MACROUSA SHIFT REPORT - MAY 1998 (Kael Hanson)

Kael Hanson (kaeld@trex.physics.lsa.umich.edu)
Tue, 2 Jun 1998 16:11:00 -0400 (EDT)

US MACRO SHIFT REPORT - MAY 1998

or

Everything you wanted to know about
machine checks,
calibrations,
and
broken phototubes
but knew better not to ask.

US/PISA PERSONNEL PRESENT AT GRAN SASSO
---------------------------------------

1) Erik Katsavounidis US PIC EK
2) Ioannis Katsavounidis Calibration Czar IoK
3) Lori Gray Semi permanent AQ resident LG
4) Sophia Kyriazopoulou Permanent AQ resident SK
5) Massimo Orsini Aka "Little Bears" MO
6) Nicola Zaccheo Tough guy from Bari NZ
7) Valeriano Marrelli MACRO acolyte VM
8) Roberto Pazzi Departed Pisa shifter RP
9) Kael Hanson Born again shiftworker KH
10) Abyssmal weather Makes Montana seem like good place to vacation

*******************************************************************************
*
* IF YOU READ ANYTHING FROM THIS SHIFT REPORT PLEASE READ THE PART ABOUT
* ERP ENERGY CALIBRATIONS SOMEWHERE DOWN BELOW. I SUSPECT WE HAVE BEEN
* USING THE WRONG NORMALIZATION FACTOR.
*
*******************************************************************************

Tip to future shiftworkers: never comment in the first days of
your shift how robust MACRO appears to have become against acquisition
glitches. The Italians say that making such forecasts brings back
luck -- "porta sfiga" -- and the only way to ward off such hexes is
to immediately touch yourself in the groin area, I won't be more
specific.
Fortunately, although there have been several crashes the total
deadtime has not been too bad. Only during the early morning hours
of May 9 did MACRO suffer a prolonged sleep. The rest of the crashes
have fortuitously occurred one after the other, or during normal
shifter hours, or have been restarted by our good friend ARE, the
auto restart batch job. See below for list of the affected runs.
Of lesser impact, yet still quite annoying, were the frequent
crashes of the MACRO gas system. Indeed, even I got a stab at
rebooting the beastie several times during my first week here.
This problem seems to have left us sometime around the 8th or 9th.

MAINTENANCE ACTIVITIES AT GRAN SASSO
Weekly maintenance on Wednesdays would be insufferable had Yanni
not brought in his espresso machine. For those in desparate need
of that caffeine buzz it's located in the 1st floor barracca. I
think that the SOGEDA people must be on strike since the coffee
machine in the "lounge" room never works.
As of May 1, 1998 there remained two tanks in MACRO without
functioning laser fibers: 1N02 and 6S01. The combined efforts
of IoK, KdH, and VM succeeded in repairing 1N02. 6S01 is a
much more tragic story, sad to say. IoK, VM, and KH have
been working for two weeks on this tank only to be disappointed
in the end.
Regarding PMTs, the techs are staying right on top of fixing
very noisy or otherwise broken tubes. EK's deamons can spot
a flasher or dead tube as soon as a run has ended and send
mail around to let people know that a problem exists. Unfortunately,
we are running a little low on spare PMTs here - the functional
form is something like
NPMT = 18 - (W - 05/15/98), W is in weeks
so only failing or failed tubes warrant tank end intervention.
There are a few other miscellaneous accomplishments of a
larger scale which I briefly describe. This does not include
the many power supplies and switchbox cards which keep our
techs busy:
1) 5/15/98 : 6W07-0 HV was dying at beginning of LED
calibrations. IoK changed to spare LED
and fixed problem.
2) 5/15/98 : WFD fix in ELN OS. Waveforms now correctly
read out on ERP N/S face hits. Original
code by C. Orth; testing and bug fixes by EK.
3) 5/22/98 : No Q's from SM6 TOHM have ceased following
swap of CAMAC crate controller.

