DEIMOS PRELIMINARY DESIGN REVIEW Tuesday, November 15, 1994 Meeting Minutes It should be noted that the tape was very hard to hear in some places, in others almost impossible, there may be gaps where conversations were inaudible. When an answer is given it is assumed it was made by the presenter unless where otherwise noted. A response indicates more dialog on the subject or related matters by the group. Present: Review Team - Dan Fabricant (chair), Peter Gillingham, Harvey Richardson, Jim Gunn, Steve Shectman, Hilton Lewis, Chris Stubbs, Jim McCarthy, John Geary, Frank Melsheimer; UCO/Lick Team - Garth Illingworth, Sandy Faber, Harland Epps, Joe Miller, David Cowley, Jack Osborne, Bill Shepard, Richard Stover, Steve Allen, Deanne Lago; Observers - Marlene Couture, David Hilyard, Matt Radovan, Dean Tucker, Brian Sutin, Gerald Luppino, Eric James, David Koo, Lloyd Robinson, Scott Trager, Katherine Wu, Erich Horne, Mike Bolte, Jerry Nelson, Neal Jern, Maureen McLean, Joe Calmes, Tom Soifer, Ed Romana, Tom Bida, Al Conrad, Lee Rottler, David Tytler, Fred Vescelus, Marc Davis. (A list with addresses and e-mail is provided at the back of these minutes as an addendum.) Introductions Of The Chairman and Opening Remarks - Garth Illingworth Garth Illingworth welcomed everyone and thanked them for their time and effort. It has been found over the years that reviews such as this one and the Critical Design Review (to be held later) are extremely valuable. The feedback from the review teams is very useful by outlining areas of concern that may have been overlooked by the design team. A detailed assessment of our project is welcome and every effort will be made to take all the comments received here and the final written report and peruse them very carefully to use them as we can to improve the instru- ment. The review team chair was introduced and each review team member made their introduction around the table. The presenters then introduced themselves. The science team was introduced last. Overview Of Instrument And Performance Goals - Sandy Faber Question: It was mentioned that the gravity vector would be less being at Nasmyth, however isn't it much worse than at Cassegrain, because the instrument must rotate if it's lost on the sky, so don't you get a very rapid change in the gravity vector? Answer: This is also the case at the Cassegrain. Response: But the change in the gravity vector at the Cassegrain is known more than at any equa- torial telescope. Although the z rotator changes (if you are close to zenith) the changes of gravity vector within the instrument pulling sideways is very little. The Cassegrain would be much easier mechanically to prevent drift on the CCDs. Answer: It wouldn't be much easier, because another factor at the Cassegrain is that you would be holding the instrument only at the topend of the instrument rotator. Here at Nasmyth it will be held with both ends. We have been noticing as we look at the mechanical design there are so many elements here that are horizontal, actually you only have to worry about gravity acting in one plane. It has actually proven to be an advantage. Response: The Australians could not eliminate drift at a Nasmyth spectrograph. Of course at the Cassegrain the spectrograph turns sideways and that does not make your spectrum drift. What you want is the slit and CCD not to be shifting. Answer: There is another problem that has crept in with the LRIS spectrograph at the Cassegrain. There the instrument rotator is attached to the telescope and is not perfectly round, it's exerting stresses on the spectrograph as you rotate. So you have this very irritating cross talk between the telescope and the instrument that can be eliminated because were doing it differently. Question: Don't you have a more fundamental problem now, isn't your choice made for you because the Cassegrain on Keck cannot support an instrument anywhere near the size you have? Answer: Yes, that's true. Question: You're talking about the stability of the flat-fields, if during the night you are using 8 or 10 different slit masks, what is the advantage there? Do you have to do a different flat for each one? Answer: That can't be answered, except to say that there are plenty of people who are going to be using the spectrograph for single object spectroscopy. So, my comments are more applicable to those folks. Response: There is also the stability over an hour or two exposure, for multi slits, you still have to have the stability. Answer: The gain is primarily for single object users. There is actually value in having stability enough that you could flat field the slit masks in the afternoon and at arbitrary positions. Unfortu- nately the Keck telescope is so big compared to the dome aperture, there are many places where you can't flat field off the dome without moving the telescope physically to a lower elevation. Therefore you want to have the spectrograph stable enough so that you don't have to calibrate it exactly the same at elevation and end solution. Question: Suppose someone were able to eliminate friction, how much would that relax things for us? Answer: It would do two things, if you look at the most critical tolerance of 8', you could exam- ine each one of these modes and we have done that (this is described in Chapter 7). It could be said that it is a wavelength stability, if the x direction went away, you would still be left with some direct steering requirements in y. For most of these modes, soon after the wavelength stability cri- teria kicks in you'll find another criteria kicking in. I will generalize and say that all the toler- ances would get easier by a factor of two, but they wouldn't go away. Question: Could you tell us a little about the flat-field procedure that you intend to use. Is this a dome illumination? How long does it take with the existing instruments to do a flat-field? Answer: It takes awhile to slew the dome, it doesn't take long to flat-field, it takes on the order of a few tens of seconds to take an exposure, we will certainly build that in. We also have an internal flat-field inside the instrument, but we have no idea how useful that is going to be. If I have to be truthful with you, I think we are going to wind up flat-fielding on the twilight sky. I am predicting that this pupil illumination issue is going to be important and if that proves to be true, I don't think any dome flat-field that we can supply will really solve our problem. Response: But that's going to become very hard with 8 or 10 masks. Answer: Yes Question: Isn't the pupil illumination going to change during the exposure anyway? Answer: That's an irreducible problem and we probably won't know until we go to the telescope and take observations. It is due to pupil illumination changes, we can make measurements and try to come to some conclusions using the Cass spectrograph (and maybe LRIS) but it may depend upon the particular spectrograph and optical configuration. Question: Could you bring us up to date on the status of the second barrel? Answer: Regarding the second barrel, not very much has changed since we have last talked to the SSC. Our energies have really been focused on this PDR and our very next item to do is to write a rather attractive brochure for our development folks who have expressed an interest in helping with funding. This is going to be our top priority this coming Spring. Response: At what point do you have to make a decision? I assume right now you are making a design assuming you are going to build two barrels. Answer: Next summer. As of early September of 1995 there will be no penalty, after that point we're going to begin fabrication. It will be much cheaper to build two rather than one. We are on a real tight schedule to find that money. Optical Design and Fabrication - Harland Epps Question: What are the typical accuracies when you start polishing? Answer: We typically go down to a few microns. Question: Regarding the drift scanning distortion, is it just a result of the camera? Answer: No, it's end to end, including the telescope, camera and collimator. Response: Where does it mostly come from? Answer: We're not sure yet, we haven't analyzed it in that much detail. We have looked at the camera really hard and we think the camera does virtually nothing, but we're not sure to what extent its the collimator or telescope, we have yet to pursue that. Question: Are you considering a astigmatism corrector ahead of the window? Answer: We have considered that and we wondered if we could make the window into the astig- matism corrector. We know how to design such a thing, but it is