Log entry time 19:19:19 on March 01, 2011

Entry number 349464

This entry is a followup to: 349435

A few suggestions about the deadtime.

If time permits, we should measure the readout deadtime of each crate
by writing a "1" to the SFI output register at start of trigger 
routine, and "0" at the end.  I think this is done in sfi4.crl 
for example.  If one crate is dominating, we may consider what to
do about it.  At some point, it may not be worth pursuing further.

I'm not too surprised that dropping TS11 didn't help much, because
although it sometimes added 5 msec of DT, this only happens every
N events.  Not sure what N is, maybe 200.  Thus the average deadtime
it contributes might be small compared to other crates.

Finally, there is a well-known "dirty trick" one may consider.  All
the DT scan runs were done with ps5=1.  If one goes to higher PS
factors the triggers become more regularly spaced in time, so it's
"like a pulser".  In the limit that's it's really like a pulser,
the deadtime is zero, provided the rate is less than 1/T where T
is the readout DT.  So, one may try, say, ps5=3 and 4 times the
beam current to gain some percentage in overall rate.