Test Results with Multianode Photomultipliers

1
Test Results withMultianode Photomultipliers

• Test beam results
• Lab test results
• Signal shape signal loss
• High magnetic field tests
• 8-dynode stage MaPMT
• Conclusions

LHCb RICH Photodetector Review, CERN, 08.01.2003
Stephan Eisenhardt University of Edinburgh
2
3 x 3 MaPMT Cluster Set-up

• 3x3 array of MaPMTs
• RICH 1 Prototype
• CF4 _at_ 700 mbar
• 120 GeV pion beam

3x3 MaPMTs
40 MHz Read-out APVm chip
Quartz lenses
3
3 x 3 MaPMT Cluster Test

• Measurements with
• MaPMT HV -1000V
• Data
• subtract cross-talk and common-mode
• Fit of photo electrons per event ?

• Cherenkov ring, 6000 events
• Photon yield in data background
• ? 6.96 ? 0.33 p.e 0.26 p.e.
• Expect from simulation
• ? 6.49 ? 0.17 p.e.

Single channel spectrum
4
Photon Yields
No Lenses
Quartz Lenses

• Demonstrates lens effect
• Photon yield ratio with / without lenses 1.55
• in agreement with simulation

5
Charged Particles

• charged particles traversing the lens MaPMT
  produce background hits

• Npe multiplicity from charged particles
• 5...10 / track for most angles
• up to 30 for angles around 45?
• RICH 1 stray tracks predominantly from the back
  (135...180?)
• total background small

6
Magnetic Field Tests

• light source LED
• geometry pin hole mask
• ?-metal shield (0.9mm)
• CAMAC readout

• MaPMT tested with
• Helmholtz coil
• B 0, 10, 20, 30 Gauss

7
Magnetic Field Results

• ?-metal
• extension d 10,13,32 mm
• reduces loss
• no structure (d 32 mm)

• B transverse
• MaPMTs are insensitive up to
• magnetic fields of 30 G
• expect mainly By ? 30 G
• B longitudinal
• sensitive to Bz ? 10 G
• ?gain loss, edge rows
• expect Bz lt 10 G

8
Testbeam Summary

• Successful test of 3x3 cluster of MaPMTs
• Close packing of an array of MaPMTs
• Quartz lenses increase acceptance
• Measured Cherenkov photon yield as expected
• Demonstrated 40 MHz read-out with APVm chip
• works in LHCb pit - charged particles magnetic
  field
• MaPMT fulfills LHCb specifications
• more expensive than HPD ? MaPMT is backup
• Further programme to keep the MaPMT viable
• Develop F/E electronics
• APVm chip not suitable for LHCb architecture
• improve signal shape measurements

9
Laboratory Set-up

• xy-scanning table
• light source blue LED
• spot size 100 ?m
• (single mode fiber, gradient index lens)

• CAMAC or APVm read out

10
Laboratory Results
single pixel spectrum (LED, CAMAC read-out)
Variation of signal gain within a MaPMT factor
3
-900 V

• Simple Gaussian signal fit
• Mean signal s / pedestal s 401
• signal loss below 5 s cut 11.5

Pixel ID
Pixel ID
11
Gain Variation

• within tube
• gain variation (max/min) lt 4 (spec lt5)
• but the two halves behave differently and in the
  same way for different tubes
• caused by different capacities of two Kapton
  cables
• gain variation most probably 2
• between tubes
• gain variation lt2

gain
old focussing
new focussing
APVm read-out magnetic field setup
12
MaPMT Developments

• Improved MaPMTs
• 2 new tubes delivered by
• Hamamatsu (Jan 2000)
• Quantum efficiency
• QE 25 - 27 at 360 nm
• 3-5 higher with respect
• to 3x3 cluster
• 14-23 more photons
• Better Focusing
• additional focusing wires
• reduced distance between
• focusing grid and entry slits
• larger acceptance at edge
• (now similar to centre pixels)

13
Photograph Comparison
MaPMT at testbeam
new MaPMT
14
Improved Focusing

• Comparison old vs new MaPMT focusing
• CAMAC Read-out

pixel scan with LED
Signal s
Lambda

• Improved homogeneity for edge pixels
• pulse height
• Collection efficiency

Relative distance 0.1 mm
15
Channel Capacitances
C(MaPMT) 1.5 pF C(Base) 1.5
pF C(Bleederboard) 2.6 - 4.5
pF C(Kapton) 14 - 30 pF
Aim Capacitance at Beetle input C lt 10 pF -gt
noise ENC 1000 e
C(MaPMT) vs. cathode
C(MaPMTlead) vs. dynodes (0 photo
cathode)
16
Signal Shape

