SVT detector electronics Mauro Villa INFN & Università di Bologna Overview: - SuperB SVT - Front End chips - Layer 0 peculiarities - SVT channels and rates - Data transmission - SVT in trigger and DAQ schemas
SuperB SVT 40 cm 30 cm 20 cm Layer0 Layer Radius 0 1.5 cm 1 3.3 cm 2 4.0 cm 3 5.9 cm 4 9.1 to 12.7 cm 5 11.4 to 14.6 cm Baseline: use an SVT similar to the BaBar one adding a Layer 0 Layer 0 options: striplets or thin pixels (40x40 um^2). Each layer has several modules, mechanically independent units (52+8) Each module has 2 half-modules, electrically independent units: Sensor, front-end chips, HDI with power/signal input and data output link Pixel half-module Strip(lets) half-module Data Pixel sensor + front-end chips HDI Power/Signal HDI Si Wafers Data Power/Signal Front-end chips Paris, 16/02/09 M. Villa
Data driven vs triggered FE chips Data Driven chips FSSR2: only minor modifications needed for SuperB MAPS: no storage of data in local high radiation environment Could allow to use SVT data for LV1 trigger Need very fast link to send all data from HDI to DAQ (layer 0 critical! See next tables) Triggered chips Sending off chip only data from LV1 evt reduces by ~ 1/10 the load to the data transmission Need to investigate if there is a good match for striplets! MAPS readout: data out of high rad desiderable. Data stored outside vertex region in custom high rad memory chips. Pixel sensor + front-end chips Off Detector HDI Power/Signal DAQ Data HDI Power/Signal Very Fast link DAQ On Detector Paris, 16/02/09 M. Villa link Data
Layer0 peculiarities Rad hard environment Large machine background ( 15 mm from beam line) Fundamental the first point of a track for vertexing Option under study: strip or pixel sensors (40 um pitch) Apsel6D – or similar 320x216 pixels a 40 um pitch. Active Area = 110 mm2 Completely data driven architecture Space time coordinates Time granularity 0.2-5.0 us (1.0 us is the goal) External Time stamp clock. Hit rate expected: 110 Mhit/s/chip (with a safety factor>5). Paris, 16/02/09 M. Villa
SuperB Layer 0 Module Plan: 2009-2010 Production of a test module 6 Chips: 1.1 cm^2 Signal lines: 180 in 1 cm width Hit Rate: 110 MHz/chip (safety>5) Module Rates: 660 MHz Bandwidth: 20 Gbit/s (design parameters) Thin aluminium bus for power and signals Main technological challenge: Line spacing (<100 um) Frequency (> 100 MHz) Plan: 2009-2010 Production of a test module With 2-3 APSEL5D chips Electric and beam tests Paris, 16/02/09 M. Villa
SuperB SVT in numbers 6 layers: Layer0 + 5 layers like BaBar SVT Modules HDI ReadOut Section (ROS) chips/ROS chips channels 0- Striplets 8 16 32 6 192 2.46E+04 0- Pixels 12.59E+06 1 12 24 7 168 2.15E+04 2 3 10 240 3.07E+04 4 64 5 288 3.69E+04 18 36 72 324 4.15E+04 total 60 120 1316 12.97E+06 Track rate @ layer0: 5.3 GHz (safety factor 5 included) fully dominated by machine bkgd Paris, 16/02/09 M. Villa
Load to the data transmission Requirements on link speed reduced by 1/10 Assuming 1 us window and 100 KHz LV1 rate Simulated Background rate x20 (cluster multiplicity and safety factor) Layer flux (Hz/cm2) 1 us occupancy HDI Link speed Read all data Data driven ( bit/s/ROS) Read only LV1 evt (bit/s/ROS) hit/evt bit/evt (with 25 bit word) 0- Striplets 1.00E+08 9.00E-01 1.73E+10 1.73E+09 2.21E+04 5.53E+05 0- MAPS 2.50E-03 1.64E+10 1.64E+09 2.10E+04 5.24E+05 1 2.00E+05 2.00E-02 4.48E+08 4.48E+07 4.30E+02 1.08E+04 2 3 1.00E+05 6.40E+08 6.40E+07 6.14E+02 1.54E+04 4 noise occu 1.00E-02 1.60E+08 1.60E+07 3.69E+02 9.22E+03 5 4.15E+02 1.04E+04 evt size L1-5 7kByte evt size L0 66kByte Present BaBar data Similar in SuperB L1-L5 Paris, 16/02/09 M. Villa
Data transmission Option A: HDI with local rad-hard ram,optical links, L1 handled in the HDI Option B: HDI with line drivers, copper cables, L1 handled outside detector Memory buffers And L1 logic Buffers and line drivers optical/copper Link Optical Link 2.5 Gbit/s FEC ROM Off detector low rad area Counting room Std electronics On detector High rad area Pro B: fewer rad-hard elements few space required on HDI full event readout (trigger) commercial optical link only Pro A: data flux reduced asap shorter high load connections Paris, 16/02/09 M. Villa
EDRO Readout board Front End Card used in Slim5 beam test (SVT par. II): Main concept: flexibility at all levels Mezzanines to decouple input/processing/output Large FPGA Several triggering schema Slink To DAQ SuperB: 60 Mhz bus clocks 20 Gbit/s input rate 2.5 Gbit/s output rate To be addressed in the next versions foreseen in 2009/10 Performances: 40 Mhz bus clocks 8 Gbit/s input rate 1 Gbit/s output rate 2.5 ME/s evaluated for L1 40 kHz DAQ rate @slim5 Data from FE chips 4 Gbit/s EPMC Stratix EPMC TTCRQ Not too far from SuperB numbers Paris, 16/02/09 M. Villa
SVT in trigger and DAQ schemas Current best FE chip candidates are data driven with an intrinsic time resolution of 0.2-1.0 us Interface between FE chips and SuperB trigger and DAQ will be provided by a board equipped with large FPGA, memory and optical links (i.e. flexible) Consequences SVT will fit both the fixed and variable L1 latency schemas SVT readout time window of 1 us or larger will be handled in the HDI or off-detector electronics. Same for fixed or variable time window. SVT might have the possibility to provide information on events before L1 trigger (L1-L5 best, L0 maybe). SVT in trigger conceiveble. Paris, 16/02/09 M. Villa
Conclusions Main SVT challenge: layer 0 rates, chips, data busses and data transmission Current best FE chip candidates are data driven Front End Card and/or HDIs with enough logic to fit different DAQ and L1 schemas Trigger rate >150 kHz, time window >=1 us ✔ Data volume: L0: 66 kB with 1 us time window and >5 safety factor L1-L5: 7 kB with 1 us time window (Babar exp.) SVT in trigger conceivable SVT electronics R&D in 2009/10 Paris, 16/02/09 M. Villa