From ajw@rana.ligo.caltech.edu Tue Jul 25 11:49:39 2000 Date: Tue, 25 Jul 2000 11:49:34 -0700 (PDT) From: Alan Weinstein To: kstrain@physics.gla.ac.uk, coyne@ligo.caltech.edu, stan@ligo.caltech.edu, kells@ligo.caltech.edu, jordan@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, ahelena@ligo.caltech.edu CC: ajw@ligo.caltech.edu, sanders@ligo.caltech.edu Subject: 40 meter upgrade core optics Hi all, Garilynn has been waiting (since May!) to know the 40 meter upgrade core optics sizes, radii of curvature, optical quality (Corning or Heraeus SV), and coatings, to start ordering. I need your expert advice to help make some of these decisions (hopefully, BEFORE the 40m advisory committee meeting on August 16). Can you please look at http://www.ligo.caltech.edu/~ajw/40m_testmass.html and tell me what you think? The biggest question is whether we need bigger (4") or smaller (3") test masses. But maybe, none of these are really important questions; they just need to be answered one way or another. I very much appreciate your advice! Alan Weinstein _______________________________________________________________________ From dhs@ligo.mit.edu Tue Jul 25 15:20:47 2000 Date: Tue, 25 Jul 2000 15:20:22 -0700 To: ajw@caltech.edu From: David Shoemaker Subject: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE Alan, I read in the weekly.... >AJW has done some more careful noise studies to determine whether we >should use old-40m-style 4"diameter 3.5"thick core optics or LIGO SOS >3"diameter 1"thick. It appears that the thermal noise in the mass itself, >which scales very weakly with mass, dominates over suspension noise (which >does depend on mass of optic) everywhere except for a few violin-mode >spikes. Advice on how to make this decision would be appreciated! Some thoughts on the size of optics: - I don't think that the radiation pressure or internal thermal or suspension thermal noise will be an important factor in the success of your experiment, so don't think this should be a consideration. - I think that taking advantage of the experience we have with the SOS and the existance of a much-better-understood-than-before suspension design for the 3" optics makes a very strong argument - we will probably use the 3" SOS-like suspension in LIGO II (for at least steering and probably mode matching), so this gives us a chance to take advantage of the parallel effort to reduce work and use the prototype applications of an incremental engineering improvement as a test for LIGO II - the optics will take less time to make; given that any estimate for delivery etc. must be multiplied by some terrible 'experience factor' greater than one, this is a strong argument - in favor of larger optics (for the 40m and LIGO applications) is the difficulty in getting a sensible local control matrix on a small optic. This might be better, probably by 4/3, in a larger optic -- not a huge margin. Bottom line, I think the 3" optic is probably a good choice. I would also use a few such suspensions in LASTI (as steering/matching), so I am also saying that I would be happy to see that size used there. then: on storage times, you may well want longer storage times than for LIGO to bring the tuning frequency down to audio frequencies. Otherwise, the signal recycling minimum would be ~200 hz * 100 = 20kHz, and it is not so easy to make the laser frequency servo etc. work up there. Maybe increase the storage time by a factor of 10 or so? David _______________________________________________________________________ From ajw@hep204.cithep.caltech.edu Tue Jul 25 16:21:39 2000 To: dhs@ligo.mit.edu In-reply-to: <4.3.2.7.2.20000725151121.02a7ffc8@127.0.0.1> (message from David Shoemaker on Tue, 25 Jul 2000 15:20:22 -0700) Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE David, What a guy! I ask for advice, and you give me some! (You're the only one). There's a bit more information in http://www.ligo.caltech.edu/~ajw/40m_testmass.html Please check it out and tell me if you have any other thoughts... > - I don't think that the radiation pressure or internal thermal or > suspension thermal noise will be an important factor in the success of your > experiment, so don't think this should be a consideration. OK! > - I think that taking advantage of the experience we have with the SOS and > the existance of a much-better-understood-than-before suspension design for > the 3" optics makes a very strong argument OK! > - we will probably use the 3" SOS-like suspension in LIGO II (for at least > steering and probably mode matching), so this gives us a chance to take > advantage of the parallel effort to reduce work and use the prototype > applications of an incremental engineering improvement as a test for LIGO II > > - the optics will take less time to make; given that any estimate for > delivery etc. must be multiplied by some terrible 'experience factor' > greater than one, this is a strong argument OK! > - in favor of larger optics (for the 40m and LIGO applications) is the > difficulty in getting a sensible local control matrix on a small optic. > This might be better, probably by 4/3, in a larger optic -- not a huge margin. Is that so? It goes like diameter, not mass? Is it the position/pitch coupling that's the problem? (Is that what's meant by "diagonalizing a suspension"?) Is it a surmountable problem? > Bottom line, I think the 3" optic is probably a good choice. I would also > use a few such suspensions in LASTI (as steering/matching), so I am also > saying that I would be happy to see that size used there. OK! > then: on storage times, you may well want longer storage times than for > LIGO to bring the tuning frequency down to audio frequencies. Otherwise, > the signal recycling minimum would be ~200 hz * 100 = 20kHz, and it is not > so easy to make the laser frequency servo etc. work up there. Maybe > increase the storage time by a factor of 10 or so? I am thinking of operating the 40m in the Signal Recycling regime (negative tune, smaller bandwidth) not the RSE regime (positive tune, larger bandwidth). This allows us to put the dip in the sensitivity anywhere we want it, eg, 500 Hz (or the 950 Hz you see in the figures in http://www.ligo.caltech.edu/~ajw/40m_testmass.html ). I believe that the control problem for SR is *identical* to that for RSE, so this is a meaningful prototype test. Comments? Thanks! AJW _______________________________________________________________________ From kells@acrux.ligo.caltech.edu Tue Jul 25 18:03:45 2000 Date: Tue, 25 Jul 2000 18:17:15 -0700 (PDT) From: Bill Kells To: kstrain@physics.gla.ac.uk, coyne@ligo.caltech.edu, stan@ligo.caltech.edu, kells@ligo.caltech.edu, jordan@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, ahelena@ligo.caltech.edu, ajw@rana.ligo.caltech.edu Subject: Re: 40 meter upgrade core optics Hi Alan, Armed with the parameters shown on the web page you point to (principly the recycling gain and input power), I have considered the optics design in more detail: I will refer to the flat ITM option (just because its simpler to discuss: no decisive reason to choose it). g=1/3 as usual. Another key parameter is the RC length, which I took to be 2.4m (close to previous). With this length it is not so accurate to describe the necessary (ie matched) RC as degenerate. That is ,with a flat ITM, the exactly matched RC has an RM ROC of 331m which can be (with some care in polishing: approaching LIGO I criteria) well specified (as distinguished from R->infinity). Starting at this "cold" configuration the thermal lens distortion due to 1 watt at a gain of 89 would be about twice that of LIGO I (design parameters) for substrates the same as specified for LIGO I (ie 10 cm thick and 5 ppm/cm absorption). This has two consequences: 1. If the RM were to be polished to match the arm cavity at "cold" conditions (ROC=331 m), then the RC would begin to be seriously mismatched to the arms at full power. Of course this can be cured by having the RM polished to match the lensing (as in LIGO I). This uncertainty in polish can be removed by ~choosing SV Silica (note that "Corning" Silica, at > 10ppm/cm is out, at least for the thick TM choice); or by ~ choosing the thinner TMs. 2. With the high Recycling gain chosen, the properly matched RC is reasonably stable (the Gouy phase advance for the mismatch mode is an order of magnitude larger that the resonance half width). This means that the SB light will be pretty much immune to the degeneracy effect that is of concern in LIGO I (of course it will still be affected as per the stable cavity matching of the last paragraph). So it seems that for the parameters you have chosen, the RC stability situation is much more robust (than LIGO I or even the 40m recycliing experiment was). On the other hand it will no longer be a good "model" of LIGO in a very essential way (the characteristic degeneracy of the RC). To fully realize this robustness, the precise 331m ROC for the RM is essential (note that for the 40m recycling exp. we approximated the RM to be flat ! A mistake I believe). This seems like a piece of cake (331 being quite curved wrt the LIGO ROC ). However, more precisely it means that the effective, local ROC must be 331m over all beam size surface patches (at least that the beam will possibly see). That is the local lumpyness much be quite small. Some quick figures tell me the residual surface distortion, on ~few mm scale (when the net mean ROC is subtracted) must approach LIGO I quality. ciao, Bill _______________________________________________________________________ From k.strain@physics.gla.ac.uk Wed Jul 26 01:41:20 2000 Date: Wed, 26 Jul 2000 09:41:11 +0100 (BST) From: "k.strain" To: Alan Weinstein cc: k.strain@physics.gla.ac.uk, coyne@ligo.caltech.edu, stan@ligo.caltech.edu, kells@ligo.caltech.edu, jordan@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, ahelena@ligo.caltech.edu, ajw@ligo.caltech.edu, sanders@ligo.caltech.edu Subject: Re: 40 meter upgrade core optics Dear Alan, Although there is not a hint of double pendulums on the suggested page ... you ask about 3" or 4" optics - we have made double pendulums with mirror masses from 0.8 kg up to 9 kg, and thickness/diameter from 0.25 to 0.8. Lighter masses tend to need very thin wires, quite hard to handle (we use either 2 loops of steel or 4 silica fibres) - so I'd suggest not going very much lighter than 0.8 kg. Also, much thinner mirrors can be tricky too. If the mirror is less than ~ 20mm thick it gets tricky to design some aspects of the suspension that scale proportional to mirror thickness. This gets harder still if the diameter is also very large (we need certain relationships between moments of inertia about different axes, for the various parts). If you want double pendulums the 3" by 1" looks like a bad choice. Clearly would not apply to any single pendulums which can be almost any shape and size. Cheers, Ken _______________________________________________________________________ From mike@ligo.mit.edu Wed Jul 26 06:49:42 2000 Date: Wed, 26 Jul 2000 09:41:47 -0400 (EDT) From: Mike Zucker Reply-To: Mike Zucker Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE To: ajw@caltech.edu, dhs@ligo.mit.edu Cc: coyne@ligo.mit.edu, sanders@ligo.mit.edu, mike@ligo.mit.edu, kstrain@physics.gla.ac.uk, pf@ligo.mit.edu, gari@ligo.caltech.edu Alan- David forwarded me your request. Definitely not an expert on DR/RSE (yet) but in the old days we found putting the waist at the input coupler (i.e., flat) made alignment and mode matching somewhat more convenient. From the standpoint of the WFS it won't matter, but it helps understand what's going on when you're manually roughing it in (this one tilts the beam, that one translates...). However this makes your PRM and SRM "almost" flat so it may be inconvenient for other reasons, unless somehow they can be dead flat too. In any case, making at least some mirrors flat has a practical advantage in the sense flats are faster/easier to get, other things being equal; and making the input couplers flats helps if/when you decide down the road to swap in another input coupling T (a fairly likely scenario IMHO, see below). My VERY STRONG preference would be to go with 3" SOS-compatible optics unless there's a serious reason not to. The worst-case noise considerations you calculate could not, in my mind, compensate the huge advantages in optic & coating lead time, cost, new engineering, learning curve, and ease of integration. Chamber footprint is also going to be an issue (if it isn't already!). And even if thermal noise becomes a motivating concern, I would argue that getting the whole optical/electronic system running as early as possible is still the best way to find out what our REAL problems are (the ones that we haven't calculated). Incidentally I feel (and I think David would agree) that the problems David cites getting good drive diagonalization with the LI SOS arise >from relatively recent implementation problems; the 'pin-compatible' forerunner of the SOS used in the PNI worked very successfully. Finally I would agree with David that going for the same T and R will move the interesting DR/RSE frequencies beyond the accessible region due to technical noise in the laser. It just isn't feasible to try and get frequency and intensity noise at the required level above a few kHz or so (or if it is, would require great heroics with no subsequent need for it). In addition, exceeding the Nyquist rate of the control and DAQ digitizers, plus the feasible time-delay limited control bandwidths, will mean that all the interesting dynamics are in fact out of scope, and your system will essentially only deal with the "DC limit". To put it another way, reducing 2L/c by 100 does not automatically scale the other critical time constants. Similarly, matching the storage times to a 4km LIGO II (reducing 1-R by 100) is probably infeasible because mirror losses will render recycling moot. David's suggestion is a pretty good wag*; it seems reasonable to expect the important cavity dynamics features to still appear, in muted form, at 1/10 the storage time, without becoming so loss-dominated that the recycling factors are useless. But it is by no means clear whether the _detailed_ scaling leaves you with a 40m optical plant so different >from a 4km that you can't extrapolate adequately. It needs to be modeled!! Given that the modeling is coming slowly, I recommend you settle the curvatures & distances and buy surplus polished substrates for the input couplers, SRM and PRM. Hold off coating (most of) them. Specify your best-guess reflectances for one initial set, at the last possible time you can still receive them for integration. Mike *btw i think David meant to say 'finesse' where he wrote 'storage time' in the attached... _______________________________________________________________________ From k.strain@physics.gla.ac.uk Wed Jul 26 07:04:02 2000 To: Alan Weinstein cc: coyne@ligo.caltech.edu, stan@ligo.caltech.edu, kells@ligo.caltech.edu, Jordan Camp , janeen@ligo.caltech.edu, gari@ligo.caltech.edu, ahelena@ligo.caltech.edu, ajw@ligo.caltech.edu, Gary Sanders Subject: Re: 40 meter upgrade core optics Dear Alan, further to the note below, and having read Mike's note