Minutes of Detector Characterization Teleconference
(January 28, 2005)
Present:
Caltech: Shourov Chatterji, Peter Shawhan,John Zweizig
Carleton: Nelson Christensen
CSUDH: Ken Ganezer
Columbia: Szabi Marka
Florida: Sergey Klimenko
LHO: Mike Landry, Greg Mendell, Fred Raab, Vern Sandberg, Daniel
Sigg
LLO: Gaby Gonzalez, Vuk Mandic, Brian O'Reilly, Natalia Zotov
Loyola: Marc Cenac, John Whelan
Louisiana Tech: Natalia Zotov
Michigan: Keith Riles
MIT: Laura Cadonati, Erik Katsavounidis
Oregon: Ray Frey, Isabel Leonor, Rauha Rahkola, Robert Schofield
Syracuse: Josh Dalrymple, Alessandra DiCredico
S3 and E11 Investigations:
- S3 Calibrations (Mike Landry, Gaby Gonzalez, Brian O'Reilly,
Patrick Sutton, Xavi Siemens)
Gaby and Mike reported that the version 3 calibrations are now final
and posted on the
calibrations home page
(Hurray!). The revised calibrations
are notable for improved modelling of servo control loops, including
non-ideal filter responses, and for improved tracking of gain variations
(alpha/beta coefficients). The Calibration Review Committee has not signed
off on this calibration release, but analysts should start using it. The
posted calibration information includes frame files of the alpha/beta
values. All of the calibration information, including matlab models and
intermediate products, will soon be placed under CVS control.
- E11 Glitch
investigations
(Laura Cadonati, Erik Katsavounidis)
Laura reported on an investigation by Lindy Blackburn of KleineWelle
triggers seen in E11 data, in which he reran the program after applying
version 1 DQ flag vetoes. As expected, better behavior is seen for the
cleaner data, but a disturbing artifact is seen in certain H1 and H2 channels.
The distribution in time-of-trigger (energy weighted) modulo 1 second
shows 16 Hz periodicity in some channels and a very pronounced peak at
zero for others (mich_ctrl, pob_q, prc_ctrl, and refl_Q for H1; BS optical
levers for H2). The glitches are quite apparent in the filtered time domain,
giving hope that they can be tracked down readily in the control room.
Since the data conditioning is common to the time domain and the triggers,
KR suggested verifying they can be seen in a DTT session with high-pass
filtering.
- S3 Data
Quality
(John Zweizig, Peter Shawhan, Keith Riles)
John and Peter have resurrected their S3 scripts and programs for
generating science mode segments and data quality statistics on those segments.
The segments will be visible to the online astrophysical search analysis
programs running under condor on Observatory Beowulf clusters..
M6
Recap (all)
Keith briefly summarized last weekend's M6 mini-run. L1 struggled
to lock during most of the 48-hour run, which impeded some of the DMT
monitor shakedown efforts, but the IFO finally cooperated during the last
12 hours of the run, allowing some useful evaluations. Bottom line on astrophysical
FOM monitors: BurstMon in good shape, using EZCalibrate; StochMon in good
shape without EZCalibrate, with EZCalibrate version imminent; PulsarMon
not doing well at all on the online machines, despite appearing to work
well in other environments. Spent part of the run resurrecting and updating
scimon web pages, directories and reference files.
E12
Preparations:
- L1 status (Brian O'Reilly)
Brian reported that L1 is now more stable than in M6 and is achieving
lower noise. Commissioners will be noise hunting between now and Tuesday
morning. A newly appreciated problem is backscatter of light from optical
components on tables that do not receive HEPI actuation. As the ground
moves w.r.t. the approximately inertial test masses, the backscattered light
produces fringes that can be seen as glitches in AS_Q/DARM_ERR. Work is
underway to commission more photodiodes to allow higher power. Only two
PD's are used now. It is hope to commission 3-4 by the time of S4. There
has been a data corruption problem indicated by checksum errors; the suspected
culprit is in the reflective memory system. Alex Ivanov is trying to track
down the problem.
