Minutes of Detector Characterization Teleconference
(December 1, 2000)
Present:
Caltech: Barish, Drever,
Lazzarini, Marka, Mours, Shawhan, Yamamoto, Zweizig
Carleton: Christensen
LHO: Gustafson, Ito,
Landry, Ottaway, Raab, Savage, Sigg, Weiss, Whitcomb
Loyola: McHugh
Michigan: Riles
MIT: Adhikari, Sylvestre
Oregon: Frey, Rakhola
Syracuse: Penn
Preliminary E2 Investigation Reports:
-
Lock Losses (K. Riles reporting)
(See preliminary report web page
for details) Periodic lock losses in E2 were expected because the coarse
tidal correction actuation was disabled. Eventually, tidal drift would
exceed the dynamic range of one of the end mass controllers. Premature
lock loss was observed, however, due apparently to saturation in the end-mass
controller in the X arm. Evidence of saturation was seen indirectly in
orientation channels and transmitted light through the X arm, was seen
more directly in the corresponding coil channels and with an oscilloscope
plugged into coil monitors on the X controller. Steve Penn suggested looking
at the side coil current, since he found a factor-of-two discrepancy from
nominal actuation gain. Under study is determining which longitudinal servo
channel was "giving way" first when lock was lost.
-
Timing Calibration (S. Marka reporting)
(See preliminary
report web page for details) This team checked the consistency of timing
signals throughout the E2 run, using GPS clock signals from different locations.
There were a couple of glitches caused by CDS disks filing up which is
understood. In addition, at one point midway through the run, a square-wave
test signal became mis-aligned by one sampling count (1/16K sec) and remained
that way. This is not yet understood. Measurements of propagation delays
for comparison with expectation are underway.
-
Calibration Studies (M. Landry reporting)
(See preliminary
report web page for details) There were two aspects of the calibration
studies: measuring transfer functions (including stability with time) and
calibration end-test-mass motion. Generally, the calibrations were found
to be stable within one day. For each arm, transfer functions for input
and end masses were quite similar, but between the arms larger differences
were seen. The shapes of transfer functions were fitted to analytic curves
with zeroes and poles. In most cases, reasonable agreement is seen with
expectation. It was discovered after E2 that an earlier calibration run
in October had led to an actuation output matrix with an incorrect factor
of two between the two arms.
-
Impulses / Transients (R. Frey reporting)
(See preliminary report for details)
Examples of transients observed during E2 were shown and advice solicited
for further investigation. It was clear that seismic noise was causing
transients in the antisymmetric port in-phase signal (AS_I) and the mode
cleaner feedback (MC_F), but other transients in those channels were just
as clearly non-seismic. It was verified directly (by driving over a speedbump
and by jumping up and down!) that it's easy to see ground disturbances
in those channels.
-
Frequency Noise (R. Adhikari reporting)
Although a preliminary report is not yet ready, Rana wanted to alert
everyone to a discovery made after E2 of a couple of bugs in the microprocessor
software that determines the Q and I phases of the longitudinal error signals.
In particular, the common-mode and differential-mode error signals have
a significant cross coupling. (See later
e-mail on this subject from Stan Whitcomb). More information will be
posted on the Web when the problem is fully sorted out.
-
Line Noise (S. Klimenko reporting)
(See preliminary report web
page for details; click on line noise, then on LNI prelim results)
Although many lines observed in the E2 data were identifiable with known
mechanical resonances (c.f. LIGO-T000020-00 by M. Landry & D. Ottaway),
some were not. There were additional lines near 17 Hz (which may be vertical
roll motion of test masses not included in the LIGO document above). "Mobile"
lines around 3 kHz were seen, which are probably due to daq electronics
(perhaps switching power supplies). The injected calibration lines were
clearly seen, and drift observed in their strengths. A line at 206.7 Hz
was found to be very strongly correlated in strength with the calibration
lines, probably because all were sensitive to arm alignment which tended
to drift during the course of a locked stretch.
-
One added note: Szabi Marka strongly urged investigating teams to make
use of the web directories available on Hanford's blue computer. These
directories are world-writeable from several Hanford computers. (See later
e-mail from KR for more information.)
