Very Forward Muon Trigger and Data Acquisition Electronics for
CMS: Design and Radiation Testing
21 Sept 2012
Jason GilmoreVadim Khotilovich
Alexei SafonovJoe Haley
TWEPP 2012 2
CMS Endcap Muon System
h = 2.4
h = 0.8
Focus on the innermost Cathode Strip Chambers: ME1/1 CSCs
J. Gilmore
TWEPP 2012 3
Overview of the CSC System
J. Gilmore
VME
ME1/1, High-Eta
ME1/1
TWEPP 2012 4
Overview of the CSC System
J. Gilmore
VME
ME1/1, High-Eta
ME1/1
New: Increase to 7 CFEBs, all with fiber links
TWEPP 2012
CSC: Frontend Trigger Problem• Out-of-time PU induces deadtime at
higher luminosity look at PU100• Particular issue is the ME1/1 “TMB”
building chamber track segments – Two aspects making ME1/1 special:
Very high occupancies ME1/1 TMBs effectively serve two chambers
(inner ME1/a, outer ME1/b)
• Need better FPGA to maintain efficiency– The algorithm is ready (V. Khotilovich)– Design of prototype TMB completed
• Improve muon trigger efficiency for |h|>2.1– Rate increase compensated by requiring 3
station coincidence for |h|>2.1 With new TMB can do w/o efficiency loss Needs firmware modifications in CSCTF 5J. Gilmore
TWEPP 2012 6
Snap 12 FiberReceiver
- fibers from 7 DCFEBs
Snap 12 FiberTransmitter
socket(used only on test boards)
Signal-level translators 3.3 V to 2.5 V
Virtex 6 FPGA + XF128 PROM QPLL
Dimensions: 7.5” wide by 5.9” high11.1 mm clearance from TMB main board
CSC TMB Mezzanine 2012
J. Gilmore
Finisar Transceiver, only on test boards
TWEPP 2012
TMB Mezzanine Location
7J. Gilmore
TWEPP 2012 8
Radiation Studies for New CSC Boards• Will the new components survive the expected exposure in the CMS
Endcap at HL-LHC?– Not just CSC trigger boards, but also for the front-end boards
DCFEBs and ODMB, as well as new TMB mezzanine Expected 1 MeV neutron fluence: 3 *1012 n/cm2 over 10-years
9 krad dose, do tests up to ~30 krad level for 3-times safety factor
• Will the Single Event Upset rates be unacceptably high?– FPGAs, fiber links, etc. used in front-end boards
Expected 20 MeV neutron fluence: 2.7 *1011 n/cm2 over 10-years Measure SEU cross sections for individual design elements
• Initial radiation testing was done in 2011– Digital components were tested with 55 MeV protons
Performed at the Texas A&M University Cyclotron facility
– Other components tested with ~1 MeV neutrons At Texas A&M University Nuclear Science Center reactor
A series of 5 exposures to test 40 different components
– Results to be published soon, paper accepted by NIMA • Additional 2012 tests completed recently at UC DavisJ. Gilmore
TWEPP 2012
Voltage Regulator Radiation Tests• Testing performed at the Texas A&M Nuclear Science Center
– 1 megawatt reactor operating at 6 kW, provides 9.9 *108 n/cm2s• Multiple samples of several COTS regulators, two exposures
– First exposure represents ~10 HL-LHC year dose (10 krad)– Second exposure adds ~20 HL-LHC years, total of 30 year dose (30 krad)– Regulator performance tested before and after each exposure
Regulators were unpowered during exposure
• Some regulators showed no ill-effects– National Semi LP38501 and LP38853– Micrel 49500 and 69502– TI TPS74901
• Others did not fare so well…– Maxim 8557– Sharp PQ035ZN1, PQ05VY053, PQ070XZ– TI TPS75601, TPS75901– No improvement seen with additional cool-down time
• More parts were tested later, all are summarized in following slides…
9J. Gilmore
TWEPP 2012 10
Summary of All Reactor Tests (1)Part/Chip Name Chip Type
10 krad Exposure Pass/Fail
30 krad Exposure
ResultComments
Maxim 8557ETE Voltage Regulator Pass Fail5 out of 6 dieat 30 krads
MIC69502WR Voltage Regulator Pass Pass MIC49500WU Voltage Regulator Pass Pass National Semi LP38501ATJ-ADJCT-ND Voltage Regulator Pass Pass National Semi LP38853S-ADJ-ND Voltage Regulator Pass Pass Sharp PQ05VY053ZZH Voltage Regulator Pass Fail Fails to regulateSharp PQ035ZN1HZPH Voltage Regulator 50% Pass Fail Fails to regulateSharp PQ070XZ02ZPH Voltage Regulator Fail Fail Fails to regulateTI TPS740901KTWR Voltage Regulator Pass Pass TI TPS75601KTT Voltage Regulator Fail Fail Fails to regulateTI TPS75901 Voltage Regulator Fail Fail Fails to regulateST Micro 1N5819 diode Pass Pass ON Semi 1N5819 diode Pass Pass 2N7000 FET transistor N/A Pass
