Recently arrived at GS: Colin Okada, Sophia Kyr...
Faithful departed: Doug Michael, Nina Lu
Soon to depart: Chuck Bower (author of this report)
Regulars: Erik Kat..., Maurizio Goretti, Paolo Saggese
(Following are miscellaneous notes drummed up while pondering
why ALL Italian grocery store checkout clerks appear to be former
prison gaurds who were fired for overly brutal treatment of death
row prisoners:)
Data taking has been a bit rocky this week, beginning early
Sunday morning. Problems have varied, but the biggest was
late Sunday night (shortly after the departure of Italian shift
workers Ivan DiMitri and Camillo Giangiordano) when tank 4c07
decided to free run (both ends in coincidence). Tohm, ERP, and
PHRASE all faithfully triggered AFAP, and 650,000 block runs
(which normally take ~6 hours) were occuring every 30-40 minutes.
By 6:00 AM on Monday, there was insufficient space on VXMACA to
begin a new run. After arriving at GS and being contacted by
Jan R., Aurelio cleared up some space. Colin turned off box
4c07, and running resumed. (On Wednesday Erik determined that
both PM's on the -1 end are toast. He will have Maurizio change
them this coming week.)
Several suggestions have been made to prevent this runaway
problem from recurring: 1) turn off MACRO and fill Hall B with
cement; 2) install a terabyte disk, 3) implement Erik's FiDEWS
(filling disk early warning system). As for 3), Eric has a program
which runs periodically (every 30 min.?) and checks the disks
to verify there is enough space to hold eight MACRO runs. If
not, it sends out e-mail warnings to disk managers (Grillo and
Parlati) shift workers, himself, and Kate S. As the disk space
decreases, the messages become more frantic. If during Italian night
and not enough space for a single run, Kate will be at a terminal (so
she says). (Recall: It. midnight = 4:00 PM on the California
beaches). If this system had been in place early Monday morning,
Kate could (and would) have turned off the HV on the out of control
tank and time would have been bought until GSers arrived to start
their Monday work day. This program has now been implemented.
A second program written by Eric checks to see if MACRO is
running. At night (for example), if there is no active run, it
would reinitialize and restart the acquisition. This would work
for the "mysterious" crashes which are fairly common in MACRO
(for example both of last Sunday's aborts). The program is
currently in the test phase (but only sending messages),
and Eric reports that it notified him on Thursday 20 minutes
after a crash.
Another procedure UNDER CONSIDERATION is a program which would
strip the waveform data from a completed run if it is determined
that a noisy trigger has filled up the disk with garbage. This
would NOT be an automatic task; someone (Grillo?) would have to
determine that such a drastic measure was necessary.
Still another potential program being considered is one which
could detect a large number of triggers from a given tank and turn
off the WFD's when a specifed threshold is reached. Apparently
code for this idea has yet to be written.
On the hardware front, the ERP Trigger Processor which has
been acting flaky for 1c13-16 has been traded with the one previously
assigned to 1n05-07. Both seem to be working at the moment....
Rich Baker will UNmodify a spare TP when he comes for shift in
October.
All attico (new and impoved!) Slow Monopole Triggers (SMT) have
arrived by way of various contraband shipments from the US. Sophia
is here to supervise installation. HOWEVER, these modules draw more
current from the CAMAC +12 V line than the crates are configured to
supply. Paolo S. has been working on this problem, but fixes progress
slowly. Unfortunately, there are currently NO spare Kinetic Systems
1502 power supplies operational. Two are on the bench with the SAME
power transistor burnt out. Replacements (equivalent substitutes)
have arrived, but their cases are LARGER than the original and
don't fit in the space left by the removed one. (Could the fact
that these "equivalent" transistor are larger indicate that the
reason the original failed was its small package??) The exact
transistor type is apparently hard to come by, but some have
been ordered from KS.
Meanwhile, Paolo is on his third try at modifying the KS 1500
crates. He currently has one working on the bench with two new
SMT modules running for three days (at the time of writing). That's
the good news. Meanwhile, a 1502 in SM2 (which was thought to be
operational holding two SMT's) has had it's +12 V output droop.
This crate also holds the LIP for SM2, so Paolo cannot debug the
problem until a Wednesday maintenance day (and/or until replacement
power transistors arrive for the two "spare" KS 1502 supplies).
The laser calibration system has reached a critical point.
The lower lasers in both SM's 4 and 6 have "weak" output and
necessitate replacement plasma cartridges. All of the spare
cartridges have been used up (including one unceremoniously
pillaged from the now defunct IU laser spec). Doug is looking
into the backordered parts.
"I'm outta here..." C.R.B.