sPHENIX HCal discussion

["Lajoie, John G [PHYSA]" <lajoie AT iastate.edu>]

Dear HCAL'ers:

    Over the past two days there has been a significant off-list discussion of production tile testing, following up on a discussion we had at this Tuesday's HCAL meeting. I wanted to bring that discussion on-list so that everyone can be involved, and additional discussion can continue on-list.  

    To make things easy I have copied the email thread below, from oldest to most recent, so people can catch up easily.

Regards,
John Lajoie

From Edward Kistenev 6/16/2016 10:20AM

Hi, everyone,

in conclusions to my presentation of yesterday I said that I suggest to build advanced mapping station for the input control of the tiles which UNIPLAST will deliver. I would like it to be available at BNL end of August and tuned testing forward tiles for winter prototype. If it will be built in a way which allows to semiautonomous running (local trigger and data to the internet) then it may in the end be located at UNIPLAST to make sure the UNIPLAST is not sending junk over the ocean (note: I strongly believe that all our testing with LED light injected through the coating is meaningless. Tile reacts differently to the light attenuated by coating and that produced on inside). It is important not to design out UNIPLAST option, the actual decision on station location may depend on UNUIPLAST ability to comply with specs in the next production cycle.

The station will in some way repeat what is known as the Tile Mapper. Box with two 100x30cm2 smart trigger  tiles separated by ~60cm vertically to produce trigger and measure the muon trajectory and shelves with 8-pacs of tiles to be mapped in the separation gap. The station is to accommodate any tile shape what drives us towards solution with 8-pack of tiles and a back plane carrying spring loaded SiPM part of SiPM holder (modified to avoid problems with connecting/disconnecting self locking pins) and preamps. 

Each of the smart tiles has 8 embedded WLS fibers, muon signal is shared between three fibers closest to impact point. The coordinate lateral to fibers  is measured to better then ~1cm, I do not have really good estimate for position resolution along the fiber. Based on just amplitude measurements it is unlikely to be better then 5cm (signal is generated by firing of 20 discreet pixels). I expect better resolution if measurements are done in timing domain but still - 1cm resolution would need timing resolution ~50ps what is unlikely. Anyway … one never knows unless tried. Currently I am working on E864 module data from our exposure at FNAL. Rough estimate - 500ps. I also asked Mike Lenz to help to build a simple prototype with 1m long fiber stretched between two SiPM’s. Fiber will be exited by light from small laser available in EMC lab. 

The total number of SiPM channels in such station will be 
			32 (smart tiles)  + nx8 (n is the number of 8packs) = 40 - 80

I suggest to use DRS readout everywhere in this setup. Existing user interface allows rather complicated trigger logic even with small 4channels evaluation modules we use. I know nothing of CAEN version but hopefully it is of similar intelligence.

We discussed this project with Steve, there could be options with copying outputs to HBD and DRS digitizers but after some thinking I decided that this will be unnecessary complication. 

This project was in the cloud for nearly a year but there were always other things to do. To become reality it still needs your endorsement, a bit of mechanical design help, a good will on part of Steve (and some of his time) and someone who will be interested in helping with all related programming and analysis.

				Edward

PS. This is the project #1. The Project #2 is the stand to rotate existing prototype around its horizontal axis (to simulate different orientations in calorimeter). That one is still waiting  for final pictures from Murat’s simulation. Subject for one more mail.

>From John Lajoie 6/16/2016 12:25PM

Hi Edward,

    Perhaps you and I were at different meetings, or we're thinking of different things?  What I recall was that after your talk I again raised the idea of testing tiles before they were wrapped and after they were coated, which was not covered in your talk.  What was discussed was a bit simpler and more straightforward than what you describe. What I had in mind was a simple light-tight box with a single SiPM, a fixturing jig, and an array of UV LEDs. The readout could be very simple.

    Before you complain about the LEDs again, it is clear that over the past several months the Colorado group has established a very clear correlation between cosmics performance (light yield) and and LEDs and they have shown this in multiple HCAL meetings.  For this reason LED's should be sufficient and a very fast way to qualify the tiles. I do like your idea of having the data automatically uploaded somewhere it could be viewed easily by everyone.

    As for timescale, I thought that at the last HCAL meeting you and I had explicitly agreed that something like this LED tile tester should be developed for the full production, it's probably too late to bring this online for the next prototype tile production.  I had suggested that ISU/Colorado or a combination of the two could develop such a box over the coming months.

Regards,
John

>From Edward Kistenev 6/16/2016 12:36PM

Yes, 

response to cosmic and external LED light injection are correlated. But the unwrapped tile response to external LED in case of coating locally damaged will change  while its response to muons when wrapped may not be visibly affected at all (foil reflects the light exiting through coating damage on scintillator  back into the tile). On the other hand - LED will not see local tile discolorations while muon map will fail template test.

