V13/V14 QCD

1. V13/V14 QCD Andrew Brandt, UTA • V13: • New dijet mass triggers (Hong Luo, Don Lincoln) • JT_95 New L2, prescaled at high lum (>55E30) • JT_125 New L1,L2 unprescaled at all lum • V14 • FPD triggers Collab TB 9/23/2004

2. inefficiency due to L2 |η|<3.2 Dijet Mass Triggers added to V13 (measured)

3. ---- L3M225 (prescale: 24) ---- L3M380 (prescale: 1) ---- Total events(660k)

4. Studied inefficiency (RecoMass>400Gev, but didn’t fire 380 trigger) L2Mass L3Eta Events have low L2 mass, large L3 η: As expected L2 |η|<3.2 cut causes inefficiency

5. V13.2 JT_L3M380 replaced with JT_L3M430 ---- L3M430 (prescale: 1) ---- Total events(100k) Raised threshold in V13.2 reduces L3 rate by 2.5 to 3

6. Weird L2 eta distribution for V13.2 runs, not understood (by us) (was fine for 13.1) 1. Created a sam dataset name: Run196243_TMB to include the thumbnail events of Run196243 2. Used d0correct v7.5(the latest version of d0correct) to create the tmbtree. >setup p16.04.00 (d0correct v7.5 is built in p16.04.00) >d0setwa >setup d0tools >rund0exe -initscript=initchunkanalyze.sh -sourcescript=chunkanalyzesource.sh -exe=TMBAnalyze_x -rcp=runTMBTreeMakerSAM.rcp -rcppkg=tmb_analyze-defname=Run196243_TMB -batch -num=100000 -nofpe -cupt=10:00:00

7. JT_95TT L2Pt L3Pt Raised L2 threshold from 30 to 50

8. JT_125TT L2Pt L3Pt Added L2 threshold of 60

9. 1x96 FPD_LM T M 1x96 LM 3x96 FPD_DFE FPD Trigger Manager Inputs • FPD_LM • Information on which detectors are hit and halo • 2) LM pass through 16 LM and/or terms • includes GapN GapS GAPSN SI etc. • DFE information from scintillating fiber detectors, • forseen to give , t now using segment information, multiplicity

10. 2004 FPD Trigger Strategy • Input information: • Currently no global run trigger capability • Vertex board is delayed (may now be finished during shutdown) • DFE boards and TM work and ready to be commissioned • Main background not from pileup (multiple interactions) but from halo spray • Strategy: • Instead of calculating bin of  and t, use fiber hit patterns to demand 2 or 3 out of 3 planes of each detector are hit. Replaces trigger scintillator, simpler algo • Use multiplicity cut to reject halo spray, code several multiplicity levels, can reject on multiplicity in adjacent detectors • NOTE fiber ADC threshold must be high enough to avoid noise, low enough to retain efficiency and allow vetoing of halo • One advantage is pot positions not needed at trigger level

11. Exposure Groups etc. • For normalization purposes each trigger is assigned an exposure group • and live time is measured for that group—different beam conditions require • different groups • Only 8 exposure groups (6 used), we have more combos so must • operate outside exposure groups—new value NOLUM (approved 1/04) • Another issue is dynamic downloading, which allows multiple global runs • This would allow us to take elastic or other monitoring data in separate • global run, so as to avoid general reco; streaming and other issues with this • Once FPD DFE/TM is ready we are in principle restricted to dedicated • special runs until these other issues are resolved

12. Interim Plan • Proceed with DFE/TM debugging during shutdown • Elizabeth indicates NOLUM work will start in ~2 months and take • 2-4 weeks to complete • Since current FPD commissioning is special run intensive and • new elastic triggers with adjacent spectrometer veto have rates • around 1 Hz, we propose adding an elastic trigger to global list in • V13.x. (REQUEST TO TB) • If we do this carefully it will only take one exposure group and allow • us to obtain good elastic sample at lums <40E30

13. Trigger Commissioning Progress • DFE/TM chain tested by Duncan with help from Jamieson and Jeff Temple, works with debugging Johnson counter firmware. • Two versions of DFE firmware developed by Mario and Wagner. Version 3: One spectrometer only, passes hit fiber info to TM. This version was simulated, implemented, and tested successfully at combined test stand but failed at platform. V3 failures traced to delays between different inputs to DFE, Tom is working on solution similar to TM, which waits for all inputs before processing Version 4: Three spectrometers on one DFE, includes some extra bits to simplify TM algorithms. This version failed some timing tests upon implementation and needs further development (proceeding in parallel), also requires merging of 3 spectrometer bits (Ricardo) 3. Jeff produced first version of TM firmware to operate with V3, Mario has version to operate with V4 (not yet implemented.)

14. V4 TM Algorithm DFE will pass word for each spectrometer indicating coincidence of two detectors tight track: exactly 1 hit with 3/3 fine segments in each detector medium : >0 with 2/3 + <20 wide segments loose: => 0 with 2/3 + <20 wide segments At TM we would form terms DIFF=any spectrometer track DIFFQ=any quadrupole spectrometer track (could be false if >1 or 2 on A or P side) DIFFD=dipole track ELAS=AU-PD or AD-PU or AI-PO or AO-PI DPOM=AU-(PU or PI or PO) or AD-(PD or PI or PO) or AI-(PU or PI or PD) or AO-(PU or PD or PO)+DI -(PU or PD or PO or PI) OVER=AU-DI or AD-DI or AO-DI or AI-DI (over-constrained track for alignment)

15. FPD Trigger List Tentative L1 FPD trigger list (V14) 1) elastic (diag opposite spectrometers) +GAPSN 2) soft diffraction (single spectrometers)+GAPS or GAPN 3) overconstrained track (pbar in quadrupole +dipole spectrometers)+GAPN 4) double pom (up-up, dn-dn etc.)+GAPSN if needed 5) CJT(2,3) + FPD Track (DIFFQ or DIFFD) +GAPS or GAPN if needed 6) CEM(1,3) +FPD track +GAP? 7) TTK(1,1,5) +FPD track +GAP? 8) MU(1) +FPD track +GAP? Monitor will trigger likely be necessary, will need to study with special runs. Also rates are not yet known. If 4 (dpom) it is not low enough to run unprescaled we would need to repeat 5-8 with two FPD tracks. Possibly 1-3+any monitors are best done in a separate global run since they will not need general farm reconstruction.