D.Websdale, LHCC referees meeting, CERN, 1-7- 2002 1 Developments since 13 May meeting Outcome of review on 24 June HPD planning MaPMT planning D.Websdale Meeting with LHCC referees CERN, 1st July 2002 ICH photodetectors – status and planning
Jan 12, 2016
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 1
Developments since 13 May meeting Outcome of review on 24 June HPD planning MaPMT planning
D.Websdale
Meeting with LHCC referees
CERN, 1st July 2002
RICH photodetectors – status and planning
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 2
Pixel chip measurements at 40MHz HPD: absolute measurements of thresholds and
photoelectron detection efficiency Bump bonding developments MaPMT – magnetic field measurements
Developments since 13 May meeting
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 3
LHCb pixel chip tests Ken Wyllie at al
40MHz lab system to be adopted as global system for all tests – chips, wafers, assemblies, anodes, HPDs
Limited functions at the moment – problems with memory access on FLIC, under investigation
Restricted event size – use LHCb mode = 1024 pixels
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 4
Threshold and Noise scans at 40MHz
Threshold ~ 1130e-RMS ~ 110e-
Noise ~ 140e-
(~130e- @ 10MHZ)
Noise and threshold characteristics satisfy LHCb RICH requirementsThreshold < 2000eNoise < 300e
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 5
Pixel HPD (480 units)
80mm photocathode window
Electron Optics:20 kV5 x demagnification
1024 super-pixels 0.5mm x 0.5mm
PIXEL HPD
PGA ceramic carrier
Kovar ring
Bump-bonded chip assembly
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 6
Response to LED – silicon bias and HV
Detector bias scan High voltage scan
´= 1.741 @ 19kV, 80V
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 7
HPD -Threshold distribution
Gaussian fit:
m = 6.76kV (<1880e) = 0.82kV (230e)
Gaussian distribution reflects the comparator threshold distribution of the ALICE1LHCb chip (without threshold adjust)
Differential number of firing pixels as a function of HPD HV (Si det. bias = 80V)
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 8
Photoelectron detection efficiency(1)
Poisson fit
= 2.6@ 19kV,80V
´= 1.74 @ 19kV,80V
Number of firing pixels per LED pulse
Back-pulse spectrum
= average number of photoelectrons per LED pulse inferred from back-pulse fit
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 9
Detached bump bonds
Responding pixels: 70%
(Am, in HPD)
Responding pixels: 94%
(Sr, on bare anode)
The principal cause of low efficiency is detached bump bonds
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 10
Photoelectron detection efficiency(2)
LED shining smaller pixel area, where bump-bonds are generally good:
Analyze event size, infer ’ from P(0)
Correct for double pixel clusters
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 11
Photoelectron detection efficiency(2)
Record back-pulse spectrum, infer from fit; problems at low ADC channels under investigation.
Compare values of ’ and ; present estimates range from 81% to 83%; not corrected for LED drift with time, LED tail, missing bump-bonds, masked pixels, photoelectron pile-up
Error estimates:LED drift: 5-10%
Fit parameters: 5%
LED tail: a few %
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 12
Bump-bonding problem
Possible causesThree main differences between Omega3 (half-scale prototypes) and
ALICE1LHCb (full-scale prototypes):
bake-out temperature lower, dwell time shorter
elongated bump shape
increased die size and thickness
elongated bump shape does not correspond to relaxed shape when bumps re-melt
expansion coefficient mismatch between Si (3.5 ppm/C) and alumina (7 ppm/C) results in electronics chip downwards bending:
calculated sagitta with 2D bimetallic strip model: ~30 m at 300 ° C for 15mm-long die; goes with the square of the die size
bump-bond diameter is 32 mBUT no major degradation of bumps after silver glass curing process (dwell plateau
400 °C, dwell time ~10 min.)
incomplete curing process of silver glass, which could possibly experience further non-uniform retraction during subsequent bake-out cycle
metallurgy problems (UBM=Under Bump Metallization) ) resulting in degraded bump adherence
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 13
New assembly measurements(1)
Example with assembly 73:
Before bake-out: 6 missing bumps
After bake-out: 33 missing bumps
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 14
Testing new bump bonded assemblies
Sr source: response table:
Number of Responding pixels from 8192 [%]
Before bake-out After bake-out
8186 [99.93%] 8159 [99.60%]
8178 [99.83%] 8179 [99.84%]
8191 [99.99%] 8191 [99.99%]
8171 [99.74%] NA
8188 [99.95%] NA
New bump-bonding process better than ever;
New, unpackaged, bump-bonded assemblies completely survive bake-out cycle;
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 15
Bump bonding – road map to solution
Bake out sensor–chip assembliesIF detach Modify bonding process (VTT)
IF survive:
Bake out anode carrier assembliesIF survive Resume HPD production (DEP)
IF detach:Test following (4 month programme)
Stiffened ceramic carrierCeramic with lower thermal expansion coeffDifferent gluePartial surface glueing
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 16
HPD project – status
First HPD prototypeSystematic measurements with pulsed LED show expected photoelectron response… with the exception of the missing bump-bonds
New HPD preparationOn hold at DEP
Bake-out tests of bare new LHCb assemblies from VTT successful
Wait for bake-out tests of packaged new LHCb assemblies from VTT to proceed
