We started to study lockloss. Using the data around lockloss at Oct 18th, we plot the time series, the spectrogram, and the normalized spectrogram for the following channels:
"K1:ASC-PRC2_Y_OUT_DQ",
"K1:ASC-DHARD_P_OUT_DQ",
"K1:ASC-DHARD_Y_OUT_DQ",
"K1:ASC-CHARD_P_OUT_DQ",
"K1:ASC-MICH_P_OUT_DQ",
"K1:ASC-MICH_Y_OUT_DQ",
"K1:VIS-ETMY_TM_WIT_P_DQ",
The attached figure shows the plots of CHARD. The lock loss happens at the right end. We found that the high frequency component starts to increase from ~6sec before the lockloss. We are going to check other lock loss time and other channels.
We checked the all lock loss during Oct 18th.
# What we found
There are 27 locklosses. Among them, 24 have durations longer than 100 sec. We find four patterns in normalized spectrum.
1. Become strong ~10sec before the lock loss ( 20 / 24 )
2. Periodically (5~10sec) in all band ( 12 / 24 )
--> related to microseismic motion?
3. Huge peak at O(10)Hz range ( 3 / 24 )
--> It seems twins at 15Hz & 50Hz.
4. Broad power at ~100Hz ( 11 / 24 )
The attached figures are typical examples of these classes. Although these classification is not rigid, but it might give us useful insight.
# What will we do next?
In order to study the reason of this patterns, we will see the upstream channels of CHARD_P, that is, REFL_QPDA1_RF17&45.
"K1:VIS-ETMY_MN_DAMP_L_OUT_DQ",
"K1:VIS-ETMY_MN_DAMP_T_OUT_DQ",
In Fig.1, there are no characteristic features in ETMY_MN_DAMP while CHARD looks messy. In fig.2, only burst-like features around 40Hz at ~40sec appear also in ETMY_MN_DAMP.
