[Halld-pid-upgrade] GlueX DIRC internal review
Justin Stevens
jrsteven at mit.edu
Thu Sep 10 08:46:42 EDT 2015
Hi Joe,
Thanks for the clarification. It sounds like what Mike proposed in terms of a table explicitly listing the magnitude of the various contributions to the SPR is consistent with what you're describing to answer some of the questions that "old DIRCies always ask". We can discuss it more at tomorrow's meeting.
-Justin
On Sep 9, 2015, at 3:14 PM, Jochen Schwiening wrote:
> Hi Justin,
>
> what we mean with the breakdown is the composition of the single photon Cherenkov angle resolution (SPR), which is part of the Cherenkov angle resolution per track via
>
> sigma_tot^2 = sigma_corr^2 + (sigma_SPR/sqrt(N_pe))^2
>
> You start out in the first section with one of the contributions to sigma_corr, the correlated term that does not scale with the number of photoelectrons (N_pe). The tracking resolution is an important part and you provide a figure for it. In addition there's multiple scattering inside of the bar, which will cause sigma_tot to have significant non-Gaussian tails, etc, and you argue that the second term, sigma_SPR/sqrt(N_pe) has to be 2.1mrad or less.
> sigma_SPR has contributions from terms like chromatic dispersion (about 5.0-5.5mrad for a standard photocathode in combination with the wavelength cut from Epotek), from bar imperfections smearing the photon angles, from the pixel size, from the focusing errors, alignment, etc.
> You address some of these in the text in several places (3mrad from the pixel size for MaPMTs, 3mrad smearing in simulation to account for the bar non-squareness, ).
> If I take the estimate of 9.1mrad and subtract in quadrature the 5.5,3.0,3.0,1.0 then I am left with almost 6mrad unexplained resolution contribution. Some of that will be mirror aberrations, etc and some will be the kaleidoscopic effect. But we should be able to account for these terms to justify, for instance, why a 3mrad pixel size is important for us. Why don't we lower the cost by combining pixels in one direction as the SuperB FDIRC planned (and we also consider for PANDA)? DIRCs are in principle simple enough that a pretty good estimate of sigma_SPR can be obtained from a back of the envelope calculation, using the stand-off distance, pixel size, refractive index ratio in the case of the pixels, etc. I assume that this is how you determined the 3mrad for the pixel contribution.
> This is the usual type of question that old DIRCies like Blair and Jerry (and I) always ask in DIRC discussions, which is why I brought it up here. I find that to be an interesting demonstration of how well a system is understood and how much attention needs to paid to each source of angular uncertainty.
>
> I hope this is more clear, but please let me know if you have questions.
>
> In terms of prototyping, I was wondering if you want to discuss any planned tests with particle beams or cosmic rays to verify the design. This is probably not a huge deal since the principle is well-known and the SLAC FDIRC is such a useful proof-of-principle demonstrator that we will probably get the TDR approved without the usual requirement of a successful prototype test with particle beams. But if you could point to future plans on the way to the full system, even the text as you have it in your mail, could be helpful.
>
> Cheers,
> Joe
>
> On 9/9/2015 8:49 PM, Justin Stevens wrote:
>> Hi Mike, Joe, Carsten,
>>
>> For the first point below, I initially thought this suggestion was to first have some discussion of the known characteristics of the BaBar DIRC bar boxes to provide some context for the ~9 mrad resolution described in the text and compared to the "SLAC FDIRC mode" simulation, which is basically the starting point that we build from for our design. This information is in the SLAC papers referenced in the TDR, but maybe it would be good to explicitly discuss it here. However, now that I reread the comment, I'm not certain what exactly is meant by the "correlated terms" here. Joe or Carsten, can you clarify what you were thinking of for this breakdown of contributions to the Cherenkov angle resolution?
>>
>> Now for the prototype planning. The schedule you outline below follows what we've loosely talked about before in terms of in-situ testing of a single focusing box, to avoid the considerable cost (both time and $$) of a dedicated test setup and get experience early with the FDIRC in GlueX. In terms of schedule, I think our best shot at a major installation like this is during a summer shutdown, which following Mike's timeline brings the question whether we could be ready for this by next summer? In addition to "Box 1" from MIT/Bates this would require: a) transporting the bar boxes from SLAC to JLab, b) having a support structure built for at least 2 bar boxes and c) having a complete electronics/readout chain installed and integrated with the DAQ (even if there is only a fairly small number of MaPMTs available).
