RUN 2B L1CAL MEETING MINUTES 23 January, 2003 Present o Arizona: E.Varnes o Fermilab: B.Kehoe, V.Pavlicek o MSU: M.Abolins, P.Laurens o Nevis: H.Evans, J.Mitrevski o Northeastern: D.Wood o Saclay: J.Bistricky, D.Calvet CalTrk Match Simulation: Erich Varnes ----------------------- o Run IIa tsim_l1l2 _ tsim_l1cal2b - signal: H-->tautau (M = 135 GeV) - compare L1cal2b jets with tracks o Tau Kinematics - tau's generally back-to-back o Trigger Response - using Jet211 bank - Et of trigger jets rather low . not sure if this is all due to invisible energy from tau - Isolation: next closest jet is usually from other tau . no Et threshold on jets (5 GeV in tsim_l1cal2b) . probably should raise this thresh. o Trigger Correlations - fairly good cor. b/w L1CTT tracks and trigger jets - Track Pt vs Et: some correlation . but track Pt is binned - highest val > 10 GeV o CalTrk Match - match = sector with track overlaps jet centroid - Et thresholds = 4, 20, 35, 50 GeV - 16 triggers 4CTTx4Cal thresh's - efficiency for any given trigger ~10% - bgrd rejection --> x>100 if require . track match + Cal threshold above lowest - QCD files used? L1CalExamine in Run IIa: Bob Kehoe ----------------------- o L1CalExamine - shifter uses while data-taking in progress - histo files kept that log different distrib's - shifter and expert plots available - can be configured online (via rcp's) o Cal Precision Readout - crucial to have comparison: trigger vs prec. readout - construct TT's from prec. readout online - ==> establish calibration, find bad TTS, etc. - plots of Trigger(TT) vs Prec(TT) o Particularly Annoying Problem - TTs go hot and dissipate in <1 minute . probably at BLS level o Shifter Plots - Made from L3 Data . rcp can select a sub-set of triggers to use - EM, HD, Tot versions + Trig, Cal, (Pulser) ver's - Et plots - No. Hits plots . rate per event especially useful for tracking noise that comes and goes - Pedestal: Mean and RMS over a few minutes - Gain: Mean and RMS over minutes or hours . Ped/Gain plots are the ways that bad towers are identified in current running ==> *very useful* - Goal is to archive these plots so that we have a record of all this - Note: pedestal/pulser runs are also used b/w runs to find these kinds of problems as well o Expert Plots - need these in various different coord syst's to track down source of problem . Physics, BLS, CTFE - Pedestal plots . ped set to 8 counts . ped's vs eta and vs phi . useful for finding > dead channels, unplugged cables, noisy channels, downloading... . EM vs HD Pedestals > rms(HD) ~ 2 x rms(EM) > rms(EM) ~ 1 count - Calibration plots (trig vs prec) . gain = TT/Prec . useful for > bad summers,drivers, improper terminations, power prob's, bad gains/DACs, timing, miscabling, bad lookup proms o Tracking Bad Towers - use ped & gain plots to generate list of TTs failing . cuts on mean and sigma - list generated online during data taking . stored to datbase . link to significant event system o Other Features - so far - most debugging/monitoring done in global runs - pulser runs haven't been used so much yet . want to add this functionality soon (next few weeks) . takes a lot of global running to collect enough stat's to make a sensible test - pulser is much faster - Pulser can be used b/w runs to verify calib scales . pulses injected just before preamps . limitation: doesn't produce the same shape signal as real data > trigger is sensitive to this shape - Hot Tower Finder . intermittantly hot towers are the dominant problem . from Examine - Monitoring (online) . this will be used more often in the future Any Other Business ------------------ o Real Time talks (Denis) - deadline for abstracts extended - submit another abstract for a L1Cal-specific poster . algorithms + architecture - Denis will write abstract & make poster - Author list . proposal: Saclay, Nevis, MSU, others??? . should let Denis know if you think you should be included o Splitters - Dan & Denis will discuss some questions by phone . shielding of cables from BLS o Next Meeting - Philippe will talk about online monitoring