User name R. Michaels
Log entry time 09:53:07 on May19,2008
Entry number 238748
keyword=deadtime decreased 6% (pedsup ROC8)
First of all, I have modified all CRL codes to have a flag
HARD_OFF, which when =1 will turn off the readout. This is
for quick test to see if this crate is a bottleneck.
We did runs 1318-1323 to check the deadtime, which had been
(from datamon) 32% for 1800 Hz L1A at 3 uA. It was clear from
raw data that ROC8 (bbite crate with 7 ADCs) was putting out a
lot of data and had no pedestal suppression. So we have turned
on pedsup for this crate and the deadtime became 26% for 1900 Hz
L1A. The event size had been 3.5 kByte and after pedsup on ROC8
it became 2 kByte with ROCs 3,4, and 8 about the same data size
and the other crates smaller.
Decided to bump up the current (during run 1324) to 3.4 uA,
and see 30% DT for 2150 Hz L1A. This is an effective readout
speed (of the slowest crate) of 140 usec, which is reasonable.
I think this is about as good as it will get unless we turn on
buffering, which I'm not sure we're setup to do on bbite. If
we turn on buffering I'd estimate we could gain ~10% at this rate.
pedsup on ROC8 uses ~a-onl/ped/ped8.dat (Kalyan's file).
Some relevant runs
1322 -- ROC4 turned off
1319 and 1323 -- ROC8 turned off
1317 -- last production run before the DAQ tests
1324 -- first production run after the DAQ tests