The CMS Level-1 Trigger System Dave Newbold, University of Bristol On behalf of the CMS collaboration

The CMS Level-1 Trigger System

Dave Newbold, University of Bristol

On behalf of the CMS collaboration

Dave Newbold, University of Bristol HEP2003 Conference - 19/7/2003

Triggering at the LHC Tiny signals, huge background

p-p inelastic rate: ~ GHz e.g. H(150) -> : < mHz …but bb ~MHz – background!

Complex events & detector Typical CMS event is >1MB Max storage rate : 100MB/s

Huge selectivity needed Trigger reduction factor: 107

[L1 o/p rate < 100kHz]

LHC collisions @ 40MHz Mean of 23 evts per BX at full

luminosity Detector response & time of

flight are > 1BX


Level-1 trigger strategy Driven by LHC physics conditions

Decays of rare and heavy particles against large “soft” QCD b/g Many decays involve intermediate W / Z; H -> also important -> Identify high-pt leptons* and photons (*including ) Low pt thresholds motivated by efficiency for W / Z / light Higgs

Trigger combinations >20GeV limit on single-lepton thresholds due to quark decay + 0 b/g Most interesting states decay to two or more trigger objects – can use

lower thresholds for objects in combination -> Find trigger objects locally, combine and cut only at last stage

Large uncertainties in background (and perhaps signal) Flexibility and control of rate are both vital -> All trigger thresholds and conditions must be programmable Trigger architecture is fixed, but this is a function of detector geometry

Must have high and well-understood efficiency -> Need to include overlapping and minbias triggers to measure


Trigger / DAQ architecture Level-1 uses muon & calo data only

Tracking data too large / complex Local pattern recognition is possible

Fully pipelined digital electronic system Physically impossible to make decision in

25ns (speed of light) All data stored on detector during fixed L1

latency, read out upon L1A Memory constraints give max latency 3.2s

(of which 2s is cable delay)

Output of Level-1 Single bit: accept / reject Triggers may be ‘throttled’ for technical

reasons – but otherwise, zero deadtime On L1A, data proceed via event builder

switch to High Level Trigger (see talk of G. Bagliesi)


Level-1 overview


Trigger system location


ECAL, HCAL cover to = 3, forward calorimeter to = 5 Trig prims group crystals / scintillators into (2 x) 32 x 72 trigger towers

Calorimeter trigger detectors


Calorimeter trigger algorithms


Calorimeter trigger performance

Full GEANT study Includes

minbias background

L=1034 cm-2s-1

Efficiencies For objects

within fiducial acceptance

Rates e/ dominated

by jet (0) background

Steep curves allow good control of rate


Muon trigger detectors

Dedicated RPC detectors Excellent time resolution for

effective BX-ID

Main DT and CSC detectors Excellent position resolution for

accurate pt reconstruction


Muon trigger algorithms


Muon trigger performance

Efficiency for any muon >3 GeV pt


Global trigger

Global trigger implements a wide range of triggers (incl. topological) Example low lumi (L=2.1033 cm-2s-1) trigger selection shown above

Total rate balanced between e/g, jets, muons for initial HLT input 50kHz Rate safety factor ~3, to account for uncertainties in background

Trigger type Thrsh (0.95 eff) Rate (kHz) Cum. Rate (kHz)

Incl. iso e/ 29 GeV 3.3 3.3

Di- e/ 17 GeV 1.3 4.3

Incl. iso 14 GeV/c 2.7 7.0

Di- 3 GeV/c 0.9 7.9

Single 86 GeV 2.2 10.1

Di- 59 GeV 1.0 10.9

1j, 3j, 4j 177, 86, 70 GeV 3.0 12.5

j && Etmiss 88 ; 46 GeV 2.3 14.3

e/, j 21 ; 45 GeV 0.8 15.1

Minbias, calib, efficiency estimation 0.9 16.0


Physics efficiencies

Typical efficiencies for preceding trigger table @ L=2.1033 cm-2s-1

Channel Efficiency % Efficiency % from each trigger type – (each) cumulative

W - > e 70 e: (70) 70

t -> eX 91 e: (82) 82 e. (62) 86 : (55) 89 jjj: (24) 90 e.j: (54) 91

Z -> ee 94 e: (93) 93 e.e: (76) 94

H(115) -> gg 99 e: (99) 99 e.e (82) 99

H(150) -> WW 87 e: (78) 78 e.(43) 81 : (34) 83 e.j: (39) 85 j: (28) 87

H(135) -> -> ej 84 e: (70) 70 e. (62) 86 e.j: (54) 91 : (38) 84 j: (34) 84

Charged H(200) 98 : (85) 85 jj: (77) 96 j.Etm: (60) 98

H(200) -> -> jj 81 : (75) 75 : (50) 79 j: (24) 81 jj: (9) 81

H(500) -> -> jj 99 : (94) 94 : (64) 94 j: (94) 99 jj: (73) 99

t -> jets 53 Ht: (39) 39 jjjj: (26) 43 jjj: (26) 46 jj: (21) 47 j: (35) 53

mSUGRA 99 j: (99) 99

H(120) -> bb 41 jjj: (12) 12 j: (27) 30 : (26) 41 jj: (16) 41

Invisible H(120) 44 j.Etm: (39) 39 j: (22) 41 : (13) 44


Practical challenges Technology

Pushing (today’s) digital processing and comms technologies to the limit Cannot afford huge outlay on custom components (-> FPGAs) System must last for 10+ years: obsolescence.

Synchronisation Time-of-flight and detector response take many BX Subdetector timing and bunch-crossing ID will be challenging

Reliability Level-1 trigger performs online selection: cannot correct mistakes System must be highly reliable, all data taking depends on it

• But some parts will fail or degrade at some time Some components on detector -> radiation, magnetic field considerations

Simulation / configuration control / real-time trigger optimisation Trigger integration for CMS begins 2004 - the real work starts now


Summary

Triggering at the LHC will be hard Leptons / photons are the key

CMS Level-1 trigger system currently under construction Reduces 40MHz BX rate to < 50KHz L1A Very large digital logic system Uses calorimeter and muon information only

Simulated performance shows good efficiency for the interesting channels

Documents

The CMS Level-1 Trigger System Dave Newbold, University of Bristol On behalf of the CMS collaboration