Minutes of Detector Characterization Teleconference
(September 26, 2003)
Present:
AEI: Bruce Allen
Calgary: Warren Anderson
Cal State Dominguez Hills:
Ken Ganezer
Caltech: Ron Drever, Albert
Lazzarini,Kaice Reilly, Peter Shawhan, John Zweizig
Cardiff: Joe Romano
LHO: Mike Landry
LLO: Vijay Chickarmane
Louisiana Tech: Natalia Zotov
Loyola: Martin McHugh, John
Whelan
Michigan: Keith Riles
MIT: Patrick Sutton
Oregon: Ray Frey, Masahiro
Ito, Isabel Leonor, Rauha Rahkola, Brian Stubbs
Wash. State: Sukanta Bose
M3 & M4 Report (see KR's web
summary
)
E10 & S3 Planning:
- Scimon shifts - KR
The first draft of the E10/S3 scimon shift schedule
was not quite ready for this meeting, but was posted the next day. Some changes
from S2:
- Weekday shifts (except Thanksgiving, Christmas, and New Years) at
LLO will be treated with 1/3 weight, since little science mode date is expected.
All of those shifts will be manned by LLO staff, who will probably spend most
of their time on other work, but will be present in the control room and
responsible for minor scimon checklist tasks not requiring a locked interferometer.
- The shifts at both sites will start at midnight / 8 a.m.
/ 4 p.m. local time.
- Shift allocation among groups is done strictly according to FTE count.
It is assumed that all groups have taken advantage (or will take advantage
soon) of trainee shifts to come up to speed.
- Invasive procedures (E10/S3)
- Calibration runs and injections (Mike Landry)
Mike summarized the calibration team's
plans
for the E10 and S3 runs (to get the most up-to-date version, go to
the calibration
home page
, then click on the "S3" link at the left). Briefly, 90% of the calibration
procedures will be completed before the start of E10, with measurements beginning
after the hardware freeze. The team requests that the first S3 shift be dedicated
to calibration measurements and setting of injected line amplitudes. In addition
to regular swept-sine autocalibrations, the team also requests mid- and end-run
calibration measurements. Recent improvements include new hardware modelling,
fast determination of unity gain points, and additional outputs from autocalibrations,
including inspiral range. Channels to be used for calibration / signal injections
a bit differerent from S2, requiring changes in LSC front-end code and RDS
channel list. Peter pointed out that the state vector bits will need to be
redefined slightly to permit excitations on the new channels without disabling
science mode. He also suggested that SenseMonitor may need revision in dynamic
gain monitoring if calibration line injections are done farther upstream
than done in S2. Mike and Patrick will confer on this last point.
- Hardware signal injections (Peter Shawhan, Bruce Allen, Sukanta
Bose)
Peter summarized the hardware injection team's plans for E10 and S3. There
will be one dedicated shift in E10 and probably two in S3, along with about
a dozen 1/2-hour sessions distributed through the S3 run. Aside from a near
repetition of the inspiral/burst injections done in S2, there will also be
five very weak pulsar injections on continuously during E10 and S3.
Bruce Allen will be working in the coming week on the waveform generation
code. There are non-trivial changes needed relative to what was done for
the 10-hour pulsar injection in S2 The stochastic group will be doing a week-long
injection during E10, also a significant change from the 15-minute injections
done in S2 Other issues discussed: 1) Peter will be using a more realistic
actuation function than a simple pendulum; 2) the problem of restarting a
crashed injection program while maintaining synchronization between the sites
needs looking into; 3) Mike requested that the excitation channel be monitored
for malfunction (alarm on presence of extra noise, not just on absence of
expected lines).
- PEM injections (Robert Schofield)
Robert plans two rounds of injections at both sites, just before/during E10
and very late in S3. He will carry out both LHO injections and the first
LLO injection, coordinating with Syracuse to get the second LLO injections
done. He will need about four hours of science mode for each injection.
DMT Software Status Review:
- Status of DMT infrastructure
- John Zweizig: Version updates, deadlines, etc.
John reported that Shourov's new linear predictor filter class is now installed,
after a problem with template libraries was fixed. The handling of minute
trends has been upgraded (spurred by Patrick Sutton's troubles). When generating
new trends, old trends will automatically be renamed, to make offline running
easier. The trend-reading class now allows reading of statistics infos and
multiple channels from the same frame. John is now testing a new class that
allows for convenient alignment of time strides in accessing data. This should
allow better synchronization of minute-trend creation and allow more efficient
once/hour running of programs like SpectrumArchiver. The current DMT version
is 2.6 and there are tags labelled M3final and M4final for what existed at
the end of each mini-run. Version 2.7 will be tagged on Wednesday October
15, two days before E10 begins.
In conjunction with M3&M4, many monitors have been fixed/improved recently.
