Main INDEX
Monthly INDEX
PREV
NEXT
Make New Entry,
Make Followup Entry
User name R. Michaels
Log entry time 19:35:26 on September 17, 2009
Entry number 290177
This entry is a followup to: 290167
Followups:
keyword=CH scaler (again)
Since this does not happen in Inj crate or Left or Right crate, it must
be a mysterious "crate problem". The CRL code is shared.
The evidence is that the chance of getting stuck is of order 0.1 and
then the scaler either stays bad for an entire run, or stays good for
an entire run. If this is wrong, please correct me because it affects
the following proposal !! (Look at entire runs to tell !!) I remind
you that 15 runs were good during day shift !
Getting stuck means the FIFO has no data because no LNE is seen.
Therefore the static array "sdata" does not update.
It seems the problem occurs in "go" where the scaler is setup and
cleared. Note, the LNE is free-running, so even in "go" the scaler
sees data. Therefore, a possible workaround is to loop N times in
"go" -- for the CH crate only -- which costs ~10*N microseconds -- and
in each loop we setup/clear the scaler and check if the FIFO has data
or not, and repeat the loop until it does. This might reduce the
probability to (0.1)^N. It sounds crazy but is the only solution I
can think of. If you want me to try this let me know.
A copy of this log entry has been emailed to: silwal@jlab.org