MACRO Shift Report for the period Aug 11 - 29
Scintillator Group Reporting:
Lori Gray (at Aug 21 meeting), Erik Katsavounidis, Ioannis
Katsavounidis, Sophia Kyriazopoulou, Massimo Orsini, Paolo Saggese, Stefano
Stalio, and Charlie Peck (your bemused reporter - why is every terminal at LNGS
different from every other one??? And why are there so many different box
naming methods??? )
During this period, MACRO has run smoothly with just one significant data
collection problem (SM3 LIP as of 8/27/97) mentioned later. However, a various
isolated problems due to PMs, cables, electronic, etc., have been detected and
many have been fixed. Furthermore, considerable work has been done on problems
associated with the LEDs. These matters are also discussed later.
On the political front, the biggest news is that the new lab director -
Alessandro Bettini - has taken a strong and, it seems, personal interest in
safety issues in the tunnel. He has demonstrated his proactive approach by
wandering around the lab, noticing things, and requiring that problems be
solved. Of course, this is overall good news since we all want to work in a
safe and secure environment, but it is also troublesome because he has made
some demands on MACRO.
The first is that we remove the oil stored in tanks in the attico. There is no
technical reason for having oil stored in the attico and since an unthinking or
incredibly clumsy person could break a tank, we have started an effort to remove
the surplus clear oil from the three filled tanks in the attico(about
4500 liters). This will require about 25 standard barrels and the mainproblem
is to obtain them at a reasonable cost. Erik has begun making inquiries at local
hardware stores, and the first estimate is $200 per empty steel barrel!
The director's second complaint was about our famous oil leaks. He had
announced some weeks ago that he would require that we empty any tanks that
leak. Of course, if interpreted in the most simple-minded manner, this would
a) decommission about half of MACRO, and b) have little effect on many of the
drips since they come from oil captured on surfaces outside of tanks.
Erik and I talked to him about the problem for about 20 minutes, taking a
positive and cooperative attitude, but explaining the situation and giving him a
copy of the oil hazard report that Doug has written (it was news to him). No
sharp conclusion came from the meeting, but he promised to read the report.
As many of you know, the fire hazard due to oil leaks is negligible so long as
we rigorously avoid allowing the oil to get on any wicking material such as
cloth, paper towels, napkins, etc. However, the amateurish and unprofessional
appearance of plastic drinking cups attached by bits of wire on our tank ends
acting as oil catchers has strong psychological effects on some people -
including some of our collaborators. Complaints about our oil
leaks would have been minimal (I think) if we had engineered small, dark PVC-
colored, metal drip catchers, permanently attached to each tank, but it's
probably too late for that now! In any case, it would be good politics to make
every effort to eliminate as many cups as possible, especially from the bottom
layer of tanks. To begin this program, we attacked the worse dripper in the
bottom layer, 2B04-11 - it was leaking at a rate of about one full drinking cup
per couple of months. It turned out that the cannister had cracked at one of
the screw holes; it has been repaired using Stefano Stalio's idea of
simply cleaning as much oil as possible out of the screw hole, then filling the
hole with black RTV, and then installing the end cover in the normal way. This
was done and it has completely stopped the leak (at least for a week so far) and
so we will now investigate the next worst leaker on the bottom layer.
Various problems, etc.:
Erik has prepared a program which detects various problems in the data just
after it has been taken and persistently sends e-mail to the shift worker until
the problem is fixed - something to look forward to on your next shift! This led
to observation of what could be either a light leak or a problem in a PMT in
5C03-0. Further investgation is in progress.
Another problem that has been with us for some weeks is that 4E02 has been
missing from CSPAM; mechanical manipulation of the cables associated with this
counter hasfixed the problem.
A WFD channel (associated with a fake box, actually - SM 3channel 26) was also
repaired by the "armstrong" technique of wiggling things -in this case rather
strenuously.
An effort was made to activate a few counters which have been missing for some
time (forever?) from the LIP trigger in SM3 during the calibration day of Aug
27. Unfortunately, something unknown was done which has caused LIP from SM3 to
completely fail. Thus starting on that date, there are no LIP triggers from
SM3. The matter will be taken care of as soon as possible.