New LHCb chip functional at 40 MHz;
Detector order approved and to be placed, mask and wafer layout to be started soon
New ceramic carrier mechanical draft design delivered; routing design underway
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 17
Longitudinal axis
MaPMT - Magnetic Field Tests
RICH 1 likely in magnetic field of 400 GaussMeasurements of MaPMT sensitivity to
longitudinal and transverse magnetic fields up to 35 mT (350 Gauss)
LED light source, APVm read-out
B [mT] vs I [A]
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 18
-metal shielding0.9 mm thicknessextension13 or 20 mm
shielded MaPMT 13 mm extension
Longitudinal B-Field
MaPMT with mumetal functions in longitudinal field up to 10 mT (100 G)
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 19
Transverse B-Field
Transverse field in x-direction
MaPMT insensitive to transverse fields up to 25mT (250 G)
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 20
RICH Photon Detector review(24 june)
June 2002 Milestone:HPD with encapsulated LHCb-ALICE pixel chip
95% working channelsThreshold < 0.4 signal (2000e)Noise < 0.15 threshold (300e)
LHCb pixel chip working at 40MHzThreshold, noise performance as above
If failure – switch to MaPMTSmall print – review if LHC delayed
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 21
RICH Photon Detector review(24 june)
Outcome of Review:- Recommendations
HPD: Solve bump bonding problem Delay HPD Milestone until end 2002
If failure – switch to MaPMT If success – produce small series of HPDs
- prepare 40MHz anode assemblies
MaPMT: Front-end chip is critical item Prepare fully functional BEETLE designs
for submission in October 2002
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 22
RICH Photon Detector history
Dec 1999: Pixel HPD chosen as baseline from 3 options“Lowest risk within budget”
Dec 2000: Milestone: Working HPDMissed: Pixel chip limited to 10MHz
Sept 2001: Milestone: HPD with 10MHz chipMissed: Delays in bump-bonding and anode
assembly
June 2002: Milestone: HPD with 10MHz chip + Working 40MHz chip
Missed: Bump bonds detached during bake-out
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 23
RICH Photon Detector history
Dec 1999: Pixel HPD chosen as baseline from 3 options“Lowest risk within budget”
Dec 2000: Milestone: Working HPDMissed: Pixel chip limited to 10MHz
Sept 2001: Milestone: HPD with 10MHz chipMissed: Delays in bump-bonding
and anode assembly
June 2002: Milestone: HPD with 10MHz chip + Working 40MHz chip
Missed: Bump bonds detached during bake-out
Months (anticipated)To LHC collisions
66
54
54
57
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 24
HPD planning – short term
Short term tasks: CERN group (3-4 FTE) + RICH team testingJune – Dec 2002:
Solve bump bonding problems:Bake out anode carrier assemblies
IF survive Resume HPD production (DEP)IF detach:
Test following (4 month programme)Stiffened ceramic carrierCeramic with lower thermal expansion coeffDifferent gluePartial surface glueing
Milestone: End 2002Demonstrate HPD with 10MHz chip satisfying all technical criteria
IF Failure switch to MaPMT
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 25
HPD planning – medium/long term
Medium term tasks: CERN group (3-4 FTE) + RICH team testingJan – March 2003:
Complete small (~4HPDs) series production:Produce and verify 40MHz anodes
April – October 2003:Complete 40MHz HPD prototypingPrepare specs/invite tenders for anodes and tubes
Milestone: Dec 2003 place orders for tubes
Long term tasks: CERN group (3 FTE) + Glasgow/Edinb. (4-5FTE)
March 2004-March 2006:Produce and test HPDsInstall HPDs
Milestone: Sept 2006 commissioning completed
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 26
MaPMT readout
MaPMT: (225k channels)
Readout options:
1. Binary (as now) 5.5CHF/channel2. Digital (as ITR scheme) 7.5 CHF/channel3. Analogue (as VELO scheme) 10 CHF/channel
Critical item for all is the front-end chip BEETLE is preferred LHCB compatible chip
BEETLE needs reduced (x ~50) gain Options to reduce gain:
Charge attenuator at pre-amp inputVoltage attenuator at shaper inputReduced pre-amp gainReduced PMT HVReduced number of PMT dynodes
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 27
MaPMT planning
Tube production and testing: March 2004-March 2006 is not on critical pathFront end chip with appropriate gain is critical component.PMT readout demonstrated at 40MHz with APV chip, but not LHCb compatiblePMT with attenuator + BEETLE pre-amp/shaper demonstratedBEETLE is preferred chip – valid for binary, digital or analogue readout schemes
Next steps: Short term tasks: Edinburgh, Oxford (+Heidelberg), Cambridge
July – Dec 2002:Planning meeting 15 JulyPrepare designs for fully functional BEETLE, adapted for MaPMTSubmission for Multi-Project Wafer run in October 2002
Medium term tasks:Decision on readout mode (costs are compared to HPD + readout chain)
Binary preferred - uses same readout chain as HPDDigital - use Inner tracker readout schemeAnalogue - use VELO readout scheme
Install HPDsMilestone: Sept 2006 commissioning completed
D.Websdale, LHCC referees meeting, CERN, 1-7-2002 28
Summary
•The 40 MHz pixel chip appears to work
•The bump bonds intrinsically survive the bake outif a problem remains it is related to the packaging in the carrier
•The prototype HPD functions as expected in all respects (modulo detached bonds)
•Sufficient time remains to complete the project provided the bump bonding is solved by end 2002
•Resources will be found to produce and test a fully functional BEETLE, with appropriate gain for MAPMT, for early 2003.
•The MaPMT remains a viable backup.