John specified an October 15
deadline for getting new monitoring code to him for running in
E10 / S3.
John has commissioned a 2-cpu linux box called alvar at LLO for offline analysis.
He has also received 3 pc's at CIT for similar uses and understands
there will soon be something similar at MIT. He thinks all of the new machines
should be operable by the start of S3.
- Status of DMT monitors
- Dave Chin / Tim Bodiya / Keith Riles: LockLoss and ServoMon/SpecMon
KR reported that the LockLoss version (condensed) of the IFO state vector
is now provided as a minute trend, not just as a DMT viewer plot, and the
online documentation for LockLoss has been updated and expanded to address
complaints from the M3 run. In addition, two bugs that affected the handling
of 2-week history files (used for DMT viewer plots) were fixed. KR exercised
the new SpecMon monitor during the M3 run and found it potentially useful,
but not yet ready for prime time. Undergraduate Tim Bodiya is working to
address the M3 complaints.
- Masahiro Ito: glitchMon
There are now four glitchMon processes running at both sites, and all seem
to be functioning correctly. Masahiro is now addressing some minor issues
raised in the M3/M4 runs. John suggested adding (for the long term) and option
that allows glitchmon to be run offline and produce histogram files on a
regular basis. At present, histograms are visible only when the monitor is
run online.h
- Sergey Klimenko: LineMonitor and WaveMon (absent -
attending burst f2f meeting)
KR commented that the LineMon programs running during M3/M4 seemed to be
functioning correctly, except that the Hanford programs were not updating
their summary html files.
- Szabi Marka: IRIG-B, TimeMon, and ShapeMon
(absent - attending burst f2f meeting)
KR commented that IRIG-B and TimeMon elicited almost no complaints from M3
& M4 evaluators.
- Tim Olson: SpectrumArchiver
KR summarized an e-mail
from Tim. Briefly, SpectrumArchiver has many new channels archived. Some
bugs were fixed, and some complaints from the M4 run will be addressed. Tim
wondered if others had succeeded in getting the DTT program running
on linux, since he would like to develop a new version, customized to pull
up SpectrumArchiver files. Peter said that he could add DTT to the LIGOtools
suite if portability is verified.
- Steve Penn / Vijay Chickarmane: BicoMon / BiLinMon
Vijay reported on a brand-new matlab-driven, medm-panel display of auto-bicoherence
that was exercised in M4. Responding to comments from Steve and KR, he is
adding some new features, including display of expected (Gaussian noise)
integrated bilinearity index with errors and user-definable bandwidths for
specified frequency pairs.
- Brian Stubbs: SuspensionMon
Problems noted in M3 with the log file have been fixed. Brian is working to
address other M3/M4 requested changes by John's October 15 deadline.
- Patrick Sutton / Mike Ashley: SenseMonitor / LineAmp and
RayleighMonitor
Patrick first reported on Mike's behalf concerning LineAmp (written by Sam
Finn and Mike) which is an offline matlab program for determining calibration
line strength via 60-second heterodynes. The program does no filtering, since
Sam had earlier done a study that showed filtering was unnecessary.
The program was recently compiled to run much faster and can go through 24
hours of data in about half an hour. Some enhancements in progress include
reading of the excitation channel too and the tracking of wandering phases.
Patrick said that no work has been done recently or is planned on RayleighMonitor.
KR suggested, though, that it be tested with the latest DMT version,
since it's not a background monitor that automatically get testesd by John.
Regarding SenseMonitor, Patrick is working to implement dynamic tracking
of DARM gain and the key input matrix element. Most of the new code is written.
The excitation channel will also be directly monitored in the the new version.
MikeL noted that in the first 15 minutes of a typical H1 lock stretch, the
input matrix changes by about a factor of five as the thermal lens sets in.
He wondered if this instability should disqualify the data from being science
mode. KR felt we should go ahead and declare science mode, and if it proves
necessary later, flag those periods in the data quality segments.
- Natalia Zotov: PTMon (report by e-mail)
Natalia reported that she had turned off triggers in PTMon for the M4 run
because of a bug, but she believes she has fixed the problem. She is now
working on a summary html file and on minute trending. She hopes to have
these features by the time of E10.
- John Zweizig: BitTest, DataQual, HistCompr,
PSLMon, and SegGener
John reported no recent work on the above monitors, but he did exercise
a new monitor in M4 for detecting inferred saturations of photodiodes (already
used as an offline monitor for S2 data quality investigations). It seemed
to give sensible results in M4, including an alarm when saturation really
did occur (and caused other problems). John will be adding more output data
products in preparation for E10. He noted that the saturation monitor is,
in principle, no longer needed because of recent checks added to the front-end
DAQ code, but KR agreed that a little redundancy would be a nice safeguard
on this issue.
A.O.B.
- Next detector characterization telecon: Friday October 3 at 2:30 EDT
(** note unusual time **)