a question cost and what the necessity is. At 3/10 of an arcsec RMS diameter at the very edge of the field, we think we are doing real well. If cost were not an object, and if someone wanted that correction we could put it in. Sandy mentioned that the problem with that corrector is that it changes the location of and the cur- vature of the focal plane. Therefore we need to have it now if we are going to incorporate it. We have had another thought since then, this astigmatism obviously is a problem mainly for the direct imaging mode for slitlets. The thought occurs that maybe we could figure a correction on the front mirror, this could take out one component of the astigmatism. Obviously it's varied below the field and varied vertically, this is something that is being discussed. Response: If you use a classical gas plane corrector it also introduces a lot of distortion and you may not be able to meet your specifications. Answer: (Sandy) We feel that the drift scanning mode is sort of a gravy thing, not something that we will make concessions for. It would be more important to cure the astigmatism of the tele- scope, but realistically we won't be able to do that now without having a lot more money in hand. Question: How do you compare the Calcium Fluoride material with that new Schott material (FK-54) which has very similar characteristics? How would it's availability and costs compare or can you get it in big enough sizes? Answer: There are several materials that you might consider as alternatives, including the FK-54. I have found that on several occasions that I have designed with it on the assurance from Schott that it was available, then when you try to buy it they say sorry, we won't sell you any. They do sell pieces a couple of inches in diameter, but I have never heard of anyone who could get a big piece. There is a material called FK-03 that OHARA is making, but it is more expensive than Calcium Fluoride. There is also a material FCD-10 that is made by Hoya, but we think that all these materials would be a compromise they are probably all more expensive, we don't know if they are available and we are happy with the Calcium Fluoride. We're especially happy that Rob Sparrow is so excited about the project and is willing to put his muscle behind delivering Calcium Fluoride. Question: Do you have a few comments about the need for an Atmospheric Dispersion Corrector (ADC) and how that might change or effect everything? Answer: Well we certainly need an ADC if we are going to do broad band imaging. Some years ago when we were doing the optical design considerations for the Keck telescope we came up with several possibilities, none of which we were very happy with. There is a possibility as Jerry Nelson pointed out recently; you make a pair of quartz prisms that are approximately the same in angle and you put them some distance ahead of the telescope focal surface. When you counter rotate them they essentially do nothing, they act like a plane plate. Then instead of rotating to articulate them, you take one and move it back towards the focal surface and what it does is more or less is to maintain the angular deviation that it had. However the leverage for correcting the color goes down, so presumably you can put this idea together with some calculations to make an ADC out of it. We haven't done it, but it sounds like a good exciting idea and it may be the key to an effective ADC on the Keck. It is something we intend to look at. Response: Is there not an optical component that you can take advantage of in any way to help improve ADC. Answer: In principal you can figure the surfaces of an ADC, I understand that has been done on the Anglo-Australian Prime Focus ADC. We are a bit hesitant to get into that because moving flats are just bad enough, but moving surfaces with power on them adds mechanical complica- tions that cost money. Question: What is the effect of some realization of an ADC at some point in the future? As I understand it there is no ADC work being done by the project. My concern is if something is done two years from now what is the effect it will have on the optical performance of this instru- ment. Answer: It could only serve to improve it. It would be relatively neutral in terms of our focal plane distortions. Response: Is there a need for someone from this project to become the driving force to make this ADC happen. Answer: Our problem is that we still need to raise the funds for the second beam. That has to be our preference at this time. We are aware of this, it should be a project of the observatory and the SSC, not DEIMOS, as it also concerns HIRES etc. Response: If nobody is pushing on it, it's not going to happen and the impression is that we are moving towards individual ADCs for the instruments. Answer: You might as well design one that would take care of the Cassegrain, LRIS and anything else at the same time by having it out in the original place if possible instead of at the Nasmyth platform. It is certainly a major project that needs looking at, it's a project of the size that it could probably have a name, PI and be developed all on it's own. It would probably end up costing approximately $1,000,000 before you get it all done. It's certainly not a $50,000 add on. Response: Does anyone know where that stands with respect to the run out budget for Keck II? Answer: (Fred Vecelesus) We talked about putting some money in the budget for this, I believe in fact that this is going to happen, what I don't know is whether or not it will be adequate. Joe Miller mentioned he remembers approximately $300,000 being reserved for Keck I. Garth men- tioned that the budget for Keck II has been carrying a comparable amount. We are concerned that it has not been enough but nobody has had time to actually to do a preliminary design to come up with another cost that could justify this area. Response: (Fred Vecelesus) The $300,000 for Keck I has now been rolled into the overall contin- gency, so it's not an identifiable piece of money anymore. Comment: (Harvey Richardson) I recently designed two for Gemini in connection with their Adaptive Optics plan. Certainly if the ADC is dedicated to an instrument it can be superior in performance and less in cost. This is because it can be placed in a collimated beam, at the tele- scope. If it is in a converging beam it could produce aberrations that are not present in an ADC internal to the instrument. Answer: (Garth) We need an ADC for the spectroscopic mode, which unfortunately needs to be in front of the slits and slit mask. Response: The DEIMOS with it's blue limit makes an ADC easier. Answer: (Sandy) The old ADC design modified to take account for the fact that we are redder on this telescope than on other telescopes actually would be great and we would be very glad to have one. Response: It doesn't matter in spectroscopic mode because it doesn't matter if you have disper- sion along the slit. Question: What is the back focal length, behind the dewar window? Answer: It's approximately .15 inches. It was discussed with the detector people and they were happy with it. Question: Did the 15 micron pixel size come first, it seems like everything comes from that? Answer: Yes, it did come first. We postulated to a 15 micron pixel size up front, anticipating that we would be able to take advantage of it. Response: Since CCDs have been made with a large span of pixel sizes going from 7 to approxi- mately 30 microns, if that were a free parameter how much would that relax some of the design? You are sort of cramming the beam into a particular footprint in some sense. Answer: We started at 15 microns and we've come up with a plan that we think is about as ambi- tious as we can manage, but we do believe we can manage it. Given that it would be silly to do anything less. We are very much opposed to the idea that you go to the limit and then back off and do only 70% of that. That's not the business that we want to be in, we want to go the limit and then do about 105%. Response: That is not the thrust of my question. A mask set for CCDs is cheap and the optics are expensive, if you leave that till the end and say all you want is for the pixels to fall within the regime of manufacturable sizes doesn't that make your job easier and cheaper? Answer: We think we are right about where we want to be. Perhaps the questions should be read- dressed, it is a question that we should be able to answer quantitatively. (Sandy) It's set by the number of pixels across the narrowest spectroscopic images you would be likely to get. With this 15 micron pixel it gives you three