Rob Feuerbach's report for July 11, 2007.

Clues from the previous week suggested the timing of the hits/events
in the ND were not fully understood.

Looking at the "blob" in the beta distribution reported at the
last meeting, showed that it was due to retimed events where the
ND common signal came earlier than expected.

1_trigger-retiming.ps is Figure 1: A diagram explaining the trigger formation
and what happens if an accidental T1 comes TOO early.

2_raw_NDtimes_DctimeL1Adep.ps is Figure 2:  showing the effective ADC and
Timing of ND hits in N1 wrt the ND common signal.  The top panel is for
"normal" T3 (coincidence trigger events) where 2705<D.ctimeL1A[0]<4000; the
lower panel is for "retimed" T3 events, where D.ctimeL1A[0] is actually in
a narrow band between 2700 and 2705.

When an accidental ND trigger (T1) comes in too early, the BB trigger
determines the time of the Coinc. Trigger and L1A, and therefore the ND
common signal.  This leads to a spread-out distribution in time for the
"accidental" ND hits relative to the ND common signal, and is the broad
band between 50 and ~100ns in the bottom panel of Figure 2.

Note that if the accidental ND trigger comes AFTER the BB trigger, we have
the normal situation where the ND is self-timed and the real signal just
comes in later.  Thus, the "REAL" coincident events from the ND are in the
top panel starting ~85ns and extend to later times.  For the "retimed"
distribution, the desired events start at ~115ns.

-------
Related are the "good PMT time" cuts that are placed upon the raw times of hits
from the vetos and ND planes.

3_raw_timedist.ps is Figure 3: shows the typical (has been checked for all
the detector planes and sides) time distribution from PMTs from the ND
planes (top left), Veto planes (bottom left).  The red lines denote the
standard 300ns window cuts.  The top right shows the correlation between
the time of hits in the vetos (y-axis) and in the ND cluster (x-axis) in
ns.

Since we have an up-to 110ns dead-time in the veto electronics, we need to
be able to look BACK in the veto at least 110ns BEFORE the ND hit.  In
addition, only those hits in the ND that can be compared against the VETO
should be used.  Thus, the "good PMT time" cuts on the veto and ND need to
be adjusted to ensure we have this complete coverage.

The effect of the present cuts permits hits in the ND to be used that
CANNOT have hits in the veto within the veto timing cuts.  This biases the
sample toward identifying more "neutrons" than charged particles.

Looking at these, I plan to adjust (in the database) these "good PMT time"
cuts to:
   -100ns < ND_time < 200ns
   -250ns < VT_time < 250ns.