- LHO running plans (Fred Raab)
Lately H1 has been running in the 7-10 Mpc range, with H2 at 3.0-3.5
Mpc. If left unattended, the IFO typically has an 80% duty cycle. For E12,
no operator owl shifts are planned and almost no scimon shifts. (Virginia
Re, a visitor from GEO, has volunteered to go through the scimon checklists
and report or fix problems.) LHO plans only minimal commissioning with much
data taking. After some discussion, it was decided to arrange during owl
shifts for LLO operators to monitor H1 / H2 IFO status and hit the "Go"
button when an IFO falls out of lock and recovers to full power on its own.
KR will talk with Dave Barker to see about setting this scheme upl Fred
expects commissioners to work mainly on improving duty cycle rather than
reducing noise between now and S4.
- Scimon shift schedule
(Keith riles)
KR posted a preliminary scimon schedule that had no owl shifts at
LLO, but during the meeting received several more volunteer requests,
enough to permit 24/7 scimonning, with lmore than half assigned to non-LLO
scientists.
- Calibration (Mike Landry, Gaby Gonzalez, Brian O'Reilly, Patrick Sutton,
Xavi Siemens)
Mike reported that he, Greg Mendell, and Rick Savage have been remeasuring
DC calibrations in preparation for E12 and expect to be in good shape by
Tuesday. They don't plan another DC measurement until the end of S4. Brian
reported that the autocalibrator seems to be working and will be run once
per week in E12 and S4. E12 calibrations are expected to be ready by Tuesday.
Fred reminded everyone that the old CARM_CTRL signal will be removed from
the data for this run (no longer needed with new common mode feedback topology).
The new GW channel DARM_ERR will take its place.
- Glitch investigations
(Laura Cadonati, Erik Katsavounidis)
Laura reported on the glitch group's plans for on-the-fly analysis
of E12 data, in preparation for an end-of-run telecon on Feb 9 to assess
whether to proceed with S4 on the nominal start date of Feb 22. Plans include
looking at glitch triggers in the database, scanning loud inspiral events,
exploring and AS_I veto, studying BurstMon trends with followup of loud
triggers, and scanning of excess power triggers, BlockNormal triggers,
KleineWelle triggers, and airplane candidates.
- Hardware injections (Peter Shawhan, Vuk Mandic, Szabi Marka)
Peter reported that they plan to test the mechanics of four injection
types during E12: inspiral, burst, pulsars, and stochastic. There are new
inspiral waveforms and some new pulsar injection code to be tested. Also,
there will be an attempt to inject stochastic signal simultaneously on L1
and Allegro, using a web page for exchanging raw waveform information. The
group requests three 2-hour injection slots during E12. They may want to
run pulsars for more than 12 hours. Erik wondered whether there would be
any injected waveforms that include an inspiral with a burst at the end.
Peter said that Patrick Brady and Steve Fairhurst would visit LLO during
the weekend to assist with injections. He thought they has such a waveform
to try. Gaby confirmed that such a waveform is available. Fred suggested
that it would be good to test injections on H1 and H2, to make sure old
infrastructure still works. Peter agreed that would be a good idea and will
arrange for the tests. For E12/S4 the calibration lines will be injected
on DARM_CTRL, and signal injections will be on ETMX_EXC. KR asked whether
the RDS should include both ETMX_EXC_DAQ and ETMY_EXC_DAQ or only ETMX_EXC_DAQ.
Peter said that only the X channel was needed. There is no plan to inject
on the Y arm. KR asked Peter to propose an injections schedule for consideration
early next week.
.
- Timing stability (Szabi Marka, Daniel Sigg)
Szabi reported that timing signals at LLO are in good shape again at
LLO, following his visit last week. Fred mentioned that LHO has been seeing
atomic clock warning messages recently. Szabi said he look into it.