recommeding 3" optics .... If you want to make double pendulums at all it will be a delicate job with the 3" optics, as they are so light (the wires will be half the area of those used with a single loop, assuming that used in the SOS is optimised). This problem is, however, comparatively trivial. It can be solved, it is probably not very hard, certainly not very expensive to solve. Given the new design territory (and that Matt Husman went elsewhere) I would add a slight reservation to my assessment of the "ease" of designing and building a double pendulum for optics of that size. It would have to be done from scratch, but using all the design experience we have accumulated. I'd have to think a bit before saying more. Cheers, Ken _______________________________________________________________________ From dhs@ligo.mit.edu Wed Jul 26 07:47:41 2000 To: Alan Weinstein Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE At 16:21 00/07/25, you wrote: >David, > >What a guy! I ask for advice, and you give me some! >(You're the only one). at least this time it was asked for -- sometimes I am too generous... > > - in favor of larger optics (for the 40m and LIGO applications) is the > > difficulty in getting a sensible local control matrix on a small > optic. > > This might be better, probably by 4/3, in a larger optic -- not a > huge margin. > >Is that so? It goes like diameter, not mass? I think so; there is a big difference between the ease of use of the SOS and LOS in LIGO I, and the resonant frequencies are similar. It could be more a question of quality control in the assembly. >Is it the position/pitch coupling that's the problem? >(Is that what's meant by "diagonalizing a suspension"?) >Is it a surmountable problem? yes, yes, and yes; in particular, the digital suspension controller being developed now for LIGO I really looks like it will allow a lot of the mechanical imperfections to be taken out in software. so I have hope! > > then: on storage times, you may well want longer storage times than > for > > LIGO to bring the tuning frequency down to audio frequencies. > Otherwise, > > the signal recycling minimum would be ~200 hz * 100 = 20kHz, and it > is not > > so easy to make the laser frequency servo etc. work up there. Maybe > > increase the storage time by a factor of 10 or so? > >I am thinking of operating the 40m in the Signal Recycling regime >(negative tune, smaller bandwidth) >not the RSE regime (positive tune, larger bandwidth). >This allows us to put the dip in the sensitivity >anywhere we want it, eg, 500 Hz (or the 950 Hz you see in >the figures in http://www.ligo.caltech.edu/~ajw/40m_testmass.html ). >I believe that the control problem for SR is *identical* >to that for RSE, so this is a meaningful prototype test. I'll read and think. I guess there is not so much difference for LIGO II? the principal question always needs to be if the experiment tells us what we need to know about the scheme for LIGO II, and if we choose RSE for LIGO II it would have to be clear that the SR mode will tell us what we need to know. d. _______________________________________________________________________ From ajw@hep204.cithep.caltech.edu Wed Jul 26 09:51:52 2000 Date: Wed, 26 Jul 2000 09:51:51 -0700 (PDT) From: Alan Weinstein To: kells@ligo.caltech.edu CC: kstrain@physics.gla.ac.uk, coyne@ligo.caltech.edu, stan@ligo.caltech.edu, jordan@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, ahelena@ligo.caltech.edu, sanders@ligo.caltech.edu, ajw@hep.caltech.edu Hi Bill, Thanks much for your quick response! - Indeed, we intend to specify the appropriate ROC on all mirrors, including the RM and SM. - Your thermal delensing calculations are comforting: thinner, SV-quality mirrors with the right ROC and LIGO-quality polishing, can handle the heat load. I would be grateful to see your detailed calculations. - I did not appreciate that the 40m power recycling cavity is more robustly stable than at LIGO (presumably due to the smaller ratio of L_arm / L_prc). The 40m also has greater tolerance to angular misalignments; I think this is true for dual recycling as well. Yes, in this way, the 40m is not a good "model" of LIGO; (it's those darn short arms!) but I don't think that's an essential requirement for a good test of the control system... what do you think? Thanks again! AJW _______________________________________________________________________ From ajw@hep204.cithep.caltech.edu Wed Jul 26 10:01:41 2000 Date: Wed, 26 Jul 2000 10:01:39 -0700 (PDT) From: Alan Weinstein To: k.strain@physics.gla.ac.uk CC: coyne@ligo.caltech.edu, stan@ligo.caltech.edu, kells@ligo.caltech.edu, jordan@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, mike@ligo.mit.edu, dhs@ligo.mit.edu, ahelena@ligo.caltech.edu, sanders@ligo.caltech.edu, ajw@hep.caltech.edu Ken, thanks much for your quick response! Could you please bring up the subject of 40m double pendulums, at the advisory committee meeting? I get the sense from many folks at Caltech and MIT that between Glasgow and LASTI, multiple pendulums are being covered, and that an effort to get scaled-down versions into the 40m is ... redundant, distracting, unneccessary. No one is making the claim that "a meaningful test of the optical control plant for LIGO II requires multiple pendula, and a scaled-down double pendula provide a sufficient test." Can you make such an argument? I'd like to hear it! In any case, if we proceed through FY00-01 with the goal of: - building LIGO I SOS 3" suspensions for the 40m upgrade - reserve the right to decide, in the next year or 2, to replace these with scaled-down double pendula, then we'd certainly be willing to get new, larger, optics. Thanks! AJW _______________________________________________________________________ From ajw@hep204.cithep.caltech.edu Wed Jul 26 10:16:21 2000 To: mike@ligo.mit.edu CC: dhs@ligo.mit.edu, coyne@ligo.mit.edu, sanders@ligo.mit.edu, kstrain@physics.gla.ac.uk, pf@ligo.mit.edu, gari@ligo.caltech.edu, ajw@hep.caltech.edu Mike, > David forwarded me your request. Definitely not an expert on DR/RSE > (yet) but in the old days we found putting the waist at the input > coupler (i.e., flat) made alignment and mode matching somewhat more > convenient. From the standpoint of the WFS it won't matter, but > it helps understand what's going on when you're manually roughing it in > (this one tilts the beam, that one translates...). However this makes Do you think this is a significant worry, with better instrumentation? (Steve Vass agrees with you). > your PRM and SRM "almost" flat so it may be inconvenient for other > reasons, unless somehow they can be dead flat too. In any case, Bill Kells argues that the RM and SM should be polished to their appropriate curvatures, for cavity stability. > making at least some mirrors flat has a practical advantage in the > sense flats are faster/easier to get, other things being equal; > and making the input couplers flats helps if/when you decide down the > road to swap in another input coupling T (a fairly likely scenario > IMHO, see below). > > My VERY STRONG preference would be to go with 3" SOS-compatible optics > unless there's a serious reason not to. The worst-case noise > considerations you calculate could not, in my mind, compensate the huge > advantages in optic & coating lead time, cost, new engineering, > learning curve, and ease of integration. Chamber footprint is also > going to be an issue (if it isn't already!). And even if thermal noise > becomes a motivating concern, I would argue that getting the whole > optical/electronic system running as early as possible is still the > best way to find out what our REAL problems are (the ones that we > haven't calculated). Very good. Given no other important driving considerations, practical issues of lead time, cost, etc make fine deciding votes! > Incidentally I feel (and I think David would agree) that the problems > David cites getting good drive diagonalization with the LI SOS arise > from relatively recent implementation problems; the 'pin-compatible' > forerunner of the SOS used in the PNI worked very successfully. And the new digital SOS will fix it all... > Finally I would agree with David that going for the same T and R will > move the interesting DR/RSE frequencies beyond the accessible region > due to technical noise in the laser. It just isn't feasible to try and I am thinking of operating the 40m in the Signal Recycling regime (negative tune, smaller bandwidth) not the RSE regime (positive tune, larger bandwidth). This allows us to put the dip in the sensitivity anywhere we want it, eg, 500 Hz (or the 950 Hz you see in the figures in http://www.ligo.caltech.edu/~ajw/40m_testmass.html ), so laser noise above a few kHz is not of concern for us. I believe that the control problem for SR is *identical* to that for RSE, so this is a meaningful prototype test. I like the idea of 40m and LIGO having the same finesse, since I am under the impression that the optical control plant challenge "scales" with finesse (the higher the finesse of the cavities, the more difficult to acquire and keep lock). Is this right? Wrong? It's just a gut feeling. > Given that the modeling is coming slowly, I recommend you settle the > curvatures & distances and buy surplus polished substrates for the input > couplers, SRM and PRM. Hold off coating (most of) them. Specify your > best-guess reflectances for one initial set, at the last possible time > you can still receive them for integration. AGREED! But I like "strawman" models to point to. Thanks much! AJW _______________________________________________________________________ From Billingsley_G@ligo.caltech.edu Wed Jul 26 10:47:44 2000 Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE >Very good. Given no other important driving considerations, >practical issues of lead time, cost, etc make fine deciding votes! A clarification: There is no difference in delivery time between 3 or 4" optics, the quote is 4 months ARO for the SV material. Polishing is 2-3 months for something of this quality, and there is currently a long line at the door of the coating house (REO). All of these depend mostly on how busy the vendor is and when we happen to get into the queue. You will pay more for bigger optics, it scales as the weight of the optic, perhaps a 5K difference per piece of SV at these sizes. About 0.5K difference for the Corning Glass, which you could use everywhere but the ITMs. _______________________________________________________________________ From ajw@hep204.cithep.caltech.edu Wed Jul 26 12:06:17 2000 To: Billingsley_G@ligo.caltech.edu Garilynn, > A clarification: > There is no difference in delivery time between 3 or 4" optics, the quote$ > is 4 months ARO for the SV material. Polishing is 2-3 months for somethi$ > of this quality, and there is currently a long line at the door of the > coating house (REO). All of these depend mostly on how busy the vendor i$ > and when we happen to get into the queue. > You will pay more for bigger optics, it scales as the weight of the optic$ > perhaps a 5K difference per piece of SV at these sizes. About 0.5K > difference for the Corning Glass, which you could use everywhere but the $ Bill argues that SV glass would also be desireable for the RM (and SM) to minimize thermal delensing, maximizing PRC and SRC stability. Do you agree? Can you remind me of the numbers? I assume that the primary difference between SV and Corning glass (besides for cost) is the absorbtion per unit length. Numbers I vaguely remember... 10 ppm/inch for Corning, significantly less for SV. Do you know? Thanks! AJW _______________________________________________________________________ From Billingsley_G@ligo.caltech.edu Wed Jul 26 13:26:49 2000 To: Alan Weinstein Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE Bill is the one with the best feel for this, though intuitively it seems that there is not much beam intensity in the bulk of the RM and SRM, so most of the absorption would come from the coating. >Can you remind me of the numbers? >I assume that the primary difference between >SV and Corning glass (besides for cost) >is the absorbtion per unit length. >Numbers I vaguely remember... 10 ppm/inch for Corning, >significantly less for SV. Do you know? > >Thanks! AJW Corning ~13 ppm/cm Heraeus 311 & 312 ~ 3 ppm/cm Heraeus 311 SV ~ 0.5 ppm/cm These are typical numbers of the LIGO glass (Corning and Non-SV) as measured by me in Claude Boccara's lab The SV values are of VIRGO glass measured in the same lab, circa October 1997 _______________________________________________________________________ From: Bill Kells To: ajw@hep204.cithep.caltech.edu, Billingsley_G@ligo.caltech.edu Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE > Bill is the one with the best feel for this, though intuitively it seems > that there is not much beam intensity in the bulk of the RM and SRM, so > most of the absorption would come from the coating. Hi everyone, In my original message on the 40m upgrade optics, the conclusions about lensing and substrate choice (material/thickness) were intended to pertain only to the ITMs (and perhaps BS). As Gari points out, intuition correctly assures us that its irrelevant for the RM and SRM. This may not have been so clear though, since I mixed discussion of appropriate RM ROC with conclusions about [ITM] material. Bill _______________________________________________________________________ From k.strain@physics.gla.ac.uk Thu Jul 27 01:39:51 2000 To: Alan Weinstein Subject: Re: 40 meter upgrade core optics Dear Alan, it is just a question of at which point the sensing/control tests are split between LASTI and 40m. Sure LASTI will do the real tests of the mass controllers with real parameters, that is necessary. As I said before any length controller for a multiple pendulum (double, triple, ...) will look very different from that for a single pendulum. This is an area where any research done and experience gained will be valuable. You will be doing lock acquisition studies at the 40m, and alignment control studies too. These will both be affected by the single/multiple pendulum choice. I agree that it is possible to fake a multiple pendulum controller response using a single pendulum (if you don't care about noise - OK for the tests mentioned). So it does not really matter whether you know about split feedback at the 40m. It is clearly additional effort to design and build a double pendulum, rather than duplicate a single one. I would suggest that this is a small addition to the whole project. I agree, however, that if you choose 3" by 1" optics they are much lighter than we have ever tried to suspend as a double pendulum, and this might need some extra design work.) I am not inclined to push very hard where there is reluctance, probably OK with single pendulums (and saves us at Glasgow some work too). Cheers, Ken _______________________________________________________________________ From k.strain@physics.gla.ac.uk Thu Jul 27 01:55:08 2000 To: Alan Weinstein Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE > I am thinking of operating the 40m in the Signal Recycling regime > (negative tune, smaller bandwidth) > not the RSE regime (positive