Level-1 Trigger CMS Week, Brussels 14 Sep. 2011 C.-E. Wulz Deputy Trigger Project Manager Institute of High Energy Physics, Vienna Prepared with slides/material from: L. Guiducci, M. Jeitler, P. Klabbers, M. Konecki, A. Kropivnitskaya, E. Perez, A. Tapper Trigger menu development currently running with “3e33 menu” − menu designed for instantaneous luminosity of 3×1033 cm-2 s-1 − small updates applied last week one more menu planned for 2011 proton run C.-E. Wulz 1 CMS Week, Sep. 2011 Recent L1 trigger developments new muon combination scheme proposed in GMT − − − − − instead of taking minimum PT of two systems take the one with the higher rank will lead to a few percent better efficiency influence on rates has to be tested new LUTs and configuration key made currently being analyzed new CSC PT assignment − based on CLCT in ME1 new features in RPC PACT − changed algorithm, to increase efficiency − HSCP trigger: 2 BX wide and 1 BX delayed wrt DT and CSC C.-E. Wulz 2 CMS Week, Sep. 2011 CSCTF PT assignment Golden eta region: 1.2 < η < 2.1 2 or 3 station tracks with ME1 Improved ME1 CLCT patterns for 2 station tracks: - curves PT ≥ 5 GeV the same for PTLUT2010 and PTLUT2011 as expected; CLCT pattern starts to work from PT > 5 GeV - Efficiency below threshold drops from 5-10% (2010) to ~2% (2011) for PT thresholds ≥ 7 GeV and up - Significant improvement in 2011 compared to the 2010 PT assignment, especially for PT ≥ 7 GeV, lowering the trigger rate C.-E. Wulz 3 CMS Week, Sep. 2011 RPC PAC trigger improvements in 2011 • In 2011 the RPC PAC trigger algorithm was changed: in the barrel it requires at least 3 layers fired (out of 6 available) to generate a muon candidate. 4 layers were required in 2010. It was possible since the chamber noise is lower than expected. efficiency in the barrel increased Offline muon pT > 8GeV/c • The optical links were arranged to obtain better geometrical coverage in the region between wheels 0 and +/- 1 additional increase of the efficiency in Offline muon pT > 7GeV/c || ~0.3 • The PAC is now able to trigger on “slow” particles, which reach the muon system in the next BX: it looks for the coincidence of hits in two consecutive BXs, the candidate BX is L1 RPC Trigger Efficiency including detector geometrical acceptance and hit efficiency vs L1 CSC or DT trigger determined by the BPTX trigger veto. C.-E. Wulz 4 CMS Week, Sep. 2011 Stability of L1RPC after automatic HV versus pressure correction no corrections from pressure changes • • • • HV(p) corrections at the beginning of each fill Variation of chamber efficiency affects the absolute trigger efficiency (difficult to follow) and pT spectrum check fraction of L1RPC candidates with given pTCode vs all. after applied modifications RPC trigger pT spectrum stable vs pressure stability better in barrel than in endcaps C.-E. Wulz 5 CMS Week, Sep. 2011 RPC data synchronisation • Only small modification of timing settings done in 2011 C.-E. Wulz 6 CMS Week, Sep. 2011 Recent L1 trigger developments (continued) new jet energy corrections have been developed − had only been applied up to 92 GeV − 1-hour test run with new corrections has been taken − modifications to be applied for next menu (“2011, version 7”) C.-E. Wulz 7 CMS Week, Sep. 2011 L1 jet energy calibration C.