs, V-L-I-Cs, out to a telescope and made what we think were the first photon-counting images ever done at an astronomical facility. Just a chapter that came and went in our history.
Austin:Not the development of this, but the development of the application of this device in something like the visible light image converter came to an end.
Our efforts to make this device, which we called the negative electron affinity device, and the visible light imaging converter were all supplanted by the advent of charge couple devices and their development for astronomical purposes and for X-ray astronomy purposes, and that put the kind of development we were doing here to an end point. But, as Jerry pointed out, the cross-grid technology and the microchannel technology remained relevant and still relevant in what we're doing today. Now, the question that we haven't addressed yet is how we make the electrons that fall upon the cross-grid charge detector in the first place and how we convert X-rays into electrons and electrons into the readout. The way we do that is with the device which I'm holding in my hand here, which is called a microchannel plate. A microchannel plate is a special kind of glass. It's a semi-conducting glass which is drawn into very small tubes, again smaller than the diameter of a human hair, and those little tubes are then packed densely together, fused together, and then sliced into thin plates.
And so if you actually could get a bright enough light looking down a microchannel plate, you'd actually be able to sort of see through it. It's a bunch of little tubelets. Okay. This microchannel plate, as I said, is made up of these very small tubelets, and what happens is that an X-ray hits the top of the microchannel plate and knocks out an electron from the inner wall of the glass, and because there's a voltage applied across the microchannel plate, electrons are drawn down into the tubelets, and they, in fact, hit the wall of the tubelet and knock out additional electrons, and this continues in a multiplication process so that one X-ray producing one free electron at the top ultimately produces between one and ten million electrons coming out the bottom. And that's actually achieved by stacking up not one but two microchannel plates, one on top of the other. They're held in this holder, which is a ceramic plate with the necessary electrodes to provide the voltages on the microchannel plates and then an exit surface so that we have the right spacing between the end of the microchannel plate and the read-out device. The actual instrument is a stacking-up of the microchannel plate detector on top of the cross-grid charge detector so that the X-rays come in, the electrons come out of the channel plate and then fall on these wires and the cross-grid charge detector. Now, just to give you an idea of how accurately one of these devices works, this is a test pattern or a test mask that we use to measure the spatial resolution of the detector. I don't know whether we can get close enough to actually — there you go. You can see in this picture some of the test pattern. You can see some very large slits and four quadrants, and you can barely make out the fact that there's a central pattern.
Austin:There's a central area right in there, a little small square right in the middle, that says in the middle of it "HEOB" or, in this one, probably, "X-ray." Or maybe we have a picture later on we can show them of the —
Murray:Try to show you a picture. Now, what we do is we take that little resolution pattern, and we stick it on top of the stack like this. Actually, we have a holder to do that. Then we illuminate the detector with X-rays. Let me see if I can tilt this so you can get a better view. We illuminate this with X-rays, and then we record the image that we get from this, and by analyzing that image, we can then make a quantitative assessment of how good the resolution is. The smallest pattern in this mask, in the very center, has components which are only twelve and a half microns in width. Twelve and a half microns corresponds to what?
Austin:A little less than a half —
Murray:A little less than a half of a thousandth of an inch. We have been able, with this detector, to resolve those features. That means that in this one-inch-square read-out of our detector, we have something like 4,000 independent image elements or pixels. So we have altogether something like 16 million image elements in this one-inch detector. That's, I think, a pretty remarkable accomplishment, which is about as good as any imaging detector that's been built to date.
Austin:It's interesting. Even though we've developed these things beyond this, as we'll see a little later on, all we've done really is to increase their size. We've never done anything substantially to be able to improve the resolution of this detector.
Murray:That's correct, and part of the limitations for that are the sizes of the microchannels themselves, and the other part of the limitation is the fineness of the read-out grid. We have not improved the resolution, not because we haven't been able to, I think, but more because we haven't needed to.
Austin:No. We haven't needed to, either.
Murray:Because of the fact that the way that these instruments are used, the resolution we achieve is actually more than adequate for the kinds of telescopes that they're being placed behind. I think it would be useful to go to the blackboard and try to illustrate some of these functions of how this works by sketching it out, and maybe we can go through some of this process again but with some sketches.
Murray:So, David, you should ask me one and then I can answer you. (laughter)
DeVorkin:All right. In profile, how does the combination of the microchannel plate and the cross-grid detector work? If you could carry the life of a photon through to becoming an electron and then being recorded.
Murray:Okay. [Drawing on blackboard.] We'll start with a cross-sectional view of the microchannel plate detector, where this represents one of the microchannel plates, and these represent the little tubelets in the microchannel plate itself. We have two of them, as I said, one stacked up against the other, and, in fact, we have the tubelets on the second microchannel plate at a slight angle to the tubelets on the first one, which helps reduce some of the internal background in the detector. What happens is that an X-ray coming from the telescope or whatever source there is comes down, and it will eventually hit on the side wall of one of the tubelets and free up an electron, and the electron jumps off. Because there's a voltage across these microchannel plates, it's attracted in this direction and hits the wall, and then several electrons come out, and then several more come out. By the time you come down through this little first plate, you've gone from one to an order of a thousand to ten thousand electrons. They spread out a little bit and hit not just one tubelet, but several tubelets on the second microchannel plate, and the same acceleration process occurs. Then you come out from the second channel plate, there's a cloud of electrons, roughly between one million and ten million electrons. Then below that are the wires of the cross-grid charge detector. So these are the wires running this way, out this plane, and then below them are the wires that are running orthogonal to that. The cloud of electrons intercepts the wires, and some of the electrons are absorbed on a wire and go off. Some go through because it's half opened and they land on these various wires, and they go off into their direction. As Jerry mentioned, we have this reflector plane, so some electrons come down and curve back up and go back to the wires so that we collect essentially all of the charge that comes out of the microchannel plate. Now, if we then take a look at what happens at the detector by looking down on the detector, here's now a top view. The electrons are coming into the blackboard.
DeVorkin:And these are wires.
Murray:And these are the individual wires, and they're connected to each other with resistors, just as we said, and the cloud of electrons is landing on the wire. So there's this much electrons coming on this wire and this much on this wire and so forth. So if this wire is connected to an amplifier, a certain amount of signal will show up. Down further, there's one more wire with its amplifier, and it will see a different amount of signal. Up here higher, there's yet a third one coming out and it sees its amount of signal. What we can do is calculate the ratio of the charge, the difference in charge between this top one and this bottom one divided by the sum of all the charges, and it turns out that that fraction, which I'll call (A-C) or (A+E+C), where this is A, B, and C, equals the position. That's something you can work out, but it's just because charge divides between the amount of resistances. So this tells us basically between this location and this location where the event occurred, and then by knowing that it was Amplifier B that was the middle, we get the absolute location. This is called the coarse-fine position. This is the fine position, and the coarse position is the Amplifier B. So now if the whole thing moves up one unit, we get a new Amplifier B, a new fine position. So we can make a detector which gives us a very accurate position in this fine coordinate, a very coarse position in this coarse thing, and then we can make the detector as big as we want. In fact, in terms of design, this was a major advantage of this approach to earlier approaches, because we could make the detector bigger. And, in fact, just as an example —
Austin:One thing you might want to emphasize is the single-photon detecting aspects of this detector. We really haven't emphasized that.
Murray:That's a good point, and let me get back to that as soon as I finish this thought. On Einstein, our detector was 25 millimeters in diameter, and we had, as I said, about 4,000 pixels. That was circa 1978. By 1990, we had built in the lab a device which had 75 millimeters diameter with approximately 12,000 by 12,000 pixels. And then, for 1995, we constructed for AXAF a device which was 100 by 100 millimeters in size, with 16,000 by 16,000 pixels. That's enormous. And all the same technology: cross-grid charge detector, just expanding the size, keeping the algorithms basically the same, and this coarse-fine extension allowed us to grow in a nice, orderly way and lead to having a detector ready for this mission. Now, Jerry mentions a very important point. The HRC, or the cameras that we're using with microchannel plates, are single-photon-counting cameras. That is, one X-ray produces one event. Just as in the Uhuru discussion we mentioned that the proportional counter puts out a signal when it detects an X-ray, these cameras put out a signal when the X-ray is actually detected. So we know the time occurrence of each event. It is not limited by anything other than the time of arrival. The other kinds of cameras which are in use today, which are the charge-couple devices, are cameras which are called frame cameras. What they do is they — [Off-tape comment.] A frame counter, or CCD camera, works by basically having an active area on which X-rays fall and produce charge, little packets of charge. For a certain amount of time, like three to ten seconds, the signals are allowed to collect there, and then they are all transferred into a read-out area and then out into a signal-processing chain. So, anything that happens in this time interval is completely lost in terms of when it happened. All you know is it happened somewhere in this last read-out frame. So these are what are called not self-triggered, but frame transfer-type devices, whereas the HRC, as soon as an event occurs, it immediately comes out. So you have photon counting, self-triggered, implies time. And you have frame transfer, which implies only a delta T, a time interval. Both have advantages. There's this very high efficiency. There's some energy resolution in these devices. They have very broadband capabilities. These tend to be good at low energy. They tend to be less efficient, but they give you this very accurate time resolution. The time resolution here is less than a microsecond. So we're talking about the difference between seconds and millionths of seconds. Quite dramatic.
So now we move on to the — I may actually have to stand up for some of this. Well, we've discussed so far the development of the actual channel plate detector/cross-grid charge detector combination, which is the heart of the instrument, but there's more to building a space-flight instrument than just the detector elements themselves. So the next logical step in the design evolution of the instrument is to try to package this detector as close to flight as possible. In the late 1970s when we were doing this, mid-seventies, late seventies, NASA had decided that they were not going to build prototype detectors any longer. They felt that the state of development was such that we no longer needed to build as many units as we used to build, and so they would not accept the notion that we would build a prototype detector. We wanted something that was more than just what would be call a breadboard, which would be a strictly laboratory version, so we coined the phrase "brass board," which is to imply a little better than a breadboard, and, actually, unbeknownst to the customer, really an attempt to come as close to the flight packaging as possible, or certainly an initial attempt at it, in order to learn the problems that go along with the packaging of a detector for flight use.