tune, larger bandwidth). > > This allows us to put the dip in the sensitivity > anywhere we want it, eg, 500 Hz (or the 950 Hz you see in > the figures in http://www.ligo.caltech.edu/~ajw/40m_testmass.html ), > so laser noise above a few kHz is not of concern for us. > > I believe that the control problem for SR is *identical* > to that for RSE, so this is a meaningful prototype test. > We are looking at the sensing for the first stage of the 10m RSE tests at Glasgow. There are subtle differences in the l- and SR mirror signals (emphasised as we have high finesse in the arms 12000, and no power recycling). Under more moderate circumstances and with most of the sensing methods under consideration, there is very little difference between a system optimised for SR and one optimised for RSE. (It is not, however, universally true.) > I like the idea of 40m and LIGO having the same finesse, > since I am under the impression > that the optical control plant challenge > "scales" with finesse (the higher the finesse of the cavities, > the more difficult to acquire and keep lock). > Is this right? Wrong? It's just a gut feeling. AS mentioned above the l- signal gets harder to separate out (from L-,L+ ) as the finesse increases. The control solution would not be the same for finesse 300 and finesse 10000. Life will be tougher if we choose sapphire substrates and hence need higher arm finesse in LIGO II. For this reason we have chosen to by T=1% ITMs for the second stage of the Glasgow tests (we have money for yet another a new mirror set if needed though) - this being somewhere in the middle of the range under consideration (3% to 0.5% approx.) This would not change the thermal problem if the arm power and coating absoption stay the same, as with 25 mm of SV you are dominated by that. Cheers, Ken _______________________________________________________________________ From k.strain@physics.gla.ac.uk Thu Jul 27 02:22:13 2000 To: GariLynn Billingsley Subject: Re: Fwd: *** REQUEST FROM AJW ON OPTIC SIZE Dear All, with 1" thick SV one can rely on <<5ppm total absorption. If one assumes the normal 0.5ppm for coating absorption we can see that SV ITM substrate absoption is unimportant for finesse >~30. (Recent SV batches have shown from <0.6ppm/cm (known) to ~1.5ppm/cm (rumour, perhaps not trustworthy, about 4th hand by now). The arm cavity finesse will be much more than this. Continuing this argument, there is no significant thermal effect in any other location (assumptions: thin beamsplitter <~1" and all coatings ~0.5ppm absorption.) (I suspect that some samples of 311, and 312 can be worse than 3ppm/cm (I assume those are about half of the price of SV, so probably not worth the saving in just a few small substrates). ) Cheers, Ken (compare substrate and coating absorption thermal lens using enhancement factor ~2Finesse/pi remembering factor 1.3 for strength of lens produced by substrate absorption compared to coating absorption of a given power.) From ajw@caltech.edu Thu Aug 17 14:51:00 2000 From: "Alan Weinstein" To: Cc: "Alan J Weinstein" Subject: SOS suspensions for the 40m upgrade Date: Thu, 17 Aug 2000 14:49:16 -0700 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook 8.5, Build 4.71.2173.0 Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V5.00.3018.1300 X-Status: X-Keywords: X-UID: 34 Hi Mark (and Dennis), I tentatively plan on using 3"x1" optics with standard SOS suspensions for all optics at the 40m upgrade (at least, as a first pass). That's 7 core optics, 3 MC optics, maybe more. I just had a chat with Sany Yoshida and Malik, who offered their (and Haisheng's) services in assembling and hanging SOS suspensions & optics for the 40m upgrade. As Sany's boss, I wonder if this is ok with you. I imagine we could need his services as early as next summer. I'm not sure how long it would take, maybe a month or 2.... Let me know what you think (and if you have opinions on the appropriateness of 3" SOS optics for the 40m, or anything else about the upgrade, I welcome them. If you want to learn more about our current plans, check out http://www.ligo.caltech.edu/~ajw/40m_081600.pdf Thanks! AJW From ajw@hep204.cithep.caltech.edu Fri Aug 18 15:18:50 2000 Date: Fri, 18 Aug 2000 15:18:35 -0700 (PDT) From: Alan Weinstein To: ajw@caltech.edu, sanders@ligo.caltech.edu, barish@ligo.caltech.edu, coyne@ligo.caltech.edu, dugolini@ligo.caltech.edu, steve@ligo.caltech.edu, ouimette_d@ligo.caltech.edu, stan@ligo.caltech.edu, RJK@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, fjr@ligo.caltech.edu, nergis@ligo.caltech.edu, kells_b@ligo.caltech.edu, ganezer@dhvx20.csudh.edu, asiri_f@ligo.caltech.edu, smith_m@ligo.caltech.edu, jay@ligo.caltech.edu, rolf@ligo.caltech.edu, jordan@ligo.caltech.edu Subject: 40m meeting X-Status: X-Keywords: X-UID: 21 Hi all, The LSC meeting brought some kind of focus to the LIGO-II dual recycling control scheme, which I believe is straightforward to implement at the 40m. I therefore think it is time to move forward, making sufficiently firm decisions on things like suspensions, optics, control electronics, etc, to start ordering parts. I wonder whether it's possible (and/or advisable) to work towards installing everything we need for a working, ligo-ii-like dual-recycled IFO by one year from now. I would like to have a ~ 1 hour meeting with all interested parties, early next week, to discuss these 40m plans. Can we do next Wednesday, 8/23, at 11am, SCR? Please let me know if you CAN'T do it, and if another time would be better. I'm sending this email to lots of people on campus, but if your name appears below, I really hope you can make it! Thanks! AJW _________ - Laboratory rehab, and OOC seismic stack Fred Asiri - PSL - Progress on assembly, delivery, layout - Peter King & Rich Abbott. - Suspensions and optics: - I need to think more about how bad the Q of a light 3"x1" optic with magnets, etc, would be; if it's not too bad, I'd like to go with - standard-ligo-i-issue SOS suspensions and optics for all optics, at least for the first go-round. (Then move on to multiple suspensions after 2002!). - Suny Yoshida et al have offered their services in assembling, hanging, balancing, etc. - Flat ITMs, specified ROC on all others - SV glass for BS, SM, ITMs; Corning for ETMs, RM. - Transmissions: I have a good LIGO-II guess now, (T_ITM = 0.5%; T_ETM = 15ppm; T_RM = 10%; T_SM = 10%). but want to wait till last minute for final specs. - MC - identical to LIGO-I. (we *may* make it longer, by adding a pipe, to be as long as the ~16.5m being contemplated for LIGO II; that shouldn't change anything in the optics, suspensions, control system, or really, anything else except for the performance.) - So we'd need 7 core optics + 3 MC optics, or 10 SOS suspensions & optics, with associated osems. BUT see discussion of output mode cleaner, below. Janeen, Garilynn, Helena - Optical layout - I'd like to make a preliminary-final review of Mike's layout, and then, by golly, start ordering parts! Mike Smith. - Output mode cleaner For DC locking, we need a small, monolithic mode cleaner like the PSL's PMC. We have to look into whether it would have to be hung. This would be interesting: hanging a coffin-shaped piece of glass, with a PZT-actuator on the end. We'll have to think about what makes sense, here. But it might mean an 11th suspension, with OSEMs for damping. - Control system, DAQ, etc; - I can imagine "cloning" the LIGO-I control system (latest and greatest, digital suspension controllers, etc), with extra equipment for the extra degrees of freedom (two pairs of RF sidebands, 180-ish MHz demodulation sensors, 7 suspension controllers, one more WFS, all ISC/LSC/ASC ligo-i-like) - I'd love to have an early discussion on what might be different for LIGO II - We must make a preliminary (but as complete as possible) parts list, and identify long-lead-time items to be ordered NOW. Rick K, Jay Heefner, Rolf Bork, Dennis Coyne - Active seismic isolation - STACIS seems like a good idea; in my mind, the issues are: - is installation, EQ safety, maintenance, etc, manageable? (Larry Jones, Steve Vass) - is the cost acceptable to the lab? (Gary & Dennis) - is there a fall-back if they don't work? (eg, an aluminum block). - I hope we can develop confidence to DECIDE and PROCEED. From ajw@rana.ligo.caltech.edu Sat Aug 19 15:57:42 2000 Date: Sat, 19 Aug 2000 15:57:21 -0700 (PDT) From: Alan Weinstein To: kstrain@physics.gla.ac.uk, stan@ligo.caltech.edu CC: ajw@ligo.caltech.edu Subject: 40m optic size X-Status: X-Keywords: X-UID: 9 Hi Stan and Ken, Sorry to belabor the issue, but... I would be grateful if you could set me straight on the deliberations below, or point me to the best person to ask. Thanks! AJW _______________________________ SOS or larger optic at the 40m? (at least, for the first go-round). - LIGO SOS optic: 3"diameter x 1.0" thick fused silica, 0.25 kg - 40m proto optic: 4"diameter x 3.5" thick fused silica, 1.56 kg Pros for smaller optic: - suspension, fixtures, etc all engineered - controller all engineered - experience with construction and operation of ~20 of them; Suny has volunteered to assemble, hang, balance them. - small footprint on chamber optical table - all suspensions at 40m would be of one type Pros for larger optic: - suspension noise is lower - re-engineering of modified suspension, controller, etc is small - easier to handle, thicker wires, easier to balance - Use 3" suspensions for MC1-3, SM, RM, BS; and 4" for four TMs. There's plenty of room for 4" suspensions in the TM chambers. - lower Q for internal thermal noise? I believe that the pros for the smaller optic all outweigh those for the larger optic, EXCEPT for the last one. Test mass thermal noise is a big problem for the 40m in any effort to expose shot noise. We'd like to keep the Q as low as possible, and this consideration outweighs all the small optic pros! So how does this work? Q = 1/loss loss = structural + coatings + attachments - I believe that structural loss depends only on material, not mass. - typical number I hear: Q = 3E7 - Might low-absorption (SV) glass have higher Q than Corning? - coating loss, I guess, depends on area, and we may coat the same area in either case. - I don't know the numbers. Maybe Q = 5e6? - attachment loss includes wires, standoffs, magnet standoffs. - I'm sure this does scale with mass, since it's a recoil effect - Wires would be made lighter in proportion (?), so the contribution to the loss would be the same in either case - but the magnet standoffs are as small as you can make them. These dominate, don't they? - If they give, eg, Q = 1E6 for a 10 kG mass, it would be Q = 2.5E4 for a 0.25 kg mass, and 1.6e5 for a 1.6 kg mass. If this is correct, it's plenty reason to go with the larger masses! From kells@ligo.caltech.edu Mon Aug 21 09:56:44 2000 Sender: kells@ligo.caltech.edu Date: Mon, 21 Aug 2000 09:55:53 -0700 From: "Bill Kells (CIT)" Organization: caltech/LIGO X-Mailer: Mozilla 4.7 [en] (X11; U; SunOS 5.7 sun4u) X-Accept-Language: en MIME-Version: 1.0 To: GariLynn Billingsley , ajw@caltech.edu Subject: Re: Fwd: 40m meeting References: <4.3.2.7.2.20000821084321.00adad90@acrux.ligo.caltech.edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Status: X-Keywords: X-UID: 2 GariLynn Billingsley wrote: > Hi Bill, > This is an excerpt from an Alan Weinstein e-mail. > Do you know how to answer the question of "how bad" is the Q of a 3"X1" > optic with magnets? > The tricky part is that I don't know - compared to what, what are his other > limiting noise sources? > > Do you have enough info to answer this or shall I try to drag more data out > of him? > Thanks! > Gari > > >- Suspensions and optics: > > - I need to think more about how bad the Q of a light > > 3"x1" optic with magnets, etc, would be; > > if it's not too bad, I'd like to go with > > - standard-ligo-i-issue SOS suspensions and optics for all optics, > > at least for the first go-round. > > (Then move on to multiple suspensions after 2002!). GAri, Here is my understanding of this issue, which "explains" the oft expressed notion that smaller optics are lower Q for fixed attatchments (eg. Stan's comments). Q^-1 is , in general meant to be the rate of mechanical mode energy loss divided by that mode's total stored energy: (dU/dt)/U Now, on the average througout the mass, the mode has some energy density u. Then the total energy stored, U, is equal to ux Volume (uxV). Supose that the energy loss is dominated by an attatchment (a glued magnet). Think of it as a small hole on the surface through which all mechanical wave energy impinging just exits, and is lost. Then the loss rate is ux (sound velocity)x Area of hole (uxsxA). Then the Q of this situation is: Q= V/(Axs). So , for a particular TM material (s fixed) the Q gets better the larger the Volume to surface area. Of course this is just a simplified model. For instance it predicts arbitrarily large Q for bigger TMs. Actually , at some point (in increasing TM size), some intrinsic internal (volume) loss begins to dominate, and the attatchments no longer matter. For specifics, though I am simply not the one who knows. If I were in Bridge I would talk with Phil Willems. Ciao, Bill From ajw@caltech.edu Mon Aug 21 10:05:55 2000 From: "Alan Weinstein" To: , "GariLynn Billingsley" Cc: "Alan J Weinstein" Subject: RE: Fwd: 40m meeting Date: Mon, 21 Aug 2000 10:12:09 -0700 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0) Importance: Normal In-Reply-To: <39A15F19.D757327B@ligo.caltech.edu> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 X-Status: X-Keywords: X-UID: 3 Bill, Thanks much for the clear explanation! It makes sense to me, and I came to the same conclusion (after a healthy email exchange with Stan), except: - the energy "exiting the hole" of an attachment depends, of course, on the mode pattern; putting the attachment at a null point minimizes it - it also depends on how "soft" the attachment is; a bad glue joint makes the energy loss much larger. - So, only a fraction of the total uxsxA is lost, and that fraction can vary dramatically So now I'm back to thinking larger masses for the ITMs and ETMs (only), and it all depends on how much of a chore or problem Garilynn, Janeen, Jay, and Sany think it is... Thanks! AJW From stan@ligo.caltech.edu Sun Aug 20 11:31:57 2000 X-Sender: stan@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Sun, 20 Aug 2000 09:25:05 -0700 To: Alan Weinstein From: Stan Whitcomb Subject: Re: 40m optic size Cc: kstrain@physics.gla.ac.uk In-Reply-To: <200008192257.PAA20187@rana> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 17 At 03:57 PM 8/19/00 -0700, you wrote: >Hi Stan and Ken, > >SOS or larger optic at the 40m? >(at least, for the first go-round). > >- LIGO SOS optic: 3"diameter x 1.0" thick fused silica, 0.25 kg >- 40m proto optic: 4"diameter x 3.5" thick fused silica, 1.56 kg I don't know any reason why you should be limited to these two choices. If it were I who whad to choose, I o with 4" diameter, 2" thick, more like a LIGO I (or LIGO II, I presume) aspect ratio. >Pros for smaller optic: > - suspension, fixtures, etc all engineered a real advantage. > - controller all engineered no advantage at all. The existing LIGO optics