1. Motivation General rule for muon triggers: Never neglect a possible backup reduction factor. It will always come back to you. Even if RPC trigger works just fine from the beginning one still wants to: Reduce rate in regions with only 4 or 3 RPC planes available. Reduce p t thresholds as much as possible. HO should be better than any pre-scale. HO Scintillators in RPC Muon Trigger Conceptual design HO Scintillators in RPC Muon Trigger Conceptual design J. F. de Trocóniz, UA-Madrid

2. Towers 8+9 represent 92% of the rate (p t > 10 GeV, |  | <1.24), but only 16% of the acceptance

3. HO Characteristics 10 mm Bicron scintillator tiles positioned between coil and MB1 RPC 1 plastic for Wheels ±1, ±2. 2 plastics separated by 15 cm iron slab in Wheel 0. Covers the full MB1 system (barrel + overlap) up to |  | < 1.24 (Tower 9) Typical cell size: 40 cm (  ) × 50 cm (  ) Granularity: 0.087 (  ) × 0.087 (  )

4. HO matches well muon system in r-  view (MB1)  : 0.087  5 deg  16 RPC strips  OK Not that well in r-   : 0.087 (HCAL standard tower size)  detailed HO – RPC map needed

5. HO Readout Standard HCAL readout: Fibers  HPD (G=2500)  QIE (  T=25 ns) 90% of energy in two samples (phase independent of HCAL) More light:  Thicker plastics,  4 WLS loops/tile,  shorter fiber path Designed to give 10 pe / mip Trigger: Energy-over-threshold bit

6. Test beam results Actual performance of HO system (Wheel 1 scintillators) measured at 2002 test beam (Jim Rohlf). 6 pe/mip/plastic Gaussian noise at normal incidence. 1.5 pe-equivalent/bucket  can be improved to 0.9 pe for “quiet” QIEs. Is this performance good enough? Can be achieved systematically at CMS?

7. HO Performance Simulated with CMSIM123 280 MeV/mip/plastic at normal incidence  6 pe 0.9 pe/bucket  64 MeV Geometrical acceptance: 93% Signal width dominated by photo-statistics. HO threshold at 1% tile occupancy  150 MeV (1 MeV deposited). Similar efficiency for 1.5 pe/bucket of noise, but 8 pe at signal peak, for E HO > 150 MeV (3% tile occupancy).

8. Backgrounds p-p interactions (10 34 cm -2 s -1 ): < 2 Hz/cm 2 Neutron-induced conversions: < 10 Hz/cm 2 (MB1 level) n-p elastic collisions: 150 MeV) Electronic Noise

9. HO-RPC Mapping Equilibrium between large acceptance and simplicity (hardware implementation)  Minimal Map Acceptance always larger than 90% (often much larger).

10. HO provides extra “RPC plane” Trigger Algorithm

11.  Require HO confirmation for low-quality RPC coincidences Built-in high efficiency (low quality RPC muons are ~30%) Remarkable threshold stability (allows tuning at CMS)

12. Rate reduction RPC noise trigger rates simulated using ORCA (50 Hz/cm 2, nominal neutrons) Large sample: 110 Mevents, corresponding to 4.4 s of LHC. High quality noise trigger fraction much smaller than 1%. For 0.9 pe/bucket, E HO > 150 MeV  Reduction factor = 100 For 1.5 pe/bucket, E HO > 150 MeV  Reduction factor = 30 Low-p t rates w/ HO comparable to high-p t w/o HO

13. ORCA Results

14. Connecting Hardware (preliminary) Processing of HO signals performed at HTR boards (4 boards/sector, 2 FPGA/board). Provide energy-over-threshold programmable bit (possibly  -dependent). All OR-ing corresponding to the HO-RPC  map also handled here  Input fibers organized according to constraints at HO end. SLB cards organize HTR bits into bit streams, and transmit to RPC Trigger Boards using GOLs (32 bits/bx)  Output streams organized according to constraints at RPC end.

15. HCAL (HO) in RPC Trigger TRIGGER BOARD READOUT BOARD SPLITTER S-link to DAQ to Level-1 trigger 90 m @ 1.6Gbit/s up to 5 m LVDS @ 80MHz QIE GOL QIE HTR (Readout) Board Optical Tx HCAL Front-end New 'Optical SLB'

16. HTR Configuration for HO P2 P1 8 8 8-way fiber in FPGA RxDeser.RxDeser. RxDeser. RxDeser.RxDeser.RxDeser.RxDeser.RxDeser. RxDeser.RxDeser.RxDeser. RxDeser. RxDeser. RxDeser. RxDeser. RxDeser.RxDeser.RxDeser.RxDeser.RxDeser.RxDeser.RxDeser.RxDeser.RxDeser. RxDeser. RxDeser.RxDeser.RxDeser. RxDeser. RxDeser.RxDeser.RxDeser. SLB FPGA Outputs to RPC Crate Total of 48 calorimeter channels per HTR DAQ out Front-end data inputs

17. Example of cabling scheme satisfying all constraints at HO and RPC ends

18. Conclusions Investigating how to incorporate HO into RPC trigger: Geometrical integration, RPC+HO extended algorithm, basic lines of hardware implementation established. If HO performance at 2002 test beam achieved systematically at CMS  RPC trigger rate reduced by 100. Efficiency O(90%) stable as a function of HO energy threshold (allows tuning). Implications much more important in case RPC noise can be reduced to 5 Hz/cm 2  consider HO to improve efficiency (less restrictive algorithms, tower 6, “classic” 3/4). HO is now part of the L1 Trigger Baseline