Minutes of Detector Characterization Teleconference
(May 21, 2004)
Present:
Caltech: Albert Lazzarini, Peter Shawhan, John
Zweizig
LHO: Mike Landry, Vern
Sandberg
LLO: Brian O'Reilly, Giovanni Santostasi
Louisiana Tech: Natalia Zotov
Loyola: Marc Cenac, John Whelan
Michigan: Keith Riles
MIT: Erik Katsavounidis
Oregon: Ray Frey, Robert
Schofield
UWM: Steve Fairhurst, Xavier Siemens
Preparing for June LSC meeting at Tufts
-
Although the upcoming 2-day LSC meeting at Tufts is devoted mostly to status
of astrophysical analysis, there will be one session late Sunday afternoon
on particular detector characterization issues common to all four analysis
groups. Tentative topics for that session include calibrations, low-byte
corruption, and PSD estimates. Other suggestions are welcome.
S2 / S3 Investigation Reports
-
Calibration:
-
Update on S3 calibrations (Mike Landry, Brian O'Reilly)
Mike confirmed that the discrepancy Xavi reported two weeks ago in
derived S3 L1 alpha/beta values between the frequency domain and time domain
methods is real. The frequency domain numbers were based on a matlab model
script which had a typo. The effect requires a modification of the response
and sensing functions, and unfortunately, the actual inspiral ranges for
L1 are about 30% worse than previously thought. The calibration team is
working to understand in detail how the error crept in, without being picked
up by the matlab-independent autocalibrator. They expect to have a version
1.2 calibration released in time for the June LSC meeting.
-
Update on time domain calibration (Xavier
Siemens)
Xavi showed the good agreement now obtained between frequency and time
domain methods for alpha/beta, with the frequency domain model parameter
fixed. There remains a discrepancy in beta at the 5% level, which may (or
may not) arise from measurement error at the reference time interval SenseMon
uses for its baseline.
-
Update
on hardware injections (Steve Fairhurst)
In trying to understand the large variance seen in effective inspiral
distances for hardware-injected inspiral signals, Steve has tried varying
alpha for injected chirps in software, applying different amplitudes and
frequencies for a sinusoidal variation. He has tested the fidelity of this
simple model by looking at not only the induced variance in effective distance,
but also the chirp chi**2. He concludes that a 1-second periodicity of
a relative variation of 5% agrees qualitatively with what is observed in
hardware injections.
-
Data quality - Investigations: (
S2 , S3
) Segment repository (S2 , S3
) (John Zweizig, Keith Riles)
John gave an updated
report on the low-order-byte corruption problem. For floating point
channels (like AS_Q!), one nominally expects an effect at the 10**-5 level
from corruption of the last 8 bits of the 32-bit word. John has tried simulating
the kinds of corruptions (seen unambiguously in otherwise quiet integer
channels), in which short clusters of spikes separated by 16 bytes
appear. The bottom line is that the effect on the AS_Q spectrum is indeed
at the 10**-5 level. Albert wondered whether check sums could be implemented
in the DAQ system to look for byte corruption. Vern mentioned that he has
in mind a more demanding test of injecting known patterns into reflective
memory (and perhaps further upstream), then checking the DAQ readout. Rolf
Bork and Dave Barker are natural candidates to work on this diagnostic,
but both are swamped with other work at the moment.
John has also been working on a new PSLMon version with line removal
(frequency domain) from the band-limited RMS measurements and has now produced
new minute trends for H1 and L1. He sees some improvement (more relative
variation) than before, thanks to removal of dominant lines. He plans similar
changes (but with time domain notch filters) in the PSLMon glitch finding.
-
Related study (Albert Lazzarini)
Albert carried out a similar low-order-byte corruption simulation study
in which he zeroed out the last 8 bits, 12 bits, 16 bits, etc. of AS_Q
data samples (using a matlab script). He too sees an effect no worse than
10**-5 for 8-bit truncation.Studies are ongoing in the stochastic group
to understand the effects on the Omega integral. Preliminary findings are
a larger effect than naively expected from above, but still safely negligible.
-
Timing (Szabi Marka, Daniel Sigg)
No update.
-
Glitches / Burst/Inspiral search group vetoes (Laura Cadonati, Erik Katsavounidis)
Erik reported that the S2 glitch studies are now essentially complete,
but that Robert and Masahiro did carry out a recent PEM study motivated
by the four low-threshold WaveBurst candidates. The glitch team is now
moving on to S3 studies, looking first at triggers stored in the database
during online DMT running and focussing on playground data. The team is
also paying more attention now to signal detection criteria
.
-
Interchannel
correlations (Nelson Christensen)
Nelson couldn't make the telecon, but provided a new table summarizing
microphone/AS_Q correlations seen in H1. The table is meant primarily for
use by the pulsar group. More tables will be provided soon for H2 and L1,
and summaries for other PEM channels, e.g., accelerometers produced.
Albert reminded everyone that the stochastic group has 2-month-long S2
and S3 coherences, where the H1/H2 coherences look much improved after
the acoustic mitigation. KR wondered if the stochastic group sees a 0.25
Hz comb in coherence that appears in 2-month-long averages of 30-minute
FFT spectral estimates (seen independently by Xavi and Vladimir Dergachev
using different PSD programs). Albert wasn't aware of such a comb. KR will
report at a future telecon on the effect seen in the PSD's.
-
Bilinear couplings (Steve Penn)
No update.
-
Correlated inter-site environmental transients & local environmental
disturbances (Robert Schofield)
Robert reported that the intersite correlation code has now been run over
the entire S3 data set and turned up some 3-sigma excess glitch coincidences
between magnetometer and voltage monitor channels at the two sites. Vern
wondered if these were seen in October, when solar flares were at an unusally
high maximum. Robert will go back and check the times.
Robert also gave an update on several environmental issues he discussed
at the April 23 telecon:
-
Dust glitches seem important only at the dark port table. Work is underway
to quiet down the dust monitors that give magnetic glitches at motor startup
(every minute or so). A prototype purged enclosure has also been built
for testing.
-
The pulsing of heater coils for the ductwork remains a problem. The pulsing
not only introduces sidebands on 60 Hz harmonics, but may eventually dominate
the 60 Hz line itself. Unfortunately, the schemes looked at so far for
reducing the pulsing effects (staged heating of coils, replacement with
coils of lower inductance, variac control of coil current, cold water with
continuous heating) are all expensive. A decision has not been reached
yet on how to proceed, but a low-inductance coil has been ordered for testing.
-
Must decide soon whether to build acoustic enclosures around electronics
racks in place or wait until after they are moved outside the LVEA.
-
The source of dewar glitches is now suspected to be stick-slip of the internal
container against bumpers attached to the outer container that are meant
to limit excursions during transport. The bumpers are not easily accessible,
or they would simply be removed. Right now one dewar has been tilted slightly
away from the suspect bumper, but more glitches have been seen since then.
Albert wondered if wrapping the dewar in insulation would ameliorate the
temperature excursions of the outer container relative to the inner. Robert
has looked into that possibility, but it's quite expensive. He mentioned
that while large dewar glitches can lead to lock loss, smaller gltiches
are a worry too. Nelson has found a small glitch that led to inspiral triggers.
Albert also wondered if something like a kinematic mount with built-in
"give" could be used to support the dewars instead of the present bolts
on concrete. Robert will look into it.
A.O.B.
-
Next detchar telecon: Friday June 11 at 1:30 p.m. EDT (DMT
software)