This is an executive summary of our latest thoughts on the AC-coupling
issues. For more detail see the email archives or the memo draft at:
http://hep.bu.edu/~macro/wfd/overshoot/.
1) We considered a (trick) fix where we REDUCE the AC cap in the WFD
d/card. This is the simplest modification. It would cause us to
lose some part of wide waveforms that aren't s.p.e. trains, but we
would keep some part (several usec) of every waveform of any size
and shape.
2) We considered a fix where we INCREASE the AC cap in the
fanout. This requires uncabling the whole detector. We would also
increase a tantalum cap in the d/card as a secondary measure.
Basically, these are the two choices. Other options such as a cutoff
diode on the positive pulse do not help: no diode drop is small
enough. It was suggested to (temporarily?) shorten the stop delay to
efficiently keep the entering pulse. We believe this is not better
than losing a fraction of each waveform as in option 1.
The decision is between (1) and (2). Initially, (1) looked promising,
but we wanted to keep ALL of every waveform for the most important range
of amplitudes and widths. This proved difficult- for the optimum
capacitor value we would start to lose some waveform data at widths >
5.5 usec (beta=3e-4 over 50cm), until the beta is slow enough to cause
spiky trains of s.p.e. Nevertheless, this is still an option if we
decide not to uncable the detector.
Initially, (2) looked discouraging, particularly the need to uncable
everything. We need very large capacitances (1000 uF), suggesting
aluminum electrolytics, which can be problematic. We have determined
the largest available tantalums (220 uF) that would fit, and Chris
calculated that this saves pulse heights >> 2x dyon level. This is
a viable option.
We talked again with our engineer about aluminum electrolytics, and
now think it might not be such a bad idea (Erik/Iaonnis were
right). The problems are bad accuracy (don't care) and short
lifetime. The short lifetime problem should be minimal since these AC
caps will spend very little time holding voltage, and they are not at
high temperature.
Either fix 1) or 2) will allow us to search for monopoles. We are now
leaning towards the large cap fix and are considering aluminum
electrolytics. If the technicians and physicists at Gran Sasso are
willing to undertake a massive fanout change, we support option 2).
Chris is in transit to LNGS now. He has a sampling of caps to try out,
and the SPICE models he has set up and checked is a convenient tool
for testing options. Ed is looking into product literature verifying
that other parameters such as frequency response, ESR (effective
series resistance), inductance, etc are acceptable for aluminum
electrolytics.
The gory details for all three options will get put in the WWW memo
soon and the choice can be made by all the interested parties.
Stay tuned, --- Ed