Minutes of E5 / Detector Characterization Teleconference
(July 27, 2001)
Present:
Caltech: Shawhan, Yamamoto,
Zweizig
Florida: Klimenko
LHO: Landry, Raab, Radkins,
Rahkola, Savage, Sigg
LLO: Adhikari, Coles,
Daw
Michigan: Gustafson,
Riles
MIT: Sylvestre
Oregon: Schofield
Portland airport: Mavalvala
Rocky Mountain NP: Whitcomb
Syracuse: Penn
E5 Preparations:
-
Introduction - KR
-
The E5 run will start at noon CDT (10:00 a.m. PDT) Friday August 3 at noon
and end 72 hours later on Monday the 6th. Daniel Sigg and Szabi Marka will
be the run coordinators at Hanford and Livingston, respectively.
-
This meeting is meant in part to review where we stand on E5 preparations
and to find out if there are any special requests from the investigation
teams.
-
It's expected that the Hanford 2K will be running in recycled mode, but
the Hanford and Livingston 4K's will not be operational (see Stan's report
below).
-
Running will be in 24-hour mode at Hanford with one operator and two scientists
on shift. At Livingston there will be one operator and one scientist on
the day and evening shifts.
-
Today is the deadline for getting new or revised DMT monitors to John for
incorporation into the standard lineup for E5 (see John's report below
on what has been received so far).
-
Szabi will be setting up the E5 web site very soon, modelled closely on
what we did for E4. The shifts and investigation lists are the only files
present in the usual place on blue at the moment.
-
Interferometer running configurations - S. Whitcomb
-
After some marathon commissioning sessions, the 2K IFO is locking robustly
for periods of up to an hour in full recycled mode with 1 Watt entering
the mode cleaner and about 0.6 W hitting the recycling mirror. A recycling
gain as high as 24 has been measured.
-
The Livingston 4K IFO is pumping down and there is a possibility that the
mode cleaner will be operating during E5. There is not much chances of
the Hanford 4K having any cavities locked.
-
Because of a photodiode shortage, not all the light at the 2K antisymmetric
port is being used, which means the shot noise limitation is similar to
what was seen in February. Nergis mentioned that a new photodiode might
be available for installation next week before the run, which would improve
performance.
-
The gate valves of the 2K will be opened up 1 day in advance of E5 to establish
stable conditions for the run. (Lately they have been kept closed when
not commissioning because of residual water vapor in the corner modules
after a recent pumpdown.)
-
There will be a few-hour run in single-arm mode at the end of E5 to permit
some comparisons with the E2E model in a relatively simple configuration
and with old single-arm data at lower power.
-
Run planning - D. Sigg
-
The data-taking configuration at Hanford will be as for previous runs,
with LDAS recording tapes for the archive and the fortress machine writing
a reduced data set (see below). There will be 6 MB/second recorded to tape.
-
Most of the old engineering run reduced data on the local raid disks
will have to be deleted to make room for the E5 data set.
-
The DMT monitors at Hanford will be running on sand and stone (see John's
report below).
-
Scientific monitoring shifts - KR
-
The current list of shifts is nearly complete,
but
several trainee slots remain to be filled at Hanford.
-
As for E3/E4, shifts will be synchronized at the two sites, with day shifts
starting at 6 a.m. at LHO and 8 a.m. at LLO. There will be a daily telecon
between the sites at the shift change between day and evening shifts, i.e.,
2:00 PDT, or 4 :00 CDT.
-
Investigations: There is a confirmed leader for all twelve
E5 investigations, but in some cases the leader is the only member
of the team. More depth is clearly needed. Each investigation leader on
the line was asked to give a brief summary of plans for E5 and make known
any special requests:
-
Intersite environmental correlations - M. Landry & R. Schofield
The same techniques used in E3/E4 will be used again. Robert will ensure
the Hanford magnetometers are working and will talk to Szabi about what's
available at Livingston.
-
Environmental disturbances - R. Schofield
Several transient-detection DMT monitors will be running (pslmon, glimon,
tid and blrms). Robert will discuss with Julien and Ed which channels and
bandwidths to monitor with tid and blrms, respectively.
