We had a quick look at the line tracker outputs, and confirmed that it is working well.
We found that DRUM1 and DRUM2 modes (likely to be of NI and WI) and BS DRUM and BUTTERFLY modes are kicked during the lock acquisition.
We also found that NE ring heater temperature changes by ~0.1 K between the locks.

Line tracker frequency outputs:
Frequency outputs from yesterday are attached. Some of them are noisy and they are maybe not from the mirror modes.

V1:PAY_{MIR}_VIOLIN#_FREQ (Good)
V1:PAY_BS_BUTTERFLY/DRUM_FREQ (Good)
V1:PAY_DRUM#_FREQ (3 at 7808.8Hz and 4 at 7808.9 Hz are too close and some times following two lines)
V1:PAY_9pt6kHz#_FREQ (1 at 9665.1 Hz and 2 at 9965.4 Hz are too close)
V1:PAY_10kHz#_FREQ (3 at 10112.0 Hz and 4 at 10112.2 Hz are too close)
V1:PAY_12kHz#_FREQ (some of them are noisy)
V1:PAY_15kHz#_FREQ (Good)
V1:PAY_26kHz#_FREQ (Good)
V1:PAY_39kHz#_FREQ (some of them are noisy)
V1:PAY_44kHz#_FREQ (some of them are noisy)

Some of DRUMs, 9pt6kHzs and 10kHzs have modes which are very close in frequency (see figures in logbook #41794), and the line tracker sometimes follows two peaks in the peak search range (see this figure).


Ring downs from lock acquisition and Q measurement:
Amplitude outputs of the line tracker are basicaly stable during the lock, but DRUM1 and DRUM2 modes and BS DRUM and BUTTERFLY modes have ringdowns during the lock. It looks like they are kicked during the lock acquisition.

We used this kick to derive the Q-values of the bulk modes. Using the 5 locks stretched for more than 1 hour of LOW_NOISE3 state in June 20, I got the following Q-values.

[BS]
DRUM      (1875.5 Hz): (6.2 +/- 0.1)e6  figure
BUTTERFLY (1256.5 Hz): (6.7 +/- 0.4)e6  figure

[Test masses]
DRUM1    (7804.5 Hz): (1.6 +/- 0.3)e7  figure
DRUM2    (7807.5 Hz): (8.2 +/- 0.4)e6  figure

Attached figures are the typical time series data for each mode used in the analysis. Variation between the measurements are also attached.
I used A*exp(-t/tau)+C for the fitting, and calculated Q-value using Q=pi*f0*tau, where f0 is the measured mode frequency (median of the line tracker frequency output during the lock).

The results look reasonable. See, also, measurements done using O2 data (VIR-0241A-18). They got Q>=2e6 from Lorentian fit of thermal peaks, and ~1e7 from ringdowns for test mass drum modes.

We note that there are some peaks which are not stable and also does not show clear ringdowns (for example, 44kHz1 often rings up and 44kHz2 often goes up and down).


Temperature dependence of drum modes:
We know that one of DRUM3 and DRUM4 is from WE from the drastic temperature change in WE (logbook #41794). We compared DRUM1/2 frequencies and ring heater temperatures, and found they both change along with the temperature change in NI and WI, but not NE and WE (see this figure; similar plot also in logbook #39767). NI and WI have a similar temperature change, so we cannot tell which is which at this point.
From the linear fit, mode frequency dependence on the temperature was derived to be 0.9 to 1.0 Hz/K, which is consistent with previous measurements (logbook #39039 and #41798).

Raw time series data used for the fit is also attached.
We noticed that NE ring heater temperature changes by ~0.1 K between the locks. Not as drastical as NE, but WE ring heater temperature also changes much, compared with WI and NI. This might indicate some scattering, point absorbers, cavity misalignment, etc.
Further investigations such as correlation between different temeprature sensors, alignment drift, etc is necessary.


Next:
 - Fit violin ringdowns and get Q-values (when violin damping is off !); we might see the effect from the ear damage?
 - Excitation tests to test if we can excite the bulk modes, identify which peaks belong to which mirrors, and measure Q-values.
 - Ring heater tests to identify the peaks, measure mode frequency temperature change, and measure the RoC change from transverse mode spacing measurement.
 - Look into NE (and WE) temperature change between the locks.
 - Look into the cause of the kicks during the lock acquisition.
 - Look into the peaks that are not very stable (and doesn't seem to be from the bulk modes)

 

As noted previously, the NE ring heater temperature sensors see larger temperature changes with lock/locklosses. To check if this is just a RH TE calibration error, I looked at the vertical drift (the blade springs sag or stiffen with temperature and thus are a best true change-of-temperature sensor). Daily ambient temperatures dominate, so I roughly subtracted this by finding lockloss during a mostly linear vertical drift time and detrending linearly. The residual confirms what the ring heater temp sensors see: NE tower sees ~2x temperature change compared to WE during lock/lockloss. The attached plot shows the end ring heater temp sensors, the linear fit on the raw vertical data (I use vertical sensors above the marionetta and in payload chamber), and the detrended vertical data. I looked also since the power increase, but we've had few locks; it looks comparable now at 25W but the ambient temp looks less linear in the lock times, making simple linear detrend less reliable. Should be investigated more with cleaner data/better subtraction. Script is in /users/hardwick/scripts/mirror_temp.py. Just need to change time and duration.

I also looked again at the west end during the large temp change when the air conditioning went off; shown here. We can roughly calibrate: 0.36 RH degC / 1 WEB degC and 235 um Vert / 1 WEB degC. Note this doesnt take into account the lock which held for the beginning part of the large ambient temp change.