So what we see here is one of two units that were built, and they were called the HRI Brass Board, produced in the middle seventies by American Science and Engineering and sent here to SAO for the scientific staff to evaluate the performance of the detector in a very much flightlike configuration. If you go back to this for a moment and remember, we didn't discuss too much about the fact that every eight resistors or every eight wires in a grid, we have one of these red wires, which is a tap that comes out of the grid to an amplifier. It turns out that for this 25-millimeter-square detector, we had seventeen amplifiers per axis, so we need seventeen of these per axis, or thirty-four total. What we see located around this little carousel here are the thirty-four individual preamplifiers. They are, each preamplifier, built on a circuit board laid into a slot. There's a yellow wire that comes out the top here up through a decoupling capacitor and through a feed-through and into the inside of the housing. Here we can see what now looks familiar to us: a cross-grid charge detector, except we see it sitting on another ceramic disc, to which the individual taps off the grid are welded onto this ceramic disc. It's really a ceramic circuit board, if you will. And then there are connections from the outboard edge of those tracers to these feed-throughs that go through the housing into the amplifier.
So this is how the connections are made from the grid off this—we call this the interconnection disc, this white ceramic disc, through the feed-throughs, down into, through the decoupling capacitors, into the front end of the amplifiers. [Off-tape comment.] Okay. So we have there the cross-grid charge detector. We have taps coming off every eight wires that are connected to pads on the white interconnection disc. The traces on the disc go to the edge of the board, and there's welded gold ribbon. You can maybe see some of it here, that makes a connection from the interconnection disc to the top of a feed-through terminal. The feed-through terminals then come through here. The ends we see here with a decoupling capacitor here and then into the input of the preamplifier through these individual yellow wires. Now, the reason we use these feed-throughs is that it's not possible to operate microchannel plates in air; they must be in a vacuum. And so you'll notice that this detector has an O-ring seal around it. It was made to bolt to the end of a test pipe here in the SAO facilities. We would evacuate this part of it in order to be able to operate the channel plates, but we did not want to operate all the rest of this necessarily in a vacuum in a laboratory, because we'd like to be able to get at this and look at some of these test points, for instance, and do signal injection and look at the outputs on an oscilloscope.
So we wanted this interior part in a vacuum and the rest of this outside. So that's why we use these hermetic feed-throughs in the bottom of this housing. Again, in the flight detector, ultimately, we maintained the interior of the detector in a vacuum all the time we were on the ground and left the rest of the instrument exposed to whatever the ambient pressure was. Ultimately, in a flight unit, these cavities were all filled with encapsulation material of silicon rubber, again because of the high voltages present here and the necessity to prevent any corona problems and so on while operating, both on the ground and in orbit. Then what we see at the back end of this unit is, again, more front-end electronics or just six circuit boards in there. These are really in a breadboard condition; they were not packaged that way for flight. The flight packaging of those boards at the time we built this had not progressed to the point that we could install flight-package boards in this unit, and we were very much interested in getting this to SAO and on a pipe and tested as quickly as possible. We ultimately learned that, in fact, the detector concept that we'd developed, starting back with the original hand-wound wires, when finally packaged in a somewhat flightlike condition in fact worked as we expected, and we were very pleased with the results. [Off-tape comment.] So after we'd tested this in a laboratory, we were also very interested in seeing how it would actually perform in a space environment behind a real X-ray optic, and we didn't necessarily want to wait until we launched Einstein. We were committed to build three flight versions of this for Einstein. We were kind of interested to know that, in fact, it was going to work. So we had undertaken a sounding-rocket program to fly these, and I can't say for sure whether this is one of the ones that flew. We had two of these, one which was outfitted for flying in the rocket payload, but subsequent additional uses of this have kind of erased all vestiges of the rocket flight. But maybe you could comment a little bit on the rocket flight program, what we learned there.
Okay. We had a rocket flight program that was designed, as Jerry said, to make sure that we understood how these detectors performed in the environment that was more spacelike. There were two particular things we were interested in. One was, we wanted to understand the background in these detectors. We were very worried about what the environment in space would do in terms of producing ions or ambient electrons that would get into the detector and produce false signals. And we also wanted to know that everything would survive in terms of the launch vibrations and the detector and the grid and so forth would actually survive. So we had a series of sounding-rocket flights out of White Sands, New Mexico, on an Astro BF launch vehicle and a Brandt [phonetic] launch vehicle over a period of several years. We finally, in, I think it was 1976, had a package together which included an X-ray telescope made out of beryllium, a very small telescope with a total effective area of about one square centimeter, and this or its cousin with a channel plate detector in the front and a vacuum cover so that we could keep it under vacuum and then open it up at the time the rocket reached altitude. We pointed the telescope at the then strongest known X-ray source in the sky, Skullex 1 [phonetic], and we collected data for about 300 seconds. During the time we were collecting data, we were also varying one of the control voltages on the detector which we thought would have some influence on the amount of background we would see.
What we were doing, actually, is controlling the voltage on a front filter directly in front of the microchannel plate, which was designed to keep ultraviolet light out of the detector because we're sensitive to ultraviolet light, and also to repel ions away from the detector. So we stepped the voltage on that shield as we were observing the source. After the rocket was recovered and we got back all the data from it and analyzed it, we found three X-ray photons from Skullex 1. This is the famous three X-ray-photon flight. We knew they were from Skullex 1 because they were all three in about the same location on the detector, which is a coincidence which would not happen unless it was from a real X-ray source. So we declared success. In addition, we monitored the total count rate over the whole face of the detector, which is the background, and we tried to bin that data up corresponding to the various settings of this voltage on the filter to see if we could see any differences, you know, one voltage give us a lower background than another, then another. In fact, we saw virtually no difference. Then we chose, therefore, to use the most neutral value for the voltage on the filter, and we communicated that back to the engineers who were building the flight models for Einstein at the time, and they actually made the very last change to the selection of voltage even as the flight models were being finalized. That occurred roughly, I'd say, within a few months of the time we actually delivered the instrument. So it was very close.
Austin:Yes. Right at the final assembly phase.
Murray:If that flight had not been successful, we would not have known for sure that we had, in fact, got it right, and we would have just guessed. I think it wouldn't have mattered in the end, but we didn't know that at the time, so there were very important engineering flights for us.
DeVorkin:You got three X-ray photons.
Murray:Yes. Right. They were named: one, Two, and Three. (laughter)
DeVorkin:But this met your expectations.
Murray:Yes, because we knew how inefficient the detector was and we knew how small that telescope was. That telescope only had one square centimeter of effective area.
DeVorkin:Oh, the collecting area was one square centimeter.
Austin:The effective collection area, yes.
Murray:The collecting area of the telescope, one square. So you've got a little thing one square centimeter in space there trying to get an X-ray through the hole. It came from Skullex 1, which is a long way away.
Austin:No, it was not a surprise that we got so little flux. We were actually expecting that kind of performance. But the beauty of imaging — I mean, in fact, that's one of the things we proved on this flight, was when you have an imaging detector and you have a telescope collecting, you don't need a lot of events to actually detect a source, because they all land in the same physical location, and that tells you that you really have something from outside, not something from inside.
Unidentified speaker:What would have happened if that flight hadn't been successful?
Murray:Well, if the flight had not been successful, we would have proceeded anyway, of course, with the Einstein activities, and we would have probably chosen exactly the same operating condition, but we would have been fearful. We would have not been nearly as comfortable that we had not missed something. I mean, the real purpose of the rocket flights was to build confidence up that what we were doing for Einstein was going to work. The investment of time and energy in Einstein was so huge that we wanted to make sure that we didn't leave anything to chance.
Austin:It was one particular parameter in the detector design that we made no provision for adjusting on orbit. Other voltages within the detector, the voltage across the channel plates, the biases in the grid, the biases between the grid and so on were all adjustable by ground command in flight. The power supply had the capability of being commanded to provide different voltages. But the bias voltage for this UV ion shield out in front of the detector had no such provisions. We had to set it ahead of time on the ground. So there was a little concern as to whether or not we were setting it at the right location or right value.
DeVorkin:This was a UV ion shield that was being supplied by NASA? You had no control over?
Austin:No, we built it ourselves, and it goes inside of the detector in front of the channel plates basically, as Steve says, to keep ultraviolet light off the channel plates. In this particular instrument it was located about an inch or so in front of the channel plate. You don't see it here in this brass board version. In a flight version you would see it, and we'll see it a little later on in a ROSAT [phonetic] flight model. We had an onboard calibration device that would shine a pattern of dots onto the detector to permit us to calibrate on orbit any distortions in the detector read-out and look for changes in the imaging properties of the detector throughout the flight. I don't think we ever saw any, but if you don't look, you don't know. So, because that was based on ultraviolet light, it was necessary to space the shield forward of the detector, and then we had this system that came in at thirty degrees off the channel plate to shine this pattern of dots. So the shield was substantially in front of the detector, more so than subsequent versions where the shields now are mounted right in front of the channel plates, but we don't use ultraviolet calibration systems anymore because we know that we don't need them; the properties don't change. So this was a shield that was sitting about an inch or so in front of the channel plate, but it needed a bias on it to determine what would happen with ions in that space between them. Would they be repelled, attracted? What should we do with them? We weren't quite sure, and so we basically, during this rocket flight, I think, we had the one positive, one negative, and one neutral value to see if we could see any difference, and we didn't. There's still some debate as to which value we finally picked, but — (laughter)
Murray:(laughter) You'll notice I didn't exactly say which value we picked.
Austin:Which one it was, and I can probably go back through the records and figure it out, but it turns out it didn't make much difference anyway.
DeVorkin:…to Einstein. How is this different? Or will you be using a different detector?
Austin:Well, we can probably see that heritage in the ROSAT instrument.
Murray:Unfortunately, we don't have an engineering model of Einstein.
Austin:Well, we should mention that and why we don't and why we happen to have the Einstein model.
DeVorkin:How is this different? Is this more similar to ROSAT or to —
Austin:No. I'll discuss that whenever you're ready to start here.
Austin:We're rolling. Okay. This, as we said, was a brass board instrument, which was flight-like packaging. We were particularly interested in flight-like packaging of the detector, the feed-throughs, the front end, all the sensitive areas of the electronics, the actual pre-amplifiers, packaged as they would be in flight. And what we don't see are some protective covers that went over this for use in the laboratory that have already been removed so we can see what's in there. This was the real heart of the detector, and we did a flight-like packaging job there. We did not package for flight from here on. These are just laboratory-grade cards, and, of course, this went on then to the actual design of the flight instrument. As I said, in this era, NASA had given up building prototypes and qualification models. We entered what was called the proto-flight era, where you built a prototype, essentially, and flew it. There was only enough hardware to fly. There were three flight instruments built, but no flight spares at all. So they were all built, loaded into the telescope, flown, and subsequently disappeared after orbital re-entry.
Murray:Well, you could say we flew the spares. I mean, we built three instruments. They were all the same.
