Tests of Fast Timing Detectors in the Fermilab Test Beam, etc

1Mike Albrow FP420, June 20th 2009T979: Tests of Precision Timing Detectors @ MTest

Tests of Fast Timing Detectors in the Fermilab Test Beam, etc.(T979) MTest : May 27th – June 2nd

Mike Albrow, Sasha Pronko, Erik Ramberg, Anatoly Ronzhin, Andriy Zatserklyaniy+ detector simulations by Hans Wenzel & Earle Wilson (student)

Motivations for ~ ps / 10 ps timing detectors

Set-up and triggers etc.

Detector configurations:

A) A – B – C in lineB) A+B transverse bars – CC) Q-bar1(A) – Q-bar2(B) – CD) Aerogel(C) – BE) Si-PMTsF) Photonis1 – Photonis2 – C in lineG) Q-bar1(A) – Q-bar2(B) – 8.7 m flight path --- C in line

Thoughts about next steps

To be explained!

Only if you ask!


hep-ex/0511057

A&R: hep-ph/0009336:


to CMS Exec Board Summer 2008(ATLAS also reviewing)

2.1mm/10psp)Δt(pc2

1TOF)Δz(pp,

?z(vertex) from pp == z(vertex) central

Pile-up reduction in FP420Want L ~ 10^34, <n> ~ 25/x

cf σ(z)vtx ~ 50 mm


At MTest, 120 GeV/c p, ~40,000/spill / 1 spill per minute

Simple trigger (schematic):2mm x 2mm scint. VETO w/hole2 PMTs in AND 2 PMTs in OR

Calibrate electronics resolution with same pulse start & stop: σ = 4.0 ps Cerenkov light in Quartz windows (~5mm, 9mm). HV ~ 4.5 kV, G ~ 5.10^5

Dark & shielded box

PHOTEK 2102 MCP, 10mm Φ

PHOTEK 2402 MCP, 40mm Φ

210

First A-B-C in line

MCP-PMT-A

ORTEC566, 567

TAC/SCA

ORTECAD114ADC

ATTENUATOR

ADC

ATTENUATORMCP-PMT-B

DAQ

ADC

CBA

Schematic DAQ :

T1


A-B-C in-line results: Cerenkov light in PMT windowsAll numbers “preliminary”, to be double-checked

ADC distributions: cut out tails and stragglers (~ 10%)T1 = tA – tBT2 = tA – tCT3 = tB – tC=======Check Ti(PH A,B)Make slewing corrections

Unfold:

A

BC

23

22

21

2

1TTTA

etc. PMT-1 (Photek-210, 4.7 kV)=10.8 psPMT-2 (Photek-210, 4.6 kV)=11.5 psPMT-3 (Photek-240, 4.2 kV)=5.1 psCerenkov light in PMT windows

9.4 mm, 5.5 mm +/- ~ 0.2 ps


Double Q-barQuartz (fused silica) bars 6mm x 6mm x 90mm PHOTEK 210Mounted at Cherenkov angle θc ~ 48 deg. on opposite sides.dz = 6 mm/sin(48) = 8.1 mm. Some light direct to PMT, ~1/2 TIR to PMTBlack “sock” over bars just to avoid light sharing

C

B

A

Unfold:σ(A) = 22.3 psσ(B) = 30.5 ps

Includes electronics (~3 ps)and 2 mm beam width smear (A,B)Δt = 2 mm x (10 ps/2 mm)

ps 28.71035.30

ps 9.711033.22

222

222

B

A

Combining [AB] removesbeam spread (later, tracking)

Difference? Optical coupling?


σ = 6.04 ch = 18.7 psUnfold C = 7.7 ps, σ(AB) = 17.0 ps

Resolution of double-Q-bar

2 mm x-spread not to be subtracted(only 3 ps electronics)

Resolution of Double-Qbar as one device

T3)3T2(T14

1C

2

BA

T1 = A - BT2 = A - CT3 = B - C ------------T1 + T2 = A - B + A - C3 x T3 = 3B - 3Cso T1 + T2 + 3x T3 = A - B + A - C + 3B - 3C = 2A + 2B - 4Cand 1/4 ( T1 + T2 + 3x T3 ) = (A+B)/2 - C


Switched on, saw signals!

A = Aerogel

B

Corrected T2 = A-B = 10.8 ch = 33.5 ps (before unfolding)


Aerogel results:Unfolding indirect because only 2 PMTs in run.A (Aerogel on 240) and B(210 in beam)T1 = t(A) – t(B) corrected for smearing:10 mm aero σ(T1) = 43.7 ps20 mm aero σ(T1) = 45.3 ps30 mm aero σ(T1) = 33.5 ps Unfold with σ(1) = 12 ps from in-line

σ (Aerogel 30 mm) ~ 31 ps

<P.H.> = 46 ch. (10mm) 72 ch. (30 mm)

Aerogel + mirror ~ massless & short (~ 5 cm), simple.Could have several in line, independent √NBUT: have large 240 tube close to beamPossibilities to focus light : smaller tube farther away, to be simulated

AEROGEL

MCP-PMT


A focusing Aerogel Counter

Normally Cerenkov light can only be focused to a ring.But if beam is very small (needle beam) you can focus to a point.Smaller PMT and farther from beam.

MCP-PMT : all light simulltaneous (from all z! isochronous)

f (can be far-ish)

p (needle beam)

off-axis parabolicmirror

conical lens(quartz)

Question is: if needle 6mm x 6mm?Image size? 40mm MCP?

e.g.


Tests of SiPMs = silicon photomultipliers

Eight Hamamatsu SiPMs, 3mm x 3mmIn beam with quartz Cherenkov radiatorsseveral thicknesses (4 – 12mm), mirrored and not mirrored.

