As reported in klog34442, there is a coincidence between glitches in OMC DC PD and sensitivity degradation.
So, I tried to systematically understand the kick of the suspension and sensitivity degradation.

As shown in Fig. 1, when glitches are very large, the OMC DC PD signals can become saturated.
In this case, the calibrated DARM signal is no longer reliable, so it should be excluded from the data analysis.

In other cases, the amplitude of each glitch is different, so I categorized them into four groups based on the maximum signal in each glitch: more than 25,000 counts (#1), between 20,000 and 25,000 counts (#2), between 15,000 and 20,000 counts (#3), and less than 15,000 counts (#4).
Figures 2-5 show the Q-transform around the glitch categorized as #1, #2, #3, and #4, respectively.
Even in category #4, some excess can be seen around 60–90 Hz.
So, it would be nice to evaluate the non-lnear effect in DARM sensitivity.
Figure 6 shows the comparison of the spectra when glitches happened (as a reference, spectrum without glitches are also shown with black lines).