Austin:Well, they were redundant. They were intended to be redundant. There were, in fact, minor differences between the instruments, principally in terms of this UV ion shield. In one case, we made a thinner one because we wanted to push for every last photon we could get, but just in case we'd made it too thin and had too much of a UV leak, we made some that were a little thicker just in case. And I can't remember; were there any other differences in the coatings? I don't think so. Channel plate coatings.
Murray:No. They were all magnesium chloride-coated channel plates, and all the electronics and selections were as identical as we could make them.
Austin:So they were redundant. The flight telescope had redundant instruments, but there were no spares. If one of the instruments failed at all during the test and integration process, the only thing you could do, in fact, is repair it. In fact, we had an accident at the Cape with the prime HRI instrument. One of the HRIs was launched into viewing position of the telescope so that if the focal-plane transport never moved, we would at least have an HRI in the field of view. And we had a commanding accident at the Cape. Certain commands that have the potential to do damage are labeled as critical commands and are flagged by the ground support equipment and brought to the operator's attention that this is a critical command; do you want to do this? And the critical command was trapped, and we were notified by TRW, the spacecraft contractor, that we had attempted to send a critical command, and we had, at the time, but the one that we sent was to turn the high voltage on. The one that they received was, in fact, to open the door. At that time, we weren't smart enough to know that when we flagged a critical command, we ought to ask which one it was. Subsequently, on ROSAT, we changed the procedures to reflect that so that when a command was trapped they told us which one it was. Anyway, the door opened, and we imploded. The detector was under vacuum. It imploded the UV ion shield into numerous small pieces, and basically had to take that detector out of the telescope at the Cape, remove the forward housing, clean it out as best we could, replace the ion shield, put it back together, and evacuate it again to get it back into the telescope, because we had no flight spares. We didn't have another one to say, well, stick this one in. It was repair that one, or it just doesn't work for flight.
Murray:Well, not only did we repair it, but it was repaired successfully, I should say.
Murray:And it was launched, and it was the first HRI we turned on during the Einstein mission. Over the course of the first month, that detector actually developed a slight abnormality, which was not very severe but was enough to make us decide to switch to one of the second, one of the other two units. I forget whether it was called Number Two or Number Three, but we switched to the second unit, which we then used throughout the entire operating lifetime of the observatory, and we actually had a third unit, which, although we turned on once, we actually never did any astronomical observing with. Every now and then we would go back and recheck the first unit to see if its condition had changed, and that slight area of higher background remained constant throughout the lifetime of the mission. But effectively we had one unit which we never had to use, we had one unit which was the workhorse, and we had one unit which had developed an early problem and then remained stable. So we had, over the two-and-a-half-year lifetime of Einstein, essentially no serious failures in the instrument, and we actually still had one backup to go when the observatory re-entered and burned up, unfortunately.
DeVorkin:Could these different units then be mechanically moved into the focal plane?
Murray:Yes. The focal plane was designed as a carousel or a lazy Susan. It could rotate around in one direction and then rotate back in the opposite direction so that it wouldn't wind itself up. There were three HRI subpositions. We had two imaging proportional counter subpositions and we had a solid-state spectrometer and a Brad [phonetic] crystal spectrometer. So there were, altogether, if I remember right, that's seven positions on this focal-plane assembly that we could move to in the course of flight. And, in fact, we did that during the mission. What we would typically do in using the observatory was put one instrument in place and get it all properly set up and all of its electrical characteristics fixed, and we would observe a number of targets on the sky, and then we would switch to another instrument and then observe a number of targets on the sky. So we'd have blocks of two or three weeks of a time where an instrument would be fixed into focus and we would do a series of observations, and then we would switch to a different instrument and do another set of observations. The way that the observatory worked is, we could see most objects on the sky for a period of two to three months before they would go out of our view because they were getting too close to the sun or behind the Earth or whatever. So it was easy to block out the observations in this manner, and we would just continue to do that over the course of the two and a half years of operation.
DeVorkin:Okay. Should we move on?
I think we move on to ROSAT. [Off-tape discussion.] What we have here now on the left here is the aft assembly of the ROSAT engineering model detector, and we have this here only because of the nature of the program that we were involved with when we built this. Due to delays in the start-up of the AXAF [phonetic] Program and the lack of any new X-ray data since the early 1980s when Einstein ceased to operate anymore, the U.S. Government decided to participate in a German program, the ROSAT Renkin Satellite Program. The Germans offered us the opportunity in the U.S. to put an imaging instrument on board their satellite, and NASA headquarters decided that it would be a good idea to basically build another Einstein HRI-type instrument and fly it. So what we have here is very nearly identical. There were only minor changes, and, in fact, in this area no changes, or most of the changes were in the forward assembly, which we can look at in a few minutes. So this is a flight-packaged detector assembly, and you can see a lot of the same elements in here.
The housing here on this one is a little shorter, a little different flange for mounting in the payload. Then this one, which was designed to fit onto our vacuum pipe, but you see here a similar section with coupling decapacitors in it, which has been encapsulated. It covers around there. That correspond to this section here, and you see a cylindrical section down here, which is exactly equal, identical to this one. The difference is that we got rid of these signal inject connectors here which I use for calibration testing of the amplifiers and routed those down inside internally so that the cover would fit on here more smoothly. But right inside of this are thirty-four preamplifiers, just like there are in that one, and the same design and same circuit layout and everything. You'll notice now that for flight, because of other considerations, the six front-end amplifiers that were in the back of this one have now been packaged in housing where there are three of them that plug in from each end here, and these are the flight-packaged versions, and the Einstein HRIs had this same what we call front-end processor box on the back of it. There are a few other features here. This one to show –- excuse me. I'll just go back to this. Here the high-voltage bias for the channel plates is basically just accomplished with a couple of high-voltage wires with a connector on the end that would plug into a laboratory power supply. Now we've packaged it for flight. Again, all of the high voltage is encapsulated. You see here they're just exposed terminals, because it's fine for ground use, but for flight we encapsulate that. We use the same kind of cable, though, and it comes down here into this high-voltage power supply that provides all of the biases necessary.
There were five different voltages required to be produced by this supply, and it runs off of a twenty-eight-volt input here and program command lines into it to adjust these various voltages. And then the voltages out of the supply are distributed around to the various portions of the detector system. If we can look inside of this now, you'll see the same familiar design here. You see the channel plate detector in the middle sitting on top of the cross-grid charge detector in a white round disk with the four flat edges on it, the interconnection disk, and the connections to the feed-through is in the housing. Same design as existed in the early brass board unit. There is a slight difference in the channel plate holder itself. You'll notice it's no longer ceramic. You see this brown material. That's a —
Austin:A Delryn AF material. We had problems with the ceramic in that channel plate holder. It was very fragile, a lot of cracking, not very robust. We found this plastic material, which is much easier to deal with, provided the same insulating properties, and we changed over to the use of that material. You also may be able to notice, if I turn this just so, a circular area about one inch in diameter in the middle of the channel plate, which is a different color. Can we pick that up at all? [Off-tape discussion.] That central area on the channel plate has a coating on it. In this particular case, it's coated with magnesium fluoride. Steve can talk a little bit about the various coatings that we use and why we use them. I'll set this back down again.
Murray:The choice of coating for the microchannel plate was itself an independent study that was begun even back in the era when we were designing and developing the Einstein detector. The bare microchannel plate itself has a very low efficiency for converting X-rays into electrons, and particularly as the X-ray energy increases, the efficiency of the bare glass decreases so that the quantum efficiency or the overall detection efficiency of the detector is going down as the X-ray energy increases. Magnesium fluoride was used for Einstein as a way of trying to improve that efficiency. We chose magnesium fluoride after doing some laboratory studies of various kinds of materials that were known to be "good X-ray photocathodes." And these are evaporated onto the front of the microchannel plate just by a vacuum evaporation depositions process. In between the launch of Einstein and the development of the ROSAT detector, we did studies on other materials that we thought would do an even better job of improving the efficiency. One of them is a material called cesium iodide, which is a material which is used in scintillation devices and has been shown to be a device which would help the efficiency, particularly at the higher energy end. We developed techniques for depositing that on the microchannel plates. I'd say a few years before the ROSAT mission actually flew — I should point out that ROSAT was supposed to fly in 1986 or '85, but because of the Challenger disaster, all of the launches in that era were delayed several years. In fact, ROSAT did not fly until 1990, and it was —
Austin:It was supposed to be originally in the shuttle.
Murray:It was removed from the shuttle and re-cast by the Germans to be compatible with a different launch vehicle, a Delta launch vehicle, and then flown in 1990. In that time interval between the time when it should have flown and the time it actually flew, we were able to not only develop the coating technology, but to get enough experience with it to feel comfortable using that coating on the ROSAT detector, enhancing the efficiency by as much as a factor of two or three, in fact. So it was significant and without any known deleterious effects. In fact, we flew ROSAT. It was launched in June of 1990. It is now, as we're filming this, April of 1998, and the HRI detector on ROSAT is still operating. ROSAT itself is still operating, so we have, in space, almost eight years of operational experience now with this coating, and we have had no degradation in the detector performance over that whole time, something we're very happy about. In fact, we're using that as the experience basis for, in fact, what we're using on the AXAF detectors, which is the same cesium iodide coatings with the same vast improvement in performance.
DeVorkin:This one has magnesium fluoride on it.
Murray:This one has magnesium fluoride only because it's the engineering model and it was not refurbished for flight with the cesium iodide coating, and also magnesium fluoride is very robust. You can leave it open in air and you can look at it, and it won't degrade. Cesium iodide absorbs moisture out of the air, and it would not be as robust for presentations and show-and-tells. We have to keep it actually in a dry, gaseous environment when we're not keeping it under vacuum. So it's more convenient.
Austin:As much as anything, that's the reason we have it in here, just so we can show people what it looked like.
Murray:What you could see if this were cesium iodide is a milky white film on the front, rather than this greenish tinge or greenish purplish tinge. Let's see if I can say those words. Instead of the greenish-purplish tinge of the magnesium fluoride, if this was coated with cesium iodide, it would be a milky white type of film that you would see on the front surface of the microchannel plate. [Off-tape comments.] I believe that on ROSAT we have zero-degree pores in front.
Austin:Yes, because we flew old Einstein plates.
Murray:That's another story.
Austin:That's an interesting subject in itself.
Murray:Manufacturing microchannel plates turns out to be somewhat of an art.
DeVorkin:And they're made on contract, or is that something you make here?