Data Monitor Tool Status (John Zweizig):
-
John added several contributed monitors to the DMT package just before
or during the E2 run, which were exercised during the run.
-
Since E2 he has been preparing for a rerun of all the data, which has involved
some fixes to infrastructure to ensure that all of the running monitors
are properly synchronized in using shared memory. Requests for reruns to
generate meta-database triggers and trend data have been made for eight
different monitors. It was decided to purge the existing E2 entries in
the meta-database, given the planned regeneration of the complete set.
Steve Penn requested that the rerun be delayed a few more days so that
his non-gaussianity monitor could be included too.
Status reports on performance characterization priority 1 tasks:
-
Line noise:
-
B. Allen, A. Ottewill (not present)
-
S. Klimenko
The code used in the E2 line noise investigation was run in foreground
as root macros. A background monitor based on the macros is in development
and will be ready for use in the control room in two months. The present
code requires one cpu on sand or stone to monitor 50 typical lines in real
time. Higher-frequency lines are more cpu-intensive.
-
A. Sintes (not present - attending GEO meeting)
-
Characterizing seismic noise:
-
Inter-channel correlations:
-
B. Allen, A. Ottewill (not present)
-
Bilinear cross-couplings
-
S. Penn
Steve's code now runs but slowly. He believes he needs a speedup of
roughly a factor of ten to make it feasible for real-time analysis.
-
Operational state:
-
D. Chin, R. Gustafson, K. Riles (KR reporting)
Dave is doing a complete rewrite of the OSC (operational state conditions)
code to make it more flexible, useful and efficient for handling hundreds
of conditions. He hopes to have the code ready by the end of December and
will exercise it thoroughly during a 3-week January visit to Hanford. The
E2 run made flaws in the existing code all too apparent. Two background
monitors, based on the OSC code, were written for and run during part of
E2 for generating meta-database trigers (lock loss and servo instability).
They worked, but had to be run "by hand" in private areas and didn't provide
as much information as would have been useful. Once the OSC rewrite is
complete, these two monitors will also be rewritten and commissioned.
Status reports on performance characterization priority 2 tasks:
-
Bandlimited RMS
-
Time / Frequency plots
-
S. Mohanty (not present - attending GEO meeting, reported by e-mail):
I have made some progress with the transient detector code since last
time. Writing of the C++ classes to emulate the required MATLAB functions
is underway. I will not try to forecast a date by which this will be finished
but the code writing remaining should, in itself, not take more than 2
weeks.
-
J. Sylvestre
-
Non-Gaussian noise
-
S. Finn, G. Gonzalez (not present)
Status reports on transient analysis priority 1 tasks:
-
Power spectral transients:
-
Servo instability:
D. Chin, R. Gustafson, K. Riles (see above)
-
Event catalog:
Status reports on transient analysis priority 2 tasks:
-
Flickering optical modes:
-
Transient detection using adaptive denoising methods
-
E. Chassande-Mottin (not present)
-
Impulse recognition:
-
M. Ito
The impulse recognition code was not running during E2, but will be
run during the data replay next week. The code will monitor roughly 200
channels. No meta-database triggers will be generated yet.
-
Magnetic field transients
-
R. Frey, R. Rahkola (Rauha reporting)
One magnetometer channel was available during E2. It was monitored
by a modified version of John Zweizig's PSL monitor. It has been decided
to use that monitor as a template for further development instead of earlier
code written by Evan Mauceli before his departure.
A.O.B.
-
Peter Shawhan announced the availability of LIGOtools, a software package
meant to give better off-site access to the data and to tools used routinely
at the sites. Much of the software can be downloaded to LSC institutes
with Solaris machines. He noted, however, that mimicking the data viewer
and diagnostic test tool locally will require setting up a file structure
like that used on the fortress machine at Hanford. More information is
available here.
-
KR noted the higher-than-usual no-show rate at this teleconference and
resolved to nag the absentees.
-
Next detector characterization teleconference: Friday January 5 at 11:00
a.m. EDT (8:00 PDT).