Here is the somewhat delayed description of the problems
encountered with the testbeam data from last run. In the analysis
we have done so far, we have not identified any DAQ related
problems in the ADC's used with the Phobos detector. We did find
some bad ADC channels which had a significant tail on the
pedestal on the negative side. This was found during the run but
all we did was verify that it was a bad ADC channel and not a bad
detector channel. We have also studied some interesting
common-mode noise problems where the whole detector (or certain
subsections) have correlated changes in their pedestals. This
analysis ended with the conclusions that the detector seemed to
be working well, the efficiency was decent, and the pulse-sharing
seemed to occur only over a very narrow range between pads.
The primary on-line diagnostics needed for the Phobos detector in
the Camac ADC's is a histogram of the signals (preferably
pedestal subtracted) for each pad as well as a 2-D histogram of
the signal on each pad versus the sum of the signals on the
surrounding pads.
For the telescope, we had much greater problems. First, there
were some selected runs where the pedestal was quite broad. We
have not done a full systematic check so I can't tell how
prevalent this problem was. The biggest problem was events in
which the data for part of the telescope appears to be identical
to the previous event. We have only looked at data separated by
sensor, not by individual VA1 chips. When a particular sensor
fails, there is a very high probability that all higher number
sensors are bad. This is why we believe that the problem was some
timing mis-match in the read-out. We also had a problem that a
significant fraction of the strips in the section of sensor#1
that we were using were not giving signals, only pedestals.
Finally, there is a very strange problem with sensor #3. When we
look at sensor #'s 1, 2, and 5 (the Y planes), we see excellent
alignment. When we look at sensor #'s 4 and 6, we also see good
agreement after we correct for the fact that sensor #4 has 100
micron strips (all others are 50 microns). However, if we try to
correlate sensor #3 with sensor #4 or #6 or with a line fitted to
the two, we see a big discrepancy. The effect is that a change in
the hit strip on sensor #3 deviates from the change in projected
distance at sensor #3. The deviation depends roughly linearly on
the location on sensor #3. This suggests that the sensor had a
different strip pitch or that it was rotated, or both. We have
looked at the data in various ways and have not yet found a good
explanation. We continue working on this problem.
We need several on-line diagnostics for the telescope, starting
with a histogram of the signals on each sensor with pedestal
subtraction (a 2-D plot with signals versus channel number).
Then, we need plots of one sensor versus the others in the same
orientation (X vs X and Y vs Y) but here we need to find the peak
and plot peak positions. Finally, we need a check for the
repeated data problem. The last check is to find tracks and fit a
line.