>>
>> -Justin
>>
>> On Sep 9, 2015, at 9:17 AM, Michael Williams wrote:
>>
>>>> *Add a breakdown of the correlated and single photon resolution terms (like the tables in the SLAC papers and our PAC42 proposal), which Jerry will want to see.
>>>
>>> Personally, I don't really see the point in separating these contributions. We have no control over the chromatic dispersion, the PMT pixel size, the quality of the bars, the presence of the old bottom inclined surface, etc. We ran our simulation in "SLAC FDIRC mode" and we get the same per-photon resolution as SLAC does, so we are confident that we are simulating things properly, and I don't see the point of not always have all effects we can't turn off in "on" mode in the simulation.
>>>
>>> What we could easily do is turn some of them off/on and work out roughly what each contributes to the total resolution. What I don't want to do is waste time trying to properly document how each one is correlated with the others. One way to do this would be to simply make a table with sigma_tot and then sigma_(tot-x) for each effect, where sigma_tot is the total per-photon uncertainty and sigma_(tot-x) is the total uncertainty in the absence of effect x. E.g., provide the resolution with no pixelation, with monochromatic light, etc. These are mostly easy.
>>>
>>>> *Add some discussion of plans for measurements with prototypes, if any. We haven't talked much about this other than the prototype focusing box from MIT. It's worth thinking through how far we want to push the prototype and what we can learn from it in time to modify the final detector.
>>>
>>> I talked with the Bates engineers about this yesterday. If our baseline is the two boxes, then each is the same size as our planned prototype. So, we think that the following is a sensible plan:
>>>
>>> * Build "Box 1" with a goal of having it complete by spring 2016. We can survey the alignment as best we can mechanically. We can then do some preliminary studies of the optics using lasers, LEDs, etc, at Bates, just to make sure things are close to OK. Also, we can fill the box and test for leaks, etc. By summer 2016, we can have Box 1 ready to take to JLab.
>>>
>>> * Install Box 1 at JLab, attach it to 2 bar boxes, instrument (at least) some of it. Do an in situ calibration/testing using GlueX data. We think the trickiest parts are hooking up to the bar boxes, and validating alignment at the <= 1mrad level and photon yield when the box is filled . It's not obvious how to do either of these things without a huge effort any way other than using GlueX itself. The goal would be that Box 1 is sufficient and will remain as part of the final FDIRC.
>>>
>>> * We should with very little data learn what we need to know about Box 1. Based on that, we can choose to make minor changes to the design of Box 2 if need be. There is no reason Box 1 and Box 2 need to be identical. If Box 1 is truly deficient, we can choose to uninstall it, and then augment it as needed during GlueX downtime. The boxes should be robust, so unless we have some catastrophic problem, it would be quicker and cheaper to simply augment Box 1; however, building a Box 3 would also be an option (and not very expensive if there is no problem with Box 1's mirrors). That said, the goal is for Box 1 to be part of the final detector.
>>>
>>> Thoughts?
>>>
>>> Mike
>>>
>>>>
>>>> -Justin
>>>>
>>>> On Sep 8, 2015, at 9:53 PM, Curtis A. Meyer wrote:
>>>>
>>>>> I don’t think that the suggestions are 100% compatible suggestions, but my take is the following:
>>>>>
>>>>> 1) in the introduction, add a bit more physics justification for the DIRC.
>>>>>
>>>>> 2) Discuss the limit of the current PID systems and at least have material
>>>>> on why we cannot do a “cheap” TOF upgrade to accomplish the same
>>>>> results.
>>>>>
>>>>> We have a really nice plot of PID in the TOF now, and we are
>>>>> basically at design specs. It is pretty obvious that kaons die off by
>>>>> 2 GeV/c and I doubt that anyone would argue that we could push this
>>>>> to 4.5 GeV/c so Mike’s comment on a paragraph, with the figure,
>>>>> probably covers this.
>>>>>
>>>>> 3) Some rearrangement of the existing material to smooth out flow. I
>>>>> don’t have a strong opinion on this.
>>>>>
>>>>> Probably also worth noting is that Jerry will probably be one of the
>>>>> most careful reviewers who questions a lot, but also very supportive.
>>>>> He wants to see the DIRC continue to do good physics.
