Recall Arrival of RPC Pad Bit at TMB

Recall: Arrival of RPC Pad Bit at TMB Note: for Ring+1/3 Ring +1/2 CSC trigger primitive found by this time Avg = 7. 3 • RMS = 0. 79 Width in RPC arrival time has contributions from: 1. CLCT pretrigger jitter CLCT pretrigger • • • 2. Chamber-to-chamber timing differences Time (bx) RPC data arrives after the CSC trigger primitive has been formed To do: Compare (CSC latency estimate + cable lengths) to (RPC latency estimate + cable lengths)… Are they consistent with this distribution? 18 Feb 2009 G. Rakness (UCLA) 1

CLCT Pretrigger Latency Estimate Starting from Jay’s spreadsheet, using current TMB documentation, removing (common) Time-Of-Flight • CFEB drift delay to last hit of 6 layers expected = 3. 0 – CMS setting assumed, CLCT lose ~10% of 6 th hits, lower thresholds help • Signal prop. on strips and CFEB cables = 0. 8 – Value is probably from Korytov • Preamp latency = 1. 0 – Buckeye latency • Comparator delay for peaking time = 3. 0 – Should study comparator and CLCT efficiency vs. this value • • • Comparator latency, clock to first triad = 2. 0 CFEB mux to triads = 1. 5 Average Skew Clear delay for ME+1/2 -3 CSC’s = 2. 1 – Computed based on cable lengths + revision type • • TMB triad demux/synchronization = 0. 0 TMB triad decoding to ½-strips = 2. 0 TMB one-shots/pattern-matching = 2. 0 CLCT pretrigger = 1. 0 18 Feb 2009 Cumulative total = 18. 4 bx G. Rakness (UCLA) 2

RPC-cable length + RAT latency • Average value of RPC RAT cable length, assuming 5. 0 nsec/m = 1. 7 bx • Used calculation of latency from receipt of RPC data to insertion in FIFO = 6 bx 18 Feb 2009 G. Rakness (UCLA) 3

Simple Sums The value of the latency in the RPC endcap Link Board (LB) path relative to the CLCT pretrigger is related by… (RPC latency + RPC-LB cable length + LB latency) + (RPC-RAT cable length + RAT latency) = position_of_RPC_data_in_FIFO + CLCT pretrigger latency From the previous slides: (RPC latency + RPC-LB cable length + LB latency) = 15. 7 What do our RPC colleagues say about this? 18 Feb 2009 G. Rakness (UCLA) 4

RPC Response… From a latency review in Nov. 2007, the RPC estimates: • 1 = chambers + FEBs • 4 = cables (max) is this right for these endcap Link. Boards? • 3 = synchronizer this is not needed for CSC. Is it possible to brake this out? • 2 = coder • 1 = slave-master transmission • 0 = data delay • 2 = muxer • 13 = TOTAL This is not consistent with my estimate… • RPC colleagues are planning to measure their firmware latency on a scope 18 Feb 2009 G. Rakness (UCLA) 5

Low Voltage Control Low Voltage control devices = Maratons and Peripheral Crates (ELMB) Recall: loss of communication to the Maraton means we have to cycle power on the Maraton to restore communication… • Using multimeter and scope to look at CANbus lines in experiment, we see: – Maraton signals do not look too bad – ELMB signals do not look too good – Some noise on both—signal analysis is differential, so it may or may not be a problem… 18 Feb 2009 G. Rakness (UCLA) 6

ELMB digging • CANbus ground was not tied to detector ground – 6 V difference between CANbus ground and CANbus tranceiver ground probably is a problem… – Have connected CANbus ground to detector ground on ¼ of detector. . . Need to see if it makes a difference… • 50 resistance through 27 AWG ethernet cables – Difference in High – Low = 2. 0 V at beginning of chain – High – Low = 1. 0 V at end of chain… – CANbus spec = 3. 5 – 0. 9 Need to see if we see a problem from one end of the chain to another… 18 Feb 2009 G. Rakness (UCLA) 7

Final stuff on CANbus • Evaldas Juska is working on a “handshake” program to write/read register on each component – Had it working on Maratons… He saw the following types of errors after several hours… • • • One Maraton who never responded Several Maratons with no errors… One Maraton who had one error One Maraton who quit responding at 3: 15 am All Maratons stopped responding when Petr messed around with the chain… – Am pushing him to get it working on ELMB’s… – I want to be able to run this overnight on Thursday night so we can get an overview of our current status • Paul Padley said, if positive progress isn’t shown by early March (when he is here), there will be a session of “lock experts in the room, don’t let them out until they come up with a solution” 18 Feb 2009 G. Rakness (UCLA) 8

What else? • Configuration database implementation waiting for ability as we had with xml files, that is… – Display results – Change values (as done with perl scripts) – Presentation to be given by Angela Brett at the online software meeting this upcoming Tuesday… • Started working at 904 to implement TMB-ALCT communication software – N. B. only have an ALCT 672 there… – Current HEAD version of online software doesn’t even compile =: -/ 18 Feb 2009 G. Rakness (UCLA) 9

To do • Get TMB + ALCT test firmware working – at 904 – At point 5 on chambers with problems • Push CANbus studies at point 5 • Combine firmware version check into configuration check moving towards a single push-button check of all configuration… 18 Feb 2009 G. Rakness (UCLA) 10

Backup 18 Feb 2009 G. Rakness (UCLA) 11

RPC trigger latency • In current version of firmwares additional latches are inserted to speed up the compilation total latency ~96 BX. • After firmware optimization it should be possible to reduce the latency to ~ 80 BX LB 8 2 1 4 3 2 1 0 2 2 19 2 3 6 11 Optimization: ~ -10 BXs (2 BX x 5) on transmissions ~ -5 BXs on algorithms Trigger Technical Coordination Meeting , 13 March 2007 flight (max) chambers + FEBs cables (max) synchronizer coder slave-master transmission data delay Muxer GOL fibber TLK Opto - PAC transmission PAC algorithm 6 PAC - TB Sorter transmission 4 TB Sorter algorithm TB Sorter - TC Sorter 6 transmission 3 TC Sorter algorithm 6 TC - HSB transmission 4 Half Sorter algorithm 6 HSB - FSB transmission 3 Final Sorter algorithm 96 Total Karol Buńkowski, Warsaw University