Finally, as described in an attached e-mail from Ioannis, investigative
modifications were made relative to the LED drive
circuits on 8/27/97. Some time ago, a modification was made to the outputs of
the switchboxes driving the LEDs. For all of the vertical channels, a 50 Ohm
resistor (cable impedance) was installed in series with the output, but nothing
was done to the horizontals. Thus, since the cable drivers have neglible output
impedance, any signal reflected from the cable mismatch at the LED is absorbed
at the source and not re-reflected toward the LED for the verticals, but these
reflections have not been suppressed in the horizontals. This modification made
it possible for Sophia to obtain nice TOHM calibration information for the
vertical scintillators, but she has had persistent problems with the horizontal
TOHM calibrations. However, it should be noted this modification reduces the
maximum signal that can be applied to the vertical LEDs by a factor of two, a
fact of significance only for LAMOSSKA calibration. (Note that the line drivers
in the switchboxes have some maximum output voltage, so just driving the system
with bigger pulses may not help).
To investigate things, we interchanged the switchboxes used for the verticals
and the horizontals. The result is that Sophia can now nicely calibrate the
horizontals, but not the verticals. Clearly, the conclusion of this exercise is
that ALL LED drivers should be terminated at the source, and work will be
starting soon to make the necessary modifications. Calibration of the LAMOSSKA
will presumably be affected by this, but our first job is to be sure we can see
monopoles! How to do LAMOSSKA calibrations is still an outstanding problem and
further developments can be expected.
I also note that Ionnis, Paolo, Massimo, and Stefano have prepared a very nice
and effective setup in the dark box to investigate LED associated issues and it
has proved very useful in understanding some questions that have arisen.
There follow two e-mails describing further things, inincluding the LED
situation as of August 26, in more detail.
Charlie Peck
______________________________________________________________________________
Date: 26-AUG-1997 15:37:56.95
From: VAXGS::STALIO
Subj: Report of tank-end interventions in the last 2 weeks
Hi everybody,
Here is a list of all the actions that were taken in the last two
weeks as regards pmt + splitter box interventions.
Stefano
- 2b04-1 Not present in runs 14516-14520
On 25-aug-1997 tank-end work on 2b04-10 was performed
in order to try to fix an oil leak fron the end-cap.
On both ends old regular bases were replaced with hv bases.
Tube 8968 (2b04-10) was replaced with 6216 because it did not match
the voltage of the other tube (the difference in the operating
voltages was 150V). It has to be considered good for use in another
tank.
new values: -10 9.8 Mohm 1380V
-11 10.2 Mohm 1390V
new splitter value is 92K (theoretical 71k)
********************************************************************************
- 3T01-1 Not present in runs 14497-14505
This counter had the 3T01-11 PMT dead (not giving any signal)
On 21-aug-1997 tank-end work on 3t01-1
3t01-10 tube 8711 replaced with 7959
on both ends old regular bases replaced with hv bases.
old values: -10 1545V
-11 1475V
new values: -10 1440V
-11 1450V
new splitter value is 95K (theoretical 67k)
********************************************************************************
- 5W01 Not present in runs 14482-14486
On 19-aug-1997 tank-end work on 5w01-1
PMT 6942 had a strong "dynode peak", replaced with 7658
Old regular base replaced with hv base.
old hv = -1620
new hv = -1410
********************************************************************************
- 6b03 Not present in run 14464
On 14-aug-1997 tank-end work on 6b03-0
6b03-00 tube 6720 (noisy, very high dark rate) replaced with 8371
on both ends old regular bases replaced with hv bases.
old values: -00 1435V
-01 1460V
new values: -00 1300V
-01 1340V
********************************************************************************
- 4T15-1
20-sep-1997
This counter had a bad splitter-box resulting into 4T15-11
not giving any signal. After fixing the splitter box looks O.K.
********************************************************************************
- 5T02-0
20-sep-1997
This counter had a bad splitter-box resulting into 5T02-01
not giving any signal. After fixing the splitter box looks O.K.
********************************************************************************
- 6B15-0
20-sep-1997
This counter had a bad splitter-box resulting into 6B15-01
not giving any signal. After fixing the splitter box, this PMT
had too high gain, and thus was reset by Stefano.
