1. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 1 Performance of CMS Cathode Strip Chambers Andrey Korytov (for CMS Collaboration)

2. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 2 Compact Muon Solenoidal Detector (CMS) Endcap Muon System is based on Cathode Strip Chambers (CSCs) 4 stations (disks of CSCs) in each endcap 6 sensitive planes per CSC 468 CSCs with total sensitive area >5000 m 2 pseudorapidity coverage 0.9<|  |<2.4 ~500K readout channels Provides: - muon trigger - muon identification and precise measurements Endcap Muon System is based on Cathode Strip Chambers (CSCs) 4 stations (disks of CSCs) in each endcap 6 sensitive planes per CSC 468 CSCs with total sensitive area >5000 m 2 pseudorapidity coverage 0.9<|  |<2.4 ~500K readout channels Provides: - muon trigger - muon identification and precise measurements

3. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 3 One of 8 endcap stations

4. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 4 What’s new Performance of CMS CSCs was extensively studied over the last 10 years during R&D and production: cosmic ray muons, muon beams, high irradiation rate conditions, etc. BUT one chamber at a time (often a small fraction of a chamber area) in lab conditions In this talk, we present the first results obtained with: 36 CSCs operated in situ as one system 400 m 2 of sensitive planes 8% of the entire CMS CSC system cosmic ray muons Presented: Track Segment finding efficiency (Level 1 Trigger) Fast Precision Coordinate reconstruction (High Level Trigger)

5. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 5 CMS Magnet Test and Cosmic Challenge ME+4 ME+3 ME+2 ME+1 60  Fall 2006 CSC scope 60  -sector of one of the two endcaps 36 chambers, ~8% of all

6. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 6 CSC design and readout Large chambers Wires groups:5 cm wide Cathode strips: 8-16 mm wide 3.3 m wire-group hits every 25 ns same information for trigger/offline Level-1 Trigger: half-strip hits every 25 ns HLT*/offline: 12-bit digitization every 50 ns *HLT – High Level Trigger

7. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 7 Track Segments for L1 Trigger 1D Track Segments (pattern recognition is implemented in firmware) wire-group hits in six planes half-strip hits in six planes 2D Track Segments are combinatorial combinations of all 1D wire- and strip-segments Efficiency requirement >99%

8. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 8 Digitized strip signals for HLT/offline Strip signals are sampled and digitized every 50 ns HLT requirements: Resolution: <0.5 mm per segment CPU: ~ ms per segment (time budget for entire HLT is 40 ms) Offline requirements: Resolution: 150  m per segment strip 1 signal strip 2 signal strip 3 signal strip 4 signal strip 5 signal strip 6 signal

9. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 9 Track Segment Efficiency measurement Magnetic field 0 T Trigger is based on ME1 and ME3 stations only ME2 station is also in readout, but not in the trigger Only one Track Segment in ME1 and ME3 ( ) ME1+ME3 provide prediction in ME2 (  ) Residual = Segment in ME2 ( ) – Prediction (  )

10. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 10 Predicted Track Segment in ME2/2 CSCs Coordinates of all predicted hits in ME2/2 Red trapezoid – chamber outline Dashed lines – “semi-dead” areas separating 5 independent plane sections Predicted hits that were missed in ME2/2 Blue sub-trapezoids – nominal fiducial area of “guaranteed’ full efficiency

11. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 11 Track Finding Efficiency of ME2/2 CSCs NO CUTS Red points – measurements Blue line – expectation taking into account “semi-dead” areas separating independent wire plane sections ONLY FIDUCIAL AREA OF FULL EFFICIENCY Red points – measurements Average efficiency 99.93±0.03%

12. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 12 Sagitta based on found track segments Scatter plot of sagitta measurements  dY is larger due to courser Y-coordinate measurements (wire groups vs half-strips) Average offset is due to iron disk misalignment during MTCC—confirmed by geodesic survey Histogram—measured residuals Line—expected residuals, simple calculations based on cosmic ray muon momentum spectrum

13. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 13 Fast algorithm for hit/segment reconstruction at HLT Reuse L1 Track Segments: can be done due to their very high efficiency and good (few mm) pointing by design, if L1 Track Segment is not found, no data are read out from that chamber by DAQ Using digitized strip signals, find x-coordinates drop calibrations (gain, pedestals, x-talks, noise correlation matrix, plane mis-alignment): can be done due to high uniformity of the system no database access is needed build x-coordinate as a function of digitized signals no iterative fitting Using six x-coordinates, find segment coordinates linear procedure (no iterations) prune up to two outliers CPU time performance: 0.45 ms per segment (Intel 2.8 GHz P4)

14. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 14 Fast x-coordinate (1) Use first two time samples to build pedestals dynamically (to reduce noise, average as new events come in) Add three samples with signal (max ± 1) Use an old method of ratio of charges to get a first approximation for a local coordinate in strip width units time charge pedestal Q1Q1 Q2Q2 Q3Q3 Q center Q left Q right strips

15. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 15 Fast x-coordinate (2) Correct for expected non-linearity using the Gatti shape of the induced charge (correction is a simple strip width-dependent function) Check occupancy for reconstructed coordinate. It is not flat indicating there is a remaining non-linearity. Fit the occupancy and reconstruct empirical correction (which happens to be almost strip-width independent)

16. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 16 Residuals in test (3 rd ) plane Plane 3 is not used in the track segment fit Residual(3) = x meas (3) – x fit (3) Plane 3 Plane 4 Plane 5 Plane 6 Plane 2 Plane 1 strip center strip edge

17. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 17 Extrapolating to the full track segment There was not a precise reference prediction for a track segment in MTCC. Hence, we just extrapolate single plane resolutions to the overall six-plane resolution CSC six-plane resolution is ~150  m by far exceeds the HLT requirement of <0.5 mm actually, very close to design spec for the ultimate offline resolution HLT requirement

18. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 18 Summary Trigger performance of CMS CSCs evaluated with cosmic rays using 36 CSCs operated in situ as one system 2d Track Segments for Level-1 trigger: efficiency 99.9% (required 99%) sagitta residuals are consistent with m.s. of cosmic ray muons (~3 mm) decision time 800 ns (firmware, by design) Track Segments for High Level Trigger localization per chamber ~150  m (required 0.5 mm) robust, no losses in efficiency (by algorithm design) decision time 0.5 ms (software, required ~ ms)

19. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 19 Backup slides

20. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 20 CSC Design Parameters Overall size: 3.3 x 1.5/0.8 m 2 (trapezoidal) 7 panels form 6 gas gaps of 9.5 mm Anode-Cathode: h=4.75 mm Anode wires: d=50  m, gold-plated tungsten Wire spacing:s=3.2 mm pitch Wire tension:T=250 g (60% elastic limit) Readout group:5 to 16 wires (1.5-5 cm) Cathode strips:w=8-16 mm wide (one side) Gas: Ar+CO 2 +CF 4 =40+50+10 Nominal HV: 3.6 kV Gas Gain: 10 5

21. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 21 Singles Rate Curve dark count rate ~ 0.04 Hz/cm 2

22. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 22 CSC aging test results Setup: Full size production chamber Prototype of closed-loop gas system nominal gas flow 1 V 0 /day, 10% refreshed Large area irradiation 4 layers x 1 m 2, or 1000 m of wires Rate = 100 times the LHC rate 1 mo = 10 LHC yrs Results: 50 LHC years of irradiation (0.3 C/cm) No significant changes in performance: gas gain remained constant dark current remained < 100 nA (no radiation induced currents a la Malter effect) singles rate curve did not change slight decrease of resistance between strips Opening of chamber revealed: no debris on wires thin layer of deposits on cathode (stinky!)— no effect on performance Anode wire after aging tests

23. Andrey Korytov, University of Florida IEEE Nuclear Science Symposium, Honolulu, 31 October 2007 23 CSC Production Sites CSC Assembly (Fermilab, PNPI-St.Petersburg, IHEP-Beijing, JINR-Dubna) On-CSC electronics (Universities: Ohio State, UCLA, Carnegie-Mellon, Wisconsin) Final Assembly and System Tests (Univ. of Florida, UCLA, PNPI, IHEP, JINR) Pre-installation tests and final commissioning (CERN)