No, we cannot make them here. They are made on contract, and the manufacturer for the microchannel plates that were used for the Einstein Observatory was a company in England called Mullard [phonetic] Limited, who have subsequently been purchased by Phillips Photo-optics of Europe, and the plant where these were manufactured, which was in England, has since been moved to southern France, in Breve [phonetic]. But in the time of Einstein, we contracted for these microchannel plates, and we got pretty much what we asked for without a great deal of difficulty, though when the ROSAT mission began, we went back to the company, which was still Mullard at the time, and asked for them to manufacture the same kind of microchannel plates they had manufactured for us for Einstein. They agreed to that. They sent us some samples, and they were not of the same quality. They were, in fact, so bad that we could not use them for flight. We then embarked on the program to try to understand what had happened in the time period between the seventies and the eighties and to recover the knowledge that had been lost of how to manufacture these devices for us.
We spent, I think, three years doing that. We bought, I won't say thousands, but certainly hundreds, perhaps a hundred or so channel plates from these people. At the end of that program, we had managed to receive from them microchannel plates which we could have flown for ROSAT, but which were still not of the same quality as the ones we had originally gotten for Einstein. In fact, on ROSAT, we are flying residual inventory microchannel plates that were from the Einstein era, and I don't know whether these are of that nature of not, but ultimately we could not get microchannel plates made as well as in the past. Now, this, for ROSAT, turned out not to be a problem because we had this residual inventory, but it immediately flagged a very serious problem for AXAF. We were going to turn to the same manufacturers and say not only do we want the same quality, but we want it in a much larger format. In the case of AXAF, what we wound up doing was having a program which was independently funded from the government, from NASA, for microchannel plate development, and we enlisted not only the people at Mullard becoming Phillips working in Breve to do this, but also a local company, Galileo Electro-optics, which were located out in Sturbridge, Massachusetts. And we had both of the companies in, effectively, a competition to who could produce flight-quality microchannel plates for the AXAF in time for the AXAF. We spent three years working with both manufacturers, and it came right down to the wire. In the end, both managed to deliver acceptable microchannel plates to us, and we elected to fly on AXAF, since we have two detectors, one detector with one manufacturer and another detector with the other manufacturer. For a while, we were very nervous that we would not be able to obtain any of these [unclear].
Austin:It's kind of interesting, actually. I ran across a fellow by the name of Jim Abrams. I don't know whether you remember him or not. I remember Jim from —
Murray:Galileo, actually, was the original supplier of channel plates during the laboratory development of these detectors in the early seventies. They were the first company we actually worked with to get channel plates, but then when it came time to get more serious about these flight plates, Galileo had a history of changing ownership and organization, and we had difficulty getting any sort of response from them. We initially started working with Varion. Channel plates at that time were being developed for use in night-vision devices. A lot of the technology was classified at that point, and so we had to go to Varion and try to work with them, not having a classified mission or any reason to have access to classified information, and still try to work with them to understand enough about the process to adapt the channel plates that they were making for night-vision devices to our purposes, because we wanted something that was different. Principally, we wanted them twice as thick as what they ordinarily made. We got plates from Varion and tested them, and they were fine, and as a backup, we decided to also go to Mullard in England, just in case anything should happen to Varion. But at the end of the life-test program that we conducted in the lab early on to see what the lifetime of the plates were, we found that the Mullard plates had substantially greater lifetime. We never, I don't think, completely understood why that was, but based upon that we made the decision to, in fact, fly the Mullard plates in Einstein based on their superior lifetime. So they started out as strictly a backup and wound up being the flight detectors. And yet we sort of came back full circle, because now for AXAF aeroplates we have once again gone back to Galileo and gotten acceptable plates from them. So that kind of closes that circle.
DeVorkin:That's a very interesting story. It also sounds like many of these manufacturers, or specialty manufacturers, are small companies that can change significantly over time.
Austin:Well, it's very much personnel-dependent. At Mullard there was a guy named Field. What was his name?
Austin:Ron Field, who nursed these Einstein aeroplates through the production, built these beautiful plates, and left.
Austin:Retired from Mullard about the time that we started building the ROSAT plates. They had made changes in the process. They'd changed the recessivity of the plates and so on for their purposes and had changed some of the composition of the materials and so on, but they did not prove to be enhancements for what we want, and they basically just forgot how to build what we wanted. They just couldn't do it.
Murray:We had a meeting there, I remember, where we discussed process, and they had described how they did things by, "Well, we use Process A now, and, before, we used to use Process B," or vice versa. And when we tried to ask what the difference was, they said, "Well, we don't really know, but we know that they're different." And Ron Field was actually hired back as a consultant to try to help them disentangle this Process A-Process B difference. He got close, and he really worked very hard to try to get us very good microchannel plates.
Austin:Well, in fact that ultimately led — it was too late for ROSAT, but —
Murray:Led to the solutions.
Austin:We realized that for AXAF we had better concentrate on learning how to build the plates again and, once we got there, not stop, but go right on into the flight plates. If we left any gap there, we were inviting disaster, because they would either change the process or forget how to do it or something.
Murray:Or change people.
Austin:So we had to get it off the ground again and keep moving right into the flight plates, and we were successful.
Murray:You made a comment about small companies. These are not small companies. Mullard or Phillips is not a small company.
DeVorkin:Not Phillips, yes.
Murray:But what happens is that specialty items are a problem. They are happy to make night-vision goggles and tank sights and all the other things in multiples of thousands, and they will do that quite well and quite repeatably, but when you come in with special orders, which they're very interested in — they have people who are very excited about doing custom devices and maybe not even making money at it, but developing technological advances and supporting the scientific community. I think they deserve credit for that, but the down side is that it's a best effort. You get what they can do, and it's very hard to get anything guaranteed. In the case of the filters we made, we'd go to a small company in the San Juan Islands off of Seattle —
Austin:Called Luxell [phonetic].
Murray:Called Luxell, which used to be a garage operation. It's expanded since then quite a bit, but not —
Austin:A bigger garage. (laughter)
Murray:A bigger garage; a shed now. But the president of that company is devoted and dedicated to producing extremely high-quality scientific-grade components, and he worked very, very hard and very long with us to ensure that we got filters that were really outstanding. Actually, it's a lot of fun to go to these companies. Generally American industry is very enthusiastic about supporting space. They don't always make a lot of money on it, but they certainly enjoy it, and I think we're very lucky that we can do that. A lot of places you can't get that kind of cooperation.
DeVorkin:Very interesting. Okay. Should we move on?
Austin:I wanted to make one more comment about why we have this at all. It's an oddity, in a sense, as a result of the nature of the program, because this was an international program, and the Germans took a slightly different approach in that they wanted to build a complete engineering model of the entire focal-plane assembly with one of our instruments and two of theirs. They flew two imaging proportional counters, or PSCs, as they call them. NASA agreed, and we built this engineering model unit, which is completely flight-designed, but not built and tested. For instance, the electronic components in here were not screened to the same high standards that we would use for the flight unit, because we wanted to get it built earlier and the lead times were shorter. But other than that, you put them on a table, you can't tell the difference between them other than looking at the serial number. So we're fortunate to have this flight-like device in our possession. This ultimately was upgraded to be the designated flight spare for ROSAT. Had something gone wrong with the flight instrument, we would have flown this one, and even though the components had not been initially screened, there was sufficient operating hours on this to have qualified the electronics and basically said that they are flight-worthy.
DeVorkin:That gives us a good history of that object.
Murray:We should probably get the other parts of this.
Austin:Yes. Now we'll go get the front. But let's leave this here for just a minute. Okay. What we have here now is the forward assembly of the ROSAT detector. This was the aft assembly. I have installed on the aft assembly a GSE cover just to keep the contents of the detector clean when we store it or are not using it. I left it here just to show how these two relate, because this portion of the cover, this little top hat, can be seen in here with this part of the housing and this flange, this flat area of the flange. That's analogous to this cover. Were I to remove this cover, I could pick this up and put it on top, and this portion in here would fit on there exactly as this does. So this kind of gives you a feel of how this fits on there. So that actually this skirt around the bottom would hang down around the edge here.
Murray:That's to keep the water off the detector during a rainstorm, right? [Off-tape comments.]
Austin:Well, if I was to take this hat off of here and bring it down to here, it would be represented by this portion of the housing. This has a skirt on the flange that this one does not, but if you imagine this skirt just cut off right here, then this much of the housing would be exactly identical to what we see over here. So that's how this would fit on. Take this cover off, lift this up and place it over, and this surface right in here, the underside of that, would essentially be at this level.
DeVorkin:Could that be done, or would that be difficult?
Austin:Yes, if you'd like to do it, I don't see any reason. I may have to stand up, because this thing's a little heavy, but let me get these screws out of here again. In fact, we can show an assembled view of it. That's a good idea.
Murray:This was the main structure on which the detector was hung, basically.
DeVorkin:Oh, you mean the —
Murray:The skirt, isn't it? It's where [unclear].
Austin:I'll go over why we have this, because anybody who looked at this would say, "Geez, what a horrible waste of metal, to do this. Why was this done?"
Murray:I have to be careful here. There's these screws that come down, right?
Austin:No. This is fine. Well, this fits on, and there's no shield here, no ion shield, so it's pretty easy to get it on.
Murray:Yes, if you had that locating screw. That's one of the rods. That's what I'm saying.
Austin:Oh, the rods are in here?
Murray:Yes. So you have to get them aligned. That's what I was trying to tell you, Jerry. [Off-tape comments.]
Austin:I'd rather show you the inside when I'm laying it on the table, because it's kind of heavy. What I'd like to do just now is put it together.
Murray:These should be located. This one's in. That one's in.
Austin:That's more or less —
Murray:Well, now it is.
Austin:Okay. This is more or less the way the assembly would look if it was ready for integration and tests, other than a couple of cables not connected. It shows the forward assembly mounted on the aft assembly. The functions that are provided here by the forward assembly are, obviously, one, the closure of the housing. But then there’s a number of other features: the internal calibration feature, the ion pumping feature, and so on, which we'll discuss when we take it apart again.
DeVorkin:Okay. So maybe that [unclear].
Austin:You want to hold the base?
Murray:I'll hold the base. You go up.
Austin:I think it's wedged in there. Usually I have to take those things — Well, I've got to do it. It's cocked this way. There it goes. Okay. Let's get that out of the way. Put this down.
Murray:Then we won't put the cover on that first.
Austin:We'll put the cover — just set it back on there to protect that.
Murray:This is heavy. Yeah, this is heavy.
Austin:Well, that's only a small part of this stainless steel.
DeVorkin:Okay. You go ahead.
Austin:We can do that later. There's no hurray to do that.
Murray:That was cute.
DeVorkin:I think we have at least a visual record of them together.
