Minutes of Detector Characterization Teleconference
(January 7, 2005)
Present:
Caltech: Duncan Brown, Peter Shawhan, Patrick Sutton, John Zweizig
Carleton: Nelson Christensen
CSUDH: Ken Ganezer
Florida: Sergey Klimenko
LHO: Mike Landry, Fred Raab, Vern Sandberg
Louisiana Tech: Natalia Zotov
McNeese: Giovanni Santostasi Michigan: Keith Riles
MIT: Laura Cadonati
Oregon: Ray Frey, Rauha Rahkola, David Strom
Penn State: Keith Thorne Milwaukee: Xavi Siemens
M6 Preparations:
- L1 Status (Mike Zucker)
Mike reported that LLO commissioners are in a phase of triage, deciding
which issues are most important and amenable to address before the S4 run.
Before the holidays, L1 achieved an inspiral range of 3.5 Mpc, close to the
nominal S4 goal of 4 Mpc. But other work since then has temporarily degraded
that sensitivity. Expect to regain it soon, however. Recent work has focussed
on changing coil driver resistors to permit lower noise at low frequencies
and trying to fix phase noise troubles at high frequencies. L1 now locks
well during the day and can even ride out a train passage. HEPI is now being
commissioned on HAM modules, which should help L1 ride out large microseismic
noise from Gulf storms (beam pointing jitter). That work may not be completed
before M6, however. There will be operators on shift 24/7 during M6, and
the calibration team plans to have calibration reference functions reasdy
before the run begins.
- Schedule and goals (Keith Riles)
KR stated that the primary goal of the M6 is to verify that DMT
monitors are functioning correctly. Unlike the M5 run, this should not be
regarded as a debugging run. DMT authors should be ensuring their code runs
successfully now at LLO. After discussion with MikeZ, the M6 schedule was
fixed to be Saturday (Jan 22) noon CST to Monday (Jan 24) at noon. DMT authors
should send KR their travel plans for the run or contact information during
the run.
S3 and E11 Investigations:
- Calibrations (Mike Landry, Gaby Gonzalez, Brian O'Reilly, Patrick Sutton)
The final S3 calibration is very close to ready (travel and jury duty interfered
with completion this week). The work should be completed by the end of next
week. The alpha's and beta's for the E11 run are being regenerated and should
also be ready about the same time (reference functions are already available).
Fred suggested that a good monitor for the future (online or offline) would
be one that tracks the unity gain frequency as a diagnostic of imminent servo
instability.
- Glitches (Laura Cadonati, Erik Katsavounidis)
Laura reported that the Glitch Team is finalizing the S3 tunings and revisiting
acoustic events. E11 WaveBurst triggers are now available and will be looked
at soon.
-
An E11 inspiral/glitch correlation
(Nelson Christensen)
Nelson and his students have started looking at glitches from E11, focussing
first on a dozen inspiral triggers. Results from looking at one of the loudest
triggers were shown. There are coincident glitches in PRC_CTRL and on a number
of environmental channels, but only PRC_CTRL is clearly associated. KR suggested
widening the time window to understand better the auxiliary channel ambient
noise.
-
E11 line Correlations
(Nelson Christensen)
Nelson's students are compiling line coherence tables from E11, similar
to those provided already for S2 and S3. A number of lines are correlated
between H1 and seismometers, microphones and magnetometers. Work is just
getting started on understanding sources.
- Environmental Disturbances (Robert Schofield)
No update.
- Strong lines
(E11 vs S3)
(Keith Thorne)
Keith described an interesting finding that came out of the data conditioning
for the BlockNormal burst analysis. Kalman fits are attempted on strong lines
seen in the spectrum (excluding calibration and 60 Hz harmonics). Keith found
that in S3 all three interferometers had a distinct comb of lines separated
by ~37 Hz, with slightly different values for the interferometers. The lines
are narrow (<4 mHz) but not quite stationary. They moved around a bit during
the run. Peter suggested that the slightly different baseline frequencies
and small variations might indicate mechanical resonances with thermal effects.