pixels across and is very good for sky subtraction. Response: That point is not disputed, but nevertheless it is part of a plate scale of the spec- trograph camera, that's the free parameter here that's picked for no good reason that I can see. Response: The assumption is that there is a practical limit to the area of silicon. This is at the large end of it, he could have gone to smaller pixels and a smaller area of silica, but I think almost certainly it would make the optical problem harder. If you could have a bigger area of silica with bigger pixels that might be very interesting. Answer: That would make the camera even bigger. Response: The back of the camera would get bigger, which might help, the front of the camera wouldn't get bigger. Answer: (Sandy) You have to consider the size of wafers, lately what may have changed our thinking is that Loral is going from 4 inch to 5 inch wafers. Response: I appreciate the ability of CCDs that plays a role here, but this is a more generic ques- tion. Answer: (Sandy) When you lay out the flow of the design like that, you try and make it simple, I have never found a reason to want to change that original function. Response: Unless it somehow opens more parameter space and cost savings. It is not clear that you are giving up very much with this design. Answer: It's not a hard problem to understand, there are a few very simple equations that were laid out three decades ago that relate pixel size information, through-put and so on and you can put those equations to use. For people who design spectrographs all the time, you can almost do it in your head. Where the actual 15 micron number came from, Sandy asked me what I thought I could do and I said maybe 15. That's where we got the number. Response: Without being obnoxious, I'm questioning that as the starting point of the whole flow of the design. Answer: (Sandy) I think the sanity check is to go back and ask yourself how many pixels do you have across the narrowest thing that you want to look at. We have done that and it's roughly three in the most stringent mode you can imagine, and we are very happy with this. Question: Won't the 3/4 arcsec slit be the work horse slit and wouldn't you be willing to accept two pixels across a half arcsec slit for optimum three pixels across? Answer: (Garth) That's what you're getting with 1200-line grating with the anamorphic factor. Response: I didn't see that in the table. Answer: Yes you are correct, there are a few more pixels across, but what is work horse for one person is not for another. Question: How much faster could the camera lens focal length design be pushed before you begin to exceed the telescope aberration? Answer: We started with the 12 inch focal length and the detector was smaller, but the field angle was the same. The 12 inch produced the first model shown. It's a little beyond what we could tackle because of the unavailability of materials, the curves are more bizarre and everything just looks a little beyond what we want to do. In the second report, I did a 14 inch focal length and a 15 inch focal length. It is surprising you can see those two cameras are considerably different. The 14 inch camera is more ambitious looking than the 15 inch. Either of those cameras could probably be built and at that point it was more a scientific choice having to do with the informa- tion, through-put and number of end pixel resolution. That's why we opted for the 15 inch. If the decision were made to go back to a 14 inch, we could probably produce it. Response: You did not answer my question, in both of these drawings, you're targeting the same spot size. The design is getting more complicated when you go towards shorter focal length to maintain the spot size. I'm saying let the spot size relax that tolerance until you are comparable to the telescope aberrations and push the focal length shorter. Answer: (Joe Miller) You have to remember you're still always imaging slits when your doing spectroscopy. (Harland) My feeling is that I don't like to allow aberrations into the camera, then you see them and I don't want to see them. I don't want anybody to realize my camera's in there at all. Question: If you did away with it would it make your life easier in getting the field flattener closer to the CCDs? Answer: Yes, I put a field flattener inside the vacuum and Jack almost died. I tried it to see what we could get from it and it didn't really help us in a way that I thought was important. I have in the past put curves on the vacuum window and here it didn't look like it was the right thing to do. It has errors - if you put an optic in the dewar side you introduce alignment problems you wouldn't otherwise have, also CCD calibration problems. The dewar will need to be handled and worked on as a separate unit before it gets integrated into the telescope. (Jack Osborne) We did this with HIRES and it was real hard to look at the CCD through the window. Response: I don't quite understand if it makes the optics easier, I think it's needed. The dewar is not going to be used for anything else. If it is optically convenient to make the window to the dewar the field flattener, you should do it. The disadvantages in doing this are very slight. Answer: (Sandy) We have a couple of things we were thinking about. One was that for align- ment purposes, it might be really useful to look directly into the dewar and see things without dis- tortion. Our other consideration was that we needed glass as UV transmitting. (Harland) Before we commit to build I am going to start with a clean page and go through the whole design again. This is one of the issues that I had uppermost in my mind. The amount of negative that the ele- ment has drives the whole choice of the glass ahead of it. It may mean by changing the rear area and I may be able to do something in the forward part. Response: You may even be able to make it out of quartz. Answer: I gave that pretty hard look in this design, it didn't go, I believe I ended up with LAKL- 9. It will be checked again. In getting to this point, our goal was to get something that optically would work and that the mechanical people could have a careful look at and comment on. They indicated that the element on the rear of the triplet is the critical item having to do with the poten- tial distortions. We have in turn looked at the image degradation and we have convinced our- selves that we are pretty happy. Mechanically we can support this lens and we think we can build it. That doesn't mean we're going to stop here, we are going to have one more hard look at it before we commit to building it. Question: Have you thought about looking at the idea of putting some power into the filter? In principal you might be able to eliminate that whole element by making it a contacted lens that goes on every filter. It's not that expensive to do if it's not aspheric. You can have a dozen of them made up pretty cheaply. Answer: I hate to put a filter in that position because the beam is very small at that point and fil- tered glass is notoriously irregular. We've got the filter in another position and I am not even happy with that I would like it more forward. On the other hand it gets bigger and thus more expensive. Response: The combination of having a curved window and maybe a curved filter might elimi- nate that last element. Question: Can you tell me about the oblique reflections (cross polaroids)? Answer: Well, we are going to have to eat whatever they do because we can't make this camera without curved surfaces. Response: No, not the camera, in the tent mirror? Answer: We will have to worry a lot about that. It certainly is a problem with polarization, if you get any oxide layer on the reflector it causes havoc. We don't have a choice, we haven't been able to come up with a different kind of a package that would eliminate that tent mirror. Response: It would be interesting to know what the magnitude of the effects might be. Answer: (Joe Miller) A freshly silvered mirror is not polarized. Response: What happens in the grate area. Answer: If you have a pure metal, you don't have a problem, it's when you get a little oxide layer on the metal that you have the problem. Response: Are you going to keep the instrument purged with nitrogen or dry air? Answer: Probably nitrogen. Response: Well you're not going to keep the tertiary and the gratings purged. Answer: That will be a problem, spectrographs all have polarization problems. I'm not happy with the tent mirror, if someone could come up with a different packaging, that would be marvel- ous, but we just haven't been able to do