Austin:That's important. I hadn't really thought of that. That's a good idea. The housing here reflects some design considerations that were imposed upon the detector. The mounting points for the detector are out here, one here, and then, inboard, ball and sleeve mount. The pattern is quite large with respect to the size of the detector, and ordinarily we would have had a much more compact mounting arrangement, but the Germans requested that we make the mounting of the HRI compatible with their detector, which was a much larger field of view, and its mounting points were basically out at these dimensions. So they asked us to extend our housing to this point. So that's the reason for this heavy skirt, is to transfer the loads all the way out to these mounting points, which are, in general, further away than they would ordinarily be. Another interesting cultural difference, I guess, in terms of design approach, in the U.S. it is customary that integration contractors who would integrate an instrument like this on a larger structure for flight basically do not make any requirements on the instrument other than to support itself, and they will take the loads elsewhere, and the detector is never used as part of the integrating structure. Germans, however, had a different approach. They had some gas tanks located adjacent to, on each side of our detector for their detectors, which are gas-flow detectors, and the loads from those gas tanks went into the same fittings as we fit onto here, and they expected us to help resist some of the moment loads at the connections. So we wound up having to add these little dog houses, we called them, on either side, weld those onto our housing afterwards in order to make these ears strong enough to resist those loads. So that was different. It's not something we would ordinarily design for in this country.
Murray:Let me interject one point here. The position of the detector relative to the X-ray telescope is very critical, that it has to be in focus. So the mounting of the instrument onto the focal-plane assembly determines that. So this mount is a very critical aspect of making sure that we are in focus with respect to the X-ray telescope. There's no adjusting mechanism in flight that positions the camera in focus, and it has to all be designed correctly and then made strong enough and thermally rigid enough so that it doesn't drift in and out of focus.
DeVorkin:What is the depth of focus, if there is any?
Austin:For ROSAT, probably a few thousandths of an inch. It was a much shorter system. Einstein was plus or minus four thousandths of an inch, and we had a larger F number there.
DeVorkin:It's still very, very small.
Murray:It's small, yes.
Austin:The shimming is accomplished by a shim here, which you can see underneath this block. There's actually a shim there. And there were circular shims that went into these recesses here. The structure that supported the detector came down in this direction so that we had circular shims here, and then there was a ball that fit into this sleeve, and then there was a shim under here to adjust the position of this. So we had to make careful measurements with regard to the focal position of the channel plate with respect to the features on the housing and make careful measurements all the way out to these surfaces so that we could cut and select those shims basically to an accuracy of a few thousandths of an inch spread over all of these measurements. As far as we know, it worked. At least it was certainly adequately in focus for the ROSAT mission.
Murray:An interesting story here is that in order to determine this focus, there were calibrations done, including calibration using an X-ray source and X-ray telescope at a facility in Germany called Panther facility. In the analysis of those data, the scientists from Germany and the scientists from the United States independently analyzed the data and independently came up with different solutions. We had a disagreement of about, if I remember right, about six thousandths of an inch between the U.S. solution and the German solution. I don't remember what we finally did in determining the resolution, but I suspect we basically took the average of them as they are. (laughter)
Austin:I can't say. I was sworn to secrecy, and all the parties are not yet dead. (laughter) So I can't say what we did. I'll leave it in my will. I'll tell you what we did.
Murray:In that case, we took the U.S. solution, apparently. (laughter) And it turned out, as far as we can tell, as Jerry said, to be all right. But we did have quite a heated discussion, I believe is fair to say, as to where the focus really should be.
DeVorkin:Shades of more recent problems.
Murray:Yes, and we were well aware of them at the time. (laughter)
DeVorkin:Could you walk us through [unclear]?
Sure. I'll tell you what some of these other bits and pieces are. Let's start with the door here on the top. The door is basically constructed — it's very much like a common laboratory gate valve, except instead of a big pneumatic or mechanical plunger out here to drive the gate back and forth, this one is motorized. There's this little DC motor here with a gear box on the end of it. This is actually the motor; this is the gear box. That drives through this gear into this gear and then subsequently into this one here, and these two gears riding along this rack here, when the motor turns, basically just drives itself along this track. There's a groove inside here with some wheels in it that will control the motion in this direction. This gear here is just an idler gear. We found that the torque required to operate this was such that when the motor went over and locked up into position, the shaft and the gear box tended to bend out of the way and separate the gear teeth, so we put this idler here to back it up so that there was something to react that moment, a lateral force when the time came.
This same design, basically, for opening the door was used on Einstein. It is capable of driving open and closed again. But there's always a concern with a mechanism like this, that what should happen if there's an electrical failure in the drive circuitry or an electrical failure in the motor or mechanical binding, a piece of debris or something gets caught in a gear here that prevents you from opening the door and renders the detector useless. On Einstein, we had a system whereby there was a latch in this position over here, which, when released, would allow the entire door to swing open, around which you can still maybe see here this axle here, this shaft here, was a pivot point, and the whole door would swing right around that pivot and out and away. It obviously won't do that here, because this is in the way, and there is no releasing latch over here. But the instructions that were given to us were to build this as much like Einstein as we could and only make those design changes that were needed to accommodate fitting the instrument into the German focal plane. And it turned out one of the problems that we had was that we didn't have enough room to swing the door open any more.
We had to come up with another way of providing the redundancy in the door, but because this had already been designed and all the features were there in the drawing, it was just easier to leave that shaft in there than it was to design it out and put something else in there. So there's a shaft in there to rotate about, but no intention to rotate it. It's just a vestige of the earlier design. In fact, over here you can see we just designed a plate to screw into the side of the thing and screw onto this block on the housing to permanently fasten the door down. So if this wasn't in the way and I took those screws out, in fact, it would pivot up, but it was never intended to do that during use. So, instead, we had to build the redundancy into the door mechanism itself. We did something that one doesn't like to do, but sometimes is forced to do, and that is to use a pyrotechnic device. There is down in here a pin-pulling device. When electrical power is applied to that, it ignites the igniter, powder flashes, and it actually pulls a pin in which — you can see it — is located down right in there, is the pin that holds these two sections together. When that pin is retracted, this spring here will retract. It's in an extended position now and will actually cause these gear teeth to separate. So that the reflected inertia of the gear box in here no longer will prevent this thing from rolling. It will now release this gear and allow the carriage to be pulled back by these two springs here that are tension — these are special springs called "negator springs," constant force springs that will just drag this carriage back.
So if there's any failure in the electrical drive circuit or anything, we can fire that and pull this back. However, when this thing is in the locked position — there is actually two sections to this carriage, and it sort of toggles over center so that when you remove power from the motor, there's no possibility of it ever coming undone, whether the reflected inertia of the gear box is there or not, and if that was to occur, if we, for instance, were to separate this gear because the door failed to open when it was in the locked position, these springs are not strong enough to cause it to pull back over center here to release this. So we put another pyrotechnic device here, a pin-pusher, which is designed to give this thing a little kick, and it's all wired in here with limit switches so that this one doesn't fire until this one's fired and the gears have separated. When one of these switches here is then activated when this thing slides down, then this one is fired to give it a little kick to move it over center. Well, we designed this and, with great interest, tried it the first time, and it was really surprising. (Murray laughs.) We pressed a button and just — bang! — right before your eyes, all of a sudden the door was back here totally open. I mean, you couldn't see anything that was going on, it happened so fast. In fact, this thing gave it such a push that it crinkled the springs. They couldn't roll up fast enough to get out of the way, and we had to replace them. It permanently bent this end member. If you put a straight edge on there, you'll find it's about fifteen thousandths out of flat. It's bowed in this direction. It's not because we didn't make it straight, but it got bent by the force of this thing pushing against that, just a reaction force.
Murray:No way this door was not going to open. (laughter)
That door's going to open. Other features that are provided here are — well, let's see. This, as Steve said, was initially intended to be a shuttle mission. One of the requirements that we had on the shuttle that we didn't have, for instance, on Einstein, was a possibility of a delay before we could open a door in the detector. On Einstein, we opened the doors, I think, within the first orbit or two, to make sure that there was no pressure build-up in the detector. We could not do that in the shuttle. You might have to spend up to seven days in the shuttle bay before they would power you up, or contingency planning in the event of shuttle emergencies. So we needed a way to vent the detector if it had to sit there for seven days not having been pumped. So these two devices here are a couple of pyrotechnic valves, basically pin-pushers that shear off an internal cup here. There are two of them, again, for redundancy. This block contains a very small orifice so that if it was necessary to stay in the shuttle bay for that length of time, we would have fired these pyrotechnic devices, knocked the cups off, and allowed the detector to bleed out for a short while before we opened the door so that we would depressurize it at a slow enough rate that we would not be concerned about exploding the UV ion shield inside of it.
So that's what this is. We mentioned earlier that we keep this thing pumped out on the ground at all times. There was a valve here, just a commercial valve, bellows-type valve, which was modified to take off the weight. We took off the handle. We didn't want to fly that weight, which is used to attach some sort of a GSE vacuum system to here. This valve is opened. You initially pump down the detector. Here is an appendage ion pump with its magnets. This is not the standard magnet that's sold with the pump. We constructed this ourselves. These are rare-earth samarium cobalt magnets that we needed to improve the pumping strength in smaller volume. When this appendage ion pump is then started, this valve is closed, the GSE system is removed. It's now on its own. This pump requires about 1500 volts to operate, and that's provided by this DC-to-DC converter over here, which takes 28 volts in on this end, and this wire out here produces 1500 volts to operate the pump. This would have ordinarily been plugged into here, but we don't keep it plugged in because it would hit on the table, and this wire is routed down along the side of the rest of the detector. We always provide enough wire in any cabling to reterminate the cable if we should have to.
If something goes wrong with this connector, you want to make sure there's enough wire in the system that you can put another one on without having to go all the way back into the unit itself and replace this wire or throw the unit away or whatever. So all the cabling here is always done allowing enough of what we call a service loop to reterminate it one or two times if we should be forced to do so. So that's the appendage ion pump system. This ceased to work. The 28 volts for this was provided by a ground support equipment item that we provided and did not operate after liftoff. It was not provided. We had made arrangements on the shuttle program, had we flown on the shuttle, to power this through what's called the T Zero plugs, which are special plugs on the shuttle that allow you to power things right up to liftoff, and at T Zero, T minus Zero, when the liftoff occurs, those things are separated. But that enables us, for instance, if there had been a late abort in the mission, ten seconds or a minute or so before liftoff they had to abort and hold for a couple of weeks, we would never have lost the pumping on the detector; it would have been maintained. The last remaining item here is the — we'll start in this direction. I mentioned the on-board UV calibration system. That's contained in this tube right here. Outboard in this end, there is a UV lamp that shines through a mask. It's a piece of UV transparent quartz with a metallization on it and holes etched in the metalization to provide a pattern of dots so that when UV light shines through there, there's a lens in here that focuses it onto the channel plate and provides a sort of double set of lines, rectangular in shape, on the channel plate. A known image can be baselined and then repeated periodically to see if there's any shift in that.
