1. Report from Fermilab: Beam Tests and QUARTZTOFs
Mike Albrow (Fermilab)
(if you heard this in September, not much new!)
Beam tests with 2 Burle 25 um-pore with in-line quartz radiator
with Photek 4 um MCP-PMT
Design and construction of two QUARTZTOF detectors.
What we learned from beam tests.
Plans … next steps

2. Argonne – Chicago – Fermilab Picosecond Timing Collaboration
Tests at Fermilab Test beam August 2008
Who: Mike Albrow, Ed May, Tyler Natoli, Erik Ramberg, Anatoly Ronzhin, Jerry Va’vra,
+ Camden Eartly, Henry Frisch, Scott Wilbur, ..
(1) Burle, head-on
(2) QUARTZTOFs
MCP-PMT
Off-axis parabolic
mirror
Half-cone radiator
Works, but cannot be close to beam
(unless edgeless MCP-PMT developed)
… even then.
Can be very close to beam (edgeless)
Several times more light, prompt.
PMT far from beam.
Still to be demonstrated.

3. Fermilab test beam is nicer than CERN’s ! 120 GeV protons.
Camden Eartly (Chicago)
Jerry Va’vra (SLAC)
Anatoly Ronzhin (Fermilab)

4. I am not going into the electronics … .high end commercial
Commercial electronics (ORTEC)
Calibrated TDC : 3.1 ps / channel
We (both Chicago and Fermilab) are developing chips that could be
actually used in FP420 (25 ns/crossing) with some waveform digitization.

5. Test electronics resolution by splitting a signal (high bandwidth splitter circuit).
= 6.4 ps
ADC1 should = ADC2
Not quite: ADC imperfections.
Probably not important.
per channel = 4.6 ps

6. = 14 ps 2x2 pads =12mm x 12mm = 21 ps = 13 ps (per detector)
Initial dt, 2 25 um Burle tubes
+ 6 mm Quartz plate + 2mm window
Cut out tails of ADCs
(some multihits)
= 14 ps
2x2 pads =12mm x 12mm
Apply small PH slewing correction
= 21 ps
= 13 ps (per detector)

7. Need optical contact or light is TIR
for TIR at back
Note: 6mm with n = 1.415
dt(F-B) = 20 ps
Need optical contact or
light is TIR
6mm
PMT p

8. ~ 15K – 20K each! Thanks to Photek (UK) for loan of two PMT210’s
2-stages:10mm diam
Photek Microchannel Plate PMT
TTS (single p.e.) ~ 20 ps
100 prompt p.e. ~ 2 ps
~ 15K – 20K each!
Thanks to Photek (UK) for
loan of two PMT210’s

9. Photek (UK) single channel MCP-PMTs 210 (2 plates, 10 mm diameter, ~4 um pores)
Loaned 2, 1 would not take HV. Put other with Burle tube already tested.
No extra radiator, just 5.6 mm quartz window.
Burle
Photek
ADC cut (Photek cleaner)
dt after ADC cuts.
(Photek + electronics) = 11 ps
(Photek alone) ~ 8.8 ps

10. We are there! & need to get PMT away from beam!
Could simply feed
signal to ~ 4 channels,
Reduce electronic
contribution
(Could be 7.4 ps
w/ tighter PH cuts)
Jerry Va’vra
& need to get PMT
away from beam!
We are there!
BUT … This geometry is not edgeless … we must have <~1mm edge

11. We had also Photonis ( “Burle”) 10 um tubes, but these had poorer
PH spectra and actually worse probably less prompt light.
 GET LIGHT!
Also we (Anatoly Ronzhin) tested some 3mm x 3mm SiPMs
with 16 mm (too much) Plexiglas radiator in front.
~ 47 ps, per device. Far to go, but interesting.
Thin and fairly edgeless, also “cheap” (cf. MCP-PMT)
Can have many (e.g. 16) in series.
But probably rate problems for FP420.

12. An isochronous & achromatic Cerenkov Counter
A possible solution to the edgelessness requirement (more light too!):
QUARTZTOF (M.G.A.)
An isochronous & achromatic Cerenkov Counter
Making Cerenkov light parallel  point focusing
New design/concept for fast timing Cerenkov counter
Much more light. Edgeless. PMTs far from LHC beam
Ray-tracing calculations done:
Expect > factor 10 more photoelectrons than either GASTOF or QUARTIC
Photons arrive promptly (< few ps) at MCP-PMT
Full (wavelength-dependent) simulations largely done.
Needed for [x,y] position dependence over [2mm x 20mm] area.
Two made for beam tests (in August at Fermilab).

13. Fast timing detectors tested in beams, in FP420 R&D report:
1) GASTOF (KP, Louvain)
2) QUARTIC ( MGA Fermilab, AB UTA, JP Alberta) ++
20 cm GASTOF with Atmospheric C4F8O  ~ 10 p.e.
nearly all collected, within few ps
QUARTIC bars ~ (inclined) 9 mm  ~ 120 pe,
but solid angle for direct light small,  ~ 3-4 pe/bar (in tests)
QUARTZTOF: Radiator mm quartz,
nearly all light collected within a few ps. > 200 p.e.
(At least for 2 mm x 2mm “sweet spot”)

14. QUARTZTOF (QT) Principle
Apex – base distance is
L is “isosceles point”
p (“needle” beam
~ 2mm x 2mm)
FP420: 2mm x 20mm L
and all emission points in between
Proton along axis of cone, with half angle = half Cerenkov angle.
All Cerenkov light emitted (all ) up to isosceles point L is totally internally
reflected out through the base and arrives simultaneously, in a parallel beam.
It can thus be focused to a point by a 90deg off-axis parabolic mirror.
Cannot use full cone in FP420, but cut in half.
 True for “lucky wavelength” … but can compensate for others.

