1.Choice of trigger source for SLAC combined TKR+CAL muon test.
3 possible trigger sources:
Calorimeter (CAL_LO)
Tracker (3-in-a-row)
plastic counters (external)
Initially tracker trigger planned to be used
It was found that there are dead regions in tracker, leaving without muons the crystals 10 and 11 in Y layers and all positive crystal ends in X layers
next choice- trigger on CAL_LO and use tracker information if it is available
CAL_LO thresholds were set torataher high value of 8 MeV to get 30 Hz trigger rate
threshold tunning was based on charge injection pulses which are 2 times shorter than muon signals
thresholds could be in reality even higher than 8 MeV
no way to measure the threshold; because
we don't know what crystal end produced the trigger
we measure slow shaper signal and do no know fast shaper signal
As muon signal is 12 MeV, these thresholds could affect muon peak position
situation is no so dramartic, because in case of cosmic muon CAL_LO trigger uses OR of signal from 8 crystals
it is enough to have one of 8 siignals above the threshold to get trigger
the distribution of the maximum of 8 Landau random numbers is significantly shifted to higher values - even 12 MeV threshold could be OK.
after having collected ~5 millions CAL_LO triggers we decided to switch to external plastic trigger for the rest of data taking, because
while plastic trigger gives 8 times smaller trigger rate then cAL_LO, the rate of selected vertical muons is only 2 times smaller
plastic trigger is independent from calorimeter energy deposition, so the muon peak positions won't be biased due to trigger thresholds
this is the way to check the effect of CAL_LO thresholds on muon calibration
we expect to collect the same number of triggers with plastic trigger as with CAL_LO
2.Choice of optimal tack delay.
tack delay scan was done to determine optimal tack delay for TKR trigger and CAL_LO trigger. From muon peak position change with tack delay in both cases optimal tack was =60.
for plastic trigger optimal tack delay for tracker was determined to be 30 and tack delay for calorimeter was set to 30+60=90.
3.New time stamps - dead time measurement.
Since end of May Byron Leas implemented time stamps attributed when event is received in PC using 10kHz clock.
the distribution of time between events showed exponential dependence (as expected) above 2-3 ms, but unexplained peaks for distances between events shorter than 2 ms
this strange behavior thought to be related to Windows operating system functioning
Recently SLAC group implemented time stamps in VME processor with 16MHz clock when event is read to VME.
the distribution of time between events is mainly the same as before: there is non-uniformity below 2.3 ms
the number of events in a peak below 2.3 ms is 25% smaller than expected from continuation of exponential distribution : see plots for 27 May and 16 Sep.
there is event losses due to dead time which could be estimated as 0.25*2.3ms = 0.6 ms
this is consistent with direct measurements done yesterday at SLAC
this dead time is explained by insufficient event buffer size on TEM - the buffer could contain only 1 full event with 4 range readout