can be controller with either our large optic or small optic controller. > - experience with construction and operation of ~20 of them; > Suny has volunteered to assemble, hang, balance them. the experience would be equally relevant to a larger TM. I discount this one to near zero. > - small footprint on chamber optical table > - all suspensions at 40m would be of one type minor advantage. >Pros for larger optic: > - suspension noise is lower > - re-engineering of modified suspension, controller, etc is small the "re-engineering" is probably minor, though there would be a number of drawings that would have to be done. > - easier to handle, thicker wires, easier to balance > - Use 3" suspensions for MC1-3, SM, RM, BS; and 4" for four TMs. > There's plenty of room for 4" suspensions in the TM chambers. > - lower Q for internal thermal noise? I believe that you mean higher Q and I believe that the evidence would support this expectation. >I believe that the pros for the smaller optic all outweigh >those for the larger optic, EXCEPT for the last one. > >Test mass thermal noise is a big problem for the 40m >in any effort to expose shot noise. >We'd like to keep the Q as low as possible, as high as possible.... >and this consideration outweighs all the small optic pros! > >So how does this work? > >Q = 1/loss > >loss = structural + coatings + attachments > >- I believe that structural loss depends only on material, not mass. > - typical number I hear: Q = 3E7 > - Might low-absorption (SV) glass have higher Q than Corning? I don't believe that there is any reason to expect the SV glass to be any better than the Corning material. Otherwise I agree. >- coating loss, I guess, depends on area, > and we may coat the same area in either case. > - I don't know the numbers. Maybe Q = 5e6? I would guess that the coating loss would be somewhat less than this. Without any real data, I would guess 2E7. >- attachment loss includes wires, standoffs, magnet standoffs. > - I'm sure this does scale with mass, since it's a recoil effect > - Wires would be made lighter in proportion (?), > so the contribution to the loss would be the same in either case > - but the magnet standoffs are as small as you can make them. > These dominate, don't they? Loss in wire is probably negligible compared with the loss in the standoffs (both wire and magnet) that are glued to the TM. The effects of these depend rather strongly on the mode shape since the the stresses at the joints depend on mode shape. Wire loss can important (as the Glasgow group discovered!) if a wire resonance happens to be degenerate with a TM mode, at first thought unlikely, but proven to be possible. > - If they give, eg, Q = 1E6 for a 10 kG mass, it would be > Q = 2.5E4 for a 0.25 kg mass, and 1.6e5 for a 1.6 kg mass. This might scale directly with mass, or maybe with the 2/3 power of mass (my personal favorite), or something completely different. Let me quote some data from real suspended masses: 3" optics measured in the modecleaner. This is a single mode at ~29.6 kHz. The Q's ranged from 3.7 to 13 E5 for 3 optics that were measured. I don't remember the Q's of different modes in the old 40 m mirrors, but I believe that a weighted average (of about 4 modes) was close to 2E6. The modes of the 2 km TMs range from ~1E5 (on one TM, probably with a not very good glue joint) to 1.2E7 (3 modes on each of 3 TMs). A reasonable average might be 4E6. stan From ajw@hep206.cithep.caltech.edu Sun Aug 20 13:38:47 2000 Date: Sun, 20 Aug 2000 13:38:46 -0700 (PDT) From: Alan Weinstein To: stan@ligo.caltech.edu CC: kstrain@physics.gla.ac.uk, ajw@hep.caltech.edu In-reply-to: <4.3.2.7.2.20000820083724.00e02140@127.0.0.1> (message from Stan Whitcomb on Sun, 20 Aug 2000 09:25:05 -0700) Subject: Re: 40m optic size X-Status: X-Keywords: X-UID: 1 Hi Stan, Thanks for the quick answer; it's very helpful, and it of course raises more questions, so my apologies for belaboring the issue... > I don't know any reason why you should be limited to these two choices. > If it were I who whad to choose, I o with 4" diameter, 2" thick, more like > a LIGO I > (or LIGO II, I presume) aspect ratio. We're not limited, of course. It's just for the sake of evaluating pros and cons. Why is it useful to have a LIGO-I like ratio? Easier to predict normal mode frequencies and couplings to attachments, because we have experience and detailed models for those? > I believe that you mean higher Q and I believe that the evidence would > support this expectation. Right, higher Q. Have I included all the pros and cons, or are there others you can think of? Higher Q is, to my understanding, the overriding consideration in this choice. > I don't believe that there is any reason to expect the SV glass to be > any better than the Corning material. Otherwise I agree. I wonder whether structural damping depends on things like impurities; but maybe the microscopic impurities relevant for optical properties are not relevant for the more macroscopic dissipation mechanism... I don't know. Anyway, it's negligible, compared with the attachments. > >- coating loss, I guess, depends on area, > > and we may coat the same area in either case. > > - I don't know the numbers. Maybe Q = 5e6? > > I would guess that the coating loss would be somewhat less than this. > Without any real data, I would guess 2E7. Good. Negligible, compared with the attachments. > Loss in wire is probably negligible compared with the loss in the > standoffs (both wire and magnet) that are glued to the TM. The > effects of these depend rather strongly on the mode shape since the > the stresses at the joints depend on mode shape. Wire loss can > important (as the Glasgow group discovered!) if a wire resonance > happens to be degenerate with a TM mode, at first thought unlikely, > but proven to be possible. I presume that we can only check that with a detailed FEA model. But we know that such accidental degeneracy does not happen with the LIGO-I SOSs. > > - If they give, eg, Q = 1E6 for a 10 kG mass, it would be > > Q = 2.5E4 for a 0.25 kg mass, and 1.6e5 for a 1.6 kg mass. > > This might scale directly with mass, or maybe with the 2/3 power > of mass (my personal favorite), or something completely different. > Let me quote some data from real suspended masses: > > 3" optics measured in the modecleaner. This is a single mode at > ~29.6 kHz. The Q's ranged from 3.7 to 13 E5 for 3 optics that were > measured. A single mode? Presumably, there are many many normal modes, and you're just quoting the lowest one...? > I don't remember the Q's of different modes in the old 40 m mirrors, > but I believe that a weighted average (of about 4 modes) was close > to 2E6. > > The modes of the 2 km TMs range from ~1E5 (on one TM, probably > with a not very good glue joint) to 1.2E7 (3 modes on each of 3 TMs). > A reasonable average might be 4E6. OK, this is relevant data! (reference?) >From it, I learn: - loss from standoffs dominates. All other considerations are irrelevant. - Q varies by factors of 4 between different instances of the same optic size. - 3" - ~1E6 4" - ~2E6 10" - ~1E7. - So it behooves us to make the masses as large as is practical - and with an aspect ratio that minimizes coupling of modes to attachments, but I don't know how to address that without FEA work. Maybe I could ask Dennis. If this all makes sense, I will learn on Wednesday >from Janeen, Garilynn, Suny, and Jay just how much extra effort is required to go with a larger mass & suspension. Thanks for any other input you might have! AJW From stan@ligo.caltech.edu Sun Aug 20 17:08:06 2000 X-Sender: stan@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Sun, 20 Aug 2000 17:09:59 -0700 To: Alan Weinstein From: Stan Whitcomb Subject: Re: 40m optic size Cc: kstrain@physics.gla.ac.uk, ajw@hep.caltech.edu In-Reply-To: <200008202038.NAA04974@hep206.cithep.caltech.edu> References: <4.3.2.7.2.20000820083724.00e02140@127.0.0.1> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 3 At 01:38 PM 8/20/00 -0700, Alan Weinstein wrote: >Why is it useful to have a LIGO-I like ratio? >Easier to predict normal mode frequencies and couplings to attachments, >because we have experience and detailed models for those? I think the LIGO aspect ratio may reduce undesireable parasitic torques. Ask me next time I see you and I will draw a sketch to support that, though I confess I haven't thought about it too deeply, and doubt that it is really a big deal. >Have I included all the pros and cons, >or are there others you can think of? I don't immediately think of any other significant pros/cons. > > I don't believe that there is any reason to expect the SV glass to be > > any better than the Corning material. Otherwise I agree. > >I wonder whether structural damping depends on things like >impurities; but maybe the microscopic impurities relevant >for optical properties are not relevant for the more >macroscopic dissipation mechanism... I don't know. > >Anyway, it's negligible, compared with the attachments. Based on very limited "data" (hearsay might be a more accurate characterization) I think that metallic impurites may have some significant effect on Q, but the SV glass differs primarily from the Corning material by having very much lower OH content. I don't think that OH affects the loss at the 3E7 level. > > Loss in wire is probably negligible compared with the loss in the > > standoffs (both wire and magnet) that are glued to the TM. The > > effects of these depend rather strongly on the mode shape since the > > the stresses at the joints depend on mode shape. Wire loss can > > important (as the Glasgow group discovered!) if a wire resonance > > happens to be degenerate with a TM mode, at first thought unlikely, > > but proven to be possible. > >I presume that we can only check that with a detailed FEA model. >But we know that such accidental degeneracy does not happen >with the LIGO-I SOSs. actually I think that it is pretty hard to predict the 60th harmonic of a wire with any great precision. The techniques we have for suspending small optics probably have 2 mm accuracy for the length of the wire (out of 300), so the accuracy with which you predict the frequency of the modes around 25 kHz is about 200 Hz, about half the mode spacing. Add the tolerance on wire thickness, and you basically have a crap shoot on any given wire. Fortunately, the wire modes are narrow so the chance of hitting a TM mode is pretty small. But it has been done once in GW history.... > > > - If they give, eg, Q = 1E6 for a 10 kG mass, it would be > > > Q = 2.5E4 for a 0.25 kg mass, and 1.6e5 for a 1.6 kg mass. > > > > This might scale directly with mass, or maybe with the 2/3 power > > of mass (my personal favorite), or something completely different. > > Let me quote some data from real suspended masses: > > > > 3" optics measured in the modecleaner. This is a single mode at > > ~29.6 kHz. The Q's ranged from 3.7 to 13 E5 for 3 optics that were > > measured. > >A single mode? Presumably, there are many many normal modes, >and you're just quoting the lowest one...? so far we have only been able to excite and measure the "drumhead" mode on the modecleaner mirrors. There should be a large number of modes, we just haven't been able to find them yet. The rather large variation among the three mode cleaner mirrors measured to date is one of the things I don't like about the small optics suspension, it just seems fussy to get the best performance. > > I don't remember the Q's of different modes in the old 40 m mirrors, > > but I believe that a weighted average (of about 4 modes) was close > > to 2E6. > > > > The modes of the 2 km TMs range from ~1E5 (on one TM, probably > > with a not very good glue joint) to 1.2E7 (3 modes on each of 3 TMs). > > A reasonable average might be 4E6. > >OK, this is relevant data! (reference?) for the first and third above, got to Haisheng Rong and Nergis Mavalvala's LSC talks for the meeting last week. For the second, the official reference is harder to get. There might be something in Torrey Lyon's thesis. > >From it, I learn: > >- loss from standoffs dominates. All other considerations are irrelevant. > >- Q varies by factors of 4 between different instances of the same optic size. > >- 3" - ~1E6 > 4" - ~2E6 > 10" - ~1E7. > >- So it behooves us to make the masses as large as is practical > >- and with an aspect ratio that minimizes coupling of modes > to attachments, but I don't know how to address that > without FEA work. Maybe I could ask Dennis. I am not sure we really understand it well enough to easily answer the questions about attachments. It would take quite a bit of modelling and comparisons with the data cited above (which sounds better in an email like this than it is when you want to analyze it), before you would have much confidence in the result. >If this all makes sense, I will learn on Wednesday >from Janeen, Garilynn, Suny, and Jay >just how much extra effort is required to go with >a larger mass & suspension. minimal I predict for the optics and controllers and assembly ( and by the way it is Sany, not Suny). Lot's for Janeen. stan From k.strain@physics.gla.ac.uk Mon Aug 21 02:14:49 2000 Date: Mon, 21 Aug 2000 10:03:55 +0100 (BST) From: "k.strain" To: Stan Whitcomb cc: Alan Weinstein , k.strain@physics.gla.ac.uk, ajw@hep.caltech.edu Subject: Re: 40m optic size In-Reply-To: <4.3.2.7.2.20000820164110.00e54780@127.0.0.1> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Status: X-Keywords: X-UID: 6 Hi, On Sun, 20 Aug 2000, Stan Whitcomb wrote: > At 01:38 PM 8/20/00 -0700, Alan Weinstein wrote: > >Why is it useful to have a LIGO-I like ratio? > >Easier to predict normal mode frequencies and couplings to attachments, > >because we have experience and detailed models for those? > > I think the LIGO aspect ratio may reduce undesireable parasitic torques. > Ask me next time I see you and I will draw a sketch to support that, though I > confess I haven't thought about it too deeply, and doubt that it is > really a big deal. > > >Have I included all the pros and cons, > >or are there others you can think of? > > I don't immediately think of any other significant pros/cons. > > > > I don't believe that there is any reason to expect the SV glass to be > > > any better than the Corning material. Otherwise I agree. > > > >I wonder whether structural damping depends on things like > >impurities; but maybe the microscopic impurities relevant > >for optical properties are not relevant for the more > >macroscopic dissipation mechanism... I don't know. > > > >Anyway, it's negligible, compared with the attachments. > > Based on very limited "data" (hearsay might be a more accurate > characterization) > I think that metallic impurites may have some significant effect on Q, but > the SV glass differs primarily from the Corning material by having very much > lower OH content. I don't think that OH affects the loss at the 3E7 level. High OH suprasil 2 holds the record, with a lack of measurements thus far on low OH SV. I'm quite sure any Heraeus/Corning synthetic fused silica will meet your mechanical and optical requirements if purchased with the correct homogeneity grade. > > > > Loss in wire is probably negligible