Minutes of Detector Characterization Teleconference
(June 1, 2001)
Present:
Caltech: Drever, Lazzarini,
Mours, Shawhan, Zweizig
Dublin: Ottewill
LHO: Ito, Landry, Savage,
Schofield, Sigg, Whitcomb
Michigan: Riles
Syracuse: Penn, Saulson
Preliminary E3/E4 Investigation Reports:
-
E4 Lock Losses (KR reporting)
(See preliminary report web page
for details and plots)
-
A striking correlation was documented between ability to acquire / retain
lock and seismic noise in the 1-3 Hz band, suggesting that many lock losses
were seismic in origin. But no smoking gun (e.g., coil saturation) was
observed in a sample of losses thought to have been seismic.
-
There were other classes of lock loss observed, however. One class was
coil saturation after tidal drift (expected and somewhat gratifying to
see).
-
Another class appeared to be a servo instability with exponential growth
in coil current of 5 ms time scale. The source is not yet understood.
-
Another class was gradual growth of a large, narrow 686 Hz oscillation
in the arm control signals that caused end-mass coil saturation. The narrowness
suggests a mechanical or electronic resonance, not servo instability. Several
BSC down tube resonances are expected in this frequency range. In an e-mail
comment after the meeting, Rana reported further study of this oscillation
that suggests that it is not a servo instability: a harmonic at
twice 686 is seen, and a change in servo gain did not change the frequency
of the oscillation.
-
E4 Frequency Noise (R. Adhikari reporting)
(See preliminary
report web page for details) His automobile's ignition acquisition
failure prevented Rana from attending the telecon. He will report on the
results posted here at the next detector characterization telecon. Highlights
include:
-
Measurement of common mode rejection of factor of ~20
-
Evidence of shadow sensor noise dominating the differential mode noise
floor
-
Preliminary investigation of seismic noise coupling into the mode cleaner
length noise
-
E3/E4 Seismic Noise (D. Greenwood reporting by e-mail shortly after
meeting)
Dick is investigating a phenomenon common to the E3 and E4 periods:
a persistent 10 Hz seismic signal measured at the EX (but not EY) station
that rose during the day and fell at night. It was also detected by the
LLO external seismometer array during E3.
In addition, the EY seismometers measured alarming ~30 Hz signals that
sometimes contributed > 50% of total seismic activity in the 1-50 Hz range.
Finer resolution revealed two sharp peaks at 29.3 and 29.7. More will be
reported at the next telecon.
E5 Preparations:
-
Likely dates and interferometer configurations - S. Whitcomb
As announced earlier, E5 is tentatively planned for the first weekend
in August (Fri Aug 3 to Mon Aug 6). It is strongly hoped that the LHO 2K
will be in recycling configuration and the LLO 4K in recombination. It
is also possible that the LHO 4K will be in recombination. If, however,
the LHO 2K is not stably recycled by early August, the run may be delayed
three weeks to Aug 24-27. This will be the first engineering run data taken
with a recycled interferometer.
-
Investigations and E2-E5 reports - KR
-
Many E2 final reports are still missing. A web
page contains links to results and final reports of the E2 investigation
teams. Only those reports that are unambiguously final are linked to it
at the moment.
-
E3/E4 investigators are asked to turn in final written reports no later
than the August LSC meeting and to make presentations at that meeting.
All teams should give preliminary reports at a detector characterization
telecon before the LSC meeting.
-
It may be appropriate to do some tweaking of the investigation organization
for E5, based on E3/E4 experience. Since E5 is the direct precursor to
the E6 upper limits run, we may want to refocus some of the investigations
to address particular issues the upper limits groups are worrying about.
Suggestions for modifying the present E4
lineup (which is an excellent starting point) are requested. An e-mail
will be sent to the entire detector characterization soliciting suggestions
and volunteers.
-
DMT software developers are strongly urged to get ready for E5. Several
DMT monitors proved quite useful in E4, but more are needed, and many of
the existing monitors need improved online documentation and information
display.
Status reports on DMT infrastructure development:
-
General - J. Zweizig
E4 preparations and the run itself took up about two weeks of John's
efforts. At LLO a 2-cpu Pentium III (1 GHz) linux box has been brought
up to run online DMT in parallel with delaronde. Despite quoted speed differences,
John finds that one linux cpu performs at about the same effective speed
as one sun E-450 cpu. The cpu power per dollar is better, however, with
the linux machine. John notes that byte swapping contributes to but doesn't
appear to dominate the linux time per job. The jury is still out on whether
large improvements in the linux performance can be achieved with software
changes.
