1. L2mu System - Status SLICs Track Finding (main data processing) Next: SLIC Upstream and Downstream DONE UPSTREAM; CICs and SFOs CIC crate; hooked into place by June 8 Boards return from stuffing ~end of May Bench tests (MIS-Eng.) done by June 8 Ship to fnal, then full cabling at MCH3 UPSTREAM fully hooked up by mid-to-late June Worst case scenario: no downstream yet – we can; Check all cabling / connections: run BIST tests over all sources, cross check data origin (headers) Exercise SLIC-run-init (SCL-init if SCL available) Receive/Dump a few events – Formats,Endianess,etc. First look at DSP-times, Event Sizes, at very limited speeds (by re-routing SLIC output to VME -- dump) DOWNSTREAM; MBTs and Alphas next;

2. Downstream - Alphas Two major problems remaining (1) DMA – data input Last week at UMich, problem solved (R.S. DØ + S.M. CDF) – standing ovation (2) PIO – board to board Mbus Communications Current understanding: an fpga programming problem, (rather than a Bus problem) – Joint effort DØ + CDF Highest priority: estimated ~2 weeks, will know soon -- Big decision next week; whether to order remaining boards prior to seeing PIO solution (?)... Circumventing PIO – OK for low rate Test Mode have single alpha performing worker + administrator takes little extra software, plus a few kluges in system Note: Current design of “Beta” approved Project (Virginia+Orsay+ Maryland) goes ahead Current boards acquire 8 “white wire” mods Upgrade incorporated into future production “future production” does not delay muons Current boards destined to Cal+Muo+Global X

3. Downstream - MBTs Last steps remaining (priority ordered) (1) Enable the SCL functions (2) Multichannel Broadcasts FW-SCL (L1+L2) into MBT-daughter board L1 decision formatted, buffered, broadcast (all SLICs) L2 decision stored in buffer accessible by MBus – raises a pin in backplane to trigger PIO-read by Alpha Other Functions; request SCL-init raise L1-busy ( L2 input buffers full, stop L1 ) raise L2 busy (L2 output buffers full, stop L2 ) Status: firmware mostly completed, D.Baden @ fnal, testing as we speak... HotLinks-to-MagicBus-to-Alphas ; Firmware done, needs testing (was held by DMA -- ~1week job) GOAL: Mu+Cal Online in August (2) Multichannel Broadcasts (3) Additional MBT firmware steps Cypress output from preprocessor to Global (MBus PIO to MBT) L2Global answer to L2 HWFW (hardware exists, being debugged)

4. L2mu System - Status (cont.) Cables all cables in place, except; CIC (14)  SLIC (16) Simulator All SLIC algorithms in place (4 stub finders, 2 L1 reporters) All Alpha algorithms in place (Central & Forward track builders) Framework done/tested, Input-to-Ntup Now adding the Lookup Tables (versions) Expect to run from CVS next week Exercise + Tune + Test Robustness of all algorithms with simulator – Online links from the same files, reads same tables, etc. SLIC: current development work is online monitoring and error handling Note: SLIC+Alpha processors do the track finding. Trigger decisions are only made by L2-Global. Development of Global needs a functioning Alpha.

5. I – Remarks on Run-II L2mu Dimuons First time we trigger on A-stubs; Fundamentally different from higher thresholds Likely one of the strengths of L2 ( inherits CFT coincidence at L1, adds Aφ granularity + PDT track residuals. Dimuons with A-stubs  a much faster collection of J/ψ ‘s than in Run-I (Eff’s, p-Calib, p-Resoln) Early run: optimize bandwidth to collect J/ψ ‘s Example: L1(μ+μ–) + L2(M&P) + L3(μ μ Mass) in // with special runs on other subdetectors (expected J/ψ visibility of 0.1Hz at 1.E31/cm 2 /s) Another (and faster) control sample: Muon-in-jet is a bona-fide guarantee of μ-quality Use higher threshold single tracks to study A-stubs Example: L1(μ+jet) + L2(M&P) + L3(μ-in-jet) (cross section large enough for a special run) (*100 – thres&η)

6. II – “Prove that L2mu is working” i.e. we get every muon (efficiency) we reject junk (rates, purity) Job: Monitor the efficiency as nearly online as possible L2 single track Efficiency =#L2 pass (for an ID-controlled sample) #L1 pass it is an approximate efficiency (ID bias) value depends on LMT of either trigger From Run-I; an effective ID control is loose muon (to minimize bias) inside jet (to certify beam origin) -- but jet introduces (L1) CFT bias... Opp. φ-hemisphere to leading jet obs: more accurate ID with less bias (eye scan, dimuon resonances etc.) come only much later A. Maciel (NIU) Muon L-2 Two examples online offline

7. II-1. L2mu Eff. Monitoring Online Possible with a L2-monitor trigger L1*L2*L3 L1; mu+jet, lowest jet threshold, 3 layer muons (4 GeV thresh.) prescale as needed (or raise jet) muon configurable as L, M, T L2; mark and pass (1mu w/ L1) L3; apply ID-control filter  loose muon (L1 tagged) inside jet - performance histos can go to Examine - prescaled full readout of L2 data to tape - difficult to monitor lowest Pt threshold A. Maciel (NIU) Muon L-2

8. II-2. L2mu Eff. Monitoring Offline Building a control sample: (1)Unbiased (virtual) trigger (2)ID-control selection Unbiased Virtual Trigger Bits, set upon reconstruction of ALL stream. MU_UNBIASED, CFT_UNBIASED, CAL_UNBIASED etc... Where MU_UNBIASED =.OR. of all triggers w/out Muons MU_UNBIASED.AND. CFT_UNBIASED is effectively an offline mark&pass sample for both L1mu and L2mu Apply ID-control filter to this sample and extract efficiencies from trigger bits Apply ID-control filter to L2µ passed events in this sample and examine purity (rejection) A. Maciel (NIU) Muon L-2