-E. Wulz Uncorr In 2010 running no jet energy corrections Jet energy corrections were derived from Monte Carlo for 2011 and tested with data (Brown University) Corrections tested up to 130 GeV (saturation value in 2010) For higher luminosity need to extend corrections up to higher PT ΔR<0.5 match between L1 and RECO jets UNCORRECTED 2011 SO FAR CORRECTED 2011 DATA 2011 DATA 8 CMS Week, Sep. 2011 L1 jet energy calibration New jet energy corrections derived from Monte Carlo using JetMET code Tests on existing data and Monte Carlo Extends high-PT region and improves low-PT region a little Test run taken and validated ready for physics NEW CORRECTIONS Monte Carlo New corrections Old corrections ±10% 2011 DATA C.-E. Wulz 9 CMS Week, Sep. 2011 Recent L1 trigger developments (continued) New beam gas triggers − triggers based on BSC not useable at current high luminosity Present beam gas trigger based on HF, uses unpaired bunches suffers from strong “albedo” after trains of colliding bunches − background from delayed nuclear reactions New triggers use special BPTX signal after “quiet” period without collisions − 500 ns, “post-quiet unpaired BPTX signal” − thanks to BRM group for supplying this signal C.-E. Wulz 10 CMS Week, Sep. 2011 “post-quiet unpaired bunches” beam 1 C.-E. Wulz 11 CMS Week, Sep. 2011 Recent L1 trigger developments (continued) new ideas for luminosity measurement − needed because of pileup effects distorting zero-counting method − proposed by Marco Zanetti and Nicola Bachetta based on Pixel cluster counting requires zero-bias trigger that is only active for a few bunch crossings in orbit − so that within rate budget (1 kHz at L1) each bunch crossing gets reasonable statistics within a lumi section (23 seconds) − luminosity depends strongly on bunch crossing − Global Trigger firmware update allows to select individual bunch crossings for a trigger algorithm C.-E. Wulz 12 CMS Week, Sep. 2011 L1 trigger rate capability rate limits at Level 1 have been rechecked with real data − by applying progressively lower prescales to L1_SingleEG5 CMS runs up to design value of 100 kHz without significant deadtime − deadtime problems sometimes observed were caused by beam conditions (PKAM events in Pixels) and not by high rate C.-E. Wulz 13 CMS Week, Sep. 2011 L1 trigger cross section monitoring - Cross-sections of selected (*) L1 triggers vs instantaneous luminosity - WBM fit is used to compute the “expected” cross-section - (*) single object triggers, with the lowest PT/ET cut which is unprescaled in all prescale columns of the menu C.-E. Wulz 14 CMS Week, Sep. 2011 L1 trigger cross section monitoring - Measured/Expected cross-section vs lumi section number for certification - Currently used by shifters to fill Run Registry - Goal is to generate GOOD/BAD LS ranges automatically C.-E. Wulz 15 CMS Week, Sep. 2011 Trigger cross section history P. Musella, J. Pela 2e33 (EG15) 3e33 (EG20) ~August ~July 1.4e33 (EG12) - Jumps due to the different selection of monitored trigger object - Can check stability of the triggers C.