1. Recall: CFEB inefficiencies Collision Run 135445 1 June 2010G. Rakness (UCLA)1 https://cms-project-csc-validation.web.cern.ch/cms-project-csc- validation/run135445/ExpressPhysics/Site/PNGS/hORecHits.png Circled chambers were targeted last week to be fixed

2. One hour of  s=900GeV Collisions: Run 136297 1 June 2010G. Rakness (UCLA)2 https://cms-project-csc-validation.web.cern.ch/cms-project-csc- validation/run136297/MinBias/Site/PNGS/hORecHits.png  Wed 26 May: CLCT’s sped up by +1bx for the circled chambers (except ME- 1/1/33, which was sped up by +2bx)… Appears to be better. Need a better way to spot this problem….

3. Discussed CFEB readout Synchronization at Timing meeting 1 June 2010G. Rakness (UCLA)3 I presented how we progressed to have readout synchronization “on the edge” for collisions… (talk in backup slides) Then we discussed “the plan…” 1.Aim for 131 in the CLCTpretrigger-L1A 5bx-wide coincidence window. This is an asymmetric window: (–1, +3) –Asymmetric allows readout of halo muons in upstream endcap Continued…

4. Continuation of “the plan” 1 June 2010G. Rakness (UCLA)4 2.Amanda makes correlation plot between online measurements of CLCTpretrigger-L1A values with the radioactivity trigger vs. ALCT in the CLCT match window for cosmic rays –Radioactivity trigger timing scans taken last Friday are suspect. Will retake measurements tomorrow… –Now we know what value of ALCT in CLCT match window corresponds to 131 3.Correct the average value of ALCT in CLCT match window from collisions to hit the value from 2. –TMB data used to determine the timing for CFEB readout for collisions… –This will give optimal CFEB readout for collisions, slightly non- optimal for cosmics. Sound good?

5. Some things to discuss Shall we try to get sub-bx corrections? Or do bx-corrections first? –Sub-bx corrections (a.k.a. “CFEB muonic timing) has never been tried at CERN… –Will need to include cable lengths… –Perhaps should be left to the next technical stop? 1 June 2010G. Rakness (UCLA)5

6. TMB configuration and L1A out- of-sync With guidance from UCLA, clct_throttle was set in TMB to prevent CLCT’s from being created for “clct_throttle” bx’s after the previous CLCT was created Since the TTCci trigger rules are no more than 1 L1A in 3bx, and the TMB L1A accept window is 7bx, we choose clct_throttle=10… Fri 14:00 – reconfigured all TMB’s –No TMB’s out-of-sync since… –We used to have ~4 per hour… 1 June 2010G. Rakness (UCLA)6

7. Otherwise… Fri 23:20:38 – Site-wide power outage for 14 seconds Sat afternoon – helped Rachel (CEO) power back on… Mon–Wed – Technical stop ongoing –Online software and DCS working all day yesterday (until 4am) and continue their work today… 1 June 2010G. Rakness (UCLA)7

8. Thoughts about shifts Recent CEO experience: CSC's run pretty well. Number of calls about CSC problems from shifters during normal running has not been overwhelming… Useful to have a CSC shifter in the control room. If things go wrong, it is good for a central shifter to have a CSC shifter physically present to handle the CSC problems. Let’s take an adiabatic approach to shift-free CSC operation… Plan: Eliminate the CSC day shift  This is the time when “playtime” happens most often  This is when the experts are typically around to do things. During the time between when the CSC owl shift goes home (07h00) and the CEO arrives at point 5 (10h00), we let the system “swim in the deep water” for a couple of hours… Do we sink or swim? Begins mid-July (next round of shifts)… Test for 2 months, then re-evaluate, possibly stopping night shifts, then evening shifts… 2 June 20108G. Rakness (UCLA) To present tomorrow at DPG meeting…

9. CEO status Five new folks to train as CEO: –Chris Farrell (UCLA) –James Dolen (UC Davis) –Antony Adair (Rice) –Lindsey Gray (UW) –Ingo Bloch (FNAL) This is a good number, because there are five CEO’s who have been around since 2008 that should probably be phased out… (not Jordan… ) 1 June 2010G. Rakness (UCLA)9

10. Training CEO’s… In order to help bring new CEO’s up to speed, I have created… –A detailed job description –A list of “Expert” questions that I think CEO’s should know the answers to In addition, I recommend to train by… –Taking CSC shifts –Following a current CEO –Becoming familiar with the CSC Operations twiki https://twiki.cern.ch/twiki/bin/view/CMS/CSCOperations –Becoming familiar with the CSC CEO twiki https://twiki.cern.ch/twiki/bin/view/CMS/CeoInformation What else? 1 June 2010G. Rakness (UCLA)10