AD8028ARHigh Speed, Rail-to-Rail Input/Output Amplifiers N/A Pass
ADM812 Voltage Monitor N/A Pass
LM41211M5-1.2Precision Micropower Low Dropout Voltage Reference N/A Pass
LM4121AIM5-ADJPrecision Micropower Low Dropout Voltage Reference N/A Pass
J. Gilmore
TWEPP 2012 11
Summary of All Reactor Tests (2)Part/Chip Name Chip Type
10 krad Exposure Pass/Fail
30 krad Exposure
ResultComments
LM1C1Z N/A Pass
MAX680CSA +5V to ±10V Voltage Converter N/A Pass
MAX664CSADual Mode 5V/Programmable Micropower Voltage Regulator N/A Fail Dead
MAX4372 High-Side Current-Sense Amplifier N/A Pass
MIC35302 High-Side Current-Sense Amplifier N/A Fail Dead
MIC37302 High-Side Current-Sense Amplifier N/A Fail Dead
MM3Z4V7C Zener Diode N/A Pass
MM3Z5V1B Zener Diode N/A Pass
PQ7DV10 Variable Output 10A Voltage Regulator N/A Pass
TPS7A7001 Very Low Dropout, 2A Regulator N/A Fail Fails to regulate
J. Gilmore
TWEPP 2012 12
Summary of All Reactor Tests (3)Part/Chip Name Chip Type
10 krad Exposure Pass/Fail
30 krad Exposure
ResultComments
SN74LVC2T45Two-bit Dual-supply
Tri-statable Bus Transceiver N/A Pass
ADM660AR CMOS Switched-Capacitor Voltage Converter N/A Pass
ADM8828 Switched-Capacitor Voltage Inverter N/A Pass
ICL7660S-BAZ Switched-Capacitor Voltage Converter N/A Fail Dead
LTC1044CS8 100mA CMOS Voltage Converter N/A Pass
MAX1044CSA Switched-Capacitor Voltage Converter N/A Fail Dead
MAX860-UIA "uMAX" Switched-Capacitor Voltage Converter N/A Pass
MAX861-ISA Switched-Capacitor Voltage Converter N/A Pass
TC1044SCOA Charge Pump DC-TO-DC Voltage Converter N/A Pass
TC962COEHigh Current Charge
Pump DC-to-DC Converter N/A Pass
J. Gilmore
TWEPP 2012
SEU Testing of COTS Components (1)• Testing performed at Texas A&M Cyclotron
– 55 MeV protons with uniform flux, collimated to 1.5” diam– Maximum proton flux ~3 *107 cm-2s-1
– 45 to 90 minute runs on each target device, 5-10 kRad in these tests
• Two samples tested for each COTS component– Reflex Photonics 3.5 Gbps Snap12 Receiver: model r12-c01001
Random PRBG data patterns @3.2 Gbps on each of six links FPGA drives data to Transmitter, fiber connects to Receiver and carries data back to FPGA
SEU cross section: s = (8.2 ± 0.3) *10-9 cm2
Also tested to ~30 krad TID at TAMU reactor: no problems– Reflex Photonics Snap12 Transmitter: t12-c01001
3.5 Gbps, tested for use in ODMB upgrade PRBG data patterns @3.2 Gbps on six links s = (7.3 ± 2.4) *10-11 cm2
– Finisar Optical Transceiver: ftlf8524e2gnl 4.25 Gbps, tested for use in DCFEB upgrade Transmit randomized GbE data packets to PC s = (1.0 ± 0.3) *10-10 cm2 13J. Gilmore
TWEPP 2012
SEU Testing of COTS Components (2)• Xilinx Virtex-6 FPGA: xc6vlx195t-2ffg1156ces
– No SEU mitigation in firmware for this study Goal is to measure cross section of individual FPGA elements Determine where mitigation is necessary
– GTX Transceiver (55% used) PRBG data transfers @3.2 Gbps s = (7.6 ± 0.8) *10-10 cm2
– Block RAM (74% used) 4 kB BRAM “ROMs” readout to PC s = (5.7 ± 0.6) *10-8 cm2
– CLB (38% used): 4 kB CLB “ROMs” readout to PC s = (3.7 ± 0.5) *10-8 cm2
• TI Bus-Exchange Level-Shifter: sn74cb3t16212– Randomized data patterns sent through all 24 signal paths– No SEU observed, s90% < 1.7 *10-11 cm2
14J. Gilmore
TWEPP 2012 15
Impact of the 2011 SEU Measurements• How would these cross sections affect CSC operations
in HL-LHC?– Snap12 Transmitter: < 1 SEU per year per link– Snap12 Receiver: ~1 SEU per week per link
These typically just affect a single data word
– Finisar Optical Transceiver: ~7 SEU/day/linkTypically just affects a single data wordLow rate, less than one error in 3 *1013 bits
– FPGA GTX Transceivers: ~3 SEU/year/link– FPGA Block RAMs: ~9 SEU/day/chip
These typically affect a single bit in a single cellNeed to investigate mitigation for FPGA BRAMs
– FPGA CLBs: ~6 SEU/day/chipNeed to investigate mitigation for FPGA CLBsJ. Gilmore
TWEPP 2012 16
Recent 2012 Radiation Studies• Testing at UC Davis Cyclotron– 64 MeV proton beam, flux up to ~1 *109 cm-2s-1
– Many of the same parts from previous SEU tests were retested using the same circuit boardsSnap12 parts are the only exceptions
New Emcore transmitters were tested in 2012All chips survived 30 kRad dose*
