Solid calorimeter readout
Contents
file dir
on dropbox
https://www.dropbox.com/sh/365le2o93vpn14e/yM3akCdLSK/readout
Summary
energy and photon
SPD PS S min energy depo from MIP(MeV) 0.7 3 30 min num of photons from MIP 35 150 1250 peak energy depo from MIP(MeV) 0.8 4 65 peak num of photons from MIP 40 200 2700 energy depo in scint from 7GeV e- (MeV) 0.8 200 1e3 energy depo in whole from 7GeV e- (MeV) 0.8 200 7e3 peak num of photons energy from 7GeV e- 40 1e4 1e5 dynamic range 2 70 35
- note
- 1. SPD is 5mm, PS is 2cm
- 2. In S, a MIP is about 300MeV including energy deposition in scintilator and lead, about 65MeV including energy deposition in scintillator only. In PS, a MIP is about 4MeV including energy deposition in scintillator.
- 3. "number of photons" is WLS output. only the entry "peak num of photons from MIP for PS" is from test,all others are from derivation assuming it's linear with energy deposition in scintillators. refer to Fig 98(b) and Fig105 from SoLID pCDR
- 4. Xiaochao confirms LHCb PS has MIP signal of 22 photoelectron from 1.5cm thick scintillator with a 15% QE PMT, so the number of photon for SoLID PS with 2cm thick scintillator and same fiber embedding is 200=(22/0.15)*(2/1.5)
- 5. dynamic range is calculated by "energy depo in scint from 7GeV e- (MeV)" over "min energy depo from MIP(MeV)"
signal
unit in uA
SPD PS S min current from one MIP 16.8 72 600 max current from one 7GeV e- 19.2 4.8e3 4.8e4 average current from MIP in PVDIS EC high inner 0.4 1.7 23.3 average current from MIP in PVDIS EC low inner 0.4 1.7 23.3 average current from MIP in PVDIS EC high outer 0.0144 0.072 0.972 average current from MIP in PVDIS EC low outer 0.0144 0.072 0.972 average current from MIP in SIDIS EC FA inner 0.0161 0.081 1.089 average current from MIP in SIDIS EC FA outer 0.0001 0.0004 0.008 average current from MIP in SIDIS EC LA inner 0.0553 0.2764 3.724 average current from MIP in SIDIS EC LA outer 0.0046 0.0232 0.311
- note
- signal is calculated based on number of photons (see table above)
- assume PMT with 15% QE and gain 3e5
- min and max current from one particle hit is pulse height based on 30ns pulse width, triangle pulse shape. For example, 200*0.15*3e5*1.6e-19/30e-9*2/1e-6=96uA for 200 photons
- average current from MIP is total charge per second based on MIP particle rate (see table below). For example, 200*0.15*3e5*1.6e-19*40e3/1e-6=0.0576uA for 200 photons and 40kHz rate.
- MIP particle rate is pi+ and pi- rate
rate
unit in kHz/100cm2
PVDIS high inner high outer low inner low outer pi+ 400 20 400 20 pi- 800 30 800 30 gamma(pi0) 0 0 0 0
SIDIS FA inner FA outer LA inner LA outer pi+ 36 0.4 96 8 pi- 36 0.4 96 8 gamma(pi0) 12 4 60 12
- note
- rate on EC
- pi+ and pi- rate includes secondary particles
- gamma(pi0) only include pi0 directly from target, no target cell wall and windows included
- PVDIS result is based on code "eicRate" with no further correction factor
- SIDIS result is based on code "eicRate" and divide by 2.5 correction factor
PMT
PMT model size price($) note H10966A 3200(small) MAPMT, possible for SPD and PS R3998-02 28mmD 400(bulk) additional socket C10344-03 $240(small),used by HAllB TPE ECAL with COMPASS module and a customized base R1166 19mmD 300(bulk) additional socket E974-17 $170 (small) R8619 25mmD 360(small) no socket FEU-84-3 25mmD 180(bulk) customized Cockcroft Walton based about $200(bulk), used by GLUEX ECAL * note: "bulk" means price for bulk order about 2000, "small" means price for a few test samples
signal estimation
preshower (PS)
- condition
- 22 photoelectrons per MIP from LHCb test in 2005 with 12x12x1.5cm module [1], 105 photoelectrons per MIP from LHCb test around 2000 with 4x4x1.5cm module and 15% PMT QE [2]. We take the later result as 22. Our 2cm preshower scintilator is 33% thicker than 1.5cm in LHCb PS, so we assume 30 photoelectrons for SoLID PS MIP
- in SoLID preshower, one MIP is 6MeV, max signal is 200MeV, (refer to the preshower response plot in SoLID pCDR). If we plan to cover 6 - 240MeV which is 1 - 40 MIP, then photoelectrons from PS range is 30-1200
- pulse full width 20ns according to scope picture from Simona's test of H8500C (slide 26 of talk). According to LHCb study, pulse shape has many variation and 25ns only covers about 85% of pluses on average. We assume 30ns.
