Minutes of Detector Characterization Teleconference
(February 7, 2003)
Present:
Caltech: D'Ambrosio, Marka, Ugolini,
Whitcomb,
Zweizig
Carleton: Christensen
Chicago: Shawhan
Florida: Klimenko
La. Tech: Zotov
LHO: Ito, Landry, Matone, Raab, Rahkola, Schofield,
Sigg
LLO: Chickarmane, O'Reilly, Taniwaki
LSU: Giaime
Michigan: Riles
MIT: Fritschel
Oregon: Frey, Leonor
Syracuse: Saulson
S2 Preparations (Friday Feb 14 08:00 PST (10:00 CST)
to Monday Apr 14 08:00 PST (10:00 CST)
-
IFO status (S. Whitcomb)
All three IFO's are working well in terms of sensitivity (100 kpc range
or better from SenseMon) and in terms of reliability. (but see P. Saulson's
comments on L1 glitches below, though.)
-
Run plans - running policies, meetings, etc. (S. Whitcomb)
-
In the week preceding the run there will be calibrations and hardware signal
injections, along with routine running of the IFO's as if in science mode.
The hope is to "coast" into a smooth S2 startup. The H2 IFO is quite a
bit more sensitive and less fussy than it was in the E9 run. It is still
plagued at times, though, by sudden increases in broadband noise (100-1000
Hz), also seen in E9. Mike mentioned this noise seems correlated with acoustic
disturbances, and one optical table is suspected to be the culprit. Robert
mentioned that new video monitors trained on photodiodes should allow monitoring
of off-center beams.
-
The calibration team (see below) will carry out a full-up calibration procedure
at the beginning of the run, once in the middle, and once more at the end.
There will also be daily autocalibration (swept sines).
-
Aside from a comprehensive suite of pre-run injections, there will be 12
half-hour blocks devoted during the run to signal injections (see below).
-
It is hoped there will be approximately 24/7 running of H1 and H2, but
it is highly likely that L1 will not be lockable on weekdays between 6
a.m. and 6 p.m..
-
At the start of the run there will be daily inter-site telecons in the
late afternoon (Stan will announce a time soon), but eventually the schedule
will probably evolve to 3 telecons / week.
-
The emphasis at the start of S2 will be achieving maximum sensitivity (vs
maximum livetime), in part to give the upper limits groups some really
good data (most analyses benefit more from sensitivity than from integrated
time) and in part to learn what price in livetime one pays for high sensitivity.
A tentative guidelines would be to tweak alignments (or break lock) if
inspiral range falls by more than about 20%. After a couple of weeks, the
emphasis may shift to maximizing livetime. There may be a couple of days
in the middle of the run devoted to realignment of optical tables to correct
for long-term drifts. Fred expressed a worry that fiddling with tables
runs the risk of making thing worse instead of better.
-
Maintaining good coincidence among the three IFOS' and GEO &
TAMA may require revisiting livetime/sensitivity tradeoffs frequently.
Fred mentioned that TAMA has arranged to stop all construction work nearby
on its campus in order to make the best of coincident S2 running. KR will
try to find out from Albert (official LIGO/TAMA liaison) whether there
is a way to see in near-real-time via the Web whether TAMA is locked.
-
Full (5 x 2) wave front sensing is not likely to be ready for any IFO for
S2, because of stability problems, although there has been significant
commissioning progress recently.
-
P. Saulson brought up the glitchiness seen in the L1 IFO during E9. High
glitch rates were observed during the run in DMT glitch monitors, and that
glitchiness is reflected in increased E9 trigger rates from burst DSO's
(relative to H1 and H2). A characteristic pattern was observed of sudden
high glitchiness occurring about every 50-55 seconds. The source is not
known, but there is some hope that the problem has gone away on its own
after recent hardware changes. Patrick Sutton did not see the glitches
in data taken two nights ago. Clearly we need to be attentive to recurrence
in the next week. If the glitches reappear in S2, then tracking down their
source should become a high scimon priority.