that yet, and maybe never will. I don't see a different way to package it. I know Jack has had a very hard look at the packaging, maybe there is a way to do it, I just haven't seen it yet. Question: Totally ball park, what are the tolerances of centering an alignment of these optical elements? Answer: Ball park - machine shop tolerances, if we put things together to machine shop toler- ances, they will work fine. Response: So 1/2 mil? Answer: Not even that, 3 mil's (ball park). Electrical & Mechanical - David Cowley Question: Is the instrument rotation part of your job or part of the telescopes job? Answer: It's part of our job, our philosophy is the same as we're using with the HIRES Image Rotator in that the instrument itself will be responsible for rotating, and receive commands from DCS. Question: Are these slopes measured on the cylinder? Answer: Yes, it's just slopes taken from the calculation of a beam with the moment of inertia of the cylinder. Response: Towards the two points. Answer: Yes Question: The idea was that there were going to be a couple of fibers that drive those astrometric CCDs that are off on the side that reside in the slit mask, how do you deal with the fibers? Answer: We haven't got to the detailed design. (Sandy) The fibers are not in the slit mask, they're at the focal plane so we know exactly where the masks are relative to the focal plane. Response: But you have no precise way to re-register a given focal plane with this do you? Answer: We do, we have these tooling balls that are fitted onto a kinematic mount. Response: If you take the focal plane out again and then put it back in how accurately do you line up again? Answer: We're hoping to realign them to a tenth of a pixel. We still have 11,000 hours of engi- neering yet to go into this instrument. (Sandy) Supposing we missed this, it's not a horrible thing. Response: It's only horrible in the sense that you can't really use dawn and dusk flats anymore. Answer: (Sandy) That's not true because if we also institute the flexure compensation system, then we can use the internal capacity to adhere any slitlet to any pixels we want. Question: Are you going to laser cut your masks? Answer: Yes. Response: They will be cut while the mask is mounted on a spring, so you'll have blanks with the skin already stretched over it. Answer: We know we are going to make at lease 40 of them, we may make as many at 100. Manufacturing is a key area. This one we made out of a solid plate and it was expensive. Question: Will it be possible to make room for more gratings? Answer: It is impossible to make room for more gratings with our slide. The way we take care of other gratings is by allowing the user to change them. (Sandy) It is worth investigating what we would do. If we move the spectrograph back a little bit, it would get the grating slide out of the elevation bearing, then maybe we could consider adding another one. Response: Quite a large fraction of users would probably want to change the gratings. Answer: We don't disagree, however with placement now, we just couldn't design for more. (Sandy) Don't forget that if you're a single object user you have four gratings. Question: Where is the center of gravity of the camera? Answer: I'm not sure, I have calculated the center of gravity of the various lenses, but never the whole camera. We just haven't gotten to that part of the design. Question: What do the blades slide in? Answer: These are just aluminium against titanium at this point. Question: The travel time across the opening is 1/10 of a second? Answer: Yes, it goes real fast, you wouldn't want to get your hand in there cause it will come off too. Question: What will be generated by this shutter over time (i.e. pieces of rubber, metal, etc.)? Answer: This is new and different so we don't know, that's why we are trying to build early. Question: Is 540 is this enough for worst case rotation? Answer: Yes, 360 is one revolution. Response: You need the 360 and 180 because of multi slits. You need the 360 for the multi slits, but in the rotation it's more than 180. Answer: (Garth) Why is it more than 180, isn't 180 worst case? Response: The question is getting the multi slits on the sky, and then having enough tolerance in either direction to go. Answer: (Garth) 180 plus 360 should be the extreme. Response: I would think that you need 180 on either end. Answer: (Garth) You're right. Response: If you're up against one end, to get your multi slits on the sky. Answer: (Sandy) When choosing your multi slits you always have the 180 of freedom. (David) Unlike the HIRES rotator, this one does have an end stop to it. Question: Will you need to heat the dewar window? Answer: (Jack) We haven't thought about that, but we don't think we will. (Dave) We have faced a couple of issues with the dewar window. Harland's design is 3/8 of an inch and we know that's right on the boarder line of what we would like as far as stress. We're going to ask Harland to go to 1/2 inch. Question: What is the thermal flux? You have it as -100C on one side and 0 on the other? Answer: We don't know, a year from now we'll know the answer, but right now we don't. Response: It's about 2 watts radiated load from the CCDs. Question: Are the cameras fixed? Answer: Yes they are. Question: Why have you put the television so far forward of the rotating bearing? Why did you choose to pick off and transmit the field all the way across the diameter? Answer: For the control of demagnetization, we need some optical path there to work with. We will need 4 to 1 (50 inches) for demagnetization. Question: It was implied earlier that we would have a choice of 6 gratings that we could use along the slit, this description implies that we can only actually look at the opposite masks is that true? Answer: (Sandy) That is true, yes. Question: If you use 3 out of your 10 slots for standard slits, is it possible to combine the func- tions of the three? Answer: (Sandy) Interesting idea, it's worth thinking about. Response: Seven per night might be close to a minimum if you have a failure, bad weather and things like that. Answer: (Sandy) I was thinking that we would need a set of grid holes. Question: What is the cycle for focusing? HIRES is focused every couple of minutes? Answer: (Garth) It is pretty stable, we may not have to do that very often. Especially if we have a temperature readout so that we can do optics based on that. (Sandy) There should be one long slit, that will probably be combined with a set of holes for focusing spectroscopically as well. I'm thinking two standard masks and therefore up to eight others. (Garth) It would be the observers choice, they could in fact take them out if they decided they didn't want them, they are all inter- changeable. Question: Would it make sense on all masks to have the option to put a long slit somewhere that you might be able to cover up? Answer: One of the problems is that this particular long slit lives right where we hope to have a rib. Question: The detail on the drawing shows that slit mask to be tilted differently is that right? Answer: Yes, there will be some sort of element, either a grating or slanted element right next to that slit, it will live on that slit mask. Response: By working a little up the axis, could you use the normal slit angle without having to special tilt that will make it incompatible with the other slits? Couldn't you have a shiny reflec- tion off a normal cut slit and have a pick-off do the work? Answer: Maybe, we would like to look at it. The whole design of this is in the early part. Response: This is a problem that you can't guide on the multi-slit, you're on the slit itself only the long slits. Answer: (Sandy) The real difference is that the TV guiders are much more rigid here and the provision of the reticle will keep you constantly oriented properly in respect to the real focal plane. This field of view, if you look directly at the slit will probably be 1' by 1'. (David) There is still lots of design to be done at this point. Question: Are you keeping the reticle illuminated at all times? Answer: (Sandy) No it will go on and off. (David) I have given a talk about the electromechani- cal and managed to say very little about the electrical. I'll try to compensate for that right now. Question: What is the total thermal range that you anticipate having to deal with, and what are the average nightly variations? Answer: The average nightly variations are a few degrees C. A check year to year is going to show you 15 degree C nighttime variation (-5 to +5). We're talking nighttime where it is nearly zero degrees. Response: So thermal issues are going to have an affect on your long term alignments and stabil- ity over a seasonal period? Question: What is your goal in mechanical stiffness of the entire instrument before you turn on your flexure control? Have you thought about what you think you can achieve in terms of stabil- ity at the detector before you turn on the flexure control? Answer: What we are designing for is a collimator tent mirror system that will have less than 8" of combined tilt. If we can stay below that, we know that the flexure compensation system is going to be minimally taxed. Response: It's more than 2 pixels for that source alone. Answer: If we move up into the sort of early arcmin area then we know that we are getting out of the range of our flexure compensation system. Garth is going to talk more about flexure compen- sation later today. Not only do you chace image motion, you start to chase aberrations at certain points. Response: It will be an important thing to develope a separate goal on what you expect the instru- ment to do that is separate from the flexure control. Answer: Our philosophy is everything that we can build into this instrument to make it stiff and not move is something then that we don't have to compensate for later. Question: Do you presumably intend to pick-off of these TV guide cameras at both the altitude azimuth and the de-rotator error signals? Answer: (Sandy) Yes, I think with two guide stars there is no question we will be able to guide really well. Response: The de-rotator is closed loop, it's locked. Answer: (Sandy) It will be closed loop, the question is what do we do with one guide star (the computations suggest that there will only be one guide star about 10% of the time) in the star fields. Response: That computation was very hard to follow because there was no magnitude limit and no exposure time. You are grossly oversampling the guider focal plane, it seems to me that you should demagnify more and get more sky. Answer: (Sandy) We can't really demagnify because we are limited by the lens that we use. Response: But you don't have any field optics so of course you're limited by the lens. You need to put on some field optics to get the light into the lens. Answer: (Sandy) We're really just interested in the square inch of the silica. Response: Nikon makes a 135 mm F-2 lens with quite a big field, this would give you twice as much area on the sky for each one. Question: You didn't say how faint you could guide, how long the exposures were so it is very difficult to figure out what was going on. Answer: (Harland) We really have not developed much in the way of our thinking in the TV guiding. Sandy came to me and said should we have the same sort of thing that HIRES did or can we do a custom lens. If so, what would it cost? I told her that it looked like it would be very expensive, however I did not look very carefully about the field lens situation. I think that we can put a field lens in, and there is an array of commercial lens available, I think that right now we will certainly not have to build a custom lens because when we get down to the point of calculating the magnitudes of the stars, we will have an array of lenses that we can buy to do the job. I think that mostly you're seeing the fact that we just haven't done the job yet. (David) The actual mechani- cal design of the TV system is being put in really late in our development schedule, because it is something that has been done before. There are a few more challenging things out there for us. Response: The other possibility that I thought about when I was looking at this is to run the guid- ers more or less straight through. In other words, not use a pick-off mirror but just use a window in the focal plane, run the guiders in, that would allow you to tie the guiders much better mechan- ically to the instrument. Answer: (Jack) You cannot slit view with that system. (Garth) The tent mirror is a big problem because the tent mirror sits bodily in the central part of the field. Question: Are we going to try for common guiders again for all instruments? Answer: (Garth) It's not clear what the other instruments are doing for guiding. Of course tech- nology is advancing, we plan to use available commercial technology/systems if they are there but they will be very different from the one's that are already in the system. Question: Why don't you use the same electronics for your guide TV's that you're using for you main CCDs, (i.e. Leach controller)? Answer: (Garth) That's rather an expensive way to do it. Response: It's certainly not more expensive than photometrics. Answer: (Joe Miller) We're also building a whole new generation of TV guide cameras for Lick. Response: What is the typical cost? Answer: (David) We don't know yet, we have a good couple of years before we really have to settle down and purchase these items. Question: That slit mask handling that you described is a very complex issue, have you explored other places that you could possibly put slits, other than that position? For instance around the side if you allowed your envelope to grow. I don't know how much clearance you've got to the floor. The thing that seems particularly difficult is that you have to flip them around and orient them into the focal plane. If you had some way inserting them in a plainer way it might simplify things. Answer: (Sandy) Maybe we can move the instrument back a little. Response: That would certainly give you some room. Question: How much freedom is there in the Keck to move the focal plane? Answer: (Sandy) We can move a few inches, but not far enough to solve this problem. (Garth) The problem is that you have the elevation. Response: Is that where you have the camera? Is that all filled with a bearing? What about where you're planning to insert the masks? If you had your cassettes arranged sideways you may be able to have a more direct insertion path than you have at present. Answer: We will have to explore this. Our plan again is to actually build a prototype by this time next year, to be able to take it through it's paces to get "mileage" on it. Question: Speaking about moving the focal plane further to get more clearance, there was also some talk about having an ADC for the field. Maybe you can combine the ADC to also correct the aberrations you could get because you have the secondary mirror other than in blue. There- fore move the focal plane the long way. Answer: (Joe Miller) One of the advantages of the segmented mirror is that if you move the focal plane back you can refigure the primary mirror of the telescope and take out some of that spheri- cal by changing the tilts of the section. Detectors - Richard Stover Question: What accuracy are you trying to get in terms of the entire array? Answer: We want 5 microns, 10 microns peak to peak over the whole array. Question: It would strike me that there's a way to achieve that without all of these feed throughs it would sure be a lot easier. Answer: Yes it sure would be. Response: Since all the elements are going to be at basically the same temperature, have you looked at the possibility of setting the thing up at room temperature in a controlled environment and then take the entire assembly cold? Answer: Not yet, we don't have an assembly like this to test. Response: That way you can see if you really need these external adjustments for the cold case. Answer: That maybe something that they can address in the model, I'm not sure. Question: What degrees of freedom to those little rods give you? Answer: There's three of them per CCD, we can do tip, tilt and piston to get them all lined up. Response: The rods in principal go up and down? Answer: That's right. Response: They don't turn or tilt. Answer: That's right. Response: Can you get pixels and columns lined up? Answer: There is no requirement to get columns absolutely lined up. There is a rotational align- ment which we can do sufficiently well. Question: In terms of the vacuum pack that is to surround this thing, I'm a little unclear as to how you're going to configure the vacuum tank so that the back end of those feed thrus is out of the back. The top of the chip of course is the window, so are you going to have this vacuum tank out the side of the dewar? Answer: It can go out the back as well. I'm not exactly sure what you mean, go out the side, there is the liquid nitrogen, but that doesn't have to be attached to the back of the CCD it can go basically anywhere that we can conveniently put it. These will almost certainly be on the back side of the dewar. Response: Where is the liquid nitrogen and how do you