15. GEANT Simulation – Kristina Yancey (summer student, +Hans Wenzel)
Enables studies of x,y dependence
Wavelength dependence of emission, absorption, reflection, QE
Arrival time distribution vs radius at photocathode
 Compensating “wedge” at MCP front face.
MCP-PMT Photocathode
Cone optimized for 689 nm (1.8 eV)
Cone angle 23.3°
Off-axis Parabolic mirror
Focal Length = 76.2 mm
Off-axis Parabolic Mirror
Absorbing Cylinder
Quartz Cone

16. Transmission of UVT Plexi sample
We used a spectrophotometer to measure the transmittance
of 200 nm – 800 nm light through 2 x 10 mm blocks, and the
two together.
Irina Kubantseva
Data sheet:
… goes into simulation.

17. Fantastic for perfectly on-axis Very position-sensitive
X = +1 mm
On axis
Here’s the issue:
Fantastic for perfectly on-axis
Very position-sensitive
X = +2 mm

18. Need larger MCP for coverage
(1,0)
(0,0)
(2,0)
Aperture of
Photek 10mm
Aperture of Burle-
Photonis 48x48
Need larger MCP for coverage

19. GEANT: Arrival time, optimal wavelength with 0,1,2 mm shifts off axis
+1 mm (in x)
On axis
σ =0.3 ps
+2 mm in x
14 ps

20. Time as a function of radius
Wiggles not real
10 ps

21. Arrival Time after correction by t(r), within r=4 mm

22. GEANT: Arrival time vs wavelength; focusing red light
~ MCP cut-off
Light at larger radii
Light at center spot
10 ps
2.5 ps
This can be compensated for with a wedge
It’s the jitter in the rise time that counts, and we know the position of the p.

23. Select light with λ > 450 nm:
GEANT
Select light with λ > 450 nm:
1 ps
σ(t) = 1.1 ps, 457 photons (1 proton)  ~ 75 p.e.
But we can use also the light with λ < 450 nm !

24. GEANT: Time front light vs x,y position wrt cone axis
10 ps
At least 4 mm x 4 mm is OK

25. CAD of Test Beam setup (John Rauch, FNAL)
Can mount Cylindrical Photek,
Square Photonis, SiPMs
Installed in black copper-lined
box with feed-throughs.
Precision alignment, & dz shifts
Design & Construction (Carl Lindenmeyer, John Korienek, FNAL)
Conical radiator Precision mechanics Off-axis parabolic mirror

26. Test beam modules with Burle tubes. 2x2 pads = 12mm x 12 mm together
Provision to adjust focus, and separation between QUARTZTOFs (10mm = 33.3ps)
All placed in Cu-shielded dark box, lined with special light absorbing paper.

27. Optical checks post-TestBeam
Observe waves on milled conical surface:
May need “optical” flatness, polishing

28. Striations from ripples (not clear in photo)
LASER BEAM
38° 2°

29. Multiple QTs
Depending on performance over full x-range (~ 20 mm)
We may want a few (~ 4?) in series.
Full depth/30 mm radiator ~ 10 cm
Optimum may be mm radiator, it scales.
Schematically:
Multiple measurements good, at least one is optimum.
Also note:
The ~ parallel beam from the parabolic mirror could (?) be
10’s of cm long, so MCP’s and electronics far from beam.

30. Remarks Fast timing is crucial for FP420 to work.
15-20 ps may be OK for L = few x 10^33 but <~ 10 ps needed for 10^34
We have not demonstrated that yet, but we can see a way and we progress.
I am optimistic about conical QUARTZTOF, even if “sweet spot” is only
a few mm^2 it should be much better than GASTOF and QUARTIC.
But we have to demonstrate it. May need optical quality cone surface, and
large area (~ 50 x 50 mm) MCP-PMTs.
Longer term there could be solutions for multiple p-measurements within
a bunch crossing … Streak Camera with pixel detectors. (KP idea).
Concept is there … should work in principle … some years to develop?
Perhaps < 5 ps timing is possible if electronics up to it (25 ns)
Perhaps multiple protons can be measured eventually.

31. FSC = Forward Shower Counters
Mike Albrow, Fermilab
What: We propose to install a set of scintillation counters around both
outgoing beam pipes, ~ 60m – 85m
Why: (a) As veto in Level 1 trigger to reduce useless pile-up events.
(b) To detect rapidity gaps in diffractive events (p or no-p).
(c) Help establish exclusivity in central exclusive channels
(d) To monitor beam conditions on incoming and outgoing beams.
(e) To test forward flux simulations (MARS etc.)
(Can also be used as supplementary luminosity monitors: E&W)
N.B. : This will provide valuable tests of radiation environment
to be expected for FP420. Terra incognita

33. Location of FSC counters
(before connecting vacuum pipe installed)
Warm accessible vacuum pipe (circular – elliptical)
Do not see primary particles, but showers in pipe ++

34. Bunches spaced by 25ns = 7.5 m dt = 3.333ns x z(mod7.5m) e.g.
z = m dt=1.9ns
z = m dt=8.8ns
Z1, bunches
Simultaneous
As at z = 0
time
Z2, bunches
separated in t z
4 shown, could be 5 or 6

35. Efficiency for detecting forward particles (Orava):
Low : > 20% for > 9.0, > 60% for 9.5 < < 11.5
(integrated over pT)

36. Summary on FSC:
We want to add FSC to both sides of CMS, before first collisions.
Either or both of Scintillators and GEMs.
Greatly improving Diff triggers, and especially central exclusive studies
Early presentation to CMS, but not yet formally proposed.
Meeting this week in CMS week.