>>>>>
>>>>> Curtis
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> ---------
>>>>> Curtis A. Meyer MCS Associate Dean for Faculty and Graduate Affairs
>>>>> Wean: (412) 268-2745 Professor of Physics
>>>>> Doherty: (412) 268-3090 Carnegie Mellon University
>>>>> Fax: (412) 681-0648 Pittsburgh, PA 15213
>>>>> curtis.meyer at cmu.edu http://www.curtismeyer.com/
>>>>>
>>>>>
>>>>>
>>>>>> On Sep 8, 2015, at 9:18 PM, Michael Williams <mwill at mit.edu> wrote:
>>>>>>
>>>>>> Today was registration day at MIT, and we had a "welcome" for the new grad students so I didn't have time go thru them all in detail yet.
>>>>>>
>>>>>>> Thanks for forwarding all the reviewer comments on the TDR. I felt that a number of
>>>>>>> the broader issues such as physics impact and expected/current baseline PID
>>>>>>> performance would be addressed in the first presentation at the review. However,
>>>>>>> we may want to revisit including some more of this in the TDR.
>>>>>>
>>>>>> I think for the TOF, we can just add a single paragraph on this. For the other stuff, I'll need to look at it a bit more.
>>>>>>
>>>>>>> Do we want to wait until Friday to discuss our response? I suspect that the simple
>>>>>>> things can just be fixed, but there are a couple of bigger changes that we should
>>>>>>> probably discuss (perhaps by email)
>>>>>>
>>>>>> If somebody (Justin?) wants to go thru and just fix minor typos, etc, that's fine with me. I'm also OK with discussing the bigger things via email in advance of Friday's meeting. It would probably help if we each at least sent some thoughts on it, so that we've each thought about it properly prior to Friday's meeting.
>>>>>>
>>>>>> M
>>>>>>
>>>>>>>
>>>>>>> Curtis
>>>>>>> -------
>>>>>>> Prof. Curtis A. Meyer
>>>>>>> Department of Physics
>>>>>>> Phone: (412) 268-2745
>>>>>>> Carnegie Mellon University
>>>>>>> Fax: (412) 681-0648
>>>>>>> Pittsburgh, PA 15213
>>>>>>> curtis.meyer at cmu.edu http://www.curtismeyer.com/
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>> On Sep 8, 2015, at 1:19 PM, Justin Stevens <jrsteven at mit.edu> wrote:
>>>>>>>>
>>>>>>>> Hi All,
>>>>>>>>
>>>>>>>> Attached are Dave's comments on the TDR.
>>>>>>>>
>>>>>>>> -Justin
>>>>>>>>
>>>>>>>>
>>>>>>>> On Sep 8, 2015, at 11:21 AM, Justin Stevens wrote:
>>>>>>>>
>>>>>>>>> Hi All,
>>>>>>>>>
>>>>>>>>> Below are comments on the TDR from Joe and Carsten. Please take a look at these and contact them if you have any questions about their comments. We'll discuss these more on Friday. We should have the remaining comments from Dave Mack later today.
>>>>>>>>>
>>>>>>>>> Thanks to Joe and Carsten for taking the time to review the TDR!
>>>>>>>>> Justin
>>>>>>>>>
>>>>>>>>> Begin forwarded message:
>>>>>>>>>
>>>>>>>>>> From: Jochen Schwiening <J.Schwiening at gsi.de>
>>>>>>>>>> Subject: Re: GlueX DIRC internal review
>>>>>>>>>> Date: September 8, 2015 7:39:14 AM EDT
>>>>>>>>>> To: Justin Stevens <jrsteven at jlab.org>
>>>>>>>>>> Cc: Carsten Schwarz <c.schwarz at gsi.de>
>>>>>>>>>>
>>>>>>>>>> Dear Justin,
>>>>>>>>>>
>>>>>>>>>> Carsten and I read the TDR draft and would like to congratulate you and the other authors on a well-written manuscript that, we think, is very close to ready for the DOE review. The text reads well, seems to be complete and the case for the DIRC upgrade is quite convincing to us.
>>>>>>>>>>
>>>>>>>>>> We have have a couple of suggestions and comments for your consideration to perhaps further improve the draft.
>>>>>>>>>>
>>>>>>>>>> We would like to suggest that you add a subsection at the start of the TDR (perhaps in section II Requirements) to motivate the 3 s.d. pi/K separation coverage to 4 GeV/c explicitly.
