Because we don't have any conviction about GAS glitches reported in klog#18246 are not mechanical issue, we investigated them at first.
Finally we found they comes from a DC filter it is set before a DAMP filter and glitches can be removed by swapping the order of the DC and the DAMP filters.
IP glitches reported in klog#18154 can be also removed with the same way.
It's better to swap the order of the DC and the DAMP filters for all suspensions, stages, and DoF.
We confirmed that LVDT sensor signals didn't reach the mechanical limits reported in klog#18155 during isolating.
All signals (F0, F1, F3, BF) were within the limits (F2 is dead) as shown in Fig.1.
- F0: -7700~-7500 (signal), -10000~-3500 (limit)
- F1: -12500~-12400 (signal), 27900~-15000 (limit)
- F3: 12040~12160 (signal), 29900~10400 (limit)
- BF: 7700~7900 (signal), -17800~14000 (limit)
After then, we noticed the glitches occurred 20 seconds (which is same as the ramp time of DC filter) after turning ON filters.
In the original implementation, DC and DAMP filters were set as "FM3" and "FM4" respectively.
Fig.2 show the error signal behavior during engaging GAS local loops.
Fig.3 show ones with swapping the order of filter implementation (DC and DAMP were set as "FM4" and "FM3").
Gliches can be removed by swapping the filter order.
Because I remembered similar situation in IP control, we tried to same things for engaging IP loops.
And then IP glitches were also vanished as shown in Fig.4 (before swapping) and Fig.5 (after swapping).