Log entry time 12:46:29 on September16,2009

Entry number 289954

This entry is a followup to: 289929

I think the problem with CH scaler is the mechanism that is
used to "turn on" the LNE at the start of a run.

Because most runs have good CH scaler, but some runs get
frozen for the entire run.  This occurs because "sdata" in 
the CRL gets a static value and never gets updated -- 
consistent with no LNE.  Scalers usually don't fail for
one run and then work for 10 runs ... can't be a scaler problem.

I could not reproduce the failure just now.  

So, just now what I've done is bypass the AND module that turns 
off LNE in between runs.  The result is that there are 2 bad
events (1st and 3rd) in the start of the run where you see
a header 0xfffca011 (0x10 = bad) instead of 0xfffca001 (=good),
then the events are good forever.  The exact same pattern
occurs on ROC25 and 26 which don't have this LNE-disable scheme.
I don't think we need it.

Let's run like this for awhile.  Please monitor the CH synch mon and
the BCM10, and look for the values to fluctuate -- at least some.
Check that BCM10 sees beam trips at same time we do, etc.