-E. Wulz 16 CMS Week, Sep. 2011 Predictions for 5e33 based on the 3e33 data Data from run 175921 (3e33) used to emulate the “5e33 column” of the current menu : lead to 80 kHz + 10% (PU from 3e33 to 5e33) = about 90 kHz Current plan for the 5e33 menu : - quite similar to the “5e33 column” that we are running currently : unprescaled: EG20, SingleMu16_Eta2p1, SingleJet128 - A new L1 menu will be deployed though, with : - L1_DoubleEG_15_5 (7 kHz) taking over from DoubleEG_12_5 - brings the expected rate to 75 kHz + 10% PU = 82 kHz - L1_Muxx_MuOpen - to be defined - SingleMu10_ETM20 (tiny rate) - DoubleJet36_Central prescaled by only 5 - will add ~ 3 kHz i.e. expect ~ 85 kHz which should be sustainable. Note that the rate in cross-triggers starts to be large. C.-E. Wulz 17 Prediction based on 3e33 data (PU increase not included) CMS Week, Sep. 2011 Effect of Pile-Up on L1 rates : trigger x-sections in latest runs Example seeds (mostly unprescaled at 3e33) that are most sensitive to PU : TripleJet_36_36_12 EG18_ForJet16 (finally p’ed a bit) Xsection increased by 40% between 2e33 and 3e33 QuadJet20_Central +75% C.-E. Wulz C.-E. Wulz Increase x 2 TripleEG_8_5_5 +20 % 18 ETM_30 + 30% CMS Week, Sep. 2011 CMS Week, Sep. 2011 Effect of Pile-Up on L1 rates : trigger x-sections in latest runs Most seeds, especially the widely used higher thresholds seeds, behave better : SingleJet92 + 5% SingleEG15 +4% SingleMu14_Eta2p1 Roughly : Seeds that are sensitive to PU account for 25 - 30% of the rate. C.-E. Wulz 19 CMS Week, Sep. 2011 The L1 menu (L1Menu_Collisions2011_v5) deployed for the 3e33 menu Prediction of rates and first definition of prescales based on earlier data ( ~ 1e33) : • emulation of the total L1 rate for a given scenario at 3e33 was obtained from the earlier data, using the L1Accept stream (“nanoDST”, ~ 5 kHz of L1A events, with only the L1 information) • neglecting the PU effect: total predicted rate at 3e33 was ~ 73 kHz. • estimated PU effect from 1e33 to 3e33 : increase the total L1 rate by 10-15%. • i.e. we were expecting a total rate of 80 kHz. Rates actually measured at 3e33: • 80 kHz in run 176163 (Sept 12) at 2.9e33. • i.e. the predictions were in the right ballpark • see breakdown in the plot • biggest are • EG15 (16 kHz) • SingleMu16 (13 kHz) • DoubleJet44 (12 kHz) C.-E. Wulz 20 Measured rates (kHz) at 3e33 CMS Week, Sep. 2011 PU effect on total L1 rate in recent runs Compare rates of two runs taken with prescale index = 2 : run 176201, LS 135 – 145, lumi = 3e33 - correct for the deadtime of ~ 10% : L1A = 84 Hz run 175921, LS 400 – 420, lumi = 2.2e33 - correct for the fact that TripleJet seed was still unprescaled in that run, while it is prescaled in 176201 : L1A = 57 Hz PU effect on the total L1 rate, from 2.2e33 to 3e33 : + 8 %. • Assuming a similar slope between 3e33 and 5e33 : - the PU would increase the L1 rate by about 10%. (Things are likely to be a bit worse though, because cross-trigger seeds (low PT objects, hence PU sensitivity) contribute more to the total rate at 5e33.) C.-E. Wulz 21 CMS Week, Sep. 2011 Muon trigger PT cuts: turn-on curves - Study per PT cut between 10-20 GeV/c with data from 1-2e32 running - Preparation of the 5e33 menu - Turn-on positions and slopes are OK C.-E. Wulz 22 CMS Week, Sep. 2011 Muon trigger PT cuts: rates |eta|<2.4 |eta|<2.1 L=5E33 - Rate measured using NanoDST data, projected to 5E33 luminosity - >70% of total rate from |eta|>2.1 - Higher PT cuts less and less effective at high - Introduced cut at 2.1 - Keep lower L1 thresholds - Some gain in plateau efficiency L1 Eta cut C.