As discussed in klog32628, we can see some breathing of the sensitivity around 60 - 100 Hz in current sensitivity.
That frequently occurred before the lock loss.
We discussed, and we suspected the DAC output of ETMX MN coil. If its value became large due to the tidal effect, the sensitivity became bad due to unknown reason.
As shown in Fig.1., there seems the some correlation between DAC output(H) and trend of the BNS range.
In this morning, Ushiba-san turned on the roll off filter to treat the tidal effect using IP control, then this breathing seemed disappeared, I think.
To test this hypothesis, I compared several sensitivity curves corresponding to different LSC-DARM_OUT values.
The time series data of DARM_OUT is shown in Fig. 1, and the spectra comparison is presented in Fig. 2.
There appears to be a clear dependence: the larger the feedback signal is, the worse the sensitivity in the 60–100 Hz range is.
The reason why a larger feedback signal leads to sensitivity degradation is still unclear.
The actuator at low frequencies is the MN rather than the TM.
Nevertheless, this observation can help us consider strategies such as offloading of the DC feedback signal to the IP.
For example, insufficient L2P and/or L2Y de-coupling of the actuators may bring this kind of things??
We carefully checked the spectrum of DARM displacement at the timing of the IP filter.
blue :
2025/02/12 09:29:00(JST)
DARM_OUT = -1.3e-7
green :
2025/02/12 09:31:00(JST)
DARM_OUT = -1.6e-7
ping :
2025/02/12 09:33:00(JST)
DARM_OUT = -7e-6
brown :
2025/02/12 09:36:00(JST)
DARM_OUT = -2e-5
orange :
2025/02/12 09:43:00(JST)
DARM_OUT = +1e-7
Actually, even we turned on the IP filter, DARM sensitivity became worse when DARM_OUT value became large. And Sensitivity would be best in case the DARM_OUT value became small.