compared with the loss in the > > > standoffs (both wire and magnet) that are glued to the TM. The > > > effects of these depend rather strongly on the mode shape since the > > > the stresses at the joints depend on mode shape. Wire loss can > > > important (as the Glasgow group discovered!) if a wire resonance > > > happens to be degenerate with a TM mode, at first thought unlikely, > > > but proven to be possible. > > > >I presume that we can only check that with a detailed FEA model. > >But we know that such accidental degeneracy does not happen > >with the LIGO-I SOSs. Even if the wire is lossy (and the Q of the mirror resonance degraded) the coupling of the two oscillators will be quite small in the GW band. The additional thermal noise won't be very much, I suspect, compared to that due to broadband dissipation due to the attachment of standoffs. (Of course this effect makes it hard to predict the level of thermal noise unless one has a very sophisticated model of the suspension & mirror.) Ken From k.strain@physics.gla.ac.uk Mon Aug 21 02:14:55 2000 Date: Mon, 21 Aug 2000 09:48:24 +0100 (BST) From: "k.strain" To: Stan Whitcomb cc: Alan Weinstein Subject: Re: 40m optic size In-Reply-To: <4.3.2.7.2.20000820083724.00e02140@127.0.0.1> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Status: X-Keywords: X-UID: 8 Can't find anything to add to what Stan said. Ken From ajw@hep204.cithep.caltech.edu Tue Aug 22 14:45:25 2000 Date: Tue, 22 Aug 2000 14:45:12 -0700 (PDT) From: Alan Weinstein To: saulson@ligo-la.caltech.edu CC: ajw@hep.caltech.edu Subject: test mass internal noise X-Status: X-Keywords: X-UID: 2 Hi Peter, Might you have some time to discuss test mass internal noise with me? Here are the deliberations so far, with much helpful input from Stan and Bill Kells. I figure you might have some deeper insight and knowledge... Thanks! AJW _________________________________________________________________ SOS or larger optic at the 40m? (at least, for the first go-round). - LIGO SOS optic: 3"diameter x 1.0" thick fused silica, 0.25 kg - 40m proto optic: 4"diameter x 3.5" thick fused silica, 1.56 kg Of course, we're not limited to these two choices. It's just for comparing the pros and cons of "light" vs "heavier" masses. Pros for smaller optic: - suspension, fixtures, etc all engineered - controller all engineered - experience with construction and operation of ~20 of them; Sany has volunteered to assemble, hang, balance them. - small footprint on chamber optical table - all suspensions at 40m would be of one type Pros for larger optic: - suspension noise is lower - re-engineering of modified suspension, controller, etc is small - easier to handle, thicker wires, easier to balance - Use 3" suspensions for MC1-3, SM, RM, BS; and 4" for four TMs. There's plenty of room for 4" suspensions in the TM chambers. - lower Q for internal thermal noise? I believe that the pros for the smaller optic all outweigh those for the larger optic, EXCEPT for the last one. Test mass thermal noise is a big problem for the 40m in any effort to expose shot noise. We'd like to keep the Q as high as possible, and this consideration outweighs all the small optic pros! So how does this work? Q = 1/loss loss = structural + coatings + attachments - I believe that structural loss depends only on material, not mass. - typical number I hear: Q = 3E7 - I gather that both Corning and Heraeus glasses of good optical grade all can acheive this. - Anyway, it's not the limiting factor here, so structural loss is irrelevant. - coating loss, I guess, depends on area, and we may coat the same area in either case. - I don't know the numbers. Maybe Q = 1e7, or even higher. - attachment loss includes wires, standoffs, magnet standoffs. - I'm sure this does scale with mass, since it's a recoil effect - Wires would be made lighter in proportion (?), so the contribution to the loss would be the same in either case - but pray that there are no accidental degeneracies between wire violin frequencies and test mass resonances! Nearly impossible to predict. - The magnet and wire standoffs are as small as you can make them. These dominate. - The effects of these depend rather strongly on the mode shape relative to where the standoffs are glue on, and on the hardness of the glue joint. - Semi-quantitative analysis from Bill Kells: loss = 1/Q = (dU/dt)/(2pi*f*U) U = stored energy = u (energy density) * V (volume) dU/dt = rate of energy loss through the "small hole on surface" presented by an attachment = u (energy density) * A (area of attachment) * s (speed of sound) * k (coupling out of the hole) k (coupling out of the hole) depends on the mode pattern relative to the attachment, the "softness" (dissipation) of the attachment, etc. It can vary by factors of 10, or even, for a fixed design, by factors of ~4. SO, Q scales with V / A and the attachment area A is already as small as you can make it SO, modulo the uncertainties in k, we can maximize Q by maximizing V (ie, larger mass). - note also that if dU/dt is independent of frequency for attachment losses, then loss is proportional to 1/f and Q is proportional to f. (as opposed to structural losses, where it is argued that dU/dt is proportional to f so that Q is roughly constant). This then implies that the loss grows as we go down in frequency from the test mass resonances to the GW band. Smaller test masses have higher resonance frequencies (inversely proportional to linear dimension) so the loss is larger in the GW band. If we argue that the loss at resonance scales with 1/V, this frequency dependence implies that the loss in the GW band goes like (linear dimension)^(-2) or mass^(-2/3). Hm. On the other hand, maybe dU/dt ~ f even for attachment losses; ie, the coupling factor k ~ f. - Experience and detailed FEA analysis of ligo-i optics might help in minimizing k; maybe this speaks for keeping the ligo-i-like aspect ratio... - Stan's guesses for realistic Q's for different size optics: 3" - ~1E6 4" - ~2E6 10" - ~1E7. - Is there any reason to choose a ligo-i-like aspect ratio, eg, 4" diameter, 2" thick? Stan says: > I think the LIGO aspect ratio may reduce undesireable parasitic torques. > Ask me next time I see you and I will draw a sketch to support that, though I > confess I haven't thought about it too deeply, and doubt that it is > really a big deal. I'll get back to Stan on that, but if (as the arguments above outline) we can minimize thermal noise by maximizing mass, and if increasing diameter introduces complications associated with the re-engineering of the suspensions, I'd question the need to stick with LIGO-I aspect ratios. From saulson@ligo-la.caltech.edu Wed Aug 23 10:02:24 2000 X-Sender: saulson@abundance X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0 Date: Wed, 23 Aug 2000 12:00:46 -0500 To: Alan Weinstein From: "Peter R. Saulson" Subject: Re: test mass internal noise In-Reply-To: <200008222145.OAA25494@hep204.cithep.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Status: X-Keywords: X-UID: 4 Dear Alan, Here is a short response to your message. It wasn't clear if you would prefer to talk on the phone; if so, please let me know. I don't have a strong opinion on the question of large versus small masses for the 40 meter. In the end it will come down to a judgment call, without a clear right or wrong answer. Given that, what considerations should weigh strongly in your judgment? Stan's opinion that the larger masses would be easier to handle ought to count for a lot. On the other hand, going with the previously engineered design of the 3 inch masses sounds good, but then again Janeen insisted that the work to re-engineer from 3 to 4 inches would be very simple. It wasn't clear from her remarks whether she would give a different answer at a different aspect ratio. As far as thermal noise goes, I agree that large masses OUGHT to have the edge. They do have a slight edge for purely internally generated thermal noise, but the dependence on mass is VERY small. The advantage should be larger for thermal noise due to external losses. But I am surprised by Stan's estimates, or at least at the appearance that he made a simple prediction. Recent experience has been quite mixed. There was a table of LIGO mirror Q's that Nergis showed at Detector Characterization at LSC (and that I've pulled up from her previous use at the Director's Review, LIGO-G000114-00-D, p. 11). It shows a wide range of Q's for large optics (much larger than you are considering), with the worst being 2.4 10^4 (!) while the best is 1.7 10^7. It is surprising and troubling how bad some of these are. At a minimum, they should give you pause in trying to predict too precisely what you will see in the 40 m. Then there is the global question of how much effort you ought to invest in trying to minimize your noise. As noble as this is, it is a very demanding job, and since it isn't your primary job it is hard to judge the cost/benefit ratio. Of course anything that is simple to do to reduce noise makes sense, so by all means you ought to pursue choices like test mass design that might make performance better or (even more important) life simpler. I hope this is helpful. Best regards, Peter From ajw@hep204.cithep.caltech.edu Wed Aug 23 10:21:32 2000 Date: Wed, 23 Aug 2000 10:21:20 -0700 (PDT) From: Alan Weinstein To: saulson@ligo-la.caltech.edu CC: ajw@hep.caltech.edu In-reply-to: <200008231701.MAA10611@abundance.ligo-la.caltech.edu> (saulson@ligo-la.caltech.edu) Subject: Re: test mass internal noise X-Status: X-Keywords: X-UID: 1 Hi Peter, Thanks much for your response. I'd like to bounce some replies back, and please let me know if you disagree. > Stan's opinion that the larger masses would be easier to handle ought to > count for a lot. On the other hand, going with the previously engineered Sany doesn't seem to think so. He is the one who actually assembles and balances the things, and he said he'd be happy to do it for us. The implication to me: after some early snafus that stan remembers, he knows what he's doing. > design of the 3 inch masses sounds good, but then again Janeen insisted > that the work to re-engineer from 3 to 4 inches would be very simple. It Not VERY simple. It's some days or weeks of work which she won't be able to spare for some months. But she's not unwilling to do it, and we're not in a hurry; the optics take ~8 months. So I now feel that we should take the 4" route for the test masses. > As far as thermal noise goes, I agree that large masses OUGHT to have > the edge. They do have a slight edge for purely internally generated > thermal noise, but the dependence on mass is VERY small. The advantage > should be larger for thermal noise due to external losses. But I am > surprised by Stan's estimates, or at least at the appearance that he made a He put in all the usual qualifications and cautions, but I left all those out and boiled them down to a few numbers, thus grossly misrepresenting his message (but giving me some feul for decision-making). > simple prediction. Recent experience has been quite mixed. There was a > table of LIGO mirror Q's that Nergis showed at Detector Characterization at > LSC (and that I've pulled up from her previous use at the Director's > Review, LIGO-G000114-00-D, p. 11). It shows a wide range of Q's for large Thanks for the reference! > optics (much larger than you are considering), with the worst being 2.4 > 10^4 (!) while the best is 1.7 10^7. It is surprising and troubling how bad > some of these are. At a minimum, they should give you pause in trying to > predict too precisely what you will see in the 40 m. Right. I'm not trying to predict, just make an intelligent decision on the optic size. > Then there is the global question of how much effort you ought to invest > in trying to minimize your noise. As noble as this is, it is a very > demanding job, and since it isn't your primary job it is hard to judge the > cost/benefit ratio. Of course anything that is simple to do to reduce noise > makes sense, so by all means you ought to pursue choices like test mass > design that might make performance better or (even more important) life > simpler. Right. Globally, if I can invest a bit of effort now to minimize thermal noise, and IF we get lucky in smoothly operating the upgraded 40m and exposing fundamental noise without enormous effort (and I know it's a big IF, but not too unreasonable), I'll be glad for the opportunity to study the shot noise. Thanks! Alan W From saulson@ligo-la.caltech.edu Wed Aug 23 11:12:55 2000 X-Sender: saulson@abundance X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0 Date: Wed, 23 Aug 2000 13:11:13 -0500 To: Alan Weinstein From: "Peter R. Saulson" Subject: Re: test mass internal noise In-Reply-To: <200008231721.KAA31260@hep204.cithep.caltech.edu> References: <200008231701.MAA10611@abundance.ligo-la.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Status: X-Keywords: X-UID: 4 Dear Alan, No disagreements. Good luck! Best regards, Peter From ajw@hep204.cithep.caltech.edu Thu Aug 24 16:37:26 2000 Date: Thu, 24 Aug 2000 16:37:25 -0700 (PDT) From: Alan Weinstein To: stan@ligo.caltech.edu CC: ajw@hep.caltech.edu Subject: test mass aspect ratio issue Hi Stan, I need to follow up on this test mass aspect ratio issue with you. You said: > I don't know any reason why you should be limited to these two choices. > If it were I who whad to choose, I o with 4" diameter, 2" thick, more like > a LIGO I > (or LIGO II, I presume) aspect ratio. and then > I think the LIGO aspect ratio may reduce undesireable parasitic torques. > Ask me next time I see you and I will draw a sketch to support that, though I > confess I haven't thought about it too deeply, and doubt that it is > really a big deal. We have argued that test mass losses are dominated by those due to attachments, and these losses scale like some negative power of the mass. So we can minimize loss and therefore in-band thermal noise by maximizing mass. I think that there will be practical complications associated with the re-engineering of the suspensions for diameters bigger than 4", so that suggests that we should continue to use relatively thick masses, like the 3.5" used at the 40m, rather than sticking with a LIGO-I aspect ratio; UNLESS you think there's a compelling reason to do so. So, can we talk more about this? I'd LOVE to be able to come to a final decision, and ask Garilynn to order glass in the next 2 weeks! Thanks! AJW From stan@ligo.caltech.edu Thu Aug 24 17:17:49 2000 X-Sender: stan@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Thu, 24 Aug 2000 17:20:10 -0700 To: Alan Weinstein From: Stan Whitcomb Subject: Re: test mass aspect ratio issue In-Reply-To: <200008242337.QAA30990@hep204.cithep.