>>>>>>>>>> My experience is that any reviewer is grateful if you summarize the main argument in favor of spending a significant amount of money on a new system right up front instead of pointing to the PAC information in the references. A couple of benchmark channels that highlight the additional physics reach would be very helpful. I think that this is the plan for the review presentation agenda anyway, so this seems a natural and useful addition. A nice eye candy figure would help sell the FDIRC idea.
>>>>>>>>>>
>>>>>>>>>> The central point of the TDR is the design of the expansion volume since it is the main new development. This could be emphasized by changing the sequence, for instance, to
>>>>>>>>>>
>>>>>>>>>> Introduction
>>>>>>>>>> FDIRC Design
>>>>>>>>>> Focusing Box Design <---
>>>>>>>>>> Photon Detection and Readout
>>>>>>>>>> Integration and Installation
>>>>>>>>>> Prototype Program
>>>>>>>>>>
>>>>>>>>>> We missed a technical drawing of the focusing box with dimensions and angles in the first part of the TDR. We feel that a larger version of Fig. 41 should appear somewhere around page 10 (in that new "Focusing Box Design section) and that the baseline design of the focusing box should be described in detail there.
>>>>>>>>>>
>>>>>>>>>> We would propose to move the discussion of the tracking resolution (last para in II, Fig 2) to a section where the expected/required DIRC resolution is discussed in one place. The draft would profit from an explicit discussion of the expected Cherenkov angle resolution for these BaBar DIRC bar boxes. A breakdown of the correlated and the single photon resolution terms with the contributions from the bar imperfections, chromatic dispersion, pixel size, and focusing uncertainties (incl. kaleidoscopic effects) would help to motivate the mirror design and should precede the estimate of the correlated terms. Jerry will want to see this breakdown to understand which effects dominate and how much effort, for instance, needs to be put into the mirror design/alignment.
>>>>>>>>>>
>>>>>>>>>> We propose to rephrase the design description in terms of "baseline design" and "design options" to make room for possible improvements to the design as a result of the ongoing simulation work.
>>>>>>>>>> Mention either the single focusing box or the two focusing boxes as baseline design, the other one and the 4 box solution as options that are being studied with the goal of cost/performance optimization.
>>>>>>>>>> This also goes for the mirror layout, where the various offsets and angles could be described as options.
>>>>>>>>>> This structure would help to keep a clear line of arguments in the TDR without too much clutter from technical details.
>>>>>>>>>>
>>>>>>>>>> The performance example in Fig 11 should be shown for a relevant momentum, 3.5 or 4 GeV/c, not 5 GeV/c, if possible. This figure is sort of a figure of merit for the design and looks less than impressive due to the unfortunate choice of momentum.
>>>>>>>>>>
>>>>>>>>>> We are missing a discussion of the plans for measurements with the prototype, if any. Do you plan to verify the design with a single-box prototype in test beams or cosmic rays? Which optical tests are planned, which resolutions or photon yields can you measure? Where can you profit from prototype tests for CLAS12 RICH, to demonstrate the photon yield and timing properties of the sensor and readout chain?
>>>>>>>>>>
>>>>>>>>>> Some other minor comments:
>>>>>>>>>>
>>>>>>>>>> General: we would prefer GeV/c as unit for momentum instead of GeV
>>>>>>>>>> Fig 1: is there a version of the drawing with the DIRC in place?
>>>>>>>>>> Fig 3: it would be useful to show Fig 20 from the BaBar DIRC NIM paper (Schematics of a DIRC radiator bar in side and top view.) here as well to establish the make-up of a bar, define the dimensions and angles.
>>>>>>>>>> Page 6: the nitrogen flow is not really to preserve angles but to prevent surface pollution and photon loss.
>>>>>>>>>> Fig 4: it is more customary to use momentum as unit for a and b, c is missing the axis label and the information on the number of particles used and the caption for b is not correct.
>>>>>>>>>> Page 7: the focusing is not really directly related to the ability to mitigate the effects of chromatic dispersion in an FDIRC, that is more of a fast timing issue.
>>>>>>>>>> Page 8, Fig 9: "local arrival time" is an unfamiliar term for me, maybe "photon time of propagation (TOP)" would be clearer?
>>>>>>>>>> Page 10: do you mean the photon detection efficiency instead of quantum efficiency? I assume you apply a weight based on the combined QE, collection efficiency, geometric efficiency/packing fraction, etc?
>>>>>>>>>> Page 11, Fig 10: The kaleidoscopic effect is not very visible in this figure.