• Improved fit routine (Tokar et al)
• Poisson distribution for npe 1, 2
• Gaussian for npe 0, 3, ...
• Hypothesis Photo conversion possible at 1st
  dynode
• Results
• Good phenomenological description of spectra
• but HV scaling contradicts
• 1st dynode hypothesis
• Still usefull to estimate signal loss

Pulse height ADC counts

• APVm read-out

17
Signal Loss Definitions

• Signal loss can be defined as
• loss of photoelectrons converted at the
  photocathode
• loss of photoelectrons converted at either the
  photocathode or the 1st dynode
• An estimate can be derived from fitted
  parameters
• photon conversion probability (?Poisson)
• integral over the 1-photon contribution(s)
• integral over the pedestal contribution

• APVm read-out

best signal resolution from HV scan series
18
1st Dynode Effect

• CAMAC read-out
• gate width 200 ns
• signal fully contained
• Signals between pedestal and single p.e. signal,
• no 2nd peak
• Gaussian fit inadequate

• Poisson fit better, but region between pedestal
  and signal still not described
• Poisson and 1st dynode fit
• best description of data
• ? (1st)/?(PC) 0.10.3
• gain at 1st dynode (K1) matches expectation
• but HV scaling contradicts 1st dynode hypothesis

19
Gain and Signal Loss

• 64 pixels of MaPMT
• CAMAC read-out
• HV -900V
• Signal loss
• for Gaussian fit,
• add 2.5 no multiplication
• at 1st dynode
• relative change
• consistent with change
• of fit method

20
MaPMT at Low Gains
HV scan with LED

• Dynode gain dVk, k 0.8
• Lower overall gain,
• keep gain at 1st dynode
• Change dynode resistors to run MaPMT at low gain
• default 3-2-2-1--1-2-5
• medium 4-2-2-1--1-2-5
• low gain 4-3-3-1--1-2-5
• Reduce gain by factor 4
• Signal width mainly due to Poisson at 1st dynode
• ? g1 s2 / ?2
• Increase loss below threshold due to signal
  broadening

Signal s e-
calibrated, averaged data
High Voltage V
Direct signal input to APVm
g1 s2 / ?2

• CAMAC read-out,
• measurements with
• APVm read-out agree

Signal s ADC counts
21
Gain Studies with APVm

• Standard Attenuation
• (passive, capacitive coupling)

No Attenuation (capacitive coupling removed)
Signal s ADC
Signal s ADC
Limited dynamic range
Signal1st dyn ADC
Signal1st dyn ADC
High Voltage V
High Voltage V
22
Signal Loss vs S/N

• Total signal loss
• Loss behaviour
• scales with S/N
• APVm with attenuation
• standard gain preferred (higher S/N ? lower
  loss)
• APVm with direct signal input
• low gain has better S/N (better pedestal) ?
  lower loss
• Direct signal input option has been discarded
• lower gain ? lower S/N ? higher loss
• very limited dynamic range

We want a S/N of 40 !!
23
Magnetic Field Tests

• RICH 1 likely in magnetic field of 400 Gauss
• Measurements of MaPMT sensitivity to longitudinal
  and transverse magnetic fields up to 35 mT (350
  Gauss)
• LED light source, APVm read-out

Longitudinal axis
B mT vs I A
24
Method
light yield in 64 pixel of MaPMT
Loss of photo-electrons ?
Yes !
Loss of gain ?
No !
Distortion?
No !
Center of gravity
5? cut

• LED light source
• with/without mask in front of MaPMT

signal spectrum of single pixel
25
Transverse B-Field
Transverse field in x-direction
90
MaPMT insensitive to transverse fields up to
gt20mT (gt200 G) without shielding
normalised light yield in whole MaPMT
26
Longitudinal B-Field
90

• Unshielded MaPMT
• gt20 G B longitudinal gt200 G B transverse
  
• Single ?-metal shield
• 0.9 mm thick, 13 mm extension
• gt80 G B longitudinal

?-metal shield 0.9 mm thick 13 or 20 mm extension
27
Longitudinal B-Field
No shielding

• Colour
• Legend
• 1 mT
• 2 mT
• 3 mT
• 4 mT
• 5 mT
• 6 mT
• 7 mT
• 8 mT
• 9 mT
• 10 mT
• 15 mT
• 20 mT
• 25 mT
• 30 mT
• 35 mT

light yield rows
light yield columns
6mT
Conclusions Agreement with previous test Top and
bottom row drop first Sizable loss for edge rows
at 3mT Decrease due to loss of p.e. Row 5 anomaly
understood from x-talk in APVm readout
CoG rows
CoG columns
28
Magnetic Field Test Summary