- Simplify pluse shape as a triangle shape, then pulse height is about 2 times pulse area
- 50 ohm matching
- at 1000V, MAPMT H10966 has gain 3e5 (comparing to MAPMT H8500 has gain 1.5e6)
- At WLS (Kurary Y11) emission peak 476nm, QE of MAPMT H10966 and H8500 15%, MAPMT H7546 15%
- both MAPMT H10966 and H8500C has anode uniformity (=gain*photocathode sensitivity) 1:3 in its spec sheet (Simona's test confirms it). H7546 has 1:3 according to LHCb test, but it has 1:5 according its spec. We assume 1:3
- result without uniformity problem
- pulse height min 30*3e5/30e-9*1.6e-19*50*2=0.005V
- pulse height max 1200*3e5/30e-9*1.6e-19*50*2=0.192V
- dynamic range 40
- result with 1:3 uniformity problem
- pulse height min 30*3e5/30e-9*1.6e-19*50*2=0.005V
- pulse height max 1200*3e5/30e-9*1.6e-19*50*2*3=0.576V
- dynamic range 120
- conclusion
- pulse height max < flashADC limit 2V
- 12 bit ADC with 4096 channel like the Jlab flasADC can cover the dynamic range about 120 if a MIP is at channel 30
Scintillator Pad Detector (SPD)
- condition
- refer to the estimation for PS
- In SoLID SPD with 0.5cm thickness, both pion and electron deposit a MIP at 0.8MeV, max deposition at 10MeV, (refer to the SPD response plot in SoLID pCDR).
- If I set SPD cut at 0.4MeV and suppose it's dynamic range needed is 50 from 0.02 - 10MeV.
- result
- With MAPMT uniformity 1:3, the range is 150
- conclusion
- 12 bit ADC should still be enough to cover it
- We would need 10 times higher gain than PS
Shower (S)
- pion deposit a MIP at 60MeV, max electron deposition at about 2000MeV (refer to the shower response plot in SoLID pCDR).
SPD/PS readout consideration
difference between LHCb and SoLID
- LHCb SPD and PS are both 1.5cm thick. SoLID SPD is 0.5mm thick, PS 2cm thick. SoLID has a lot low energy background and thus prefer thinner SPD.
- LHCb SPD ans PS are in trigger. SoLID PS is not in trigger, SPD is. flashADC can't be used for LHCb if it were available (?)
- LHCb has to use 25ns time window to match its 40MHz bunch crossing time, only 85% of its signal is integrated and it needs very front end and front end electronics to compensate. SoLID has more freedom to choose longer time windows like 30ns or longer and thus accept 100% signal into flashADC to do pulse analysis at offline (?)