KR reported that he was able to confirm the comb's presence by going back
to some spectra produced by Vladimir Dergachev last year. He also mentioned
that Robert Schofield had been queried on a possible source for the lines,
but couldn't think of anything plausible with such a narrow width. The LHO
crew on the call couldn't think of a plausible source either. KeithT reported
that a look through E11 data showed the comb still to be present in H2 but
not in H1. Apparently, the problem was fixed somehow.
- Data Quality (John Zweizig, Peter Shawhan, Keith Riles)
KR mentioned the E11 version 1 DQ release made in late December and promised
to provide an S3 version 4 DQ release as soon as the final S3 calibration
with revised bad alpha/beta interval listings is available. Anyone having
other flags to add to this release should get them to KR soon.
DMT Software Status:
- John Zweizig: Infrastructure status
John fixed a problem found by John Whelan when running on linux, in which
decimation via the data accessor was absurdly inefficient. He also found
and fixed the likely culprit behind a slew of apparent floating point underflow
traps in E11 for LineMon and BurstMon. The problem was due to a change in
the data accessor shortly before the run, where assignments weren't preserving
number type. He also found and fixed some memory leaks in PulsarMon. The
deadline for new DMT code to be run in M6 is next Friday (January 14). John
plans to tag a version 2.10.2 the following Monday. He will be at Livingston
for the run from the evening of January 19 until January 24.
- Keith Riles: DMT trend channel names
(strawman proposal)
Following up on a suggestion Patrick Sutton made after the E11 run, KR proposed
a new standardized naming convention for DMT trend channels, to avoid the
frame builder confusion seen in E11. KR will iterate with John and Daniel,
and Alex Ivanov to finalize the convention. KR will request Alex to allow
the frame builder to recognize longer names (now limited to 39 characters
total, excluding file extension).
- Patrick Sutton: Calibration infrastructure
Patrick is helping John Whelan and Marc Cenac with an EasyCalibrate compilation
problem and discussing with Sergey Klimenko and Igor Yakushin the possibility
of doing burst group MDC simulations in the DMT. He would need to implement
the reading of old alpha's/beta's from an xml file for those simulations.
That support has been planned for some time, but is not yet ready. Patrick
confirmed that the extraction of calibration line amplitude is once again
supported by SenseMonitor.
- John Whelan / Mark Cenac / Brian O'Reilly: StochMon
John reported by e-mail that their DMT installation is finally working under
linux, after some help from John., and Marc is working hard to get StochMon
compatible with EasyCalibrate. Another Loyola student, Thair Judeh, has started
work on a coherence monitor to complement StochMon.
- Ken Ganezer: seis_blrms
Ken has been trying to understand apparent filter ringdowns at LHO and LLO,
following data dropouts. He finds ringdowns to be more pronounced at LHO
than at LLO. He finds long data dropouts (~1 minute) lead to large ringdowns.
KR asked for some examples to be shown at next week's telecon. Ken presently
sees apparent dropouts at LLO every 10-15 minutes. John said he would try
upping the priority on the data pusher.
- Sergey Klimenko: BurstMon
Sergey is glad to have the number trap problem fixed in BurstMon and LineMon.
He doesn't have any major modifications planned before M6, but promised to
revisit some lingering problems KR found in E11.
- Giovanni Santostasi: PulsarMon
Giovanni is back from vacation and working at LLO to understand the discrepancy
between online and offline behavior of PulsarMon.
- Rauha Rahkola: SuspensionMon
Rauha plans to increase some trigger thresholds and investigate a slow memory
leak reported by JohnZ in E11. At KR's request, he will verify by next week
that all trends are once again being written to disk, now that the trend
channel names are believed to be fixed.
A.O.B.
- Next detchar telecon: Friday January 14 at 1:30 p.m. EST