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 3 At 04:37 PM 8/24/00 -0700, you wrote: the most importan part of my remarks was: > > though I > > confess I haven't thought about it too deeply, and doubt that it is > > really a big deal. > >We have argued that test mass losses are dominated by those >due to attachments, and these losses scale like some negative power >of the mass. So we can minimize loss and therefore >in-band thermal noise by maximizing mass. Right, but my instinct says that the BIGGEST difference you could expect might be a factor of 2 in Q between 2" thick and 3.5" thick (i.e., 40% in thermal noise), and might not be much more than 1.5. Not a big difference. The parasitic torques that I am talking about come from sideways forces on the magnets. These combine with the moment arm of the magnet to give a torque and with the 40m aspect ratio, the moment arm is about the same as the one for the intended torques. Thinning the masses by a factor 2, makes difference between the intended and unintended torques a factor of 2 larger. >I think that there will be practical complications >associated with the re-engineering of the suspensions >for diameters bigger than 4", >so that suggests that we should continue to use >relatively thick masses, like the 3.5" used at the 40m, >rather than sticking with a LIGO-I aspect ratio; >UNLESS you think there's a compelling reason to do so. Compelling? absolutely not. >So, can we talk more about this? I don't hit campus again until after Labor day (actually pass through briefly on Saturday, but mostly to pick up mail drop off stuff). >I'd LOVE to be able to come to a final decision, >and ask Garilynn to order glass in the next 2 weeks! I don't see any reason not to decide in the next two hours. I think that the optimization function is so weak that it is impossible to tell which is best. stan From ajw@caltech.edu Thu Aug 24 19:12:12 2000 From: "Alan Weinstein" To: "Stan Whitcomb" Cc: "Alan J Weinstein" Subject: RE: test mass aspect ratio issue Date: Thu, 24 Aug 2000 19:12:38 -0700 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0) Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 In-Reply-To: <4.3.2.7.2.20000824170739.00e0e980@127.0.0.1> X-Status: X-Keywords: X-UID: 10 > the most importan part of my remarks was: OK, I get the message! > I don't see any reason not to decide in the next two hours. I think that > the optimization function is so weak that it is impossible to > tell which is best. I could spend the next two hours flipping coins. But really, it only takes a second to flip a coin. > The parasitic torques that I am talking about come from sideways forces on > the magnets. These combine with the moment arm of the magnet to give a > torque and with the 40m aspect ratio, the moment arm is about the same as > the one for the intended torques. Thinning the masses by a > factor 2, makes > difference between the intended and unintended torques a factor > of 2 larger. I'm trying to get the sign, here... Are you saying that the 40m aspect ratio (4" by 3.5") gives the intended torques (good) while thinning it makes it worse by a factor 2? Is this something that Janeen can engineer into the modified suspension? Does she know how? Because in either case, she has to spend "weeks" on the modified (for 4") design. Presumably less work for her if she sticks with 4" by 3.5", since she did that for the 40m years ago and can pull out those drawings. I just want to make as intelligent and informed a choice as I can, even if it doesn't make a lot of difference; and I appreciate your advice. If you had 2 hours to decide, what would you choose?? AJW From stan@ligo.caltech.edu Thu Aug 24 20:25:29 2000 X-Sender: stan@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Thu, 24 Aug 2000 20:27:47 -0700 To: "Alan Weinstein" From: Stan Whitcomb Subject: RE: test mass aspect ratio issue In-Reply-To: References: <4.3.2.7.2.20000824170739.00e0e980@127.0.0.1> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 12 At 07:12 PM 8/24/00 -0700, you wrote: >I could spend the next two hours flipping coins. >But really, it only takes a second to flip a coin. It's not much better than that..... >I'm trying to get the sign, here... >Are you saying that the 40m aspect ratio (4" by 3.5") >gives the intended torques (good) >while thinning it makes it worse by a factor 2? Actually the other way around. That sideways forces give a larger torque in a thicker mass than the same forces would in a thinner mass. The intended torques are due to longitudinal forces, assumed to be the "same" in the two cases. Not really sure that it is a big deal. >Is this something that Janeen can engineer into the >modified suspension? Does she know how? >Because in either case, she has to spend "weeks" >on the modified (for 4") design. Don't know that there is much that she can do to "engineer " us out of it. >Presumably less work for her if she sticks with >4" by 3.5", since she did that for the 40m years ago >and can pull out those drawings. Maybe she has the drawings, or maybe not. We had another engineer who did a lot of the 40 m stuff, and I don't remember how deeply Janeen got into it. >I just want to make as intelligent and informed >a choice as I can, even if it doesn't make a lot of difference; >and I appreciate your advice. >If you had 2 hours to decide, what would you choose?? I came awfully close to recommending the 3.5" thickness in my last email, for a terrible reason, namely that I would be less culpable for going with the flow if it turned out to be flawed, than I would if I was the one to push for thinner and that turned out wrong. In the end, I would probably choose the thinner one (2.5"), for an equally bad reason, namely, all the GW detectors now being built are using a roughly 2 (dia) to 1 (thickness) ratio, presumably the result of deep thought, whereas the 40 m ratio was chosen in the "old days" before people really knew much. stan From smith@ligo.caltech.edu Fri Aug 25 08:40:32 2000 X-Sender: smith@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Pro Version 4.2.0.58 Date: Fri, 25 Aug 2000 08:39:09 -0700 To: GariLynn Billingsley From: Mike Smith Subject: Re: 40 m wedges Cc: smith_m@ligo.caltech.edu, Alan Weinstein Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 1 Gari, Here is a first estimate of the 40m wedge angles: RM 2.5 deg BS 1.0 deg ITM 1.0 deg ETM 2.5 deg SM 2.5 deg Mike At 05:10 PM 8/23/00 -0700, you wrote: >Hi Mike, >Can you get me what you know about wedges to date? >Thanks! >Gari From ajw@hep204.cithep.caltech.edu Fri Aug 25 09:32:25 2000 Date: Fri, 25 Aug 2000 09:32:24 -0700 (PDT) From: Alan Weinstein To: stan@ligo.caltech.edu, janeen@ligo.caltech.edu, gari@ligo.caltech.edu, coyne@ligo.caltech.edu, kells_b@ligo.caltech.edu CC: ajw@hep.caltech.edu Subject: An' a 3, an' a 4 ... X-Status: X-Keywords: X-UID: 2 Hi Stan, Janeen, Gari, Dennis, Bill, I can't nail down Stan on the 40m test mass aspect ratio... > I came awfully close to recommending the 3.5" thickness in my last email, for > a terrible reason, namely that I would be less culpable for going with the > flow if it turned out to be flawed, than I would if I was the one to push for > thinner and that turned out wrong. > > In the end, I would probably choose the thinner one (2.5"), for an equally bad > reason, namely, all the GW detectors now being built are using a roughly 2 > (dia) to 1 (thickness) ratio, presumably the result of deep thought, whereas the > 40 m ratio was chosen in the "old days" before people really knew much. Sigh. I don't know who put in that deep thought; do you? Who else can I bother with these endless emails about inconsequential matters? OK, Janeen and Gari, how much difference does it make to you if we (randomly?) choose 4" dia, 3.5" thick (old 40m TM, large mass) 4" dia, 2.5" thick (different aspect ratio, less mass) 4" dia, 2.0" thick (2:1 aspect ratio) 5" dia, 2.5" thick (larger diameter, 2:1 aspect ratio, more mass) In fact, if the last option is not too troublesome (I think there's plenty of room in the 40m TM chambers), it satisfies both (a) large mass so hopefully lower thermal noise. (b) 2:1 aspect ratio that is the result of somebody's deep thought. But maybe it's more troublesome to engineer the suspension, or maybe the time and money for ordering, polishing, coating is more problematic... Thanks for your input! AJW From Billingsley_G@ligo.caltech.edu Fri Aug 25 09:41:39 2000 X-Sender: gari@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Fri, 25 Aug 2000 09:41:13 -0700 To: Alan Weinstein From: GariLynn Billingsley Subject: Re: An' a 3, an' a 4 ... In-Reply-To: <200008251632.JAA31062@hep204.cithep.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 1 >OK, Janeen and Gari, how much difference does it make to you >if we (randomly?) choose >4" dia, 3.5" thick (old 40m TM, large mass) >4" dia, 2.5" thick (different aspect ratio, less mass) >4" dia, 2.0" thick (2:1 aspect ratio) >5" dia, 2.5" thick (larger diameter, 2:1 aspect ratio, more mass) > >In fact, if the last option is not too troublesome >(I think there's plenty of room in the 40m TM chambers), >it satisfies both >(a) large mass so hopefully lower thermal noise. >(b) 2:1 aspect ratio that is the result of somebody's deep thought. > >But maybe it's more troublesome to engineer the suspension, >or maybe the time and money for ordering, polishing, coating >is more problematic... > >Thanks for your input! AJW From an optics perspective it doesn't make much difference in the grand scheme of things. Perhaps a slight increase in material and polishing cost. I think that if it can fit, and if there isn't a show stopper in suspending it, we should go with the last option. G From coyne@ligo.caltech.edu Sun Aug 27 23:53:20 2000 Date: Mon, 28 Aug 2000 00:12:39 -0700 From: Dennis Coyne X-Mailer: Mozilla 4.7 [en] (Win98; I) X-Accept-Language: en MIME-Version: 1.0 To: David Shoemaker , Stan Whitcomb , Bill Kells , GariLynn Billingsley CC: Alan Weinstein Subject: TM aspect ratio Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Status: X-Keywords: X-UID: 14 David, Stan, Bill and Gari, Alan is finalizing the optics sizes for the 40m lab. He asked me about the aspect ratio. I've looked into the LIGO-1 documentation and have found very little and it doesn't make a completely coherent picture: 1) The COC DRD, LIGO-950099-04, section 3.2.1.1.2 states that "The aspect ratio is chosen to insure sufficiently high internal mode frequencies and Q." Appendix C indicates the variation of the first few modal frequencies with thickness for a set diameter. However, no optimum is indicated and no criteria for "sufficiently high" is made. 2) The COC FDD, LIGO-E980061-00, section 2.1.3 indicates that the minimum first resonances are < 6 kHz for the RM, ITM, ETM and < 3 kHz for the BS in order not to complicate the LSC controls (no analysis or reference is given). This requirement coupled with Appendix C of the DRD would indicate a 8 cm thick TM, as opposed to the 10 cm thick optics chosen for LIGO-1. 2) Kent used the Gillespie & Raab methodology to calculate contours of total mirror thermal noise as a function of radius and thickness (for a range of mirror dimensions around the LIGO-1 test mass and for LIGO-1 beam waist size). This analysis appears to only be documented in a brief mention in P960031-C. One could conclude from Kent's analysis that the aspect ratio should have been less than 2.3 for LIGO-1 (subject to fabrication considerations). Thus minimization of thermal noise would argue for a thickness of 11 cm. Do any of you have any other insights or recall any analysis in the literature for aspect ratio optimization? Alan also asked why the test mass for LIGO-1 was doubly-wedged, i.e. why wasn't one face set perpendicular to the cylindrical axis? In fact the DRD requires that one face only be wedged, whereas the design (FDD and dwgs) have the BS, ITM and RM doubly-wedged. I found no justification in the documentation for this 'violation' of requirements. I suspect that the double wedge is very slightly better for balancing. Is there also a fabrication ease argument? Dennis From dhs@ligo.mit.edu Mon Aug 28 09:14:04 2000 X-Sender: dhs/18.120.0.82@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Mon, 28 Aug 2000 12:13:11 -0400 To: Dennis Coyne , Stan Whitcomb , Bill Kells , GariLynn Billingsley From: David Shoemaker Subject: Re: TM aspect ratio Cc: Alan Weinstein In-Reply-To: <39AA10E7.1123C434@ligo.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 19 Dennis, my recollection from the dark ages is that looking at mode frequency plots it 'seemed like a good point of diminishing returns' to have 'about' the ratio we ended up with. Another set of considerations for thickness is the absorption, scatter, and inhomogeneity of the substrate, all pushing for thinner substrates. The thermal noise for LIGO I should be re-calculated using Geppo Cagnoli's approach, and should be used for thermal noise calcs for the 40m if it is an important measure of the performance. I think that getting high internal resonance frequencies is the thing that would most likely have a tangible effect for the 40m in choosing an aspect ratio. This eases servo design, and locking. I'll admit that I thought our optics for LIGO I _were_ singly-wedged! I don't think I know of problems with a double wedge, though. Interested to hear from others on this. d. From romie_j@ligo.caltech.edu Mon Aug 28 09:16:07 2000 Sender: janeen@ligo.caltech.edu Date: Mon, 28 Aug 2000 08:59:15 -0700 From: Janeen Hazel X-Mailer: Mozilla 4.5 [en] (X11; I; SunOS 5.6 sun4u) X-Accept-Language: en MIME-Version: 1.0 To: Alan Weinstein , Gari Billingsley CC: janeen@ligo.caltech.edu Subject: Re: An' a 3, an' a 4 ... References: <200008251632.JAA31062@hep204.cithep.caltech.edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Status: X-Keywords: X-UID: 20 Dear Alan, Here's my two-cents. Changing the diameter is more troublesome to me than the thickness. So, I prefer options 1, 2 and 3 over option 4. However, both can be accomodated fairly easily. I prefer to go with the status quo only because the design is done. Janeen From Billingsley_G@ligo.caltech.edu Mon Aug 28 10:18:48 2000 X-Sender: gari@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Mon, 28 Aug 2000 10:18:00 -0700 To: Dennis Coyne , David Shoemaker , Stan Whitcomb , Bill Kells , GariLynn Billingsley From: GariLynn Billingsley Subject: Re: TM aspect ratio Cc: Alan Weinstein In-Reply-To: <39AA10E7.1123C434@ligo.