Of course, as we said earlier, we never saw any, and we've ceased to fly these sort of things. This last remaining article over here is the power supply for the lamp. The lamp also takes about minus 1500 volts to operate it — plus 1500 volts, rather. This is a DC-to-DC converter, 28 volts in, 1500 volts out. Because a lamp itself, once the arc is struck, becomes virtually a short circuit, there's a ballast resistance in here to avoid shorting out the power supply once the arc is struck. And that's about it. This cable here mates with a corresponding cable on the aft assembly. All the electrical connections to the instrument in terms of the interface were on the aft assembly. The one last thing on here is this little black block down here, which is also a sumeriam cobalt magnet. The Germans made a requirement on us that there be no net magnetic moment out of this detector. In this case, it was not feasible to go in. On Einstein, we actually went in and removed the magnets on the ion pump before flight. It was not practical to do this on ROSAT. So this flew, and because it flew and represented a net magnetic moment, we had to put a magnet over here with essentially that same strength-oriented 180 degrees, a way to balance that one out.
Murray:That's needed because the ROSAT observatory operates using the Earth's magnetic moment to push against when it's working on maintaining its pointing direction or actually when it's unloading the momentum that's accumulated in reaction wheels that control the observatory. So you want no extra magnetic moment in the instrument, because that tends to produce a force which is constantly pushing the observatory in one direction, and that means that the reaction system has to operate against it, and that's what they didn't want.
Austin:The measurements were kind of interesting. To make these magnetic dipole measurements, a classic method is to bring a compass up near this and observe the deflection of the needle. Actually, you put the compass down first and point it north and move the instrument in and observe the deflection of the needle, and by knowing where you are on the Earth, you can calculate what the magnetic moment was. So when it was necessary to make these measurements, we would go downtown to Boston. On the dock there, there was a company — I think the guy's name was Robert White, maybe White Instruments or something like. He would loan us — actually, I think we rented it for like twenty bucks or something — a great big nice liquid-filled Navy compass that he used to calibrate all of the compasses that he sold to the boatowners. He'd let us borrow it for a day. We'd go out in the middle of the parking lot, away from all cars and metal objects and so on, and set this compass on the ground and orient it and then put this a known distance away and measure the deflection of the needle and calculate the magnetic moments. We'd get a lot of strange looks out in the middle of the parking lot, hovered over this thing on the ground.
Murray:High tech. (laughter)
Austin:(laughter) Yes. Then after we'd calculate what we'd need for the trim magnet, then we would make the trim magnet, put it back on, and go out and measure to make sure that the net residual moment was within limits. [Off-tape comments.] Now, there's a couple of other interesting features here to be seen on the inside. First of all, there is this black ring, which is actually two concentric rings with a labyrinth space between them and these little spring fingers around the edge. This fit onto the holder that contained the UV ion shield so that no light could go around the shield, and any light coming in here had to go through the shield. It couldn't go around it, and yet we needed to be able to vent out the door. So we have these two concentric cylinders with this labyrinth path so no light could get through, but it could vent any pressure. Also you see here a pocket in the door, which held a radioactive source, a curium 244 source that fluoresced an aluminum target to produce X-rays so that we could make sure the detector was still functioning properly on the ground by essentially shining real X-rays on it and observing the pulsite distribution and so on to check the gain of the instrument and its detection efficiency.
DeVorkin:And that's part of the door.
Austin:That was part of the door, then, which would be removed during flight, and the source just went off to the side and shown on the side of the detector but had no influence then. You also see three little glass windows there. We have a fiducial light system, and we have in our presence here the great fiducial light system designer. (Both Austin and Murray laugh.) There were three little LEDs on the detector which would shine. When the door was open, of course, you didn't need those. It would shine into the aspect system and tell us something about the relative position of the detector and the star trackers and the X-ray optics. But we also wanted to use that system on the ground, just to make sure it's working and the lights are working and that they can be seen by the star tracker. So we put three little quartz windows in there for that purpose on the ground.
Unidentified speaker:Before you put it back, can you show me the aperture for the target, the X-ray?
DeVorkin:You mean the UV?
Austin:Yes, I can show you here. This is the inboard end of the UV calibration source, which was sort of a subassembly that stuck in on this flange in the housing, came in here. So it would produce this beam. There's a lens very near the front end there that would project this thing on the channel plate, which is essentially sitting about at this level. These pins here are just alignment pins that we used to engage the aft housing, and when it's down and installed, they've got flats on them so they can be turned out and removed from below and the screws put back in. But that just enables us — we had to be very careful of spring fingers here that engage a cylindrical feature on the UV ion shield support, and if you don't hit that just straight, you'll bend these fingers, and you have no way of knowing it short of taking it off and saying, "I didn't bend it. I'll try it again, and did I bend it? Well, I don't know." So we use these alignment pins to just make sure that it goes on correctly, and then they come out afterwards, and are six screws that go up through the bottom flange and screw into these threaded holes. You can see here the lubrication pattern that's left from the O-ring when it was joined together. This is the cable that goes down and mates. Takes all the electrical connections for the motor. The UV cal source, the ion pump, everything requiring electrical power on the front end comes through this connector to a corresponding connector on the aft assembly, then out to the interface.
Murray:Now we're moving on from the Einstein-ROSAT era to the AXAF era, and what we have in front of us are some of the pieces of the detector that was developed for the AXAF high resolution camera. This is a read-out device for what we call the spectroscopy detector. It's a long and narrow unit that's 300 millimeters long and about 20 millimeters wide. Behind it, upside down right now, is the cross-grid read-out device for the imaging detector. It's a 100-millimeters-by-100-millimeters-square device. Sitting alongside of that is one of our square 100-millimeter-by-100-millimeter microchannel plates. Then here on my right is one of the intermediate steps. This is a proof-of-concept detector which we built after the Einstein-ROSAT devices but before we built the AXAF devices. It's a 75-millimeter-diameter detector with a microchannel plate and holder and a demonstration that we could actually build these larger grids. So this is about three-quarters of the size of the final high-resolution camera read-out. What we did here was basically take the same technology that was used for the Einstein-ROSAT detectors and expanded the size of the ceramic block. We found a place where we could get the laser ruling on the edges done over this three-inch, almost, area rather than the one-inch area. The block that we used to bring out the individual amplifiers was done on a PC card. You cannot see underneath here, but these thin film resistor blocks that we showed before, now we have gangs of them glued onto each side. Rather than one, we had to use several of them. The microchannel plate holder is the same basic design: an insulating block holding the microchannel plates in the proper orientation and making all the electrical contacts, the top, the middle, and the bottom. That sits on top of the read-out position by steps at the corners, and then that whole assembly then sits on top of a transfer device which would then connect to the outside world.
DeVorkin:With the larger detector area, can I assume that the entire detector, the preamplifier stages, all of those stages are proportionally larger as well?
Murray:The numbers of preamplifiers goes up, and, in fact, on AXAF that became, for us, a serious problem. In this device we have 63, 64 amplifiers on each axis, so we have 128 amplifiers altogether as compared to the 34 we were using for ROSAT. And, believe it or not, the weight and power restrictions on AXAF were as severe as they were in any of the other missions. So we actually went to the development of hybrid electronic components. That is, if you remember, on ROSAT we had these nice little circuit cards with all the discrete components, and we had them lined up in this nice little carousel. For the purposes of power and space and weight, we went to a design where we took the amplifier design and shrunk it down into basically a smaller number of components using integrated circuit techniques and then mounting them all onto a tiny little card, which is put inside of a little circuit can. So the about four-inch-by-two-inch real estate associated with an amplifier for ROSAT because about a one-inch-by-half-inch sealed can with a bunch of pins on it.
Austin:Actually, there were four amplifiers in the can, too.
Murray:And each one is a quad. And so we had much fewer numbers of elements and much smaller components. I believe that the total power on the AXAF front end is like 4 watts. That's all 128 channels. I think the equivalent power, probably, on the ROSAT must have been about 10 watts, not that there was so much power difference, although we have 128 versus 4, but certainly the size and weight were changed considerably. That was an interesting development activity. It went fairly well. Our own engineers did all the design, did all the layouts. We went to a company that then turned that into hybrids. We had test units. We went through an evaluation process. We had to make some changes, and then we eventually got those final units, and then the engineers went through and did a calibration selection. So we actually graded all of these electronics, and we took the best of the best, if you like, and put those on primaries. The way we do AXAF is we have complete redundancy in the detector; we have two full sets of electronics. We put the best of the best in the first set, which we'll call our primary set, and we took the rest of the best and put those in the second set. And that's how we will fly the instrument. Jerry can talk a little bit more about the details of how we managed to go from this nice square design to this very elaborate long and skinny design. It's probably not possible to tell from the camera angle, but this is not a flat surface. This is a detector which has a flat central region, and then from here to here, and here to here, there's a slight tilt-up like little gold wings, and that shape was designed to approximate a very flat circular shape to better match the focus of the telescope when it was being used as a spectrometer.
DeVorkin:Is the dispersion such that the spectrum is actually back?
Yes. The transmission gradings are inserted into the beam. The transmission grading are located behind the flight optics. They swing into place behind the optics and cause a dispersion of what would otherwise be a single focused beam in this direction. So you get little blobs of X-rays at different distances, depending upon their energy. So even though the instrument itself doesn't have very good inherent energy resolution, we now basically measure energy by measuring distance from the zero position and can now more precisely determine the energy of the X-rays just based upon their dispersed position. And the low energy — [Off-tape comments.] The X-rays are imaged into little groups — I don't know what you want to call them — in principle, lines in the spectrum according to their energy. And so we can measure their energy by measuring their position, but we need to have a long, thin detector. As Steve said, in order to match as closely as possible with a piece-wise flat detector the roll and circle geometry of the gradings, we had to elevate these two outside sections slightly, a degree and a half or so. It's there. It's not real perceptible, but you can see it. So that presented an interesting problem from the classical standpoint of making a cross-grid, because we've always wound wires under tension in two directions. And while we can wind wires in this direction, and have, we didn't know of any way to wind wires in this direction with a kink in the middle.
