User name R. Michaels
Log entry time 10:03:13 on October 6,2002
Entry number 87448
This entry is a followup to: 87401
keyword=re: synch problems
Follow-up to halog 87401 which was mostly about SYNCH errors.
(See also ~adaq/synch/README).
I checked runs from 1133 to the present run 1166. Some runs I analyzed
completely, the others only ~50 K events. Result is that runs 1148 and 1149
have a lot of SYNCH errors due to missing gates. This affects the 1877s
mostly in ROC1 and ROC2, but sometimes in ROC3, yet never in ROC4.
This is very bad !
In the other runs, there was only very rarely a missing gate (e.g. once in
run 1135, twice in run 1166). Conclusions:
1. There is always 1 extra gate at the start of run in ROC1, but none after.
Only the 8th and 25th slot are affected. This is a bit dangerous for ESPACE
because it only reports the 1st 40 bad events; 40 events are always considered
bad because of the buffering by 100 (maybe less is bad, but this is conservative).
If extra gates happen after that, ESPACE doesn't report. Luckily it doesn't happen.
2. ESPACE does report the missing gates, always. (Note, "missing" gates are a
different problem from "extra" gates). That's what Xiadong noticed.
3. For this experiment, UNBUFF mode is fine. It is safer because fastbus is
cleared each event. I will switch to this and make it the default, starting run 1167.