MAPMT crosstalk
- info
- MAPMT has crosstalk between pixels. It's a effect combines optical and electronic leaks
- Hamamatsu gives spec for H8500/H10966 max 3.5%, H7546 typical 2%
- LHCb used fiber to test it, Hamamatsu used a lamp with filter, They both include the anode uniformity effect already
- For H7546 with R5900-00-M64, LHCb test around 2000 [3] shows 20-25% crosstalk and LHCb test in 2005 [4] has max 2.5%
- For LHCb Hamamatsu H7546 with R7600-00-M64MOD, LHCb test in 2007 [5] shows 2% after 21 out of 200 unit replaced by Hamamatsu
- SoLID case
- SoLID PS signal are very localized and dominated by pions
- SoLID PS signal has 1MIP of Pion and electron peak at 10MIP and goes up to 40MIP, (refer to the preshower response plot in SoLID pCDR)
- Assume H10966 with 3.5% crosstalk, electron peak deposition gives 10*3.5%=0.35MIP, electron max deposition gives 40MIP*3.5%=1.4MIP. If we are cutting on 3-4MIP for PS, it should be a small effect.
- comparing to LHCb, it has a range of 1-50MIP by and covered by 10bit ADC with 1000 channels [6]
- background is at a MIP level (refer to the background response plot in SoLID pCDR) and it shift both pion energy and electron energy up by the same amount, the dynamic range becomes smaller and we just need to shift the cut by same amount
- channel arrangement
- choice 1, same radius on same MAPMT to have same background level (preferred?)
- choice 2, same sector on same MAPMT and match calorimeter module arrange because modules in a cluster will be summed anyway, for PVDIS, total 1800 channels and 30 sectors, 1800/30=60, so we can have one sector on one 64 channel MAPMT
MAPMT info
- file location
- LHCb Study
- LHCb structure before ECAL is 1.5cm SPD + 2.5X Lead + 1.5cm PS, MAPMT H7546 with special front end electronic for readout, 10bit ADC for PS, 1bit for SPD
- LHCb Hamamatsu H7546 with R5900-00-M64 (12 stages) test in 2005 [10], 2000 [11]
- LHCb Hamamatsu H7546 with R7600-00-M64MOD (8 stages) mass characterization in 2007 [12]
- H7546 as a package may have involved from R5900-00-M64 to R7600-00-M64MOD inside
- Hamamatsu info
- MAPMT assembly from hamamatsu http://www.hamamatsu.com/jp/en/product/category/3100/3002/index.html
- H7546 is a package name and its current spec sheet mentions R7600-M64 PMT inside[13]
- The new MA-PMT, H12700 [14], is still under development. We hope to be able to release it by the end of this year
- H8500 and H10966 uniformity map in Figure 3 on page 2 of its datasheet [15] and it's similar at different HV setting [16]. By standard spec, the non-uniformity has to be less than 1:3. It reflects variation in anode sensitivity, which is gain x photocathode sensitivity. The 1:3 means that we record the relative output of all anodes: the lowest output must be larger than 1/3 of the highest one.
- other
- CLAS12 RICH H8500C test, including crossstalk result (Hoek_H8500_Characterization.pdf)
- CLAS12 RICH has some H7546 test done also (Montgomery_NPESeminar_150911.pdf)
- CLAS12 RICH frontend (130626_RICH_Tech_Rev_FrontEnd.pdf)
- CLAS12 RICH DAQ (HallB_RICH_DAQ_June2013.pdf)
- CLAS12 RICH review report (Rich_Tech_Review_report.pdf)
- SBS coordinate director, similar to SoLID EC SPD (CDet-talk-SBS-MeetingJune2013-Sarty.pptx)
PMT tube info
- Consideration
- WLS output wavelength is peaked at 475nm (Kuraray Y11 for PS and SPD), 496nm (Bicron BCF91A for S)
- Our simulation shows SPD:PS:S signal ratio is roughly 1:10:100
- Assume LHCb SPD/PS test gives 22 photoelectrons per MIP, then our PS has 30 photoelectrons per MIP. Assume similar QE about 15-20%, the gain we want is for SPD,PS,S would be about 1e7,1e6,1e5 respectively to have signal within 2V flashADC limit
- With x10 amplifiers on SPD, we may use same PMT with 1e6 gain for SPD and PS (?)