So we, in fact, couldn't use discrete wires, and we took this ceramic block and, in this case, we actually deposited a layer of gold on the block, very similar to the reflector plane that we deposited on the other blocks, and then we just etched very fine cuts through that to separate it into individual segments so that now we had mostly gold but just isolated by these small cuts in the gold so that they form the combination reflector plane, if you will, but more importantly, the lower grid is now what we call a solid-state grid. It's deposited on the block and is no longer individual wires. But it functions just the same. There is still a resistor block here for the wires that are running in this direction with connections, and, again, every eight resistors there's a tap-out that comes out through this cable, out to the preamplifiers. In the other direction — and I will turn this over carefully, but I don't want to damage this. This, by the way, was a prototype unit. I think it sits on that all right. Here now we see resistor strips again down the side for the wires that are running in this direction. See the wires coming up over the edge, held in place with these two ceramic clamping strips trimmed in here with the laser on both edges to remove the continuously wound wire, to separate these into about — I'd say there must be about —
Murray:About 1,500 wires.
Austin:About 1,500 individual wires there, each one with its own resistor. I don't know whether it shows up at all in the — if you get in really close, you can probably see the tap-outs that are going every eight wires that go to this circuit board, which then take those signals out again through these wires out to the preamplifiers. But in between there, for every one of these little wires there is a single stitch with the 1-mil gold wire that stitches the wire to a resistor on that network. You'll notice that this ceramic here is white. These ceramic strips are white. This is this sort of orangey color. That's because this piece was what we call HIP'ed — H-I-P — Hot Isostatic Pressing, which causes the ceramic to become even more dense than it otherwise is. It's basically—I don't remember how many atmospheres of pressure and temperature, but we put it in this special facility, and it was done to get rid of problems with voids and so on and to improve the strength of the material. The flight units were HIP'ed with a slightly different process, to the point that the ceramic almost becomes translucent.
Murray:The ceramic is basically alumina, aluminum oxide. Aluminum oxide is the stuff that sapphire is made out of. But the HIP'ing process basically brings us closer and closer to the sapphire state. So this is a sort of a semi-sapphire, if you like, and the phi units are even closer. We actually thought about making the whole read-out out of sapphire, and for cost and production reasons, it was easier to do it with this HIP ceramic. You can crudely machine the ceramic before you do the high-pressure treatment, and that makes it easier to machine, and then you do the high-pressure treatment and do the final machining, and you get the best of both worlds, basically.
Austin:It wasn't possible to get this large a one piece of ceramic that we could machine and not open up a void somewhere on the surface, and then you always have a stress riser there and you be concerned about fracturing it during liftoff or any other vibration test. So this was an effort to get away from that sensitivity. [Off-tape comments.]
DeVorkin:In a way, it's analogous to a photographic plate.
Murray:Absolutely. A bent photographic plate.
DeVorkin:A bent photographic plate, exactly. I've seen X-ray spectrometers where they would send a discrete detector along the rolling circle using a motor. Did you consider that as an alternative, or that doesn't work here?
Murray:We didn't consider that, and it doesn't work here because the spectrum from the source is filling all of the detector all the time, and if you were motorized, you would —
Austin:One photon at a time.
Murray:— you would not know when to have it in the right place at the right time.
Austin:Go back and forth very fast, and then you still wouldn't necessarily get it.
Murray:Got to do it very quickly. And the derivation of this design, basically, was, as Jerry is pointing out, we actually started making the large square detector. Actually, in our original design, original plan to do it was this 100-millimeter-square detector. In fact, originally we were going to have two of these.
Austin:We were going to have two of these oriented sort of like this.
Murray:Diagonal to diagonal.
Austin:As close as possible. In fact that's why this — I'll turn it around this way. That's why this corner is cut as close as it is. It's different than the other three because it was designed to be built next to another one that was as close as possible to minimize the dead area.
Murray:And then we were going to disperse the spectrum along this way, and we'd get about 250 centimeters of detector. As the AXAF mission went through a lot of evolution from its original planning in 1984 to its current flight status, the amount of space, the weight, our thoughts about what was required changed, and we realized that we could take one of these large microchannel plates and actually slice it into three pieces by scribing it and cracking it, basically, and then take those three pieces and align them one along the next, and have three segments. That led us to this design of the three-segmented spectroscopy array where we get 300 millimeters of length, rather than 250, and we have basically optimized one detector, then, for spectroscopy. There's no dead area in the middle. And we have optimized one other detector for imagery, and the combination of the two of them when they're actually flown sort of, it's this geometry. It has this geometry. The imaging detector is still on a diagonal relative to the dispersion axis of the grading detector, and we did that so that, in case of a failure, we get at least some of the spectroscopy with the maximum distance, rather than if we had oriented things this way, we would only have this much distance. So we get a factor of about 40 percent greater coverage with essentially no cost except in making the design that holds it a little more complicated.
Austin:Plus it also serves, then, as a backup for the medium and high-energy transmission gridding, which is normally read out with a shorter CCD array in the ASIS [phonetic] detector. But that pattern falls entirely on this detector as well. So it can serve as a backup for those observations, as well as can this one.
DeVorkin:So there's a lot of redundancy built in.
DeVorkin:Was the limiting technical concern the microchannel plate at this point? In other words, you said that when you realized that you could break this into three, that's when you decided you could build this.
DeVorkin:If you thought you couldn't make the microchannel plate, then you wouldn't have —
Well, the initial limiting concern was twofold. The AXAF telescope is a ten-meter telescope. It has a focal length of ten meters. So the first thing that happens is that the focal-plane scale is such that you needed a detector this large to cover the half-degree field of view that the telescope is capable of providing. So the first technological challenge was to get the manufacturers to go from these generally smaller detectors to this large one. That turned out to be made possible by the fact that the manufacturers of channel plates were already making fairly large, actually rectangular, channel plates to be used as face plates for high-speed oscilloscopes. It turns out that's another application for these devices. And so some of the technology we needed had been developed, and what we needed to do was have them sort of customize that technology to our requirement. That led to the possibility of this very large square detector. The evolution of the square detector to the spectroscopy detector, as I said, was in part because we could get the channel plates cut, and that meant we didn't have to go through a separate channel plate development for pieces that would fit.
The second part is harder to pin down. It's part of this evolution of AXAF. Originally AXAF was conceived as a very Hubble Space Telescope-like system with these very large instrument compartments, low Earth orbit, in orbit, in-space servicing. It eventually evolved into this higher-Earth orbit, two-instrument system with much smaller compartments for the instruments. As we were repackaging and repackaging and redesigning — we did this about three times during the course of the early part of the mission — the idea of a long spectroscopy detector as an alternative just sort of evolved. I can't actually put my finger on the instant when we said, "That's the right thing to do," but it just sort of evolved as one of the many things we studied, and then the realization of how we could actually accomplish it evolved. What we were left with was, as Jerry said, the main new technology was in this process of putting the wires on the curved surface down and then wrapping the wires. This is sort of a hybrid of what we at one time had hoped we would build, which was a higher solid-state detector. We never quite achieved that, but we learned enough from that work to realize that this one dimensional set of stripes would actually work. We actually built a one-third model of this detector early on, which we called "the dogbone" — it was just a center component with these two big wings — to prove the concept. Then having done that, then we committed ourselves to the full-scale model. So there is actually a logical and orderly transition, if you like, from one kind of technology to the other, with some stops along the way to make sure you could do it. Had we not been successful, then the fallback would have been to have these two side-by-side devices that would still provide essentially all the needs of the mission, but not the extra performance that we've now squeezed out of it.
Unidentified speaker:At one point, I think, while I was moving the camera, you were referring to this and showing where the lines —
Unidentified speaker:— the spectra go. And why it needs to be so wide.
Austin:You mean so long?
Unidentified speaker:So long, yes.
Murray:Okay. When the gradings are at the back of the X-ray telescope and become a dispersive telescope, what happens is that a point source is still imaged at the center of the detector, but the dispersed spectra disperses to the left and to the right, depending on your perspective, right and left, so that as a function of energy, as the energy becomes lower and lower, the dispersion becomes greater and greater. So we start here with the high energies and disperse to low energies in this direction, or high energies to low energies in this direction. The length of the detector determines what the lowest energy is that will still land on the detector. If you make a detector this size, then all that energy is gone and so on. Our objective was to be able to go from the highest energy to about one-tenth of a kilo-electron volt or 100 EV, or about 140 angstroms of wavelength. You can use any unit you like. Since the dispersion, as I said, for the grading was about 1 angstrom per millimeter, we needed to have about 140 or 150 millimeters of dispersion, and that led to this size. [Off-tape comments.]
DeVorkin:So the objective grading produces a zeroth-order and first-order spectra, nth-order spectra, but they're redundant, are they not?
Murray:Yes, they are.
DeVorkin:The first-order spectra are mirror images of one another.
DeVorkin:So why did you want two of them and not offset your zeroth order so you get more of the first order?
Murray:That's a good question, and there are two parts to the answer. The first part is that, as we discussed, in X-ray astronomy we never have enough signal. So had we only put the zero order at one end and looked at one first order, we'd lose half of the photons. And losing half the photons, in our case, is serious business. The second part of the answer is that at a certain point in energy, our efficiency goes away because we have a window, this thin plastic window, that's preventing the UV light and the ions from getting onto our detector. And that window, which is sitting up on top of the channel plate, becomes basically non-transmittive at around 150 angstroms or longer. It just doesn't transmit any more X-rays.
Austin:We wouldn't have had anything out there anyway.
Murray:So our X-ray detection efficiency goes to zero. We actually worked very hard to make the windows as thin as we could. They're, in fact, for spectroscopy, thinner than they are for imaging just to allow that extra little bit of efficiency. I could also tell you that there's a political statement, which is that AXAF is an X-ray astronomy observatory, and X-ray astronomy is defined, if you like —
Austin:Stop it. (laughter)
Murray:— as beginning at 100 EV and energy and working its way up towards about 120 kilovolts, and going longward of that length, it moves us into the extreme ultraviolet regime, which we are not. Not that scientifically there's that sharp a distinction, but that was part of the politics of what we could get away with.
Austin:There are also some other subtleties. We offset point the zeroth order a few millimeters with respect to the center of this, which allows us to better match the roll and circle in some regions on one detector and one on the other, to try to get a better match to that circular geometry.
Murray:And there are other things, like, you know, the fact that we are so limited in signal means that if we make two detections and they both show up, you'll believe the result. If you make one and not the other, you'll have an issue. There are all kinds of reasons why, but mainly it's that we just don't have enough photons. In an optical world you can get away with throwing away some of your light and not lose your ability to do science. I should mention, for AXAF, the typical observation is probably going to be on the order of a day. That is, about 75 to 85,000 seconds looking at one object, doing one observation will not be atypical. And even with that length of time, the number of actual source photons we get will be measured as, you know, typically a few thousand. So we're not really in a situation where we can afford any loss. Every few percent means a lot to us.