-
Calibration studies - M. Landry
In the week preceding E5 Mike will put together a swept sine control
sequence (amplitude envelope that gives good coherence without destroying
lock) that can be done in situ in a few minutes. It is planned to do a
swept sine just before the start and near the end of the run. If the procedure
can be done quickly and reliably enough, swept sines will also be performed
at periodic intervals during the run. Several calibration lines will be
injected throughout the run, but not as many as were used in E2.
-
Tidal studies - F. Raab
Fairly long stretches of data are needed for good model comparisons
(see E3/E4 report below). It's not yet clear whether the new feed-forward,
common-mode tidal servo is working properly (controls temperature of laser
reference cavity). It's hoped that data taken this weekend will answer
that question.
-
Lock losses - KR
The lock transition and servo instability dmt monitors have been recently
upgraded and will be run during E5, this time allowing real-time display
of lock history in the DMT viewer. The usual on-the-fly investigations
with the data viewer and DTT will be carried out during the run, followed
up after the run with more focussed studies on unexplained losses. A list
of all lock transitions will be available on the web in real-time, and
a table of lock stretches longer than a minute will be posted shortly after
the run.
-
Timing precision - D. Sigg
The same timing ramp monitors will be used in E5 as in previous runs.
-
Data integrity - J. Zweizig
The usual data-checking DMT monitors (bittest, slicemon, framecompare)
will be running in E5. John hopes also to implement a new script that checks
every 100 seconds or so that the CDS and RDS data streams are identical.
This would require some network reconfiguring subject to Dave Barker's
okay.
-
Data merging - P. Shawhan
A catalog of both raw and trend data arriving from the sites to the
CACR archive will be assembled, and a server set up to retrieve any of
the data using getframes or lidax when it's available. A number of data
consistency checks will be performed. Peter is working with John to incorporate
trend data produced by the DMT monitors along with standard trends. He
hopes to have it ready by E5.
-
Line noise monitoring - S. Klimenko
Having studied line noise at Hanford during E2, Sergey is eager to
see whether any changes are observed nine months later in E5. He is working
with Ed Daw on a revised background line monitor which he will run remotely
on sand from Florida. This is a background version of the monitor that
has been run in foreground in E2-E4. KR asked that Sergey add periodic
updates to the Hanford elog of calibration line tracking plots during the
run. John pointed out that Adrian Ottewill's multitaper line tracker now
has a DMT viewer interface which can be used for this in realtime.
-
Reduced data sets
Robert and Daniel will circulate a proposed RDS channel list for Hanford
early-to-middle next week for comment. Szabi will do the same for Livingston.
The Hanford list will add channels to what was used for E2, and the Livingston
list will be a pruned version of the E4 list.
-
Special running requests
Only the calibration group plans any special running at Hanford during
E5. KR mentioned that Anthony Rizzi had expressed interest in testing seismic
transfer functions at Livingston, but since only the mode cleaner will
be running there, that work may be done at a different time. Anthony and
Rana will get together to discuss this.
Preliminary E3/E4 Reports:
-
Tidal modelling - F. Raab
The Livingston E3/E4 data have been examined and tidal derivatives
computed for comparison with the model used successfully with the Hanford
E2 data. A plot
of the single-arm E3 result was shown for a period of four days. The agreement
is qualitatively good, but two problems are seen:
-
The phase difference between data and prediction is small but non-zero
and changes sign. Fred believes the sign change indicates a systematic
problem in the data. He suspected a temperature variation in the laser.
A check of the data from the temperature sensor, however, showed it was
very constant, perhaps too constant, suggesting invalid data. This will
be investigated further.
-
The data on day 70 of the UTC year (roughly E3 Sunday local time) is all
over the map, presumably because of short lock stretches preventing accurate
derivative measurement.
Similarly, the E4 data was found to be nearly useless for tidal studies
because of the short lock stretches.
-
Calibration studies - M. Landry
Mike summarized two documents posted recently on the calibration
studies web page:
-
A global scaling error has been found in the sensitivity curves produced
from the E2 and E3 data. The sensitivity curves were originally computed
to be about 10 times lower than was correct. Revised curves are now posted
and are in reasonable agreement with independent determinations. The E4
calibrations were not affected.