11. To do Tomorrow Clean up configuration according to 12 May –is problem_mask=0x7d correct? Make sure timing scans get done properly  Amanda Elog ME-3/1/10 and other cable lengths Record TTCci configuration information Send “Expert system” analysis to Misha, then to Evaldas and Jinghua… Analysis Look at AFEB time bins ON for timing scan runs Point 5 Continue to encourage automated Hard Resets automated… Documentation Look at/input transitions into TriDAS/emu/doc/html/EmuFSM.xhtml To do at 904 Online software –Put ALCT firmware database check into ALCT firmware write routine –Resurrect ALCT self-test –ALCT hot-channel mask in xml file/config DB –Expert config check –Configuration check with separated DMB and CFEB Managerial stuff Talk to Valentin Sulimov and Sergey Vavilov regarding responsibilities w.r.t. 904 and calibration runs 25 May 2010G. Rakness (UCLA)11

12. Backup slides 1 June 2010G. Rakness (UCLA)12 Talk given at CSC timing meeting 28 May 2010 http://indico.cern.ch/conferenceDisplay.py?confId=96284

13. Recall: CFEB readout timing condition CFEB data are read out if the CLCT pretrigger at the DMB is followed by an L1A 131bx later CRAFT 2008 (LHC setting): ±1bx. –This means 130, 131, 132  readout Before CRAFT 2009: ±3bx –128, 129, 130, 131, 132, 133, 134  readout 1 March 2010: – 1, +3bx (asymmetric) – 130, 131, 132, 133, 134  readout –(See later in this talk…) 1 June 2010G. Rakness (UCLA)13

14. Online timing scan to set CFEB readout timing  The delay of the CLCT pretrigger timing is determined by an online timing scan using radioactivity trigger of tmb_lct_cable_delay (in the DMB) in order to “hit” 131 on average. An example scan result looks like this… Best AFF to L1A Counter[126] = 0 Best AFF to L1A Counter[127] = 1 Best AFF to L1A Counter[128] = 1 Best AFF to L1A Counter[129] = 4 Best AFF to L1A Counter[130] = 18 Best AFF to L1A Counter[131] = 33 Best AFF to L1A Counter[132] = 40 Best AFF to L1A Counter[133] = 0 Best AFF to L1A Counter[134] = 0 Best AFF to L1A Counter average = 131.115 1 June 2010G. Rakness (UCLA)14 We had the system “perfectly” timed in for cosmic rays in CRAFT 09…. What happened to get some CFEB’s out of time for collisions 2010?

15. 18 Nov 2010: Shifted CLCT timing 1 June 2010G. Rakness (UCLA)15 A. Deisher (UCLA) Made bx-level adjustments to the CLCT (in TMB) in order to align CLCT’s with ALCT (L1A timing) This is a step towards “CLCT muonic timing”

16. Step 1: I didn't remeasure tmb_lct_cable_delay Elog: "Make the corresponding change in tmb_lct_cable_delay to keep the L1A-CLCTpretrigger time be 131. Since we have lined up CLCT and ALCT at the TMB, this amounts to setting the value of tmb_lct_cable_delay=0 on all chambers." I justified this to myself because… CLCTpretrigger-L1A window is 7bx-wide The collisions might have different timing from radioactivity/cosmic rays, anyway… Point of note… Rachel (CEO) just measured the CLCTpretrigger-L1A position for most chambers in both endcaps… 1 June 2010G. Rakness (UCLA)16

17. Step 2) CLCTpretrigger-L1A window shrinks by 2 asymmetrically 1 March 2010 - put in DMB+CFEB firmware to gain back 50nsec SCA timing. Elog: "Tests at 904 indicate that the CLCT pretrigger - L1A coincidence window has shrunk from 7  5 bx wide." These same tests were not conclusive as to whether the window was asymmetric or not…  Tests at 904 yesterday proved it is indeed ASYMMETRIC, just like Ben said Measurements of tmb_lct_cable_delay at point 5 last Friday indicated that step 1) (imprecise synchronization) together with step 2) (asymmetry in the window) put many chambers on the lower edge of this window (but still within the window) for cosmic rays. Q: For collisions, is 131 the right number to aim for? 1 June 2010G. Rakness (UCLA)17

18. Cosmics Run 135432 1 June 2010G. Rakness (UCLA)18 https://cms-project-csc-validation.web.cern.ch/cms-project-csc- validation/run135432/ExpressPhysics/Site/PNGS/hORecHits.png Hard to see the effect in cosmic rays…

19. Step 3) CLCT's are 0.5 bx later for beam than for cosmic rays 1 June 2010G. Rakness (UCLA)19 A. Deisher (UCLA) Step 3), together with steps 2) and 1), push some CFEB's out of readout for collisions. The worst have been fixed by applying a coarse bx adjustment. Q: Now that we have the L1A timing set, should we try to address the rest?

20. Proposed plan 1.Decide on the value of CLCTpretrigger-L1A to aim for in 5bx-wide window –131? –Symmetric or asymmetric? 2.Establish the correlation between online measurements of CLCTpretrigger-L1A values with the radioactivity trigger with Amanda's analysis of ALCT in the CLCT match window for cosmic rays –Now we know what value of ALCT in CLCT match window corresponds to our “goal” value from 1. 3.Correct the ALCT in CLCT match window from collisions to hit the value from b). –This will give optimal CFEB readout for collisions, slightly non- optimal for cosmics. Comments? 1 June 2010G. Rakness (UCLA)20