-
Declaring common and science modes (all) - KR mentioned an ongoing lal-discuss
controversy over the exact definition of "science mode data" and
the use of SenseMon calibration drift corrections that appear to be unreliable
in the first minute of some E9 lock stretches, presumably because of excess
noise from recently changed filters that interfere with measuring injected
calibration line strength. Some discussion ensued on whether to ask operators
to hold off on declaring science mode until it's clear that things have
settled down. Stan felt strongly that there would be too much ambiguity
in such a procedure. One should declare science mode as soon as all nominal
control parameters have been set in the rampup to full common mode feedback.
Downstream analysts will be encouraged either to veto blindly the first
minute or two of locked stretches or to use archived info such as SenseMon
coefficients to determine which data to veto.
-
Scimon shifts (K. Riles)
Shift block schedules have been posted
in which all expert scimon shifts are filled, along with a subset of trainee
shifts. Shift contact persons from the groups are responsible for filling
in the names of the shift takers.
-
Scimon procedures (K. Riles)
The scimon worksheets
used in E9 provide the base for the S2 worksheets. Nearly every worksheet
has been updated in the M2 (Dec 2002) or E9 (Jan 2003) runs. A couple more
updates have been promised before S2 begins. KR will upgrade the lock statistics
worksheet to include the daily posting of more than just lock statistics
and lock histories. 24-hour trend versions of the present DMT viewer figure
of merit displays will be put together as standard data viewer parameter
files. Volunteers to help in this effort are welcome.
-
Investigation teams
- E9 Reports and S2 plans
-
Calibration stability (M. Landry, B. O'Reilly)
There is now a convergence of previously divergent calibration procedures
/ data handling at the two sites. The LHO auto calibrator now works at
LLO too, and the modelling used in LLO data handling is now used at LHO
too. The auto calibration was tested successfully in E9 (although maintaining
coherence throughout the swept sine proved troublesome on L1), and tests
of the modelling are underway. It is expected that preliminary calibrations
will be carried out this weekend, allowing Patrick Sutton to test SenseMon
thoroughly on Monday. Mike and Patrick are trying to track down a small
residual discrepancy between the posted calibrated power spectra and what
SenseMon finds. Calibration lines are now injected routinely during commissioning
at LHO; they will be next week at LLO.
-
Violin modes (S. Klimenko)
Nothing to report on violin modes; focussed on improved tracking of
calibration lines (see below)
-
Steady-state correlations (N. Christensen)
Adrian Ottewill spent time at LHO during E9 where he made improvements
to the CorrMon DMT monitor for tracking inter-channel correlations.
Although he initially enabled triggers, he found that the rates were too
high and had to be disabled. He is now going through the E9 data to determine
reasonable thresholds. A few artifacts will receive attention in S2, including
correlations between AS_Q and optical lever channels. The S2 correlation
investigations will be spearheaded at LLO by Nelson and at LHO by Dennis
Ugolini.
-
Glitches (J. Giaime) - Had to leave early - from discussion above, it's
likely that the glitches team will have some work to do in S2.
-
Bilinear couplings (V. Chickarmane)
Vijay now has an offline matlab-based program for zeroing in on upconversion
of calibration lines. This was used in E9 to set reasonable ampliltudes
on injected lines. But it also provides a monitor for scimons to check
from time to time. Vijay agreed to write up brief scimon instructions on
checking the upconversion monitor, since it's not DMT-based.
-
Correlated inter-site environmental transients (R. Schofield)
Robert is making sure that the necessary PEM hardware is in place,
primarily coil magnetometers, voltage monitors, and RF monitors. It looks
likely that the RF monitor at LLO will not be back in operation by the
start of S2. Robert will make a short trip to LLO next week.
-
Local environmental disturbances (R. Schofield)
There will be a series of PEM disturbance injections at both sites
in the week leading up to S2. Robert would like to recruit scimons to help
track down noise sources. KR suggested updating the scimon line-tracking
worksheet to include a link to Robert's document of known lines and updating
the document itself to include lines that have been seen but whose source
is not yet known, so that scimons have guidance in where to focus efforts.
Robert agreed to do so.