DeVorkin:You haven't turned over the 100-centimeter-square one. Are you still having problems?
Austin:Well, I left it this way on the table mostly because the sensitive things on here that can be easily damaged by handling, are these small stitch bonds here. This is a functioning, working detector that we do use in the laboratory, as is this one. So we don't want to do anything to destroy them here. But this shows the more normal pattern that we've seen evolve now from the initial Einstein era up to this size, the 100-millimeter size — the same two sets of crossed wire-wound grids. It's the same wire. It's the same spacing. The only real development that we've made here, other than increasing the size of it, is in the resistor networks themselves. We've now been able to go to thin-film technology that enables us to get all the resistors in a smaller space so that we don't have to mount them on opposite sides, and it greatly simplifies the stitch bonding. But other than that, there have been no significant changes. The technologies just keep pushing it larger and larger.
Murray:I would expect that we've now reached about the maximum sizes that we will get, mainly because there's nothing pushing the microchannel plate manufacturers to go to larger sizes. There's nothing that stops the technology of the read-out from being larger. As you see, we can do it. We've gotten the amplifiers down to such small size and low power that we can grow those. But the microchannel plates themselves are now going to become a limiting factor, from the point of view of size, and I would also say, for X-ray astronomy, that the telescope fields of view aren't going to grow any bigger.
Austin:Aren't going to grow any bigger.
Murray:And the focal lengths are going to be — it's going to be hard to make telescopes with ten-meter focal lengths.
Austin:So there's no real need to have a bigger detector.
Murray:That, I'm sure, will change over a long period of time, but certainly in the next phase of instrument development we're not looking at going larger than this. We're actually looking at similar-size types of detector, even in the one case reverting and going to even small detectors.
Austin:You don't want to be the guy that's known as the one that said, "We will never need a larger X-ray detector." (laughter)
Murray:I'm careful not to say never. Never say never.
DeVorkin:In the case of the microchannel plate, you mentioned that high-speed oscilloscopes drive the availability of them. What about the cross-grid technology? Has it found application in other areas?
Murray:I'll try to answer that. The cross-grid technology has not, to my knowledge, been used in any other application other than X-ray photon counting. There are a number of alternative read-outs, schemes, for microchannel plate-type detectors that use different technologies, and some of them have been in use in other areas. For example, for extreme ultraviolet, there's a read-out technology called "wedge and strip," which is a continuous resistive anode that's used. For some other applications, people have been using what are called "resistive sheet anodes." All of those technologies are roughly four times worse in their ultimate spatial resolution than the cross-grid charge technology, but they all have the advantage of being relatively high speed; that is, they can work at count rates of 100,000 photons per second. This technology is fairly limited in its speed. The highest rates we can work at are about 3 to 10,000 events per second, based on the way we've designed our electronics and in order to get this very high spatial resolution. If there were applications, and we've talked about the possibility of medical applications, industrial process applications, where the requirement for very high spatial resolution supersedes the requirement for very high count rate, these technologies could be applied. But, (A), they haven't come up; and, (B), from the perspective of the Smithsonian [Institution], at least, we're in the business of doing science research, not necessarily trying to foster industrial applications. But all of this information is public domain. That is, the technologies have been published, there are no licenses or patents held on any of these technologies, and so they are freely available for someone who wants to use them. [Off-tape comments.]
Murray:The size that we're working at now, which is about 100 millimeters squared, is pretty much at the limit of the technology and the limit of the interest that we have in developing the technology.
Unidentified speaker:Why is that? [Off-tape comments.]
Murray:The detector we built now, which is 100 millimeters by 100 millimeters, is about as large a detector as we think we have to make. In part, that's because the microchannel plate manufacturers are going to be limited to the size microchannel plates that they need to make for their applications, and we would not be involved in anything larger than that unless they suddenly became available. Besides that, the X-ray telescopes that we use these for are not demanding the larger sizes. Ten-meter focal length, which is the size of AXAF, which caused us to build this size detector, is about as long as a focal length as we expect for high-resolution-type X-ray telescopes. One has to always be careful never to say never, and surely the technology, as we've shown, can grow from the tiny sizes to this 100-millimeter size to this 300-millimeter size. So we know how to make very large read-outs. If large detectors become available from the channel plate manufacturers, we know how to hold them, and if an application arose for using them, we would pursue it. But I don't actually think that that will happen, and I think that the next generation of missions that we're actually thinking about are all geared towards different goals which don't involve very, very large detectors; involve improvements in detector performance.
DeVorkin:Okay. That's good, but you brought up that magic word of "goals." Your goal, of course, is to do more than build the detectors, but it's to use them. I know launch is sometime at the end of the year.
DeVorkin:December third. You must have some game plan as to what you're going to be observing.
Absolutely. Absolutely. The way that AXAF was conceived was that the instruments themselves were not proposed. What were proposed were scientific investigations supported by instruments. In fact, the team that's put together the high-resolution camera doesn't just consist of people building the instrument, but also has a whole cadre of scientists waiting with baited breath for us to get in orbit. As part of the arrangement with NASA, the team has what is called "guaranteed observing time." In the first ten months of operation, the HRC team receives a certain fraction of time, which amounts to about one and a half million observing seconds. We have submitted a list of targets that we want to look at, which range from stars and stellar coronae sources to a very deep survey of a region of the sky to the maximum sensitivity that we can reach with AXAF in any reasonable time. I'm devoting 350,000 seconds to staring at one tiny part of the sky with the camera to make the most sensitive measurements of X-ray sources that we can possibly make.
That program will be carried out by, first, two months following launch we will be doing what's called "orbital activation and calibration." We'll be turning on the various instruments, making sure they work. We'll be internally calibrating the systems, and then we will be looking at a few well-known objects in the sky to characterize the on-orbit performance of the instruments. After that two month period is over, there'll be a two-month period where all the observations will be carried out in support of the instrument guaranteed-time observation teams. I'm right now in the process of selecting which of the forty-two different targets I have in my observing program I want to do in those first two months. Then in the following nine months, AXAF goes into a general observer mode where 70 percent of the observing time is dedicated to the entire astronomical community in the United States and, in fact, in the world. We've just finished a peer review for that in selecting which targets will be looked at. The remaining 30 percent is spread out amongst all the guaranteed-time observers, giving us the rest of the allocation that we had for this first year of operation. Then AXAF goes into its second normal year of operation. There'll be another round for general observers. Again, 70 percent of all time will go to the general community, 30 percent to the guaranteed observers. Then starting in the third year, the guaranteed time drops down to 15 percent, the general time goes up to 85 percent, and we go into a steady-state mode. AXAF's supposed to run for five years. We hope that it can continue to run for at least an additional five years beyond that.
There's every reason to believe that in terms of the way the observatory's been designed and in terms of what we need to use up as consumables, if you like, and, in fact, given the orbit and given the success we've already seen with ROSAT in terms of the longevity of the instruments, it's conceivable AXAF could have a fifteen-year operating lifetime. Now, what we do scientifically in that is virtually everything we can imagine, from, as I said, real detailed spectral studies of the coronae of stars, which has never been done, observations of active gallactive nuclei, quasars, clusters of galaxies, globular clusters, supersoft sources, neutron stars, pulsars, binary systems. I mean, it just goes on. We have some unique capabilities on AXAF. Besides the high-resolution camera and the low-energy transmission grading, which we sort of talked about, there's a CCD camera which has been built by friends at MIT [Massachusetts Institute of Technology] and Penn State, and a high-energy grading which extends the spectral resolution with very good resolution to the higher end of the access spectrum, which goes from a tenth of a kilovolt to 10 kilovolts. So there's an awful lot to be done scientifically, and there's an awful lot of expectation that we will do as much for X-ray astronomy with AXAF as Einstein did when it first was launched.
Austin:And Uhuru before that.
DeVorkin:If we were to have a sound byte, something that – I don't think I'm asking for predictions, even speculation. What is it that you'll be looking for that will be answering fundamental questions about the structure of the universe, its origin?
Okay. We have two programs that I can talk to, that address that. One program is a program which we're trying to use, to use X-ray measurements to measure the Hubble constant, and, in fact, not only the Hubble constant, but the deceleration constant. There are two constants that basically describe some cosmologies: how fast the universe is expanding and how fast or how quickly that expansion rate is changing as you look back in the universe. With X-ray astronomy, we are hoping that we can measure what's called the Sunaev Dolvich [phonetic] effect. It's an effect that occurs as microwave background, which is produced when the Big Bang passes through clusters of galaxies on its way to the Earth, and by making measurements of the properties of clusters of galaxies that the microwave background passes through and measuring very carefully what their temperatures are and what their electron densities are, and comparing those measurements in the X-ray wavelengths with measurements that are made in the microwave wavelengths, you can deduce what the distance of these clusters are from the Earth.
By knowing their distances from that measurement and knowing their red-shift distances from the shifting of the wavelength of their radiation, you can make a comparison, and that gives you ultimately these two constants. You need to have the sensitivity of an AXAF and the spectral resolution of an AXAF in order to make the X-ray measurements, and on the ground there are microwave observatories that are making the database needed to compare against. That has implications immediately for those two constants. The deep surveys that we'll do at AXAF, I told you about one that I'll do with my instrument. There will be another done with the CCD camera, and then probably three or four others done as part of the course of the mission. We'll be looking there for the most distant X-ray sources we can detect, because we're looking at the faintest, we're looking at the most distance. So we're going to be asking questions about what is the time of onset of X-ray activity in distance objects. When did the nucleus of a galaxy suddenly become active? What produces that phase in a galaxy evolution? When does a galaxy begin to produce stars that begin to produce supernovae that begin to produce X-ray emission? Those kinds of questions relate immediately to the whole early history of the universe in terms of the formation of these kinds of objects.
As we accumulate a lot of observations with AXAF, sampling different parts of the sky, we'll be asking questions about the uniformity of the universe. If you look in this direction, does it look the same as when you look in that direction? When you start comparing what are called the two-point correlation functions, that is, what is the average separation of sources on the sky for given red-shift distances, you begin to learn something about the distribution of matter as it separated from the radiation-dominated universe into a matter-dominated universe. These are kinds of things which deal with what are called large-scale structure questions. Why did galaxies line up in these clusters and then why do clusters gather in these superclusters, and why do we have these voids and walls and so forth that people have been seeing in the optical? In the case of X-ray astronomy, what we hope to do is push the frontier back in terms of X-ray source counts to see if we see these same kinds of structures and whether we can relate them to the optical structures and ultimately to these early universe density fluctuations. So those are examples of how we anticipate using AXAF. Now, I can't tell you the results. I wish I could. But certainly those are all going to be exciting areas for us.