- PEM injections (Robert Schofield, Josh Dalrymple)
Robert and Josh hope to verify that magnetic and acoustic couplings are
no worse now than in S3 at LHO and LLO, respectively. Robert has been monitoring
couplings at LHO and expects no surprises, but the situation at LLO is not
yet known. Josh wants to start injections early in E12 and expects to need
four hours.. KR asked Josh to propose a schedule for consideration along
with signal injections. Unfortunately, some accelerometers at LLO are not
connected. Fred mentioned a new environmental coupling site at LHO. Noise
on the thermal compensation system (TCS) laser intensity shows up visibly
in the AS_Q spectrum with a 1/f shape up to 500 Hz from heating of the ITM's.
It is easy to magnify noise in AS_Q by a factor of 10 by disturbing the TCS
laser. He suggested that glitch hunters pay attention to the TCS power channel.
- Reduced data set generation (Greg Mendell, Isabel Leonor)
Isabel and Greg are sorting through a variety of requests for new channels
in Level 1 and Level 3 RDS frames. Not all channels (old and new) are present
in the current data streams. So some iteration has been necessary. Greg
mentioned that it should be possible to exchange Level 3 RDS frames between
the sites. KR asked Daniel whether the new fast channels for looking at strain
at the FSR and 2xFSR frequencies should be included in the real-time S3 for
immediate distribution over the network. Daniel pointed out that only H1
channels are commissionied at the moment and suggested waiting until after
the S4 run to create a special L3 RDS with those channels.
- Data Quality (John Zweizig, Peter Shawhan, Keith Riles)
John is dusting off his S3 segment statistics infrastructure. He has added
new DMT alarms on microphone channels and now keeps track of microphone trigger
counts.
Peter's conlog-driven segment generation scripts now trim off the last two
seconds of science segments, since there is a 1-2 second delay for conlog
receiving notification of state vector changes.
DMT Software Status:
- John Zweizig: Infrastructure status
(hardware & software at LLO)
John and others moved several DMT machines at LLO into racks immediately
before the M6 run, and to everyone's relief, everything went smoothly. He
has moved Szabi's SPI page software into the ops directory so that others
can make modifications when needed. The WatchDawg monitor has been modified
to use config files and will run on all online DMT machines. The web emulator
of the DMT viewer has been repaired. KR wondered if the DMT data dropouts
at LLO are understood yet. John said he does not understand it, but has been
investigating it. KR reported that most (but not all) new DMT trends are
visible with data viewer and ligo_viewer; he will work with Ashfaq Khan in
the next few days to get the remaining channels visible.
- Patrick Sutton: Calibration infrastructure
No update.
- John Whelan / Mark Cenac / Brian O'Reilly: StochMon
Joh reported that the new StochMon with EasyCalibrate compiles and runs on
fortress. It will be tried at Livingston shortly.
- Ken Ganezer: seis_blrms
Ken reported that the November 2004 version of the program has been running
since that time at both sites and recording DMT data dropouts that correlate
strongly with SenseMonitor dropouts. A new version is being tested that suppresses
potential filters ringdowns for three minutes after each dropout.
- Sergey Klimenko: BurstMon
Sergey reported that the BurstMon version running at LLO was updated a few
days before the M6 run and seems to be working well. KR asked about a hang
that occurred late Sunday night during M6. JohnZ said he suspected that the
monitor was hanging after resumption of science mode data after a long hiatus.
Sergey said he has a new version of BurstMon for E12 testing that reports
several new figures of merit: the frequency of the loudest recent cluster
and the cluster's SNR (recent means within the last 1-minute time stride),
along with the rate (Hz) of clusters with N or pixels.
- Giovanni Santostasi: PulsarMon
No report.
- Rauha Rahkola: SuspensionMon
SuspensionMon trends are now visible at LLO after some config file changes
on the frame builder machine.
A.O.B.
- Next detchar telecon: Friday February 11 at 1:30 p.m. EST (E12 recap
and S4 preparations)