get cold? Answer: Again where we put the liquid nitrogen is not a terribly big problem. Response: What's the dewar look like around this thing? I don't see the front and back, I don't actually see where the dewar is now. Answer: We don't have a drawing for the dewar yet. Response: Is this something you can get at with your hands. Answer: Yes, to make the adjustments. Question: Is the bottom plate at room temperature or at liquid nitrogen temperature? Answer: At room temperature. Response: So then liquid nitrogen doesn't come in contact with the columns? Answer: These are thin wall stainless steel. Response: I don't see why you have to do this at all. You should be able to build something that you bolt together and you're done. Before resorting to this for specs at the 1/2 mil level you should be able to make pieces of metal with no adjustment that you bolt together and you're fin- ished. Answer: (David) That may be true, we haven't done the engineering, where we are at is sort of a preliminary level and one of the preliminary questions is how would you ever adjust it. Response: The answer is you don't. Response: I second that, I think this really needs a hard look before you go to this much compli- cation. Answer: (David) I agree, we have many hours of engineering to do. (Richard) This is basically a scaled up version of the thinking that went into what was going to happen on HIRES. Response: At the very beginning you don't need to worry alot about CCD packages, you're going to have some freedom in this. If you design them mechanically so that you can get them in vari- ous ways, and that they are rigid enough so that you don't have to worry about effects of silica etc., then you really should just be able to just bolt the array down to an invar/covar plate. Or at the very least be able to shim it. Response: My suspicion is that you won't have any trouble at that level of flatness factor. Question: What defined the 1mm gap? Answer: We can make that gap a little smaller, however we can't make it 100 microns in the other direction. There are no structures on the top of the CCD, but there are along its sides. (Garth) What do you think might be practical? Response: The top set's hard, but can be done, 1mm is easy. Total 600, (300 per side) is about enough to do it. Especially because this device does not need frame transfer alterations, that makes the bussing a little easier. Response: You have got to keep the slit masks rigid in trapezoid too. You're not giving up very much compared to what you have already. Answer: (Sandy) I'm quite happy with what we have. I'm trying to understand the motivation for making it half as large. Response: I don't think there is one. Question: Wouldn't you be better off targeting -90 or -100 vs. -120 for the red response? Answer: I don't know, we haven't looked at what response is as a function of temperature, I'm sure there is some effect there. That's one of the things we'll be looking at. Question: Is the thinning done mechanically? Answer: No, it's chemical, there is an initial stage mechanical thinning and the final is done chemically. Question: How thick is the boron layer? Answer: Well we have two boxes of these new wafers made by Lawrence Semiconductor, one with 4 micron and one with 2 micron thick boron layers. Response: Is that thin enough so that you can get a big enough gradient? Answer: No, you have to thin that down until you get to the peak of the distribution, so you have a good gradient going into the device. That's why we have two different thicknesses because we don't know exactly how that spreading is going to happen. Question: Does the boron layer make fringing better or worse? Answer: It won't have any effect on the fringing, but we are doing other things that may address that. Question: What about crystals? Answer: There are no crystals. Response: You just have to grow the normal oxide. Answer: Yes, I have high hopes that this is going to work out. We don't know in fact these dou- ble epi layers are good enough to make CCDs. CCDs put the most possible constraints on the device of anything in the world. Question: Using high resistivity CCDs that are thick how does it look? In the pixel size where do the electrons go, what happens to the MTF's? Answer: The reason you want high resistivity is because you do get the fields extending down into the silica. The claim is that the velocity of the electrons going forward towards the front of the device (where your well is), is much higher than the transverse philosophy. While you do get some loss of MTF, it's not going to be critically bad. Question: About fringing, the photons that you are measuring are not in the regime where they get absorbed by the metallization layer, you're in the regime where you're losing them because your seeing right through, so why bother with thinned devices. It seems to me that if you're going to push on something, you should push on the completion devices. Response: If you build completion devices that aren't thin you get creamed with cosmic rays. Response: With the device Lincoln is making for the x-ray community now they can't make at these deep depletions, so they make bulk deep depletion wafers. If they don't thin them the cos- mic rays look like massive galaxies because you get this minimum ionizing radiation traffic that's going all the way through your wafer. Answer: (Sandy) Are you indicating an intermediate approach? Response: I'm wondering if there is a compelling reason to go for classic thin CCDs? That's my fundamental question, there are numerous alternatives, the one that suggests itself from what Richard said was to push on the deep depletion side, not on the thinning side. Answer: Actually, we're going to be doing yet another kind of CCD using a technique called epi replacement that they've been working on up at Davis, which will make an intermediate thickness device. They can bond a bulk silicon high resistivity wafer to a standard wafer just by contact, no adhesives and then thin down the high resistivity side to 50 microns and then build the CCD on top of that. It's like making an epi layer, you can't successfully grow epi layers much thicker than what we are already using, 50 microns is pushing the technology right now. Response: Part of my point is that you are trying to step back and look at the broad view. If deal- ing with this fringing is such an awkward issue regarding the flat fielding it isn't worth it. Response: But if the cosmic rays in long exposures are going to be a serious contaminate, that's a problem. Answer: (Garth) Fringing is a concern in the red. There is a large part of the spectral region that we will be working in, where through-put efficiency is important to us. It is not clear that we haven't explored all the ways to minimize fringing, even in devices. (Sandy) I'm still not quite sure on exactly what your program runs. You don't want a back illuminated device. Response: I'm just drawing to your attention the fact that there are trade-offs. Answer: (Garth) It's real hard to step away from that. I think we worked real hard to find ways of dealing with the fringing. Response: You've got a lot of that back with these deep depletion parts though. You've got to get the light through the poly, even if it's long wavelength. Response by Sandy to talk about National Laboratories vs. private: Orbit has expressed con- cern over our involvement with Lincoln Labs to the point where they will not sully their hands anymore with people who exploit National Laboratories in competition with industry. Response: It was one individual, not company wide. Answer: (Sandy) You're right, but it was a very important individual. Response: There are two viewpoints on the whole issue. One is that the companies pay taxes and those tax dollars are used to support national facilities like Lincoln Labs, who then turn around and put them out of business because they can undercut them in prices or develope technology that they just don't have access to. On the other hand our tax dollars pay for the same facilities and we're forbidden to go to them for technology that only they right now have. Question: One company that really delivers big CCDs now is SITe, you said you eliminated them because they could not meet your specifications on curvature. But I wonder if that was wise in the sense that you might be able to design a camera with a curvature that was acceptable from the point of view of SITe chips? Response: They're curved, but the radi are within a fairly tightly constrained limit and you could just fit those to the curve of the focal plane. So I don't think you should