CSC LCT Occupancy per LHC Bunch

---

Results based on ~1 million minimum bias events in full ORCA simulation

• Equivalent to ~50,000 bunch crossings, assuming ~17 minimum bias events per BX
• Data volume estimate assumes 100 kHz rate of random Level-1 triggers
• No Level-1 algorithms are applied, and 1996-era LCT algorithms from UCLA are assumed
• Correlations with Level-1 "good physics" triggers are NOT INCLUDED
• This is our best-estimate from ORCA as of September 2001

• First possibility: assume that only CLCT coincidence required with L1A  (16-time samples unless noted)

• Second possibility: assume that ALCT.and.CLCT required with L1A  (16-time samples unless noted)

• The above results are based on simulated minimum bias events only!   The EMU data rate increases when track correlations with good L1A physics triggers are included!
  • Preliminary estimates suggest that the addition of simulated "good physics" events and applying the L1 trigger algorithms could increase the average event occupancy by ~60%.  See this link for updated results, but basically add 60% to the red table entries above for the best estimate.
  
  
• A single DDU reads out data from the CSCs within one 20-degree wedge in each of the 4 muon stations for one endcap. It is expected that the wedges connected to a DDU will be rotationally separated in phi by ~60-degrees from station-to-station. In this way, all DDUs share the readout load equally.

 

• There are 36 DDUs read out via 8 DAQ links. We expect that 4 DAQ links will each carry data for 5 DDUs, while the other 4 DAQ links will carry only 4 DDUs. The DAQ load IS NOT carried equally by 8 DAQ links with this layout.



Notes:

1. The number in [brackets] is the number of Readout S-LINK/DCCs required for our planned DDU configuration
• There is one DMB per CSC, and up to 15 DMBs per DDU
• L1 occupancy enhancement of ~60% is NOT considered here, but should be
• EMU plans to have 3 or 4 FED crates in the counting house, each with up to 12 DDUs and two DCCs
• This crate structure requires an integer multiple of 3 or 4 DCCs (Data Concentrator Cards)
• one or two readout S-LINK per DCC
2. Average event volume is 5500 bytes per triggered CSC for 16-time samples, 3100 bytes for 8-time samples
• Click here for the single CSC event size details.
3. Total EMU Average Data Bandwidth is determined by
• Bandwidth = Occupancy * L1A rate * DMB Event Size + Overhead
• L1A rate assumed to be 100 kHz; assume that each L1A originates from a proton bunch
• Average DMB Event Size = 5424 bytes (16-time-samples)    OR   3024 bytes (8-time-samples)
• Overhead (due to S-Link-64 Headers and Empty-Events) = L1A Rate * 24 bytes/DDU * 36 DDUs = 86.4 MB/s nominal
4. Average EMU Trigger Rates
• These are the AVERAGE rate estimates
• The maximum instantaneous "burst" rates are certainly higher (but not simulated yet)
• Average DDU trigger rate = Occupancy * L1A rate / 36 DDUs = 7500 Hz  (FULL EMU design)
• Average DMB Trigger rate = Occupancy * L1A rate / #CSC-DMBs
• 540 DMBs in FULL EMU design (assumes ganged ME1/A)
• 468 DMBs with descoped ME4/2
• 432 DMBs with ME4 descoped entirely
5. Our old estimate for CSC occupancy was only ~0.8 per 3 BX
• The old estimates used a naive approximation model for neutron and charged particle LCT backgrounds
• Old average LCTxL1A for all CSCs in a DDU was 2205 Hz
• Old Occupancy = 2205 Hz * 36 DDUs / 100 kHz = 0.794
6. Occupancy here refers to the average number of CSCs with an LCT in any given bunch crossing
• 17.3 minimum bias events per crossing is assumed
• 25 ns beam crossing period

---

Last updated 9 December 2002.

gilmore@mps.ohio-state.edu