Here are the old -> new values:
6B15-00 1680 -> 1700 V
6B15-01 1630 -> 1580 V
Splitter 333 -> 813Kohm
********************************************************************************
Date: 27-AUG-1997 19:24:28.23
From: VAXGS::KATSAVOU
Subj: PMT splitter boxes fixed (2B04-0, 2B12-0)
To: KATSVNDS, PECK, SOPHIA, GRAY
CC: STALIO, SAGGESE, ORSINI,KATSAVOU
20 records
Attributes: None
Ciao a tutti,
During today's maintainace day we discovered and fixed two more
broken HV PMT splitter boxes.
- 2B04-0 : the "usual" cold solder joint for the series resistor
- 2B12-0 : a very difficult to find (thanks Massimo !) problem, since
one of the two HV connectors had its center pin too low and
thus was not making contact.
For 2B12-0, there was also the need to reset the gains for the two
PMTs - please ask Stefano for more details on this.
Please make a note of these channels that may turn hot for PHRASE
and ERP-GC and also make an entry on the wrong list. We hope that we
have fixed all bad splitter boxes by now - let's keep a close eye
for those that may break in the future.
Regards,
Ioannis Katsavounidis
********************************************************************************
Date: 26-AUG-1997 12:20:33.26
From: AXPGS3::KATSAVOU
Subj: LED suggestions
Ciao a tutti,
Following last Thursday's group meeting, we invested some more time on
testing the whole LED setup, trying to find ways to improve it. To simulate
the existing situation as much as possible, we used the dark box, providing
the high-voltage from the spare LeCroy mainframe. The portable HP-pulser
was used through the old/new BU fanout + switchbox + 40-mt. long cable to
drive the LED in the dark box and the PMT signal was again fed through
the (positive overshood modified) new BU fanout before looking at it with
the digital scope. An important addition was that of the light-concentrating
cone and the placement of the LED at exactly the same position as they
are in (most of) the detector.
The entire technical team here at Gran Sasso (Stefano, Paolo, Massimo
and myself) have contributed in this project. Here are our findings.
HORIZONTAL LEDs
===============
These consist of the usual red HP-LED, in parallel with a 50-Ohm
resistor. The polarity of the LED is such that it requires negative
signals to turn on.
These LEDs suffer from severe reflections, coming from poor
termination (the LED has 20-30 Ohms when ON). This is clearly
visible for all types of input signals.
Suggestion: using a trimpot, we determined that a resistor in series
at the switchbox side with a value~=18 Ohm can virtually eliminate these
reflections. At the same time, this value of 18 Ohm (which is also
commerically available) reduces the maximum light output of the LED
by an amount that still allows the big-charge calibrations (LaMOSSKa)
to be performed succesfully. The same applies to the LED/TDC calibrations.
What remains to be seen is if this modification can also make the fine
tuning required by the low-light TOHM calibration easier.
VERTICAL LEDs
=============
These consist of the same red HP-LED, in series with a 33-Ohm resistor.
The polarity of the LED is such that it requires positive signals to
turn on.
Sometime back in Feb. 1996, the switchboxes that correspond to the
vertical counters had been modified by adding a 50-Ohm resistor in series
at the switchbox side, presumably to reduce reflections and other ugly
side-effects.
Our study showed that these LED, having 30(resistor)+~20(dynamic
resistance of the LED when it is on) are effectively terminated. Thus,
the addition of the 50-Ohm resistor in series at the switchbox side
simply halved the dynamic range of voltages for driving the LEDs,
without contributing at all in reducing reflections.
That explains why Lory's big-charge calibrations (i) can not produce
big charges for these counters (ii) never see any reflections.
That also explains why Sophia can sample with much more accuracy
the efficiency region of the TOHM for these counters (half the dynamic
range implies double resolution).
We think that this 50-Ohm in series is not necessary.
If there was need to fix reflections back in Feb. 1996, that was
for the horizontals and not for the verticals. Given the low light
seen by LED/TDC and the LaMOSSKa calibrations, removing it will
improve the situation overall. Low-light calibrations will suffer
slightly by this removal of the 50-Ohm resistor, but in a way that
is easy to compensate for: just re-adjust the voltage range corresponding
to the efficiency region, something that needs to be done, anyway (see
item below).
SWITCHBOXES
===========
The switchboxes we use, unfortunately, have a less-than-unity gain.