-
Calibration parameters are now posted for the E2-E4 data, with fitted poles
and zeroes, along with instructions for producing sensitivity curves in
the DTT or in Matlab. Click here
to see sample sensitivity curves for E2-E4. There is dramatic improvement
from E2 to E4, as hoped(!).
-
Frequency noise - R. Adhikari
Rana summarized information from a web
page discussing two issues:
-
Plots
were shown from E4 of common-mode and differential-mode arm control signals,
the differential-mode error signal, and shadow sensor voltage noise. There
is an apparent common mode rejection factor of about 20 for the frequency
noise as seen in the DM control signal, lower than design but not surprising
at this point. There is qualitative (but not yet quantitative) evidence
that the sloped noise floor in the DM signal is dominated by shadow sensor
noise (1/f) suppressed by the pendulum to become 1/f3. There
are several low-frequency peaks in the CM and DM signals, suggesting non-optimal
local damping parameters (too light where pendulums visible, too strong
where stacks visible).
-
A plot
of seismic noise and the mode cleaner error/control signals shows some
coupling of LVEA seismic noise to mode cleaner length which introduces
frequency noise into the main interferometer. The main source of the coupling
is suspected to be substantially unequal damping parameters for the different
mirrors that form the mode cleaner.
-
Line noise - S. Klimenko
-
Sergey gave a brief update on calibration line tracking from E4. Plots
of the amplitude and phase of eight lines tracked in the GW channel were
shown for a 5-minute time interval when the IFO was losing and reacquiring
lock. dqthat included. An unexpected phase drift is observed for lines
near 270 Hz in the X arm and not in lines at lower and higher frequencies.
Amplitude drifts are observed on all lines shown.
DMT Software Issues:
-
General - J. Zweizig
-
Several new or revised DMT monitors have been submitted to John recently
for running in E5: multitaper line tracking (Ottewill), earthquake alarm
(Rahkola), transient ID (Sylvestre), lock losses and servo monitoring
(Chin), and Rayleighness monitor (Sutton). John is also expecting a revised
timing stability monitor from Szabi soon.
-
New or revised infrastructure packages include histogramming support (Ito
- see below), an IIR filter class (Daw - see below), and operational state
conditions (Chin).
-
Altogether there will be 11 background monitors running during E5 plus
at least three foreground monitors run privately in root sessions. John
plans to carry out an early test run of all background monitors Tuesday
at noon PDT. Monitor authors should make sure John has up-to-date config
files before Tuesday.
-
Both sand and stone will be needed at Hanford to run the monitors because
of the large frames being produced there by the DAQ system with the
present framecpp version. At Livingston only delaronde will be needed,
but the new unix box will be used in foregroun for Steve Penn's bispectrum
monitor.
-
Histogramming - M. Ito
Masahiro has given John the code needed to support histogram creation in
the DMT and transportation of the histograms to the DMT viewer. He
is working now on the GUI that displays the histograms in the viewer and
expects its essential features to be ready in time for E5. A special test
program based on Masahiro's existing glitch monitor will be run during
E5 to demonstrate the new infrastructure for other monitor developers.
-
IIR filter class - E. Daw
Ed provided a documentation
web page describing the calling sequences for defining and using IIR filters
in the DMT. One can specify a list of poles and zeroes. The code uses a
bilinear transform algorithm to compute filter coefficients and has been
tested on a variety of challenging filter shapes. It presently does not
alert the user to gaps in the time series passed to it, but will in the
future.
-
Line noise - E. Daw, S. Klimenko
Sergey's original foreground line tracker has been converted to a background
monitor and will be run remotely on sand by Sergey from Florida during
E5. He and Ed are also working on a DMT viewer interface which is not expected
to be ready before the run.
A.O.B.
-
There will be no detector characterization telecon in August. The next
meeting of the group will be in parallel sessions at the Hanford LSC meeting
(Aug 13-16). Persons wishing to speak in those sessions should contact
KR very soon if they have not already. By default all E3/E4 investigation
leaders will be placed on the preliminary agenda, which will be posted
within the next few days.