-
Timing
precision (S. Marka, D. Sigg)
During E9 timing was quite stable for H1 and H2. There were three apparent
jumps for L1, but it's not yet clear whether those were real. Since E9
there has been a major revamping of microprocessor code to keep precise
track of timing, including how long it takes to process data. The code
is now smart enough to eliminate processing on a given cycle if in danger
of falling behind. There are many more diagnostics and automated controls
in place, which should make the front-end system more robust. Stan cautioned
people not to hook up test points that needlessly place a drag on the processors.
This can be done inadvertently by looking at test points with the data
viewer.
-
Data quality
(J. Zweizig)
John has started looking at the many E9 trends produced by the DataQual
monitor. In general, things look better than in S1. He sees the same glitchiness
in L1 seen by others, but has no insight into their source. Silent alarms
have been installed on the band-limited RMS part of DataQual, but thresholds
need re-tuning. These alarms might prove useful in heading off an unintended
lock loss.
-
Data access (P. Shawhan)
Peter will set up a web page devoted to S2 data access issues, including
pointers to reduced data sets. The page will hang from the main S2 web
page.
-
Data set reduction (I. Leonor)
Isabel has circulated a proposed
list of RDS channels for S2, incorporating various recent requests
from upper limits groups and investigation teams. More suggestions will
be solicited in the coming week. These channels will be automatically
written by LDAS scripts. The signal injection team will separately be writing
an RDS during injection times before the run begins, using the DMT frame
writer. Mike suggested testing that early, since he has had trouble with
it recently.
-
Signal injection
(S. Marka, P. Shawhan)
A detailed signal injection plan with extensive pre-run injections
and more limited in-run injections, as noted by Stan above. A web page
will provide detailed information on the injections, as was the case for
S1. Some testing has already begun, using burst injections. For reasons
not yet understood, not all of the injections were found by the offline
DSO's. The team would like to start Monday night with reasonable calibration
info in hand. Peter is working on better automation of the procedure
for the 12 half-hour injection periods during the run.
DMT Software
-
Infrastructure status (J. Zweizig)
John will recompile all DMT code tonight with a new (but not final)
version of FrameCPP. He does not plan to wait to use the final S2 version
of FrameCPP because it is not expected to be ready before next Wednesday.
All final DMT code changes for running under the process manager must be
in John's hands this weekend. Only serious bug fixes will be allowed, starting
next week.
-
New or developing monitors:
-
LineMon (S. Klimenko)
During E9 there were jumps seen (link
one link
two) in the phases of the calibration lines as seen in the injection
and AS_Q channels. It appears those jumps were due to lost frames in the
DMT data stream (memory buffer problem). To get around this problem and
address a concern by the calibration team, Sergey has rewritten and is
now testing LineMon. The program now allows multiple channels to be monitored,
providing a trend file for the phase difference between the excitation
and AS_Q channels.
-
SenseMonitor (P. Sutton - out sick)
-
ShapeMon (S. Marka)
ShapeMon has been running smoothly under the process manager since
before E9. It has proven quite useful in identifying channel hopping, not
to mention earthquakes.
-
SuspensionMon (R. Rahkola)
Rauha is working on a PSLMon-based backup for SuspensionMon in case
the primary program (B. Stubbs) cannot be debugged in time. A trial config
file exists, based on a threshold of 3 x nominal RMS on large optic and
recycling cavity sensors. Rauha will be testing and tuning before S2 begins.
No coding changes are required for this stripped-down backup program, only
definition of the PSLMon config file.
-
SpectrumArchiver (K. Riles for T. Olson)
The monitor ran successfully at Hanford during E9 and will continue
during S2. Disk space has now been allotted at LLO where the monitor will
also run during S2. For the moment, the RDS channel list has been used
for guidance in choosing channels, and fixed-bin spectra have been used.
In the long term, a more refined channel list and variable-bin spectra
will be used. It is still cumbersome to retrieve the archived spectra from
the DTT import window. Tim and students are working on a better interface
for integrating into the DTT.
S1 Report:
Steady-state
correlations (N. Christensen) - deferred until next detector characteriation
telecon
A.O.B.
Next (regular) detchar telecon: March 7, 2003