rule out SITe at this point. Working with SITe is not smooth sailing, but on the other hand they do build CCDs that work and deliver them packaged. Question: Have you ever asked them how much it would cost to make flat ones? Response: I don't think they know how. The question is motivated by the idea that there is some- thing you could get maybe you could figure out how to use it. Answer: (Garth) What do you mean gets? Do you mean 2048's? Response: Those are much larger CCDs that you're talking about. Answer: (Garth) We waited two years to get one for HIRES from SITe. Yes, if they started delivering 2048 by 4096's and it looked like they were doing that on a fairly reasonable time scale at a reasonable price, sure we would do business with them. Response: Design a camera with a curved focal plane. The amount of curvature you have to fig- ure in is relatively small compared to what that field flattener is already doing. Answer: (Garth) As long at it's well defined. That's one of the areas that we're not sure about. Response: I don't know what the answer would be in terms of what they could do. Response: The variance on the curvature is about 10% and the radi of curvature are approxi- mately 2 meters. Response: If they can only afford $50K for detector I'm afraid SITe may simply be unacceptable. Question: Regarding the sawing of the collimator into two pieces, would the coarsest adjustment be on the slit mask itself? Answer: Then you have to guide the telescope too. Response: You have to guide the telescope period. Answer: (Harland) That doesn't move the pupil though, you have to move the pupil to have any track. Response: If I move the slit, the spectrum will move I promise. Answer: (Harland) But the pupil is important. Response: If you design the spectrograph right the pupil is not going to move due to flexure enough to matter, so it doesn't matter if you move the collimator. Answer: (Harland) You haven't calculated that quantitatively, you may turn out to be right, but I can't calculate in my head, I need to use a computer and find out how much things move quantita- tively. The point is well made that we should look at it, but I don't think we can adjudicate it here because we need some quantitative numbers as to how things really do move. Question: My understanding is that this spectrograph will cost $4,000,000, you could get a new secondary mirror for the telescope for a tenth of that. (For about 10% of the cost of this one instrument you can have a new secondary mirror for the Nasmyth focus.) Then you can move your focus out, (you're so crowded against this main telescope bearing you can't have anything forward of the focus). This means you have your axis of de-rotation and vertical. If there is no change in gravity vector whatsoever, then you also have one more reflection, but you can proba- bly do without the tent mirror. So instead of having a high angle tent mirror in it, you just have two 45 mirrors, one stands half the field up, one stands half the field down. You have loads of space here in the telescope and then you have your vertical de-rotation. You could probably save enough money on the instrument to maybe make up for the cost of having a second secondary mirror for the Nasmyth. Answer: (Joe Miller) I did propose a vertical spectrograph at one point, but the question is, are all the clearances there that you think are there? (Garth) Can we go down below the Nasmyth platform? Response: It looks like there's lots of space. Response: You could find a hole big enough for the spectrograph. Response: I would love to get rid of that tent mirror. If you're splitting it and sending half up and half down the dividing line is going to rotate. Answer: This flexure control system doesn't look very expensive. I know it will work, I don't think that will be a problem. Detector Electronics - Richard Stover Question: You are planning on using both outputs for CCDs, which would be 16 amplifiers/cali- brations. What if you changed your design strategy to one amplifier per CCD and just read it out faster? Answer: That's certainly possible, but we don't know if we can get it. (If we can maintain CTE and that sort of thing.) Response: Play speed is not going to be the issue, it's pixel rate I think. Answer: It all relies on whether you can get the CTE, (especially CTE in serials) that's really important. Response: Not at a few hundred kilograms. Answer: If we can reduce the integration with these new generation of high beam low noise amplifiers. Response: Your system will allow you to do that because with the new Leach Controller you can go to 2 micro seconds per AD. Answer: The fastest readout rate comes from shrinking everything down as far as you can. We can easily do 200 kilopixels out of 1 amplifier instead of 100 out of two. Response: If you plan that now, you reduce the number of electronics you'll need. Answer: We're not ordering any of this stuff right now. The second generation Leach board does not even exist now. I think we'll have a better handle on whether we can in fact run fast with one amplifier and keep the noise down. Response: The issue is more from a users standpoint, a calibration standpoint and a cross talk standpoint. It might be nice to have half as many readouts. Response: I will never again build an instrument that readouts multiple amplifiers from a single chip. Answer: This is a whole point that has to be truly investigated. (Sandy) What difference does a single chip mean vs. multiple chips? Response: It's just more chips except for the cross talk issue. Response: Well it's two issues, one is cross talk and the other is that the analog stages are not free. The cross talk issue is really non trivial. I have yet to see a multi amplifier single chip sys- tem that does not have cross talk at some level. Your specifications says that you want less than an electron of cross talk and I don't think that's going to happen with amplifiers that share a chip. Because when a saturated pixel goes out this side the reset drain structure that underlies the whole thing knows about it. If you are adamant about using multiple amplifiers, I strongly suggest you consider a low resolution 8 bit signal cap that has the full well of dynamic range. A parallel low resolution 8 bit digitization signal path would work, if you are forced to use multiple outputs to get the through-put that you want. Answer: If this arises because of the way a CCD is made, maybe we should be thinking about making the CCDs differently. Response: I think it's something you're going to have to deal with the question is, how hard it is to do. The fundamental thing you can do is not make the serial register continuous, split it in the middle. The reset drains are then separate, they do not connect to each other. They are sitting on the same piece of silica, but do not connect to each other. Response: My understanding is that reset drain structure underlies the whole array and in fact sets the potential of the spectra that you get. Response: It may cost you a pixel or two to do this correctly, you have to do it through the whole device. Essentially you're printing two devices that look like one but actually they have a division of some microns. One thing you have to think about is how often and with what kind of spectra is that important. Software - Steve Allen Question: Is an astronomer at any campus going to have a hardware configuration to come back from a run using DEIMOS and be able to deal with his data? Or is he going to have to utilize some hardware at some place like Santa Cruz to be able to process his data? Answer: This will not be a problem, I have a feeling that you'll find many people will have PC's which can handle the job by that time. It may be slower on a system you have at home rather than the mountain top, but it will work. (Sandy) I think you should be able to go home with spectrum. Response: At the worst you'll go home with 60 images from an exposure. Question: Are you proposing to control the rotator using a Galil? Formerly we have had a lot of trouble with them. We eventually abandoned that approach altogether because Galil didn't give us the kind of control we needed. You mentioned that you regarded the module from an instru- ment rotation as part of the instrument itself. Maybe it should be looked at like all the other instrument rotators on the telescope, instead of using this system. It could use the standard one we're planning to use for the Cass instruments. Answer: (Dave Cowley) From my point of view we would like to use the Galil system because it can be tested more easily here in Santa Cruz. I've used