More precisely, they have a gain of ~0.8. Paolo found the data-sheet
for one of the two stages of buffering that takes place and it is
quite clear why this is the case: these buffers have a unity gain
when driving 1K load. They have 0.9 gain when driving 50-Ohm.
So, two stages of these 0.9 gain buffers result into the 0.8 gain
we get for these switchboxes.
Of course, we can not avoid driving 50-Ohm at the final stage (=the LED),
but it is not clear to me why we have a 50-Ohm load at the intermediate
stage.
It is quite important to have a unity gain switchbox, in particular
for the LED/TDC and the LaMOSSKa calibrations.
You see, the maximum output out of the HP-Pulsers used for calibrations
is 8V. Thus, when going through the switchboxes, this becomes 6.4V, and
thus we can not get the maximum light out of the LEDs (40m cable, resistors
in series e.t.c.) We saw, for example, that we can compensate for the
reduction in amount of charge we get from the horizontals when we put
the 18-Ohm in series by increasing the pulser output by 2-3 Volts.
Thus, it is quite important to provide the maximum possible voltage
to the LEDs, ideally all 8V that come out of the HP-Pulsers.
Possible fixes:
(1) Take out the 50-Ohm load from the Old-BU fanout that is currently
used to create 8-copies of the pulser output. This way, we are effectively
doubling the pulser output (+-16 Volts).
(2) Modify/change this Old-BU fanout with another system that has gain ~1.2
in order to compensate for the 0.8 gain of the switchboxes.
(3) (Not recommended) Try to modify all switchboxes in an attempt to change
their gain to be as close to 1.0 as possible. Note that from the data sheet
there appears to be no external compensation for adjusting the gain on these
buffers. So, it is going to be very difficult to modify the switchboxes to
achieve a unity gain.
(4) Do nothing; just accept lower light from the LEDs.
PULSE DISTRIBUTION SYSTEM
=========================
Currently, there are 2 Old-BU fanouts creating copies of the pulser
outputs for each supermodule (=inputs to the switchboxes). These fanouts
have different gains for different polarities of input signals. More
specifically, they have
(i) More-than-unity (~1.3) for negative signals
(ii) Less-than-unity (~0.8) for positive signals
They also provide different terminations for the two different polarities,
effectively changing also the shape (including overshoots/undershoots)
for it.
As is, this actually makes the situation worse:
- For the horizontals(=negative pulses), where we have a hard time fine
tuning low-light calibrations, this increases the dynamic range and reduces
resolution.
- For the verticals(=positive pulses), where we have the 50-Ohm in series,
this decreases the dynamic range and results in getting even less light.
One alternative is to use the new BU-fanout instead of the old BU-fanout.
This should equalize the two polarities, since the old BU-fanout has an
almost unity gain (=0.97) in both cases.
Another alternative is to use another distribution system that has
higher-than-unity (~1.2 ?) gain to compensate for the 0.8 gain of the
switchboxes. But, we *may* have just been lucky, having this asymmetry
for these old-BU fanouts. See item below.
AN EASY (??) FIX
================
Given the overall situation, here is a solution that *may* just solve
all our problems in one.
(1) For the switchboxes, just swap the vertical ones with the horizontals,
changing the 50-Ohm series resistor with an 18-Ohm resistor. It is
understood, that since we have 6 horizontal for every 4 vertical switchboxes,
we will need to further modify 4 switchbox cards per supermodule. Thus,
4*6 = 24 cards - 4 (North + South) = 20 cards that need to be modified
from scratch (cutting traces and adding 18-Ohm in series). In doing
this swapping, we can double-check for all the other modifications
that were done in the past (10K in parallel to avoid voltage pile-up,
removal of various resistors/capacitors from the logic part of them e.t.c.)
(2) That's it ! Now, we have the (almost) right terminations for all LEDs
and the fact that the old BU-fanout gives more voltage to the negative
pulses than it does for the positive gives us the additional charge we
need for them, after the 18-Ohm resistor in series addition. As for
the verticals, since they (more or less) worked before, they should
work even better now, without the 50-Ohm in series. Needless to say,
this modification requires all monopole calibrations to adjust the
Pulser settings - but I'm afraid this is unavoidable if we do any
modification to the existing system.
If we all agree on the above, we can go ahead and try it on one
supermodule tomorrow (Wed., maintainance day). Give us feedback ASAP.
Regards,
Ioannis Katsavounidis