Best conditions σ(t) ~ 33 – 37 ps

10-15 photoelectrons

Channels

Between SiPMs and C. Slewing correction appliedSiPM on end of bars?


World’s Best Beamline Time-of-Flight System?

24 psec resolution positron peak,Using average of A & B times

Can measure momentum of a proton with 2 MCP-PMTs! (if you know it’s a proton!)

Start = Double-Q-barStop = Photek 240Start-stop dist. = 8.7 m Predictions of proton positions


Possible Next Steps

For FP420 a σ(t) = 10 ps edgeless detector we learnt a way Need to include CMS-compatible electronics/DAQwith reference time signals (jitter <~ 5 ps)

θc = 48deg Q-bars onto PHOTEK 240 MCP-PMTs

40 mm diam. MCP

6mm x 6mm barsTIR: isolated

p

20mm

I

“beam” only 6mm vert.,20 mm horiz.

MCP

MCP

Should get < ~ 8 ps

+ More aerogel? To test in Fall?


Design by John Rauch & Carl Lindenmeyer 5 x 3 QUARTIC

Protons go through 5 6mm bars(inclined so ~ 8.4 mm)3 bars wide ~ 20 mm 40 mm Φ pmt

Why better than old QUARTICs?Better, single channel 40mm MCPAll light isochronous (1st idea)Bars touch nothing on sides (precision boxes)

Pressure on PMT with optical grease


Needs work!! But CAD GEANT4 presumably and CMSSWEnsemble QUARTICs : two 5-6 bars in line, UP & DOWNInter-bar gaps not in line: Some particles in only one (good)UP & DOWN good for checking resolution, independent of yTwo single bars out to side, one covers full 8mm x 24mmand (maybe) one retracted to cover 8mm x outer 20 mm (e.g.)

John Rauch, sent yesterday!


Earle Wilson (student) and Hans Wenzel doing some GEANT4 simulations of QUARTIC.First results (not checked)

Photoelectrons: Hamamatsu MCP-PMT R3809U-65

Photoelectrons: Photek 240


Timing and Timing-Resolution vs. Angle Incident Beam

-Timing and timing resolution obtained using DCOG Method-Cerenkov Angle: 48.2-Jitter: 30 psec-Gain: 100-Each data point is taken over 1000 events.


Photoelectrons: Photek 240Photoelectrons: Photek 240


Earle Wilson (student) and Hans Wenzel

I’m not sure about abs values, but 48deg gets light fastest, butmore acute angle more p.e. and better time resolution. Unfortunately did not test this!


Reference timing. We are continually asked for a practical demo.OK yes we should! Nobody else has volunteered; I hope a FNAL-LLNL combined effort can do this. I will focus a bit more on it! Brian Chase is our best Fermilab person, he presented something to us in Feb:

From his slides:



Possible future development: GHz streak camera (concept only):

In principle, timing up to several protons/bunch crossing. MCP (if needed) low gain.All in vacuum tube of course:

~ 400 MHz RF4 x 400 = 1.2 GHz

= 830 ps period

Cerenkov photons

Window& photocathode

MCP(maybe)

~ 10 kV acceleration tube

Focusing tripletMagnetostatic orElectrostatic

RF sweep(both x and y) CMS Pixel

detector, 25ns readout

Say ~ 10 p.e. MCP gain ~ 100 (?) 1000 e’s in beamAt focusing lens ~ 1000 x 10 keV electrons. Good signal in pixel detector.Time spread ~ r/L can be kept to < 5 ps (L ~ 20 cm)Say sweep circle r = 10 mm, 830 ps, 100 μm pixels 1.5 ps/pixel.Depending on aberrations etc, maybe measure >= 2 times if dt >~ 20 ps.

Use precision space for precision time

MGA + Vic Scarpine (FNAL) interestedand expert … so far just thinking.Hope for full simulation.


Summary (IMO):

We have (at least) THREE promising phase 1 detectors:GASTOF --- QUARTIC’ --- Aerogel (if focused)different characteristics: G – 1 channel, Q = multichannel, A = thin & short

I think actual < 10 ps detectors will be demonstrated this Fall (Oct?)

Electronics is another issue, want σ [detector+electronics] <~ 10 ps(Actually ~ 15-20 may be acceptable initially : future upgrades but we really want future upgrades to be 6 ps, 4 ps, 2 ps!

Reference timing is another issue to be demonstrated. Want itto give negligible contribution so σ <~ 5 ps. Will anyone else take it up?

Unknown: Stability of system. day-to-day variations can be calibratedout, probably hour-to-hour, but not minute-to-minute.


Back-UpMotivations


Motivations:Timing on single particles σ(t) typically >~ 100 psA factor 10 – 100 improvement likely to have unforeseen benefits.We know of some (foreseen), e.g.:

Particle ID in beamsE.g. at 25 GeV/c, over 15m:Δt(π-K) = 20 psΔt(K-p) = 50 psor at 10 GeV/c, over 30m:Δt(π-e) = 10 ps==================Areas ~few cm2, want thin.

Particle ID in large detectors(~CDF-like or ILC)E.g. at 6 GeV/c, over 1.5m:Δt(π-K) = 17 psΔt(K-p) = 43 ps==================Areas ~several m2, want thin.

Pile-up reduction e.g. in FP420:Extensions to CMS & ATLAS in prepn.p + p p + H + p + nothing elseMeasure p’s M(H), J, C, P, Γ

CMS

Hp p240m … 420m240m … 420m

Argonne-Chicago-(Henry Frisch et al.)

Fermilab group

3

21

PET-TOF β+e Δt = 10ps : Δ z = 3mm

4

Documents

Tests of Fast Timing Detectors in the Fermilab Test Beam, etc