Abstract:

I found three types of lockloss.
One is the lockloss during changing the CARM_SERVO_IN1GAIN.
Another is the lockloss after finishing the INCREASING_LAS_POWER state, and the lockloss caused just after breathing IMC_SERVO_MIXER_DAQ.
The other is the lockloss after finishing the INCREASING_LAS_POWER state, and the lockloss caused just after breathing IMC_SERVO_MIXER_DAQ.

Detail:

I checked 5 lockloss events during the silent run on Octorber 3 and found the three types of lockloss.
What I checked is the lockloss happened around the following UTC time.

1. 2025/10/03 09:55:18 UTC
2. 2025/10/03 10:12:53 UTC
3. 2025/10/03 12:37:58 UTC
4. 2025/10/03 13:14:28 UTC
5. 2025/10/03 13:27:25 UTC

Lockloss during changing CARM_SERVO_IN1GAIN :

The first type of lockloss is the lockloss happens when changing CARM_SERVO_IN1GAIN.
Figure 1 (corresponding lockloss 1) and 2 (corresponding lockloss 5) show the several LSC related channels when the lockloss happens.
Both lockloss happened when CARM_SERVO_IN1GAIN was changed from -23 to -22, so it might be due to the glitches that was induced when CAR CMS gain was changed.

I'm not so sure this kind of lockloss happens in the different timeing such as changing gain from -22 to -21, -21 to -20, and so on, so it is important to check all the similar lockloss.
This lockloss loss might be aoidable by changing the lock acquisition procedure.

Lockloss after finishing the process in INCREASING_LAS_POWER state and after breathing of IMC_SERVO_MIXER_DAQ:

The secound type of lockloss is the lockloss happens after finishing all processes in INCREASING_LAS_POWER state.
So, no IFO setting change happens during this time.

Figure 3 (corresponding lockloss 2) and 4 (corresponding lockloss 3) show the several LSC related channels when the lockloss happens.
In these lockloss, IMC_SERVO_MIXER_DAQ signals becomes large just before the lockloss, so it might be due to the frequency noise.
One important note is this breathing often happens and some of them does not cause lockloss.
So, it might be some threshoulds of the frequency noise or some other missing triggers.

Lockloss after finishing the process in INCREASING_LAS_POWER state and after the glitches of IMC_SERVO_MIXER_DAQ:

The third type of lockloss is the lockloss after finishing all processes in INCREASING_LAS_POWER state.
So, no IFO setting change happens during this time.

Figure 5 (corresponding lockloss 4) show the several LSC related channels when the lockloss happens.
In this lockloss, there were several glitches in IMC_SERVO_MIXER_DAQ before the lockloss.
So, the lockloss might happen due to these glitches.
Figure 6 shows the enlarged view around the glitch timing but no significant features could be seen with this resolution.
So, it migh be necessary to monitor with more fast data like Moku:Lab.