I have looked at the data quality of timing reconstructions for almost
the entire attico era (runs 7477-14999) and I thought a couple of
observations should be passed on to you at GS.
First of all, boxes 478 and 479 have performed rather poorly for the
entire 3 1/2 year period. When I plot tError (the difference between
measured time of flight and expected time of flight), these boxes have
a distribution centered at 0, as expected, but the sigma is about two
or three times worse than for normal boxes (3 ns versus 1.5 ns).
When I put a cut at sigma=2.7 ns, there are a few other boxes that
fail it for a week or two, but these two boxes fail it most of the
time throughout the running period. I expect this is a problem with
the ERP S/H, which should probably be replaced.
Also, the quality of the S face timing reconstructions are worse
during the period that Ioannis just calibrated (13969-14999) than they
were before. I don't recall the exact history, but I seem to remember
that for many of the earlier weeks the standard CALMOD database
contains S face constants that somone else (Colin? Teresa?)
calculated rather than the ones ERP_CALIBRATOR produced. At any rate,
when I use whatever is in CALMOD, the fraction of weeks the S face is
unusable is much greater during this latest period (13969-14999). I
don't know if that is due to hardware changes or software changes.
Bob