the larger Galil motors for lifting very heavy items and had no trouble at all. We're planning on turning the model sometime early 1995 to determine whether it is something we can indeed do. Response: Where I'm coming from is if we use the same system for DEIMOS as the other instru- ments, then we simply have the same control system everywhere for all rotations. Now we have to figure out how we're going to get the rotational information to you and how are we going to handle the feedback for the guiders, it becomes suddenly a much more complicated issue. What is the benefit of doing it this way at all? Response: The instrument it guiding the telescope rather than the telescope driving the instru- ment blind. Response: Instrument rotations have nothing to do with the telescope. Answer: (Garth) Well it does in this case because we do have two guiders and the intent is to close the loop on the rotation from the guider signals which is something that we don't currently do with LRIS. Response: It falls on axis here, correct? So if you rotate, you need to re-point the telescope, you do things like picking up initial stars in the offset guiders. Does DCS have to know about that? Response: In the review materials you mention there were concerns about using friction couple encoders for the rotator. You might resort to using an optical alignment system similar to what we're using on the telescope, a precision reference mark system, but they don't work particularly well it's a very difficult job to make it work properly. Conceptually it's very simple, but not in practice. Answer: (David) We use optical solutions all the time and get a very high precision out of them. Again, we are going to physically try it out early next year. Sometime in February/March of 1995 we plan to start rotating the model. There is a lot of things we want to do, flexure being one of them. (Sandy) We have to be able to rotate this instrument in our shops. Response: I was just wondering what the degree of precision was. Answer: (Sandy) 8". (David) My overall plan is when it gets delivered to Hawaii, it works and we've tested it. The fewer things that are new and different when we get to Hawaii the better it's going to be. So if I run everything and check it all out here, then really all I have to do is reconfigure it in Hawaii. (Sandy) Well I don't know, we have to come to grips with this interface problem. Because the TV has got to talk to the positioner. The TV also has to talk to the tele- scope. Response: I think it depends on where we're going to be doing this guiding and rotation correc- tion. Answer: (Sandy) Why don't precision reference marks work? Response: They do work, some of them, the capability is quite good, it has something to do with thermal grading. If you're sitting on the Nasmyth platform you're in a much more benign envi- ronment than the telescope which is raised to the night sky. Answer: (Sandy) I think we would very much appreciate feedback on how to get accurate open loop tracking on this position angle because our estimates suggest that we will lack two guide stars about 10% of the time in sparse fields. Slit Mask Laser Cutter - Bill Shepard Question: How long does the loading process take to load the slits into the slit mask changer? Answer: 20 minutes. Question: Where is this cutter going? Answer: The slit mask cutter is going to be in a separate room. The actual location is TBD. There are advantages to both Mauna Kea and Waimea. One will be the ability to make last minute changes, which could become a disadvantage because then that invites poor planning. There is also the issue of serviceability. Question: Is it anticipated that slit mask blank making would be something the astronomer would do or would this be done by a member of the observatory staff. Answer: (Garth) We are anticipating that the observatory will interface with the astronomer, but provide an experienced operator with the system. Question: Why are you placing a .02" accuracy on your slit positions? Answer: We asked them for this specification and they said fine. Question: Do you have any numbers on the material removal rate? Answer: What we have is a linear estimate on a first pass on the lowest power type of system. We have someone meeting a 35 minute per mask goal with stainless steel. That's also allowing for 72 slitlets. This figure needs to be looked as some more. Question: Could you repeat what the tolerance was for matching the focal surface. Answer: 15 mils. (Sandy) It's probably tighter than it needs to be, it's quite tight. Question: If you use a cone and section it to the surface you can actually do your machining on a flat surface, then later wrap the cut thing onto the frame. Answer: (Sandy) You could be causing trouble for yourself. Question: Are these cutters all just 2 and 3 axis machines or are they 5? Answer: The 5 axis system would be more complicated than what we need, because of the spher- ical shape. One of the cutter vendors we are looking at is talking about just pivoting the output mechanism for the beam, they would solve the problem between the pivoting, spherical surface and long beam weight. Question: Is stainless better than other metals? Answer: The thing that's convenient about stainless is that you can cut it to a very smooth edge with the laser and also because of how thin it is it allows for some of the lower powered laser sys- tems. (Sandy) We want to cut these things when they're installed in the mask, we don't want to lay them on a flat and then install them. Response: If you made a cone, you could make the whole thing big and then wrap it. Answer: (Sandy) I understand, but we want to locate these slitlets extremely accurately with respect to the same tool involved that we're going to use to work with the mask. Question: What about the cost of the lamps? Answer: The lamps typically on the medium size systems (50-100 watt output beams), are approximately $100-$200 each. Question: How do you go about making a rectangular cut? Answer: These are vector scan systems. Typically the vector scanning is done for the precision we need it is done by an x, y table. You can cut to any shape you want. You can program what- ever the graphical interface will generate. (Sandy) You spoke about training periods, from what I understand the typical training period to get an employee up to speed to operate the machine and do the regular servicing is approximately 1 week. Question: By pushing this to the thin stainless are you trying to get it to conform to the focal sur- face of the telescope? Answer: (Sandy) I would say that the mask material is very much open, and my preference is to go with plastic material which would allow us to keep a single mask in one piece. Response: Why can't you just use aluminum and cut it with a milling machine? Answer: A milling machine would be a lot slower. Response: I don't think so at all, and it gives you a fall back for using any machine shop. People can make their own and bring them, you can do it there, you don't necessarily have to buy your own. Answer: The driving issue is that we need to be ready to go after producing the masks. Schedule - Dave Cowley Question: What is the function of the Critical Design Review (CDR)? Answer: This review will cover most of the remainder of the project. Question: Are you planning on buying the slit mask machine before the CDR? Answer: Yes. Response: That's a large dollar investment. Answer: We want to get some mileage on the machine before it ever goes to Hawaii. Response: What is the time between CDR and shipment to Hawaii? Answer: Three years. Response: That sounds like enough time to test the slit mask machine. Question: You show that you're deciding on the detectors in March of 1996 and beginning to test in July of 1996, doesn't that sound optimistic? Answer: Richard tells me that when we have one detector we can then start testing immediately. Budget - Dave Cowley Question: This is for one beam? Answer: Yes, it is for one beam, we need an extra $1.5 million for the second beam. Question: How much overhead is built in to this budget. These numbers don't come out any- where what they would be in a real-world situation. Answer: (Sandy) We do not charge overhead on Keck projects. To calculate this we estimated the number of hours and took the senior person from each of the shops salary range and multiplied by certain factors (i.e. that people don't work 100% of the time, etc.) and came up with our num- bers. Question: Do you have a process for tracking progress against expenditures? Answer: My approach is to ask what has to be done, not what has been done.