Monitored power for signs of latchup (none observed)
• Some FPGA tests included mitigation this time– Enabled native ECC feature in Block RAMs
BRAM test used Read & Write under software control Software designed to distinguish each failure mode
– CLB tests based on triple-voting systemCLBs were implemented as a system of shift registers
Given common inputs and checked against each other Error counts were recorded in registers and monitored by software
J. Gilmore
TWEPP 2012 17
SEU Test Results 2012 (1)• Reflex Photonics 3.5 Gbps Snap12 Receiver: r12-c01001
– Random PRBG data patterns @3.2 Gbps on each of eight links– These SEUs only caused transient bit errors in the data– 2012 SEU cross section result: s = (6.4 ± 0.2) *10-9 cm2
– Combined 2011+2012: s = 9.5 *10-10 cm2 per link Similar to 2011 result, about 40% smaller
• Emcore 3.3 Gbps Snap12 Receiver: EMRS1216– Same PRBG test as above– 2012 SEU cross section result: s = (9.8 ± 0.2) *10-9 cm2
This gives s = 12 *10-10 cm2 per link Similar to Reflex Photonics combined result, about 30% larger
• Emcore 3.3 Gbps Snap12 Transmitter: EMTS1216– Same PRBG test as above; tested two of these parts– These SEUs only caused transient bit errors in the data– 2012 SEU cross section: s = (1.7 ± 0.2) *10-10 cm2
This gives s = 2.1 *10-11 cm2 per link Nearly double the 2011 result for Reflex Photonics transmitter Still very low rate of SEUs, so not a concernJ. Gilmore
TWEPP 2012 18
SEU Test Results 2012 (2)• Finisar Optical Transceiver ftlf8524e2gnl: Transmit side
– Gigabit Ethernet packet transmission tests to PCI card, 4 kB @ 500 Hz Bad or missing packets received at the PC are “transmit” SEUs
– These SEUs caused lost GbE packets and rare “powerdown” events– 2012 SEU cross section result: s = (4.3 ± 0.3) *10-10 cm2
About 6 times the 2011 result; consistent with *6 increase in link duty cycle
– Correcting for real CSC transmitter duty cycle: s = 8.2 *10-9 cm2 per link We expect to see ~1 SEU per link per day during HL-LHC running
• Finisar Optical Transceiver ftlf8524e2gnl: Receive side– New test in 2012, load the BRAMs with data and read them back
Errors read back twice the same way are “receive” SEUs
– These SEUs only caused transient bit errors– 2012 SEU cross section: s = (7.5 ± 0.1) *10-9 cm2 per link
We expect to see ~1 SEUs per link per day
– *Three Finisars tested: one died at 33 krad, another at 41 krad The third chip survived with 30 krad
• TI Bus-Exchange Level-Shifter: sn74cb3t16212– Still no SEU observed, 2011+2012 result: s90% < 4.0 *10-12 cm2
J. Gilmore
TWEPP 2012 19
FPGA SEU Results 2012• GTX Transceiver (55% used in FPGA)
– Random PRBG data patterns @3.2 Gbps on each of eight links– These SEUs only caused transient bit errors in the data– 2012 SEU cross section result: s = (10 ± 0.8) *10-10 cm2
Similar to 2011 result, ~50% larger, consistent with additional active links
• Block RAM (74% used in FPGA)– Built-in ECC feature was used to protect data integrity– Software controlled write and read for BRAM memory tests– No errors were detected in the BRAM contents: mitigation at work– 2012 SEU cross section: s90% < 8.2 *10-10 cm2
• CLB (43% used in FPGA)– Most of the logic is a shift register system with voting– Some of it was unvoted logic for control and monitoring
This masks the “mitigation” effect of voting somewhat– 2012 SEU cross section result: s = (6.0 ± 0.5) *10-9 cm2
Much smaller than 2011 SEU result, factor of 6 better: mitigation at work With this we expect ~1 CLB SEU per FPGA per dayJ. Gilmore
TWEPP 2012
Conclusion• TMB Mezzanine development coming to a close– We have a design and production plan for new CSC
electronicsThis will maintain a high level of efficiency for the foreseeable future
– Prototypes have been built & testedGood results from radiation tests
– We have found satisfactory COTS parts to meet all our design requirements
– Development work still needed in SEU mitigation firmware
• Final CSC ME1/1 Electronics production begins soon– Need over 500 DCFEBs, plus spares: starts next month– Need 72 each for new TMB and ODMB, plus some spares
Start producing these early in 2013
– Installation in CMS from June-August 201320J. Gilmore