To me mapping wrapped tiles with cosmic (JH proposal was to take all tiles to test beam at FNAL and …) is the only way to produce a quantitative estimator which will serve multiple purposes - 

(1) tile rejection if map disagrees with template; and (2) SiPM-to-tile-to-tower matching. Such station may easily incorporate light leak test too. This is not a small job.  If we want it done before tiles are shipped - then data collection must be done at UNIPLAST with analysis and conclusions anywhere else in the world (Colorado/IOWA/BNL …). Then tiles are  shipped  only if approved.  Single point cosmic measurement will be enough to verify matching before tile is installed into detector.

One station production rate (two Outer or 4 Inner loads/day ) will be sufficient for the whole project.

If better solution exists Iet’s discuss it.

Edward 

>From John Lajoie 6/17/2016 8:02AM

Hi Edward,

    As I see it we're really talking about two different things.

    In a mass production like this I really believe you need a QC step at every stage.  The LED testing box I have described will test the light yield and uniformity after the coating is applied before it is wrapped, all with a common SiPM. Ideally we would build this box and provide it to Uniplast, and this step would take less than a minute per tile, providing a first look with quantitative data at the tile in the production line.

    What you are discussing is a cosmics check of the wrapped, assembled tile - also a key, important step! However, as the tiles are assembled at this point I don't see any point in doing this test at Uniplast.  I think that the assembled tiles should be shipped to BNL and tested in a cosmics station as you describe, *with the unqiue SiPM that will be paired with that tile*.   At this point we will be testing/characterizing the tile and SiPM as a *system* and the cosmics MIP peak for a given tile+PM is the beginning of the calibration of the HCAL.  Doing this with a common SiPM would be a wasted step, in my opinion, as it would need to be repeated with the SiPM that will be paired with a given tile.

    So, I think I agree with the cosmics testing/characterization that you describe, but I don't favor doing it at Uniplast for two reasons:

-> The tiles should be tested with their paired SiPM and I don't favor shipping the SiPM's to Uniplast; the less handling the better.

-> I view what you are describing as a part of a final acceptance check of the tiles.  If all the other QC checks in the system are working properly (including the LED test) then any problems that crop up at this stage will not be straightforward, and it will require sPHENIX physicists to evaluate the impact.  I don't think we want this final step/decision made by Uniplast.

John   

>From Edward Kistenev 6/17/2016 8:20AM

There is no purpose in mapping tile with unique SiPM assignment - mapping is to a very large extent equivalent to calibrating tile light yield from a given tile and it is convenient to do it with unique set of welll known SiPM’s totally decoupling SiPM installation from testing and characterizing tiles. SiPM’s are delivered with rather precise parameter definitions made by Hamamatsu - we are unlikely to compete in that field. Verifying stability of SiPM gain in the test station is simple - noise run once a day. Maintaining it stable requires temperature control (argument against locating mapper oversees). There is a bit of uncertainty in optical coupling between two coupler components - hope we took care of it in the current design of the coupler. As a safety - we’ll need final preinstallation measurement done  on a tile with selected SiPM and wide acceptance cosmic trigger.  

Decopling tile mapping and SiPM association means that we may delay SiPM purchase till late in the project and still be certain that tiles are of really good quality. Locating mapper anywhere around or at UNIPLAST is just a matter of manpower - if we have it let’s do it in US otherwise let UNIPLAST take data, they will be available for analysis even before the file is closed. 

Edward

>From John Haggerty 6/17/2016 8:23AM

John and Edward,

John's right, there are several questions here, and probably testing has 
to occur in several phases.  To break it into pieces by manufacturing 
step, we would like to:

Q- verify that the raw extruded scintillator is "good quality," i.e., 
its light output and attenuation length and surface quality are all 
about the same.

A I would think Uniplast already tests this, somehow, maybe by visual 
inspection, but I think Anna said that they had some kind of simple test 
station.  We should find out.

Q- verify that the grooved tile with the fiber glued in is "uniform." 
We still have not quantified how uniform is uniform enough.  1% 
variation? 10%?  25%  100%?  Indeed, maybe we don't even care about the 
absolute scale of uniformity as long as the tiles are all "the same." 
(But again, to what precision? 10%? 50%?)

A This is what an array of LED's might do quickly and efficiently; I 
imagined that we'd illuminate the tile maybe a few tens of locations and 
image the fibers--could be done with pulsed LED's and SiPM's, but I 
would think you could get away with a much simpler thing to measure the 
light output from the fiber (i.e., the moral equivalent of an iPhone 
camera).

Q- verify that the wrapped, dipped, tile is good.

A This is Edward's complicated cosmic test... I don't think that's for 
Uniplast, that's for either a local physics institution or one of our 
other collaborators.   I think we here also have to determine what is 
acceptable.

In the end, the result of this is some kind of yield, which will come 
into the cost; if we reject 90% of the tiles as unacceptable when 
complete, Uniplast would have to charge us more unless they want to lose 
money.  So it's in everybody's interest to keep the quality high at the 
start of production.

John Lajoie

Professor of Physics

Iowa State University

 

(515) 294-6952

lajoie AT iastate.edu

Contact me: john.lajoie