please forward to the appropriate folks in SuperK (fwd)
Jeff Wilkes (wilkes@dumand.phys.washington.edu)
Wed, 26 Apr 1995 17:58:32 -0700 (PDT)
Date: Wed, 26 Apr 1995 17:58:32 -0700 (PDT)
From: Jeff Wilkes <wilkes@dumand.phys.washington.edu>
Subject: please forward to the appropriate folks in SuperK (fwd)
To: SuperK DAQ <skelec@dsirae.lampf.lanl.gov>
Forwarded at the request of JGL.
---------- Forwarded message ----------
Date: 26 Apr 1995 14:25:49 -1000 (HST)
From: John Learned at the University of Hawaii <JGL@uhhepb.phys.Hawaii.Edu>
To: Wilkes@phys.Washington.edu, Young@phys.washington.edu
Subject: please forward to the appropriate folks in SuperK
Dear SuperK Collaborators:
Sorry that my tolerance level was low this AM... I get frustrated at our
rehashing of what I thought was clear (but apparently was not) from our
meeting in Irvine and then the last phone call. Let me try to summarize the
situation:
1) We all recognize that the primary job of the OD is tagging of ingoing
particles, muons in particular. This was studied and reported upon at UCI by
at least: John Hong, Bob Svoboda, Mika Masuzawa, Todd Haines, and measurements
by Ken Young and company, plus others I have probably forgotten (sorry). What
everyone finds is that photons will straggle in over a long period, with a time
constant of order 70 ns (or thereabouts). This is very different than in
DUMAND and somewhat I believe (though not sure) in MILAGRO, where there are not
the long reflection paths.
2) A second feature of the OD is poor geometry for doing energy reconstruction
of events, particularly on the top and bottom where there are lots of beams.
It is a lousy calorimeter. Thus we can forget about measurements of muon
track length via energy deposition, measuring the muon energy via knowing
the track and using the dE/dx, etc..
3) However, some studies show that entering/exiting muons produce hot spots,
which can be used to track muons helping to nail the trajectory for muon and
neutrino astronomy purposes. We also know that entry/exit point fitters can be
written to be fast, remarkably accurate, and efficient for use on-line. How
much this adds to the inner detector information remains to be explored, but I
think it may be quite useful, particularly in the event of showers in the inner
detector or multiple entering tracks. The lever arm is, of course,
advantageous.
4) We have not discussed some other phenomena which may be important. For
example, there will often be multiple muons entering the detector. The inner
detector reconstruction of multiples will surely be poor, as we know from
trying this with IMB... all ID PMTs will get blasted. So also will outer
detector PMTs, when integrating. It will be useful to note early large pulses.
5) Another physics question not investigated yet is the importance of low
energy sensitivity for catching particle exiting the ID. This has little
impact on the debate at hand, but argues for minimizing the threshold.
6) Yet another piece of physics might arise from a supernova in our galaxy. In
this case we can expect a huge number of events. Should something happen to
the ID data recording, we might save the day with the OD records. While they
will not be great in terms of reconstructing individual events, we can get the
time profile from this 10kTon detector (if our electronics does not saturate).
7) Another piece of physics to be investigated is the existence of significant
rates of UHE neutrino interactions from such as AGNs. These would be manifest
in UHE muons, radiating like hell when entering the detector. We know, again
from IMB experience, that the inner detector will be saturated by such >100x
min ionizing events, so the outer detector can make a big contribution by
telling us whether it was one barn burner going in or a gang of entering
particles. There are long standing hints for such phenomena, extending back to
Kolar Gold field data from 30 years ago. There are also some hints of entering
neutrino-like events having more particles than should be the case (should be
mostly single muons). The outer detector can be critical in untangling such
things.
I can go on, but you see that it is important to be able to do the best
possible job to have a low threshold and to tag multiple particle events. I
note that the MCs until now have not touched upon most of the physics issues
discussed above. The more general policy of building a detector with redundant
data, when you can afford it, hardly needs reiteration. Having more
information opens up the possibility for catching and/or compensationg for
unforseen problems, and more interestingly for discocery of new physics.
So, what to do: we could imagine waveform recording of all hits, but we can not
afford this now (right?). Thus we must compromise. It was my understanding in
previous discussions, that we had all know the things I wrote above, more or
less, and that we desired to make the best compromise. In particular we want
to prevent finding false hot spots caused by recording TOT only in the presence
of delayed photons. The proposals at UCI were for MILAGRO and DUMAND circuits
to be investigated, and BU wanted to look into using a real charge integrator
(LeCroy) chip as well.
The proposal of Ed Kearns on the phone conference today was that we have only
fast time plus an integrated pulse, Q, from the LeCroy chip. I objected to
this because it seems to me that it works in the wrong direction, giving away
information which we may need, for the reasons touched upon above in #1-7,
and the more general reason of more information being better.
Thus I propose that we fall back to the general technique of digitizing both a
TOT and a second piece of information. This second piece can be a Q from Mr.
LeCroy, a la DUMAND, or a la MILAGRO. I think it makes little difference which
we choose, but I realize this is an issue upon which reasonable folk may
disagree. Also, debatably, I think that if we have both we should not low pass
the prompt pulse, allowing us to catch multiple particles to the extent allowed
by the PMT pulse fall time.
To be explicit, for the sake of debate, I propose we use TOT straight from the
primary timing discriminator, no stretching, and Q delayed by 500 ns. (I am
suggesting 500 ns, because that is well past all physics and light propagation
times we expect, and still not so long as to cause trouble with overlapping
noise pulses or getting into afterpulsing). How the Q is achieved I leave to
designer's (and budgeter's) choice.
Let us get this settled ASAP! Regards,
John