CALIBRATION RUNS
Several calibration "specials" performed during this shift
period deserve attention. LG has been working for some time
on improving the efficiency of monopole calibrations. These
runs simulate slow monopole passage through individual tanks
by flashing the pulser. They are complicated by fact that
each tank has its own unique response to a given pulser
input voltage. The improved routine groups tanks with similar
responses and processes them simultaneously.
IoK revived the LED/ADC calibration during the SM3/4
calibration pass. This run involves flashing the LEDs
varying the voltage instead of the delay time. The intent
of this run is to use spe statistics to obtain an actual
map of adc -> photoelectrons.
KH completed ERP pedestal calibrations on all supermodules,
including the N/S faces. These runs use the EX_GATE input of
the ERP supervisor to generate an MBT on all channels independent
of activity in the tank.
As always, details on the complete history of calibration
runs can be found in the text file
DISK$MACRO:[MACROCAL.DOCS]CAL_RUN_LIST.TXT
on the LNGS VMS cluster.

MACROUSA OFFLINE
IoK and NZ are currently bringing the CALMOD database up to
date. IoK has spent the last few months hacking at ERP_CALIBRATOR
and is finally ready to release the results to the public. The
hottest news on this has just developed today - IoK has fixed the
puzzle of the poor quality of timing data from the S face. He and
NZ are rerunning the Jan 98+ cals with this fix; ETA next week.
KH is developing analysis of laser calibration data. The laser
runs have not been analysed since 118?? several years ago because
of a host of difficulties. This means that ERP ADC (software)
pedestals and timewalk constants have been frozen for two years.
Of particular interest to some of the fast monopole analysers may
be the fitting of higher-order terms in the adc->pe function,
certainly important in highly ionizing events for correct
reconstruction of the ERP energy, occassionally relevant in some
tanks around muon energies. The ERP energy reconstruction in
CALMOD has nominally been trusted to a maximum of roughly 100 MeV.
Unfortunately, CALMOD only provides for at most a quadratic
polynomial. To maintain compatibility I am therefore releasing
two versions of the map: one in the standard CALMOD KUMAC which
only contains pedestal, gain, and quadratic terms, another
in simple columnar text tables refit with higher orders. Those
interested check the macro-cal mailing archives. Of more
importance to the neutrino people might be the new timewalk
constants. I have to admit, however, that there are some peculiar
features in the laser timewalk data. This is also probably better
discussed in
LG continues work on analysing the waveform calibrations.
The WFD readout on Lamosska runs allows Lori to detect
problems with PMT bases that are otherwise difficult to spot.
These bad bases are the primary cause for WFD buffer overflow
because, while the PMT still afterpulses enough to cause
buffer fillup, the base isn't able to hold the large charge
and so the Lamosska circuit which causes the early stop does
not trigger.
EK has been compiling data from his RAREDSTs, looking for
statistics on WFD buffer overflows prior to implementation of
the Lamosska fix. His aim is to analyse the complete set of
lost waveforms and see how many of those events are good monopole
candidates using a sophisticated time-of-flight cut gotten from
stopmaster times.
SK's recent monopole works, from the source:
I have been using slow streamer tracking in my
analysis to filter out events that pass my software
trigger. Using the slow streamer chain in order to
perform tracking takes a long time because we have
around 150 hits in every microvax. The streamer
monopole analysis has been using the time infor-
mation provided by the monopole streamer trigger to
narrow down the hits only to the ones in the time
interval indicated by the beta slice firing. Since
no such trigger is required in my analysis I am
using all hits in the slow chain when I try to track.
I have tried to compare the tracking efficiency of
the slow chain with that of the fast chain. Thus,
I have chosen events which fired Bari and ERP
(two face ERP). In a sample run there were 4400
of those events which gave at least one track in the
horizontal fast view (requiring 5 hits in the wires
and 5 hits in the strips). The events were also
tracked in the slow view. For those events the
scintillator boxes that were intercepted by the
track(s) in the fast view were compared with the boxes
intercepted by the track(s) ,if any, in the slow view.
We required that all boxes intercepted by the track
in the fast view were intercepted by the track(s) in
the slow view. 95% of the events conformed with the
above requirement, which I think is something that
allows me to use slow horizontal tracking in my
analysis.

