Hi Rich,
Well, I'm waiting for the output of a detailed check of all the events
in the last burst... the ADCZ vs TDCZ check should definitely be in
the code but possibly there's some bug. In the meantime, I can make a
few comments:
1. Note that TDCL is always used for TDCZ reconstruction if there
is an under or overflow in TDCH, and this TDCZ is what is used to
compare to ADCZ. (I assume that's what you do too?)
2. I strongly suspect that the problem with the 6th hit is a
whacked-out timewalk correction with the constants I'm using (run
10884-- I usually lag a few iterations behind Alice): for this tank,
one of the TDCH values was underflow, so TDCL values were used for
reconstruction. Since TDCL1 is much greater than TDC0L, clearly the a
z value of 77.5 near the center is wrong; it should come out near side
1 (or way past side 1). The only way the z could end up getting
reconstructed this way is if there is some crazy timewalk correction,
which is not unheard of with our current timewalk correction curves at
very small ADC values.
I think you are right that this event was accidental coincidence of
radioactivity, and I suspect that with a reasonable timewalk correction the
monitor would have thrown it away. Anyway, I'll check this explicitly
and get back to you.
The question of quality of timewalk corrections for low pulse heights
an important one that I've been worrying about for a while: basically,
for a lot of tanks, the timewalk corrections at small ADC values suck.
If you scan through a the TW correction curves, a lot of them (20%?)
go completely berserk at low ADC values, sometimes within the regime
where we're reconstructing low energies: the corrections shoot up to
infinity or even turn over and go way negative. This is a real
problem, and I'm not yet sure what the best way to solve it is. I've
thought about using some kind universal TW correction function (the
physical shape of it shouldn't vary too much from tank to tank);
however, the problem is that the way the ERP calibrations are done now
(tuned for muons), the TDC offsets and timewalk constants for the ERP
are enmeshed and don't individually reflect their physical meaning
(i.e. the real physical time offset for a particular box is partly
contained in the timewalk function normalization). So, if you want to
use some universal timewalk function, you have to recalculate the all
offsets. On the other hand, we might be able to do a better job of
low-ADC individual box TW correction. It's a bit of a sticky
situation, but probably fixable with some thought.
A quick and dirty fix that should help is just to put some limit on the
magnitude (and sign!) of the timewalk correction -- I should do that.
Both the TDCH and TDCL correction functions have equal probability of
being whacked out for low pulse heights (I looked through a bunch of
them when I was worrying about whether it would be OK to have
differing MBT and TDC thresholds, to see whether TDCL reconstructed
info would be much worse that TDCH info. The answer was that they're
equally bad.)
3-- As for the time differences: well, I'm not perturbed by a few
millisecs. The GC buffer times are calibrated using a few tens of ERP
muons per SM at the beginning of a run, and I've measured in the past
that this calibration can drift up to several millisecs after some
hours-- usually it's 1-2 ms in 5-6 hours but can sometimes be more.
The amount of drift is consistent with the errors in the UT vs
ERPtime fit. This is perfectly OK for a GC burst search. Now that we
have the Bari box, we don't need really good inter-SM time matching to
veto a lot of the inter-SM muons anyway (I have a 4-ms inter-SM coincidence
cut window for that anyway). We could do better by looking for more
ERP muons at regular intervals during the run to refresh the ERP clock
calibration, but I don't think it's worth the effort (at least, there
are plenty of things higher up on the to-do priority queue).
The burst events happened several hours into the run, so they could
be off some millisecs -- I don't think it's a problem.
Anyway, I'll get back to you after I've looked at the events in detail.
Kate.