-E. Wulz 23 CMS Week, Sep. 2011 Muon trigger: efficiency losses and rates L=5E33 Threshold 10 12 14 16 18 20 Eff loss for pt>30GeV 0% (*) 0.3% 1.1% 1.4% 1.6% 2.0% Rate [kHz] (|eta|<2.4) 44.5 35.1 28.4 23.2 20.9 18.9 Rate [kHz] (|eta|<2.1) 16.0 11.7 8.22 6.48 5.65 4.86 (*) efficiency loss computed wrt L1_SingleMu10 C.-E. Wulz 24 CMS Week, Sep. 2011 Total RPC trigger rate • We do not expect significant non-linear effects in scaling from 3e33 cm-2s-1 to 5e33 cm-2s-1 • RPC rate on plots below is doubled RUN 176201 (peak lumi: 3.1e33cm-2s-1) C.-E. Wulz 25 CMS Week, Sep. 2011 Conclusions Improvements to most L1 trigger systems have been made. Rate capability has been rechecked with data. Predictions for rates at 5e33 have been made, taking into account pileup. Preparations to run at 5e33 are well under way. C.-E. Wulz 26 CMS Week, Sep. 2011 BACKUP C.-E. Wulz 27 CMS Week, Sep. 2011 PU effect from 2.2 to 3e33: Jet seeds L1_QuadJet28_Central L1_QuadJet20_Central L1_TripleJet28_Central L1_TripleJet_36_36_12_Central L1_DoubleTauJet44_Eta2p17 L1_DoubleTauJet36_Eta2p17 L1_DoubleTauJet40_Eta2p17 L1_DoubleTauJet32_Eta2p17 L1_DoubleJet64_Central L1_DoubleJet52_Central L1_DoubleJet52 L1_DoubleJet44_Central L1_DoubleJet36_Central L1_SingleJet128 L1_SingleJet92_Central L1_SingleJet92 L1_SingleJet80_Central L1_SingleJet68 L1_SingleJet52_Central L1_SingleJet52 L1_SingleJet36 L1_SingleJet16 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 These plots show the ratios of L1 rates measured In two runs with different PU conditions (see slide 21), normalized to the same luminosity. C.-E. Wulz 28 CMS Week, Sep. 2011 PU effect from 2.2 to 3e33: EGamma seeds L1_TripleEG_8_8_5 L1_TripleEG_8_5_5 L1_TripleEG7 L1_TripleEG5 L1_DoubleEG_12_5_Eta1p39 L1_DoubleEG_12_5 L1_DoubleIsoEG10 L1_DoubleEG10 L1_DoubleEG5 L1_DoubleEG3 L1_SingleEG30 L1_SingleEG20 L1_SingleEG15 L1_SingleEG12 L1_SingleEG22 L1_SingleEG18 L1_SingleIsoEG12_Eta2p17 L1_SingleIsoEG12 L1_SingleEG5 0.9 C.-E. Wulz 0.95 1 1.05 29 1.1 1.15 CMS Week, Sep. 2011 1.2 PU effect from 2.2 to 3e33: Muon seeds (note: also GMT changes) L1_TripleMu0 L1_DoubleMu_5_Open L1_DoubleMu0_HighQ_EtaCuts L1_DoubleMu0_HighQ L1_DoubleMu5 L1_DoubleMu3p5 L1_DoubleMu3 L1_DoubleMu0 L1_SingleMu25 L1_SingleMu20 L1_SingleMu16_Eta2p1 L1_SingleMu16 L1_SingleMu14_Eta2p1 L1_SingleMu12_Debug L1_SingleMu12 L1_SingleMu10 L1_SingleMu7 L1_SingleMu5_Eta1p5_Q80 L1_SingleMu3 L1_SingleMuOpen 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Recent GMT changes increased the muon rates in latest runs. Hence the rate ratios do not correspond only to the PU effect. C.-E. Wulz 30 CMS Week, Sep. 2011 1.6 PU effect from 2.2 to 3e33: Global Sums L1_ETT220 L1_HTM50 L1_HTT75 L1_HTT150 L1_HTT100 L1_HTT50 L1_ETM70 L1_ETM50 L1_ETM30 L1_ETM20 L1_ETM100 0 C.-E. Wulz 0.2 0.4 0.6 0.8 31 1 1.2 1.4 CMS Week, Sep. 2011 1.6 "PU effect from 2.2 to 3e33: Cross-triggers" L1_HTT50_HTM50 L1_HTT50_HTM30 L1_Jet52_Central_ETM30 L1_Mu10_Jet36_Central L1_Mu7_Jet20_Central L1_Mu3_Jet28_Central L1_Mu3_Jet20_Central L1_Mu3_Jet16_Central L1_Mu0_HTT75 L1_Mu0_HTT50 L1_Mu10_Eta2p1_DoubleJet_16_8_Central L1_ETT300_EG5 L1_DoubleEG5_HTT75 L1_DoubleEG5_HTT50 L1_EG5_HTT125 L1_EG5_HTT100 L1_EG5_HTT75 L1_EG5_DoubleJet20_Central L1_EG18_ForJet16 L1_MuOpen_DoubleEG5 L1_DoubleMuOpen_EG5 L1_Mu12_EG5 L1_Mu7_EG5 L1_Mu5_EG12 L1_MuOpen_EG12 L1_Mu3_EG5 0 C.-E. Wulz 0.2 0.4 0.6 32 0.8 1 1.2 1.4 1.6 CMS Week, Sep. 2011 1.8
© Copyright 2026 Paperzz