The CBC2 ASIC for 2S Modules at HL-LHC Davide Braga IOP 2014 Joint HEPP and APP Groups Annual Meeting, Royal Holloway, May 2014

Outline Tracker upgrade & detector module The CMS Binary Chip (CBC) v.1 & 2 CBC2: architecture & performance CBC2 testing Beam test TID test IOP 2014 MeetingDavide Braga, IC2

IOP 2014 Meeting3 Phase-II upgrade of the CMS Strip Tracker Baseline design: Barrel+5Endcaps Based on 2 module types only Provides at the same time: readout data upon receipt of L1 trigger trigger (10 trigger hits up to η=2.5) 2S strip-strip double-layers ~8400 modules ~34M channels ~155m 2 PS strip-strixel double-layers ~7000 modules ~230M channels ~62m 2

Basic trigger module concept Pass  StubFail R  ~200μm ~100μm ~2-4 mm outer sensor inner sensor High-PT tracks (stubs) can be identified if cluster centre in top layer lies within a search window in R-Φ (rows) p T cut given by: module radius (z), sensor separation and correlation window Davide Braga, ICIOP 2014 Meeting4

Davide Braga, ICIOP 2014 Meeting5 4 mm 5 mm 7 mm 11 mm CBC1 CBC2 2 versions have now been produced - both in 130nm CMOS CBC1 (2011) 128 wire-bond pads, 50 mm pitch front end designed for short strips, up to 5 cm DC coupled, up to 1mA leakage tolerant, both sensor polarities binary unsparsified readout pipeline length 6.4 msec chip worked well in lab and test beam no triggering features CBC2 (January, 2013) 254 channels ~same front end, pipeline, readout approach as CBC1 bump-bond layout includes triggering features CMS Binary Chip (CBC)

r  beam offset window width top layer clusters bottom layer clusters zero offset Cluster width discrimination (CWD) logic exclude clusters with hits in >3 neighbouring channels wide clusters not consistent with high pT track Offset correction & correlation logic for a cluster in bottom layer, look for correlating cluster occurring in window in top layer window width controls pT cut stub found if cluster in bottom layer corresponds to cluster within window in top layer window width programmable up to ± 8 channels offset defines lateral displacement of window across chip programmable up to ± 3 channels n n+1 n-1 1/2/3 strip cluster on channel n programme cluster width to accept channel comparator outputs CWD logic Stub finding logic 6Davide Braga, ICIOP 2014 Meeting

CBC2 architecture pipe. control FE amp comp. pipeline shift reg. v th 256 deep pipeline + 32 deep buffer test pulse bias gen. fast control slow control stub shift register offset correction & correlation cluster width discrimination MHz diff.clock signals in blue are single-ended trig’d data out stub shift reg. O/P trigger O/P I2C 254 channels: 127 from each sensor layer channel mask: block noisy channels from trigger logic CWD logic: exclude wide clusters >3 correlation logic: for each cluster in lower layer look for cluster in upper layer window trigger output: 1 bit per BX indicates correlation logic found one (or more) stubs triggered data out: unsparsified binary data frame in response to L1 trigger nearest neighbour signals T1 trigger fast reset test pulse I2C refresh 4411 chan. mask 4411 nearest neighbour signals reset OR_254 OR_stubs Ck Davide Braga, ICIOP 2014 Meeting7

8 CBC2 performance All core functionality meets requirements Correlation functionality verified with test pulses, cosmics (backup), and in test beam Analogue performance close to simulation and specifications e.g. 1000e noise for 5 cm strips (~8 pF) achievable for total channel power of 350 uW noise & power vs. external capacitance noise [e RMS ] power/chan. [uW] external capacitance [pF] electrons mode / measured / simulation S-curves channel offsets tuning counts comparator threshold [I2C units] holes mode before tune after M. Raymond Davide Braga, ICIOP 2014 Meeting

CBC2 testing activities Wire-bond CBC2 Useful to develop wafer probe procedures X-rays TID testing 2xCBC2 hybrid Hybrid characterization and chip integration Bump-bonded ASICs Inter-chip links & logic Davide Braga, ICIOP 2014 Meeting9 2xCBC2 mini-module + sensor Sr-90 source Cosmic rays Beam Test

10 Pt module beam test at DESY TOP VIEW: strip direction into page positron beam low angular divergence (small angular error) scintillator module #4 on rotation stage module #3 on fixed assembly Mimosa26 MAPS (DATURA telescope) ~3 GeV e+ Datura telescope planes pT module December GeV positron beam Datura telescope + 2 pT modules (1 rotatable to simulate B-field effect) + 2 different strip sensors Custom control and DAQ Davide Braga, ICIOP 2014 Meeting

S. Viret Davide Braga, ICIOP 2014 Meeting11 Beam test results Pt Selection cut: simulationmeasured efficiency Other measurements include: Study of cluster width System noise Beam profile avg. cluster width vs angle of incidence EXP. p T cut: 2.14 GeV/c FIT: p T cut: 2.17 GeV/c σ: 0.1 GeV/c M.Pesaresi reconstructed p T cut of r=75cm layer Incident Angle Cluster Width

Total Ionizing Dose test First xray irradiation to 10 Mrads CBC2 operated continuously during irradiation monitored currents, biases, pedestal, noise no significant change in performance, moderate increase in current before annealing Next: TID test up to ~50 Mrads Davide Braga, ICIOP 2014 Meeting12 Xray tube CBC2 Bias/readout

Davide Braga, ICIOP 2014 Meeting13 Milestones so far: CBC2 working to specs Stub finding logic functioning TID dose irradiation test under way First prototype version of 2S module extensively tested Beam test at DESY successful Pt selection cut and stub logic demonstrated DAQ working well Future work: Continue with 2S module prototyping (8chip hybrid from CERN) SEU test of CBC2 CBC3: final prototype version with full stub readout First results to prove the track-trigger concept a very new idea never before implemented  CMS ambitious plans for a track trigger look promising Conclusions and Future Work A. Honma

Backup slides Davide Braga, ICIOP 2014 Meeting14

Davide Braga, ICIOP 2014 Meeting15 M.Pesaresi Expected Spectrum

Davide Braga, ICIOP 2014 Meeting16 < 3e+05 Gy = 30 Mrad NB: calculated for 3000 fb -1 but with present Tracker BRILRadiationSimulation HL-LHC dose

Davide Braga, ICIOP 2014 Meeting17 2S Module Assembly From Antti Onnela:

127 bottom top 8 bits to describe cluster address in bottom layer 5 bits to describe correlating cluster address in top layer window final choices for front end Optimized for 5cm strips (possibly longer), AC coupled, n-on-p stub data definition ½ strip cluster resolution Increased max acceptance width from 3 to 4strip clusters 8b address (for ½ strip resolution) of cluster in bottom layer 5b for stub bend information (rough Pt) stub data formatting & transmission 13b/stub, up to 3 stubs/BX => 39 bits +1 bit unsparsified L1 triggered readout data => 40 bits / 25 nsec e.g. 10 lines at 160 Mbps (per chip) Designed for final trigger rate & latency Priority encoding of Pt stubs (if desired) other useful features e.g. slow ADC to monitor bias levels … concentrator CBC 10 lines / CBC CBC3 – the “final prototype” Davide Braga, ICIOP 2014 Meeting18