Present at GS - Chris Orth (author of this report), Erik
Katsavounidis, Ioannis Katsavounidis, Sophia Kyriazopolou, Stefano
Stalio, Massimo Orsini, Paolo Saggese, Lori Gray, and Doug Michael (for
the first week)
The focus of the past two weeks (at least as seen by the
author of this report) has been the Waveform Digitizers. As most are
probably aware, the WFDs have been running at a reduced sensitivity
because capacitive coupling on both the PMT fanouts and the WFD
Discriminator circuits lead to long overshoots for large charges seen
at the PMT fanout. These overshoots can be substantial enough to
trigger the WFD's +/- 2.5 mV threshold for 100usec (the maximum time
depth allowed by our readout window with no zero suppression).
The first week of this shift was spent in discussion by Erik,
Ioannis, Chris, Doug, and Lori in attempt to converge on a solution to
this problem. Prior to the meeting there were two proposed fixes.
One idea was to make the capacitor on the PMT fanout much larger, from
2.2uF at least 1mF. The capacitor on the WFD would also be increased
from 1uF to 220uF (the maximum value available for surface mount
capacitors in the right package). An alternative proposed fix was
simply to make the WFD capacitor smaller.
The arrived at solution is actually a hybrid of the two. We
intend to increase the fanout capacitor to 1mF and reduce the WFD
capacitor to .5uF. In a nutshell, the advantage of this hybrid
approach is that the 1mF limits the positive overshoot of 10V, 15usec
pulses to 1mV. But for large pulses pedestal would actually be shifted
upward by 1mV for quite a long time, about a tenth of a second. The
reduced capacitor on the WFD serves to differentiate this effect away.
The net fix is expected to give us full sensitivity (0-10V signals) up
to about 15usec width. Beyond this our expected sensitivity drops
linearly with charge (pulse width times height).
Unfortunately this fix alone is not enough. For large pulses
where we would regain sensitivity in the WFD, there can still be
sufficient afterpulsing in the PMT and/or Scintillator to cause the
WFD to digitize more than 100usec of "stuff" before it is stopped and
read out 1ms later. So an additional fix involving the creation of a
new highly ionizing event trigger (UHT?) which would serve to stop the
WFDs at 100usec is required. The PMT signals for this trigger will be
obtained from the spare CSPAM fanins.
Chris and Stefano looked at pulses through a PMT fanout channel
and WFD daughter card with the above fixes at a grid of values (pulse
heights of 5V, 2V, .5V, and .2V and pulse widths of 50usec, 40usec,
30usec, 20usec, 10usec, 5usec, 1usec, and .5usec). They observed no
unexpected effects (e.g. spurious discharge of the electrolytic 1000uF
capacitor), and the observed sensitivity was consistent with expected.
Lori has been working to understand our current sensitivity
through LED WFD calibrations. This Wednesday her sensitivity
measurements were given top priority as we attempted to see how
quickly we could find our cutoff sensitivity in the WFDs for 1usec,
2usec, and 5usec wide LED pulses. This is critical because once the
WFDs are repaired we will not be able to measure our sensitivity for
the past year and a half. The biggest constraint was that only 1
large pulse can be fired per PMT fanout or per WFD channel to reduce
the likelihood of crosstalk. To be on the safe side, Lori and Erik
ran jobs this Wednesday that fired only two boxed at a time, one on
each of two supermodules. They were able to complete a calibration of
the first input of each tank-end-1 wfd channel on SM 3 and SM 4 in
about 5 hours (45 minutes per face). It is hoped that this can be
sped up next Wednesday by firing more tanks at a time (up to of
course, the given cross talk constraint), using this week's data as a
crosscheck against cross-talk.
Lori has noticed that the WFD for 3c01-1 has a strange
behavior for her tests. The WFD signal appears to be out of
calibration. The digitized pulse appears to be 10V while the input
pulse is much smaller.
Paolo also entered the WFD universe when he attacked a problem
reported by Ed Kearns' WFD Monitor software. 2B07-0 apparently had no
discriminator hits in the WFD. The problem turned out to be a minorly
damaged FADC chip. The adjacent card had been removed the previous
Wednesday when this problem appeared. Apparently when this adjacent
card was replaced, the replacee inadvertently knocked the card into
2B07-0's FADC chip, causing two of the delicate pins to short
together. The problem was easily fixed. Paolo took this opportunity
to undo the positive overshoot fix that had been performed on this
card previously.
In other WFD news, Alec Habig's DREAM WFD matching code is in
its alpha stage here at Gran Sasso. About 1 event per run is getting
lost or mangled by this WFD matching software. The problem manifests
itself in Erik's statistic files, which show a slight decrease in the
rate of every trigger in the post-combined data vs. the un-combined
data. Erik's DST production code also shows missing events from the
list of runs that meet his first pass monopole trigger criteria.
Debugging this problem is potentially very difficult, because
at the rate errors are occuring it is likely caused by some other
DREAM routine unrelated to the WFD Matching that is handling its
arrays or ZEBRA banks improperly. In fact Alec stamped out a larger
scale problem by debugging the lateral tracking software while he was
developing this code at GS in December. Nevertheless we hope to
implement this code soon in the copy procedure of MACRO so that
everyone can easily analyze the WFD data without having to take into
account ID=2 records. We have therefore invited people to look
carefully at the new data format and verify that the new software
won't break any analysis code that the authors weren't aware of.
The Rest of MACRO
=================
Paolo also repaired some broken CSPAM channels. The symptom
was a loss of sensitivity in 4E14 and 3B07 as seen by Erik's end of
run monitors. Paolo traced the problems to broken AMP cables. Paolo
reported that these cables are quite fragile. One of the CSPAM
problems was traced to a bad (oscillating) fanout, which Paolo fixed
on the fly, since there are no spares.
Stefano reported about an EMILY error received 6 times (once
every two hours) about the HV on 5W08-1 being about 2% less than
demand. When EMILY ran just prior to Stefano's investigation of the
problem, it went away. Stefano looked at this channel directly with
the oscilloscope to look for some oscillation, but found nothing. He
also checked adjacent channels for problems.
Last Wednesday, Stefano and Massimo inserted new LED
switchboxes which are used to fire the North and South face tanks
independantly. The calibration software has yet to be modified
(Stefano is working on this) but through the use of expeditious
doubling up on addressing for the new switchboxes with the EAST face,
good calibration data was obtained for the first time in a very long
while for the North and South faces. No-one here is quite sure why
switchboxes weren't installed for the N/S faces earlier.
Stefano and Ioannis looked into at least one hardware problem
with ERP South Face. Both 6S03 and 6S04 gave ADC Underflows for 90%
of events. They tested the hardware by providing a muon-like pulser
signal to both tank end inputs to the ERP S/H Module, and examined the
output to the Trigger Processor. Both channels were found to have
missing signals on the end-1 line to the Trigger Processor for their
simulated events. 6S03 was repaired by the replacement of a
transistor (2N3906) in the "TRANSCOND-AMP-1" section of the circuit.
The same fix did not work on 6S04, which exhibited strange sawtooth
noise on its output to the trigger processor, which oscillated in
amplitude. Eventually they decided to leave the card in place as it
was, because they were worried about damaging the newly obtained
calibration constants for the other three boxes.
Another ERP problem Ioannis reported about was that on SM3 the
ERP muon rate is about 10% higher than the other SMs. He plans on
comparing this rate to other triggers to track down the scale of the
problem.
Ioannis has also been working on refining the LUTGEN software.
That's all that has been going on these past two weeks.
-Chris