Motivated by an interesting microphone pickup in E4 (see lock loss
report above), John got a program running on the LLO suns that converts
any data channel to audio for the listener's pleasure.
John has also been documenting and upgrading parts of the DMT infrastructure.
The current development version now uses root 3.01.03 (old was 2.25.03)
and uses a new gcc compiler version too, which is reputed to give better
code performance.
John and Daniel have been preparing for next week's DMT camp for which
seven persons have signed up. Daniel has written up lectures that will
be posted on the DMT home page (and linked to the DetChar bulletin board).
-
Histogramming - M. Ito
Masahiro is creating a new histogram class within the DMT which will
be used to transport information from the DMT background processes to the
DMT viewer program (monserver/monclient infrastructure). A header file
has been given to John and Daniel for comment. On the DMT viewer end, the
DMT histogram class will be converted to a root histogram class so that
root's powerful display code can be exploited. Masahiro has looked at what
is required to create a histogram gui and believes he can have a first
version of the entire package ready for E5.
Status reports on performance characterization priority 1 tasks:
-
Line noise:
-
B. Allen, A. Ottewill (Adrian reporting)
Adrian has been working mainly on the multi-taper line removal code
(already in the DMT) for the LDAS data conditioning API and will be attending
the LDAS camp next week. He has worked a bit on getting the dmt viewer
running on his computer at Dublin to see plots generated by background
monitors running at Hanford. He succeeded in getting the viewer program
running, but had trouble seeing the plots, a misunderstanding which was
cleared up during the telecon. The line monitor code will be the first
to be integrated into the dmt viewer, followed by the correlations monitor
code.
-
S. Klimenko (not present)
-
A. Sintes (not present due to conflicting GEO meeting)
-
Characterizing seismic noise:
-
Inter-channel correlations:
-
B. Allen, A. Ottewill (see above)
-
Bilinear cross-couplings
-
S. Penn
Steve's monitor is now runnable on the online suns (developed on his
linux laptop). The executable and source code can be found in the spenn
directory on delaronde. Called xbispec, the monitor allows real-time display
of either bispectra for a single channel or cross-spectra for 2-3 channels.
Steve is working on a version to display normalized bi-coherence. He will
give instructions to interested commissioners and operators on using the
program (KR offered to make a link on the DetChar bulletin board page for
any web instructions). John suggested this program, which is cpu-limited
at high requested bandwidths, would be suitable for running
on the LLO linux. Steve reported being unable to ssh directly to delaronde
from Syracuse. KR believed that to be a deliberate security decision, but
John and Daniel thought that decision to be inappropriate and volunteered
to look into it.
-
Operational state:
-
D. Chin, K. Riles (KR reporting)
Dave has added several new operational state conditions, designed to
flag simple global changes to channel behavior, but has spent more of his
time on integrating the lock loss monitor into the dmt viewer. He has a
private version working now that allows display of lock history in the
viewer and will be adding some bells and whistles in the next few weeks.
New versions of the lock loss monitor and of a dmt-viewer-supporting servo
monitor will be ready well before E5 and exercised on E4 data for shakedown.
Status reports on performance characterization priority 2 tasks:
-
Bandlimited RMS
-
Time / Frequency plots
-
S. Mohanty (not present due to conflicting GEO meeting)
E-mail report: not much progress since last month, working on test
of prototype background monitor for transients. Search algorithm applied
recently to E3 data and report in progress.
-
J. Sylvestre (not present)
-
Non-Gaussian noise
-
S. Penn (nothing new reported)
-
S. Finn, G. Gonzalez, P. Sutton (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:
-
J. Sylvestre, R. Weiss (not present)
Status reports on transient analysis priority 2 tasks:
-
Impulse recognition:
-
M. Ito
Masahiro has focussed solely on histogramming work lately (see above).
KR noted, however, that Masahiro's glitchmon monitor had generated
some very interesting triggers in the meta-database during E4 (see lock
loss report above) and encouraged others to look (with the guild program)
at those triggers.
-
Magnetic field and seismic transients
-
R. Frey, R. Rahkola (not present due to Ph.D. candidacy oral exam,
which Rauha passed )
Not much progress on magnetic field monitoring, but Rauha has developed
a seismic alarm for control room operators, which is almost working now.
A.O.B.
-
Next detector characterization teleconference: Friday July 13 at 11:00
a.m. EDT (8:00 PDT).
(NOTE: This will be the second Friday in July, not the first.
KR will be travelling on the 6th.)