>>>>>>>>>> I think we have nicer examples from Roman's G4 sim now and/or maybe a zoom insert would be helpful and a brief explanation of the effect. Jerry is always very interested in a proper discussion of this aspect.
>>>>>>>>>> Page 12: the BaBar DIRC geometric reco method does not store Cherenkov angles but the photon propagation vector "kBar" in the look-up tables.
>>>>>>>>>> You could refer to the time-based imaging methods here that are being studied for the Belle II TOP and the PANDA Barrel DIRC.
>>>>>>>>>> Fig 11: please explain the term "ROC".
>>>>>>>>>> Figs 14, 16, 17, 19 are very (too) small.
>>>>>>>>>> Page 13, Fig 15: how is the PMT are defined? Do you require a certain percentage of photons to be detected for each radius value?
>>>>>>>>>> Page 13, 14: A focusing mirror is usually better than the approximation via a segment mirror. The reason why the three-segment mirror performs
>>>>>>>>>> better for some angles should be discussed in the text.
>>>>>>>>>> Page 14, Fig 18: we were a little confused by the use of the word "dispersion" when it comes to the mirror resolution. Do you mean image distortion or imaging errors?
>>>>>>>>>> Page 16: mention that the difference between measured and expected photon propagation time can be used to reject EM background photons (as was done in BaBar).
>>>>>>>>>> Fig 21: can you simply rotate the figure to agree with new bar box orientation?
>>>>>>>>>> Page 19: did you consider PHOTONIS Planacon MCP-PMTs? Those are definitely ready for use and have many attractive features (but a high unit cost). Jerry will probably ask why not if you don't mention it here.
>>>>>>>>>> To me the main improvement between H8500 and H12700 is not so much the QE but the clearer separation of the single p.e. peak from the pedestal. At least one should mention that.
>>>>>>>>>> Fig 23 and 31 are not mentioned in the text, please add the reference.
>>>>>>>>>> Page 21: do you really mean an occupancy of 0.7% per MaPMT, not per pixel?
>>>>>>>>>> Page 24: add "Mechanical Design, ..." to the title?
>>>>>>>>>> Page 25: It would be nice to add a few words and a drawing about the mechanical design and support of the photon camera/focusing box here. More detail would be useful.
>>>>>>>>>> Fig 38: from the text it is not clear to me what significance the color of the round image blob has and if the diameter of the blob is a function of the imaging properties or of the setup (mirror radius and distances).
>>>>>>>>>> Fig 39: do you want to comment on the poor signal to background ratio obtained for the 345nm LED compared to the blue LED? Also, I am not a fan of a zero-suppressed y axis, which hides such problems.
>>>>>>>>>> Page 31: the discussion of the box design, as far as it concerns the GlueX design and not the prototype design, should come earlier.
>>>>>>>>>> Page 32: I think the box will be mounted vertically, no?
>>>>>>>>>> Page 35: do you really want to supply continuous N2 flow? It seems unnecessarily complicated, a double-bag with N2 fill should be sufficient, no?
>>>>>>>>>> We agree that additional drawings and photos should be added since the transport of the precious bar boxes will be a major concern for DOE and for Jerry.
>>>>>>>>>>
>>>>>>>>>> Please don't hesitate to contact us if you have any questions about our comments.
>>>>>>>>>>
>>>>>>>>>> Best regards,
>>>>>>>>>>
>>>>>>>>>> Carsten and Joe
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Halld-pid-upgrade mailing list
>>>>>>>>> Halld-pid-upgrade at jlab.org
>>>>>>>>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>>>>>>>>
>>>>>>>> <FDIRC_TDRmackcomments.pdf>_______________________________________________
>>>>>>>> Halld-pid-upgrade mailing list
>>>>>>>> Halld-pid-upgrade at jlab.org
>>>>>>>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Halld-pid-upgrade mailing list
>>>>>>> Halld-pid-upgrade at jlab.org
>>>>>>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Halld-pid-upgrade mailing list
>>>>> Halld-pid-upgrade at jlab.org
>>>>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>>>>
>>>
>>>
>>> _______________________________________________
>>> Halld-pid-upgrade mailing list
>>> Halld-pid-upgrade at jlab.org
>>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>>
>>
>> _______________________________________________
>> Halld-pid-upgrade mailing list
>> Halld-pid-upgrade at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
>>
> _______________________________________________
> Halld-pid-upgrade mailing list
> Halld-pid-upgrade at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-pid-upgrade
More information about the Halld-pid-upgrade
mailing list