• Measured photon yield and CoG up to 350 Gauss
• Loss of gain? ? No! (CoG)
• Distortion of pattern? ? No! (masked measurement)
• Loss of photons? ? Yes!
• Loss of photons lt10 for unshielded MaPMT
• gt20 G B longitudinal
• gt200 G B transverse
• Loss of photons lt10 for single ?-metal shield
• 0.9 mm thick, 13 mm extension
• gt80 G B longitudinal

29
MaPMT - 8 Dynode Stages
Hamamatsu data sheet gain G

• MaPMT with lower gain
• run at lower HV
• fewer dynode stages
• MaPMT - 8 dynode stages
• discussed with Hamamatsu some time ago
• would have to pay development costs
• Summer 2002 M64 with 8 dynode stages now
  available
• Data sheet
• gain G 60000 at HV 800V
• Good signal/noise to be verified

107
12-stage
106
G 300000
8-stage
105
G 60000
104
103
800V
1000V
500V
30
MaPMT - 8 Dynode Stages

• Procurement
• 2 tubes M64 with 8 dynode stages on loan from
  Hamamatsu
• arrived at Edinburgh Nov 22 2002
• Bleeder resistor (ratios)
• 3 - 2 - 2 - 1 - 1 - 1 - 1 - 2 - 5
• Connect to single bases
• adapt single base
• cut NC pins of tube
• First measurements
• CAMAC Readout LeCroy ADC, Phillips 10 x Amp
• Brief test
• APVm Readout, without attenuation

31
8 Dynode Stages Spectra I
32
8 Dynode Stages Spectra II
33
8 Dynode Stages Spectra III
34
MaPMT Gain Model
Gain G Vk at each dynode Tuned to measured
gains (CAMAC Readout)
Beetle MaPMT
Beetle 1.2
MaPMT with 8 dynode stages looks promising!
35
8 Dynode Stages Spectra IV
8-stage
12-stage
CAMAC read-out
12-stage
8-stage
APVm read-out (without attenuation)
36
Mu-Metal Shielding

• 4x4 array of mu-metal
• produced at Edinburgh
• Features
• 10 Single pieces with half slits
• Welded at the outside
• Ni alloy brazing powder for inside
• Heat treatment
• Heating under vacuum
• at 300 0C / h, leave 1 h at 1070 0C
• Annealing, degaussing
• Brazing
• Simulation of array underway

Extending from Photo cathode 20 mm and 13
mm Thickness 0.9 mm Precision 0.2 mm
37
Beetle Read-out Board

• Beetle 1.2 test read-out (12 channels)
• Edinburgh received a Beetle 1.2 mother board and
  daughter board from Heidelberg, set-up of test
  system in 01/2003
• Plan to mount Beetle-MaPMT chip on such a
  daughter board
• Plan to connect MaPMT to this read-out

Interchangeable board, allows test of different
Beetle/BeetleMap versions.
Picture of board fully operational at Oxford
38
Photodet. Testing Facilities

• 500 HPD or 4000 MaPMT to serial tested for
• functionality within specs
• individual characteristics
• working parameters
• for 340k channels
• full automation needed
• selection of detectors according to test results
• position in detectors wrt. occupancy
• to be operational in end 2003
• if HPD use the UK L0-L1 demonstration system
• if MaPMT use Beetle hybrid prototype
• Time estimate for all measurements scans
• for one tube 24hrs
• (including handling and resting in the dark)
• 2 test facilities needed for 1 1/2 years
• (Edinburgh Glasgow)
• Need to know which detector to test

MaPMT test setup
ODE
MaPMT
xy-table
39
Conclusions

• Successful test of 3x3 array of MaPMTs
• Close packing, Quartz lenses, photon yield as
  expected
• MaPMT fulfils LHCb RICH specifications
• Improvements in QE and light collection
  efficiency
• Measured MaPMT signal shape and signal loss at
  low gain
• gain, width, signal loss from lt600 to 1000 V
• We want S/N of 40 to keep signal loss at O(10)
• Measured sensitivity of MaPMT to magnetic fields
• longitudinal up to gt80 G with 0.9 mm ?-metal
  shield
• insensitive to transverse fields up to gt200 G
• 8-dynode stage option
• first measurements with CAMAC and APVm look very
  promising
• plan to measure with Beetle read-out