The situation becomes bad though when we talk about
lateral tracking. There we have no strip information
for the lower lateral view and attico lateral strip
info is not yet used in the code. There are 2800 events
having at least one lateral track in the fast view
(5 hits required ). For those tracks we compare the
lateral tanks the y-z projected fast track intercepts
with the ones the slow track intercepts. Only 46% of the
events had slow lateral tracks that intercepted the same
lateral tanks as the fast lateral tracks. Obviously this
is not a nice result, thus for the moment lateral
tracking can not be used the way it is for slow tracking.


**************************************
* What is the true muon <dE/dX> ?? *
**************************************
This parameter has apparently changed at least once during
the lifetime of MACRO. Let me be specific so that there is
no ambiguity to the question. Ionization losses by muons
are Landau distributed so there are two relevant statistical
averages: the mean and the mode, or most probable value. In
scintillator, even at energies up to 1 TeV the dominant energy
loss mechanism is ionization so we needn't include radiative
components. The energy normalization portion of the ERP
calibration software fits the pathlength-corrected energy
to a Landau using 1.8 MeV/cm as the target value. This is
is where it starts getting foggy: the comments actually say
1.8 MeV/(gm*cm^2). The results of a montecarlo using GMACRO
materials and a realistic down-going muon flux indicate that
the comment is correct, not the code. Using the value 1.8 MeV/(gm*cm^2)
and then multiplying by the oil density (0.87 gm/cm^3) gives
dE/dX (most probable) to be 1.56 MeV/cm, 13% lower.
I am currently grappling with mathematics that I left
behind a decade ago to understand the nature of this problem
analytically. The Bethe-Bloch equation should give the
correct answer for the mean of the muon dE/dX (which I calculate
as 2.5 MeV/(gm/cm^2) assuming a monoenergetic beam of momemtum
approx. 106 GeV/c). Folding in the Landau fluctuations and a
non-monoenergetic flux is proving somewhat more difficult, however.
Comments welcome on this point, folks.

**************************************
* MACRO US SHIFT REPORT "Highlights" *
*******************************************************************************
* Gas system shutdowns : Always complains "Tank C pressure out of tolerance" *
* 1) 03-MAY-1998 21:05 *
* 2) 04-MAY-1998 08:20 *
* 3) 04-MAY-1998 08:45 *
* 4) 04-MAY-1998 13:56 *
* 5) 04-MAY-1998 16:50 *
* 6) 06-MAY-1998 09:00 *
* 7) 06-MAY-1998 17:00 *
* 8) 07-MAY-1998 09:00 *
* 9) 07-MAY-1998 17:00 *
* 10) 07-MAY-1998 22:00 *
* 11) 08-MAY-1998 08:30 *
*******************************************************************************
* ACQ crashes *
*******************************************************************************
* 1) 08-MAY-1998 22:15 Run #15782 Power failure in PHRASE *
* SM 6 Rack 9 Crate 3 *
* 2) 08-MAY-1998 23:53 Run #15783 Machine check MV5 - autorestarted *
* 3) 08-MAY-1998 23:59 Run #15784 Dies 09-MAY-1998 2:32 *
* 4) 09-MAY-1998 08:25 Found 15784 dead - restarted *
* 5) 09-MAY-1998 08:33 Run #15785 15 sec junk run *
* 6) 09-MAY-1998 08:56 Run #15786 Good run after reboot of all uvaxen *
* 7) 15-MAY-1998 12:01 Run #15812 Unknown reason *
* 8) 15-MAY-1998 12:54 Run #15813 "" *
* 9) 15-MAY-1998 13:12 Run #15814 "" *
* 10) 23-MAY-1998 07:41 Run #15849 System error in MV4 *
* 11) 23-MAY-1998 08:20 Run #15850 MV2 never started *
* 12) 23-MAY-1998 15:29 Run #15852 Unknown reason *
* 13) 24-MAY-1998 16:43 Run #15857 "" *
* 14) 24-MAY-1998 23:53 Run #15859 "" : 15860 auto restarted *
* 15) 25-MAY-1998 07:47 Run #15861 WFD Crate on SM1 lost -6 V supply. Zero *
* suppression malfunctioned as a result *
* 16) 26-MAY-1998 09:41 Run #15867 Unknown reason *
*******************************************************************************