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 1 >Alan also asked why the test mass for LIGO-1 was doubly-wedged, i.e. why >wasn't one face set perpendicular to the cylindrical axis? In fact the >DRD requires that one face only be wedged, whereas the design (FDD and >dwgs) have the BS, ITM and RM doubly-wedged. I found no justification in >the documentation for this 'violation' of requirements. I suspect that >the double wedge is very slightly better for balancing. Is there also a >fabrication ease argument? There is no manufacturing requirement which called for the double wedge. My recollection from the dark ages is that it was a requirement to eliminate a "retro-reflective glint" which can happen between the front surface and the 90 degree cylinder. From kells@acrux.ligo.caltech.edu Mon Aug 28 11:07:13 2000 Date: Mon, 28 Aug 2000 11:21:31 -0700 (PDT) From: Bill Kells To: dhs@ligo.mit.edu, stan@ligo.caltech.edu, kells@ligo.caltech.edu, gari@ligo.caltech.edu, coyne@ligo.caltech.edu Subject: Re: TM aspect ratio Cc: ajw@hep206.cithep.caltech.edu X-Sun-Charset: US-ASCII X-Status: X-Keywords: X-UID: 3 > From coyne@ligo.caltech.edu Sun Aug 27 23:51 PDT 2000 > Date: Mon, 28 Aug 2000 00:12:39 -0700 > From: Dennis Coyne > X-Accept-Language: en > MIME-Version: 1.0 > To: David Shoemaker , Stan Whitcomb , > Bill Kells , Hi Dennis, Let me try to clarify the issues of TM size raised. In general I subscribe to the comments of DHS in answer to this. Bill > > David, Stan, Bill and Gari, > Alan is finalizing the optics sizes for the 40m lab. He asked me about > the aspect ratio. I've looked into the LIGO-1 documentation and have > found very little and it doesn't make a completely coherent picture: > > 1) The COC DRD, LIGO-950099-04, section 3.2.1.1.2 states that "The > aspect ratio is chosen to insure sufficiently high internal mode > frequencies and Q." Appendix C indicates the variation of the first few > modal frequencies with thickness for a set diameter. However, no optimum > is indicated and no criteria for "sufficiently high" is made. > > 2) The COC FDD, LIGO-E980061-00, section 2.1.3 indicates that the > minimum first resonances are < 6 kHz for the RM, ITM, ETM and < 3 kHz > for the BS in order not to complicate the LSC controls (no analysis or > reference is given). This requirement coupled with Appendix C of the DRD > would indicate a 8 cm thick TM, as opposed to the 10 cm thick optics > chosen for LIGO-1. With respect to the original COC DRD, a strong guiding inclination was to proceed with the Pathfinder dimensions (25 cm dia x 10 cm thick) unless specific significant calculations indicated otherwise. At this stage there was no exact LSC design, and the global thermal noise trends not so well described (next item), so constraints from these were not emphasized. It has now been shown (from LIGO I commissioning) that the internal mode resonances can significantly impact the servo operation (requiring detailed filtering design), so it is probably worth factoring this into the decision. > 2) Kent used the Gillespie & Raab methodology to calculate contours of > total mirror thermal noise as a function of radius and thickness (for a > range of mirror dimensions around the LIGO-1 test mass and for LIGO-1 > beam waist size). This analysis appears to only be documented in a brief > mention in P960031-C. One could conclude from Kent's analysis that the > aspect ratio should have been less than 2.3 for LIGO-1 (subject to > fabrication considerations). Thus minimization of thermal noise would > argue for a thickness of 11 cm. Some global thermal noise mapping (such as this or Cagnoli) vs aspect ration (and other) is the way to go. My recollection is that for ratios anywhere near those being considered the dependence is mild (and one would wonder wherther the effective mode Qs achieved might vary with a. ratio in some dominating way). > Do any of you have any other insights or recall any analysis in the > literature for aspect ratio optimization? There is also the nice, consise, though probably not universally complete, paper by Bondu, et al. which at least indicates the basic trends. > > Alan also asked why the test mass for LIGO-1 was doubly-wedged, i.e. why > wasn't one face set perpendicular to the cylindrical axis? In fact the > DRD requires that one face only be wedged, whereas the design (FDD and > dwgs) have the BS, ITM and RM doubly-wedged. I found no justification in > the documentation for this 'violation' of requirements. I suspect that > the double wedge is very slightly better for balancing. Is there also a > fabrication ease argument? See the recent e-mail from Gari on this > > Dennis > From coyne@ligo.caltech.edu Mon Aug 28 13:16:17 2000 Date: Mon, 28 Aug 2000 13:35:47 -0700 From: Dennis Coyne X-Mailer: Mozilla 4.7 [en] (Win98; I) X-Accept-Language: en MIME-Version: 1.0 To: David Shoemaker CC: Stan Whitcomb , Bill Kells , GariLynn Billingsley , Alan Weinstein Subject: Re: TM aspect ratio References: <4.3.2.7.2.20000828120635.0361d400@127.0.0.1> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Status: X-Keywords: X-UID: 7 David, I doubt that internal resonances are likely to be an issue for the 40 m lab: LIGO-1 Small Optics Alan's proposed "big" 40 m optics: 3.0" dia x 1" thick 5.0" dia x 2.5" thick saddle 2 x 20.2 kHz 2 x 15.0 kHz 1st drumhead 28.6 20.6 in-plane comp. 2 x 37.4 2 x 22.1 Unless (in a perverse sort of way) we make them thin so as to get lower frequencies and contend with the narrow notch filters near the controller bandwidth -- more representative of LIGO-2? Dennis David Shoemaker wrote: > I think that getting high internal resonance frequencies is the thing that > would most likely have a tangible effect for the 40m in choosing an aspect > ratio. This eases servo design, and locking. From dhs@ligo.mit.edu Mon Aug 28 13:37:11 2000 X-Sender: dhs/18.120.0.82@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Mon, 28 Aug 2000 16:35:47 -0400 To: Dennis Coyne From: David Shoemaker Subject: Re: TM aspect ratio Cc: Stan Whitcomb , Bill Kells , GariLynn Billingsley , Alan Weinstein In-Reply-To: <39AACD23.87531D9B@ligo.caltech.edu> References: <4.3.2.7.2.20000828120635.0361d400@127.0.0.1> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 8 Dennis, I wonder if the BW required at the 40m might be 5x-10x greater to lock up the system, so the 'scaling' might have already been performed by the environment. Alan, do you have the velocities at LLO and Caltech to compare? d. At 16:35 00/08/28, Dennis Coyne wrote: >David, >I doubt that internal resonances are likely to be an issue for the 40 m >lab: > > LIGO-1 Small Optics Alan's proposed "big" 40 m optics: > 3.0" dia x 1" thick 5.0" dia x 2.5" thick >saddle 2 x 20.2 kHz 2 x 15.0 kHz >1st drumhead 28.6 20.6 >in-plane comp. 2 x 37.4 2 x 22.1 > >Unless (in a perverse sort of way) we make them thin so as to get lower >frequencies and contend with the narrow notch filters near the >controller bandwidth -- more representative of LIGO-2? > Dennis > >David Shoemaker wrote: > > > I think that getting high internal resonance frequencies is the thing that > > would most likely have a tangible effect for the 40m in choosing an aspect > > ratio. This eases servo design, and locking. > From kells@acrux.ligo.caltech.edu Mon Aug 28 14:16:07 2000 Date: Mon, 28 Aug 2000 14:30:24 -0700 (PDT) From: Bill Kells To: coyne@ligo.caltech.edu, dhs@ligo.mit.edu Subject: Re: TM aspect ratio Cc: stan@ligo.caltech.edu, kells@ligo.caltech.edu, gari@ligo.caltech.edu, ajw@hep206.cithep.caltech.edu X-Sun-Charset: US-ASCII X-Status: X-Keywords: X-UID: 4 > From dhs@ligo.mit.edu Mon Aug 28 13:35 PDT 2000 > X-Sender: dhs/18.120.0.82@127.0.0.1 > Date: Mon, 28 Aug 2000 16:35:47 -0400 > To: Dennis Coyne > From: David Shoemaker Hi David, Dennis, Certainly at the "old" 40m (recycling days, with CM servo feedback to the PSL), such much wider BW was the case. We seem to be succeeding at LIGO I (2k) with remarkably low BW (<200 Hz !) in the all feedback to TM scheme. Probably mostly thanks to the quiet environment improved SUS/seismic and low noise PSL. Not clear (to me) what the minimum REQUIRED BW would be for a new PSL'd 40m would be. > Dennis, > > I wonder if the BW required at the 40m might be 5x-10x greater to lock up > the system, so the 'scaling' might have already been performed by the > environment. Alan, do you have the velocities at LLO and Caltech to compare? > > d. > > > At 16:35 00/08/28, Dennis Coyne wrote: > >David, > >I doubt that internal resonances are likely to be an issue for the 40 m > >lab: > > > > LIGO-1 Small Optics Alan's proposed "big" 40 m optics: > > 3.0" dia x 1" thick 5.0" dia x 2.5" thick > >saddle 2 x 20.2 kHz 2 x 15.0 kHz > >1st drumhead 28.6 20.6 > >in-plane comp. 2 x 37.4 2 x 22.1 > > > >Unless (in a perverse sort of way) we make them thin so as to get lower > >frequencies and contend with the narrow notch filters near the > >controller bandwidth -- more representative of LIGO-2? > > Dennis > > > >David Shoemaker wrote: > > > > > I think that getting high internal resonance frequencies is the thing that > > > would most likely have a tangible effect for the 40m in choosing an aspect > > > ratio. This eases servo design, and locking. > > > > From ajw@hep204.cithep.caltech.edu Tue Aug 29 15:48:26 2000 Date: Tue, 29 Aug 2000 15:48:14 -0700 (PDT) From: Alan Weinstein To: stan@ligo.caltech.edu, kells@ligo.caltech.edu, gari@ligo.caltech.edu, dhs@ligo.mit.edu, coyne@ligo.caltech.edu, janeen@ligo.caltech.edu CC: ajw@hep.caltech.edu Subject: 40m test masses Hi all, Thank you all for your input on this burning question! They, of course, bring up more questions, so please bear with me. First off, some tentative conclusions: _______________ Size and aspect ratio: - I asked Gari and Janeen if it gets more difficult or more work to engineer optics and suspensions with different radii and aspect ratios. They both said that (a) if we go with 3x1 SOSs, all the engineering and drawings are done. (b) Anything else requires some engineering and drawings, but it didn't really matter to them what radius or aspect ratio was chosen. - If we accept that (a) maximizing the mass minimizes the loss due to attachments and thus maximizes the overall Q, minimizing thermal noise (b) LIGO I aspect ratios are desireable for some reason (c) no other factors are important, then we can go with 5x2, which has the same aspect ratio as the LIGO 10x4 COC, is a hefty 1.4 kg, and can be engineered by Gari and Janeen without too much pain. - But see below, where I struggle to understand your comments! _______________ Double wedges: I figure that the SOS optics should be identical to what was built for the LIGO-I IO, so when Gari gets drawings from Reitze, we'll just go with that. as for the four test masses: Gari says: > My recollection from the dark ages is that it was a requirement to > eliminate a "retro-reflective glint" which can happen between the front > surface and the 90 degree cylinder. Well, the ol' "retro-reflective glint". I don't know what this means for us. It remains unclear to me, whether the test masses should be double-wedged, and solicit your advice. _______________ NOW for my questions: >From david: > my recollection from the dark ages is that looking at mode frequency plots > it 'seemed like a good point of diminishing returns' to have 'about' the > ratio we ended up with. WHAT mode frequency plots? WHAT point of diminishing returns? Of what? FYI, I have used Kent's code to plot resonant frequencies and effective mass coefficients of test masses of various sizes (0 circumfrential nodes); see http://www.ligo.caltech.edu/~ajw/rfreq.pdf >from which I conclude that the differences between, say, 4x2, 4x3.5, 5x2, 5x2.5, are not significant. The 3x1 optic is different because it's so thin, I guess, so that I should be looking at modes with >=1 circumfrential node... > Another set of considerations for thickness is the absorption, scatter, and > inhomogeneity of the substrate, all pushing for thinner substrates. OK! > The thermal noise for LIGO I should be re-calculated using Geppo Cagnoli's > approach, and should be used for thermal noise calcs for the 40m if it is > an important measure of the performance. Can you point me to a reference for "Geppo Cagnoli's approach"? Doesn't he do suspension noise? I model suspension noise ala Gillespie&Raab; it seems to be under control... I model internal noise using Gillespie&Raab mode sum, Yu Levin's infinite-half-plane, and using Bondu's formulas (in bench.m). They all give around the same answer, for a given Q. So it all comes down to the Q. > I think that getting high internal resonance frequencies is the thing that > would most likely have a tangible effect for the 40m in choosing an aspect > ratio. This eases servo design, and locking. See http://www.ligo.caltech.edu/~ajw/rfreq.pdf. Resonant frequencies for all test masses under consideration are over 15 kHz. Surely the servo bandwidths are less than that, even at the 40m! >From Bill: > It has now been shown (from LIGO I commissioning) that the internal mode > resonances can significantly impact the servo operation (requiring detailed > filtering design), so it is probably worth factoring this into the decision. LIGO I test masses have resonances as low as 3kHz, right? Are the servo bandwidths as large as that? > Certainly at the "old" 40m (recycling days, with CM servo feedback to the PSL), > such much wider BW was the case. We seem to be succeeding at LIGO I (2k) with remarkably > low BW (<200 Hz !) in the all feedback to TM scheme. Probably mostly thanks to > the quiet environment improved SUS/seismic and low noise PSL. Not clear (to me) what the minimum REQUIRED > BW would be for a new PSL'd 40m would be. OK, we will install STACIS active seismic isolators at the 40m, in the coming months. This reduces the v_rms of the test masses by a factor of ~ 5-10. > There is also the nice, consise, though probably not universally complete, > paper by Bondu, et al. which at least indicates the basic trends. The Bondu,Hello,Vinet paper (PLA 246, 227 (1998)) report a weak (~ 10%) dependence on aspect ratio, only getting significantly worse when thickness < radius/2. We're not considering anything that extreme. They also ignore any dependence of Q on the aspect ratio. That paper also makes the statement "it seems clear that provided the ratio thickness/radius is about unity, there is a benefit in reducing as much as possible the size of the mirrors." But if you look at their fig 1, if you keep the ratio thickness/radius fixed (to, say, 4/5 as in LIGO I TMs), there is only very weak dependence on the size of the mirrors. And if, as we've argued, the losses are dominated by attachments and thus Q increases with mass, one comes to the opposite conclusion that they reached. >From David: > I wonder if the BW required at the 40m might be 5x-10x greater to lock up > the system, so the 'scaling' might have already been performed by the > environment. Alan, do you have the velocities at LLO and Caltech to compare? My crude calculations give v_rms ~ 0.1 um/s without STACIS and < 0.02 um/s with STACIS. I have no idea what the numbers might be for LLO, or even how they are calculated; I developed my own calculation, and have no idea how it compares to anything done by anyone else. (but I'd like to know, if you can point me to the right person). I can repeat my calculations with the nominal Hanford spectrum, and with LIGO-I-like stacks; and I get numbers that are very similar to 40m with stacis (that was some months ago; I can go back and check it if you think it's worthwhile). ____________ I would be grateful for any further wisdom you may have! AJW From dhs@ligo.mit.edu Tue Aug 29 16:49:07 2000 X-Sender: dhs/18.120.0.82@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Tue, 29 Aug 2000 19:48:30 -0400 To: Alan Weinstein From: David Shoemaker Subject: Re: 40m test masses In-Reply-To: <200008292248.PAA28880@hep204.cithep.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 2 At 18:48 00/08/29, you wrote: >Can you point me to a reference for "Geppo Cagnoli's approach"? >Doesn't he do suspension noise? yes, I just meant using bench -- that should cover both suspension and substrate thermal noise nicely if the scaling laws are not violated and approximations ok. d. From dhs@ligo.mit.edu Tue Aug 29 16:49:04 2000 X-Sender: dhs/18.120.0.82@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Tue, 29 Aug 2000 19:47:31 -0400 To: Alan Weinstein From: David Shoemaker Subject: Re: 40m test masses In-Reply-To: <200008292248.PAA28880@hep204.cithep.