- SPD/PS only connect about 2 fiber heads, small PMT is enough and cheaper(?)
- S connects 100 1mm fibers and needs larger PMT to cover at least 1cm2
- info for vendors
The SoLID calorimeter has three parts, scintillator pad(SPD), preshower(PS),shower(S). http://solid.physics.umass.edu/~seamus/solid_precdr.pdf On page 106 of the file, there's a sketch drawing showing the preshower and shower part. The SPD is just a thin plane of scintillator before PS
1. Scintillation wavelength: All of them are readout by embedded Wave length shifting fibers which has peak wavelength 475-496nm
2. Min. and max. number of incident photons per channel per event: a rough estimate, SPD: 20-100, PS: 150-6000, S: 1k-10k
3. Active area per channel: The SPD and PS has 2 of 0.5mmD or 1mmD fibers for readout by PMT. The S has 100 of 1mmD fibers for readout by PMT
4. Event rate: 1MHz level for low energy event 0.2kHz level for high energy event
Hamamatsu info
refer to page 34-43 of [17]
choose PMT with headon type, range 300-650nm, peak at 420nm,Bialkali(BA) cathode,Borrosilicate(K) window
- for SPD/PS
- R1635, 10mm
- R647, R4124, 13mm
- R1635 has only 8 stage, others has 10 stage. R647 has larger anode current
- prefer R647 due to larger gain and higher anode current
- for S
- R9800,R7899,R4998,R1924A, 25mm, R9800 has 8 stage while others has 10 stage
- R6095,R6094,R6427,R7205-01,R3998-02,R7111,R7525, 28mm, R7525 8 stage and R3998-02 9 stage while other 10 or 11 stage
- prefer R9800, R7525, R3998-02 due to small gain
- from Ardavan Ghassemi (AGhassemi@hamamatsu.com, Phone: 908-252-7632)
- It would be most cost effective if you can select the same PMT for S, PS, and SPD.
- R1166 could be a suitable and cost effective option for you; although, it's pulse linearity is within ±2% deviation of 4mA and ±5% of 7mA anode peak currents. Would you be able to correct for these deviations? Here is R1166's specs and Its budgetary unit price estimate at qty. of 1800 is around $300.
- For SPD, Even if gain is decreased to 1E+05, anode current is still too high. In order to decrease incident light intensity onto the PMT, would it be possible to reduce count of WLS fibers per channel or try another countermeasure?
- For PS & S, another suitable economical candidate is R8619, please find its datasheet and linearity
- Furthermore, the required pulse linearity is borderline for both R8619 and R1166 when a voltage divider with linear ratios is used, so we recommend a divider with tapered ratios instead. Please refer to the attached pulse linearity data for R8619. For R1166, typical pulse linearity a peak anode current of 25 mA at +/-2% deviation using a tapered divider.
- R3998-02 spec. in that qty. range is $400 with standard bialkali photocathode and $500 with super bialkali photocathode (higher QE), it has a rather limited pulse linearity. R3998-02 includes a bare socket E678–14C with no built-in divider or power supply. We offer C10344-03 socket [18] with built-in power supply and voltage divider, in 1K-4K units qty. range is $240. At qty. of 20 units, its price is $392.
- price quote [19] for 2 units each of R1166, R8619, and H10966A. Out of these 3 PMTs, R1166 has a socket E974-17, H10966A has a built in socket. I expect R8619's pricing to be far more attractive than R1166
FEU info
http://www.melz-feu.ru/catalog/?cat=5
- for SPD and PS
- FEU-60, size 15mm and window Sb-Cs, http://www.melz-feu.ru/products/?id=15
- for S
- FEU-85-1, size 30mm and window Sb-Cs, http://www.melz-feu.ru/products/?id=14
HallD use FEU-84-3 for ECAL about $180 each, use Cockcroft Walton bases for these tubes at about $200 per base in large quantities, it's spec here [20]
contact:
info@melz-feu.ru
S.S. Yakushov 995-02-33(phone) 963-51-83 fax
Davyd Kravchenko <kda.pmt@gmail.com>