Minutes of Detector Characterization Teleconference
(July 2, 2004)
Present:
Caltech: Laura Cadonati, Szabi Marka, Patrick
Sutton, John Zweizig
CSUDH: Ken Ganezer
LHO: Mike Landry, Fred Raab,
Vern Sandberg, Daniel Sigg
LLO: Brian O'Reilly
LSU: Gaby Gonzalez
Michigan: Keith Riles
Oregon: Ray Frey
S2 / S3 Investigation Reports:
- Fast fluctuations in calibration (
pdf
,
ppt
) (Gaby Gonzalez)
Gaby went over some investigations concerning fast calibration fluctuations,
summarizing findings to appear in the final S2 calibration document. She
reviewed the history of injected line amplitudes and resulting changes in
precision of alpha*beta determination. In separating spread due to statistical
fluctuation and that due to true calibration fluctation, she looked at the
diffference in alpha*beta between adjacent time intervals and at standard
deviations of 1-minute determinations for 2048-second "chunks". In summary,
over the course of the run real fluctuations of 5-10% were seen in alpha,
with measurement errors of a few percent or better (except the early
H1 data with low amplitude). For L1, there are fluctuations at the 0.6% level
for 1-30 minute intervals. A good estimate of expected background effects
on line amplitude measurement is sqrt(2*S0*fsamp) with S0 the power spectral
density of background noise and fsamp the sampling frequency. Gaby
defines an intrinsic SNR rule of thumb as amplitude / sqrt(2*S0*1Hz) so that
a 1 Hz sampling gives relative measurement error equal to 1/SNR. For S2 data,
these SNRs (high-amplitude for L1,H1) gave about 70, 8.5 and 7 for L1, H1,
and H2, respectively. To obtain 10% resolution per measurement in S2, one
would have needed 50, 0.7 and 0.5 Hz, respectively. Gaby recommends increasing
the relative strengths of injected lines for H1 and H2 in the S4 run.
Another measure of calibration fluctuations comes from the strengths of sidebands
on the calibration lines, which are visible in L1, but not in H1 or H2, with
hour-long averages of 4-second PSD estimates. The sidebands result primarily
from alignment fluctuations. Power spectral estimates of the alignment fluctuations
indicate more cleanly the contributions from the micro-seism and stack resonances.
DMT Software Status & Plans:
- Planning for "Detector Investigation Camp" following August LSC
meeting (Keith Riles)
The detector investigation camp will be August 20-22 at LHO, following the
LSC meeting. The tentative plan is to cover a variety of analysis software
tools, including DMT, Matlab, DTT, and data access tools on the Grid.
- Status of DMT infrastructure:
- John Zweizig: Version updates, etc
A new blade 1000 DMT machine called travertine has been installed at LHO.
The new DMT version 2.9 is working fine at LHO and will automatically be
installed at next machine reboot at LLO. A new trigger table that allows
DMT monitors to store more information is now implemented. Decimation now
has a compensation for delay. A new standard destructor message is printed
at the end of DMT jobs that displays use of vector during execution. The
new feature should help code optimization.
- Patrick Sutton: Calibration infrastructure
Progress has been held up recently by uncertainty on how to package reference
info that will include reference functions and various constants. Patrick
had planned to use frames, but has learned that some of the generic features
in the frame spec aren't actually supported by our frame-reading tools. He
is now considering xml files as an alternative, for which some tools already
exist in the DMT. Daniel promised to send a note with specific suggestions
on defining and parsing xml files. Patrick will learn more about XML
- Status of DMT monitors under substantial development / revamping
/ debugging:
- John Whelan / Mark Cenac: StochMon
Brian O'Reilly reported that he is helping Mark (SURF student at LLO from
Loyola) modify SenseMonitor to create a stochastic background monitor. They
will use a standard spectrum from the "other" IFO to compute sensitivity,
to be read in at startup. A preliminary FOM is working now, but sending the
info to the DMT viewer is not yet implemented.
- Tim Bodiya / Keith Riles: SpecMon
Tim has run into problems this week with type definition changes and the
root library, probably associated with John's recent upgrade to DMT 2.9.
Brian commented that he found it hard to get a compiler-compatible, DMT-compatible
root combination going at LLO.
- Ken Ganezer: seis_blrms
Ken is modifying the existing code to catch time step discontinuities. He
will send the code to Keith to get advice.
- Sergey Klimenko: BurstMon
No report.
- Tim Olson: SpectrumArchiver
No report.
- Steve Penn: BicoMon and BicoViewer
No report.
- Giovanni Santostasi: PulsarMon
Giovanni reported by e-mail that he has been able to obtain a strain h and
produce a FOM plot, using an external matlab script. He hasn't tried sending
plots to the DMT viewer yet.
- Rauha Rahkola: SuspensionMon
Ray reported no recent work on SuspensionMon, but mentioned that the Oregon
group is working on a new DMT monitor for the output of cosmic ray scintillators,
based on glitchmon.
- Natalia Zotov: PTMon
No report.
- John Zweizig: LSCMon et al
John is working on support for multiple triggers/stride in PSLMon and to
take advantage of the new trigger format (see above).
- At the calibration team's request, Peter Saulson has recruited
a review committee for future calibration releases. Serving will
be Albert Lazzarini, David Shoemaker, and Rai Weiss. The scope of
the committee's charge is under discussion. The calibration team
is happy with this heavyweight committee and looks forward to
working with them.
- Next detchar telecon: Friday July 23
at 1:30 p.m. EDT (S2/S3 investigations)