I recently scanned a few events looking at what is going on with
channels that hit the bye limit. There seem to be a few things: BTW, I
copied a few postscript files to LNGS and even printed them out so you
can find them this morning. The files are in my home area on VAXGS.
1) channels with something clearly wrong. For example, 0504 (that is
SM-5 channel 4 counting from 0) is always at an ADC count of 100. The
culprit seems to be input 3. You can check the resistors for that input
on the d/c and see if one is disconnected. It's probably easier to
change cards (swap the whole motherboard with one with calibrated
d/c's). Don't forget to update the WAVFRM log.
2) channel 0210 has a very noisy baseline. This seems to be on input
three. This is possibly noise on the cable. Look with a scope. If you
can't see, for one run, just swap two input cables and see if the
problem moves. I am trying to be as non-disruptive as possible. Be sure
to record which run.
These two channels are noted in FIXME.PS. See also BYTLIM6SM.PS which
are some occupancy histograms based on run 010322.
3) There are some channels where one input fires more than the rest, but
not 100% of the time. Sometime this week you fixed one, know it or not:
channel 0400 hit the byte limit all the time in run 010186, but is fixed
now. Two examples of this are in HOT-INPUT.ps (0400 before the fix, and
0423). We may have to tweak the thresholds on a few channels. This is
mostly FYI, I suggest we keep running at the same threshold everywhere
for now.
4) The main new problem seems to big amplitude muons. Somewhere, perhaps
the fanout, there is sufficient capacitive coupling that the positive
overshoot goes over the WFD threshold and turns on the discriminator for
many usec. This is bad and we have to figure out what to do. You can
look at CUTOFF.PS to see two ends of a tank with such a muon. Yes, you
have to grok my funky display where the time scale is due to
concatenating each 5 ns sample- see the TIMEWD plot for
realtime. Anyway, the effect of these overshoots is more dramatic on
this distorted scale.
How do I know it is big muons? Well, first, it is a sensible hypothesis.
But also, I looked at some waveforms where the channel almost hit the
byte limit, but not quite. See ALMOST-CUTOFF.PS where I plot the muon
both in ADC counts and Volts. This pulse is almost -8V in amplitude!
One possible solution is to raise the positive threshold a bit, while
leaving the negative one at -2.5 V. We can accomplish this by changing
992 resistors. This is what we in MACRO excel at! But let's study this a
bit and make sure we understand what is going on- particularly, where
the capacitive coupling is. I know the fanouts have a 2.2 uF AC
coupling. However, I'm reluctant to try to modify the fanouts- the
cabling and uncabling is much too disruptive.
That's all for now. -Ed