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 1 At 18:48 00/08/29, you wrote: > >From david: > > my recollection from the dark ages is that looking at mode frequency > plots > > it 'seemed like a good point of diminishing returns' to have 'about' the > > ratio we ended up with. > >WHAT mode frequency plots? >WHAT point of diminishing returns? Of what? Macmahon. integrated into the Gillespie stuff. point of diminishing returns: The lesson is one you mention several times in your email, that although 1:1 is close to ideal, the loss as you head toward thinner optics in thermal noise performance or resonant frequencies is slow, and the optical problems with the thicker mass and cost and delay in fab get better. where to stop making it thinner? 'about' where we have for LIGO I, I think; we will do about the same for LIGO II. d. From ajw@hep204.cithep.caltech.edu Tue Aug 29 17:48:27 2000 Date: Tue, 29 Aug 2000 17:48:19 -0700 (PDT) From: Alan Weinstein To: dhs@ligo.mit.edu CC: ajw@hep.caltech.edu In-reply-to: <4.3.2.7.2.20000829194746.02a4ead8@127.0.0.1> (message from David Shoemaker on Tue, 29 Aug 2000 19:48:30 -0400) Subject: Re: 40m test masses David, Thanks for the quick response. OK, I think we're converging on the aspect ratio stuff! As for thermal noise; good. I've adopted all the bench technology (bondu, thermoelastic, etc). But bench ignores violin noise, internal resonance peaks, etc. I try to do those a bit more correctly. Thanks! AJW From stan@ligo.caltech.edu Wed Aug 30 07:26:18 2000 X-Sender: stan@127.0.0.1 X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Wed, 30 Aug 2000 07:25:03 -0700 To: Alan Weinstein From: Stan Whitcomb Subject: Re: 40m test masses Cc: coyne@ligo.caltech.edu In-Reply-To: <200008292248.PAA28880@hep204.cithep.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 11 Alan, You have to bring this to some conclusion immediately! All of these people have other things to do than answer your emails. Just make a decision and get on with it! Dragging this out just delays the 40 m program, as well as drawing away resources >from other LIGO I and LIGO II activities. stan From Billingsley_G@ligo.caltech.edu Wed Aug 30 17:16:48 2000 X-Sender: gari@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Wed, 30 Aug 2000 17:12:27 -0700 To: ajw@hep.caltech.edu From: GariLynn Billingsley Subject: Weekly Report - Billingsley Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 2 Material specifications have been completed for the ITM, ETM, RM and BS. Material specifications and drawings exist for the three mode cleaner optics. Material drawings have been generated for the RM and BS. Dave Reitze has graciously offered 5 MCCM blanks to get us started in our 40 Meter acquisitions. This will help the schedule by several months. From Billingsley_G@ligo.caltech.edu Wed Aug 30 18:09:54 2000 X-Sender: gari@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Wed, 30 Aug 2000 18:07:55 -0700 To: coyne@ligo.caltech.edu, ajw@hep.caltech.edu From: GariLynn Billingsley Subject: 40M COC Specification and Drawing Review Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 3 I invite your comments and corrections on the specifications and drawings available through the following website. http://ligo.caltech.edu/~gari/40M I am ready to go out for a formal quote and place an order. I would appreciate your comments by Wednesday September 6. Thanks much, Gari Date: Thu, 31 Aug 2000 07:47:15 -0700 (PDT) From: Alan Weinstein To: Billingsley_G@ligo.caltech.edu CC: coyne@ligo.caltech.edu, ajw@hep.caltech.edu In-reply-to: <4.3.2.7.2.20000830180326.00b24ef0@acrux.ligo.caltech.edu> (message from GariLynn Billingsley on Wed, 30 Aug 2000 18:07:55 -0700) Subject: Re: 40M COC Specification and Drawing Review Garilynn, GREAT! I'll thank Reitze profusely. I will call you today to ask stupid questions like, are all those drawings and specifications more-or-less identical? OK, I see loosening of the absorption and homogeneity spec for the ETM and MCC. _____ I note that the RM and BS drawings have no wedge angles. Is this intentional? Why? The MC drawings have no wedge angles, either. If that's the way it is at LIGO-I, fine. I remain unsure about wedge angles for the test masses. I presume that we should go with double-wedged TM optics, since that's the result of deep thought at LIGO-I. _____ Stan says: > You have to bring this to some conclusion immediately! All of these people have > other things to do than answer your emails. Just make a decision and get on with it! OK! Janeen has no problem with engineering a scaled SOS for a 5"diameter by 2"thick test mass optic, and such a one seems to satisfy all my vague concerns, so let's go with it! _____ I will prepare a document which details: - justification for the optic sizes and aspect ratios - radii of curvatures and tolerances - polishing spec - preliminary coating specs _____ Regarding polishing, I don't know how to specify the low-spatial-frequency variation or the high-spatial-frequency variation without doing detailed FFT runs. I have asked Ken Ganezer to do these; maybe he can accomplish it before we send the optics out for polishing. _____ Regarding ROC tolerances: My simple modeling says that ROC tolerances for the RM, SM, BS, flat ITMs are all quite loose. The ROC tolerances for the ETMs are quite tight!: 57.375 +- 0.75 meters. Is this doable? Does this make sense to you? _____ THANKS! Talk to you later. AJW From Billingsley_G@ligo.caltech.edu Thu Aug 31 08:53:43 2000 X-Sender: gari@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Thu, 31 Aug 2000 08:53:12 -0700 To: Alan Weinstein From: GariLynn Billingsley Subject: Re: 40M COC Specification and Drawing Review In-Reply-To: <200008311447.HAA13404@hep215.cithep.caltech.edu> References: <4.3.2.7.2.20000830180326.00b24ef0@acrux.ligo.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 1 At 07:47 AM 8/31/00, you wrote: >Garilynn, > >GREAT! > >I'll thank Reitze profusely. > >I will call you today to ask stupid questions >like, are all those drawings and specifications more-or-less identical? >OK, I see loosening of the absorption and homogeneity spec >for the ETM and MCC. yup, more or less. The absorption, homogeneity, inclusion specs and drawing numbers change. everything else stays the same. >____ >I note that the RM and BS drawings have no wedge angles. >Is this intentional? Why? >The MC drawings have no wedge angles, either. >If that's the way it is at LIGO-I, fine. These are for blank material, the wedge will be put on at the polisher's where it is machined. >I remain unsure about wedge angles for the test masses. >I presume that we should go with double-wedged TM optics, >since that's the result of deep thought at LIGO-I. You won't need that information for a couple of months yet. We won't have the test mass material for a while. >_____ >Stan says: > > You have to bring this to some conclusion immediately! All of these > people have > > other things to do than answer your emails. Just make a decision and > get on with it! > >OK! Janeen has no problem with engineering a scaled SOS >for a 5"diameter by 2"thick test mass optic, >and such a one seems to satisfy all my vague concerns, >so let's go with it! > >_____ >I will prepare a document which details: >- justification for the optic sizes and aspect ratios >- radii of curvatures and tolerances >- polishing spec >- preliminary coating specs > >_____ >Regarding polishing, I don't know how to specify the >low-spatial-frequency variation or the >high-spatial-frequency variation >without doing detailed FFT runs. >I have asked Ken Ganezer to do these; >maybe he can accomplish it before we send the optics out for polishing. Perhaps I can help with this... You won't get a better super polish than <1 angstrom without a serious r&d program. You've got a fairly small beam, so if we ask for the same specs we had for LIGO 1 (which our polishers know how to meet) over the 25-30 ish mm central area we should be quite safe without having to spend any development money on technology or analysis... So I would suggest... Low - sigma (rms) with tilt, power and astigmatism removed < .8 nm rms over the central 30 mm High - sigma (rms) <1 angstrom >_____ >Regarding ROC tolerances: >My simple modeling says that ROC tolerances for the >RM, SM, BS, flat ITMs are all quite loose. > >The ROC tolerances for the ETMs are quite tight!: >57.375 +- 0.75 meters. yikes! I don't have much experience in the short ROC regime... that's ~40 - 100 nm tolerance in sag over the piece, attainable from a polishing standpoint, but I'm not sure how accurately one can measure something of that radius.. that's 8 fringes over 75mm (a lot!). We could specify radius over a smaller area, 30mm for instance, that's a sag of ~8 - 25 nm. That's about 2 fringes... possible, but I'd have to look into it. One approach is to get several spares and have them try a bunch (6 or so.. this material isn't that expensive) then measure them here and choose the best pair for coating. Clarification though... does the ETM ROC have to be as stated in an absolute sense or.. do the pieces just have to match to within .75 meters at roughly 57.3 M roc? That's a very important distinction, and one that can save you lots of time and money. I've asked in more detail about the delivery rate that GO could accomplish, it looks like it's closer to 2 a month, so we're better off than we were! From ajw@hep204.cithep.caltech.edu Thu Aug 31 11:46:19 2000 Date: Thu, 31 Aug 2000 11:46:09 -0700 (PDT) From: Alan Weinstein To: Billingsley_G@ligo.caltech.edu CC: ajw@hep.caltech.edu In-reply-to: <4.3.2.7.2.20000831082713.00adfc80@acrux.ligo.caltech.edu> (message from GariLynn Billingsley on Thu, 31 Aug 2000 08:53:12 -0700) Subject: Re: 40M COC Specification and Drawing Review X-Status: X-Keywords: X-UID: 2 Hi Gari, > >Regarding polishing, I don't know how to specify the > >low-spatial-frequency variation or the > >high-spatial-frequency variation > >without doing detailed FFT runs. > >I have asked Ken Ganezer to do these; > >maybe he can accomplish it before we send the optics out for polishing. > > Perhaps I can help with this... > You won't get a better super polish than <1 angstrom without a serious r&d > program. You've got a fairly small beam, so if we ask for the same specs > we had for LIGO 1 (which our polishers know how to meet) over the 25-30 ish > mm central area we should be quite safe without having to spend any > development money on technology or analysis... > > So I would suggest... > Low - sigma (rms) with tilt, power and astigmatism removed < .8 nm rms over > the central 30 mm > High - sigma (rms) <1 angstrom That sounds good to me. I was under the impression (from you and Dennis) that the <1 angstrom spec is the "super-polish" that requires significantly more time/money/effort than, say, <3 angstrom, so that we needed to ask ourselves whether we can't live with the latter. I gather that this spec is driven by scattering, the low-angle kind that stays in the beam. Our beams are smaller and thus have more divergence than in LIGO, but we also don't have output telescopes focusing it onto a PD. I think those two effects cancel at some level... If it's not a big deal, my gut tells me that if <1 angstrom is good enough for ligo, it's good enough for us (in terms of scattered light). > >_____ > >Regarding ROC tolerances: > >My simple modeling says that ROC tolerances for the > >RM, SM, BS, flat ITMs are all quite loose. > > > >The ROC tolerances for the ETMs are quite tight!: > >57.375 +- 0.75 meters. > > yikes! I don't have much experience in the short ROC regime... that's ~40 > - 100 nm tolerance in sag over the piece, attainable from a polishing > standpoint, but I'm not sure how accurately one can measure something of > that radius.. that's 8 fringes over 75mm (a lot!). We could specify radius > over a smaller area, 30mm for instance, that's a sag of ~8 - 25 nm. That's > about 2 fringes... possible, but I'd have to look into it. > One approach is to get several spares and have them try a bunch (6 or so.. > this material isn't that expensive) then measure them here and choose the > best pair for coating. > > Clarification though... does the ETM ROC have to be as stated in an > absolute sense or.. do the pieces just have to match to within .75 meters > at roughly 57.3 M roc? That's a very important distinction, and one that > can save you lots of time and money. Ah, you're right; they need only match between the two arms. > I've asked in more detail about the delivery rate that GO could accomplish, > it looks like it's closer to 2 a month, so we're better off than we were! GREAT! Thanks... AJW From Billingsley_G@ligo.caltech.edu Thu Aug 31 12:49:28 2000 X-Sender: gari@acrux.ligo.caltech.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Date: Thu, 31 Aug 2000 12:48:54 -0700 To: Alan Weinstein From: GariLynn Billingsley Subject: Re: 40M COC Specification and Drawing Review In-Reply-To: <200008311846.LAA27448@hep204.cithep.caltech.edu> References: <4.3.2.7.2.20000831082713.00adfc80@acrux.ligo.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Status: X-Keywords: X-UID: 5 At 11:46 AM 8/31/00, you wrote: >I was under the impression (from you and Dennis) that the ><1 angstrom spec is the "super-polish" that requires >significantly more time/money/effort than, say, <3 angstrom, >so that we needed to ask ourselves whether we can't live with the latter. > >I gather that this spec is driven by scattering, >the low-angle kind that stays in the beam. >Our beams are smaller and thus have more divergence than in LIGO, >but we also don't have output telescopes focusing it >onto a PD. I think those two effects cancel at some level... > >If it's not a big deal, my gut tells me that if ><1 angstrom is good enough for ligo, it's good enough for us >(in terms of scattered light). The low spatial frequency spec (microroughness) is driven by high angle scatter, which does not stay in the beam, it is a direct loss and scales as 1/rms^2 (and may be exactly lambda over the square of the rms microroughness, if I recall correctly) I think, as you say, it's not a big deal, so we'll go with that. Gari From ajw@hep204.cithep.caltech.edu Thu Aug 31 14:52:38 2000 Date: Thu, 31 Aug 2000 14:52:37 -0700 (PDT) From: Alan Weinstein To: janeen@ligo.caltech.edu CC: ajw@hep.caltech.edu Subject: 5x2 suspensions X-Status: X-Keywords: X-UID: 4 Hi Janeen, OK, 5x2. This means we have two types of suspensions at the 40m - existing design for the SOS suspensions (3x1 optics) - scaled up design for 5x2 optics. Now, the beam should lie on one horizontal plane, everywhere. (Maybe that's not necessary, but desireable). SO, can you engineer the 5x2 suspension to put the optic center at the same place as the SOS? OR, can we simply put the smaller suspension on a platform, to keep all optic centers on the same horizontal plane? Also: do you have a guesstimate of the footprint for a 5x2 suspension? Better, all 3 dimensions? I presume we want to keep the pendulum frequency the same (1 Hz) in all suspensions. Thanks! AJW