Worth noting - we had a strong EQ yesterday (an M6.5 off the coast of Japan) arriving at ~19:40 UTC (Friday July 11).
Of the two SEI chambers actively isolating that day:
- ETMX, which was fully isolated (both stages at level 3), saw both stages trip simultaneously (independantly) on the actuators at the time of the quake (ST1, ST2)
- ETMY, isolated at level 1 (see alog 12706), rode the EQ out and did NOT trip
A trend of the ground BLRMS for the day of July 11 is attached (made with ligoDVweb because BLRMS trends not accessible via nds recently - units in nm/s not counts).
Just one example, but could be worth exploring.
This is kind of what we expect, nice to have it confirmed, assuming that the ETMX was running some form of the 40mHz blends and the ETMY was running with a Start blend( 750mHz) which is usually how level 3 and level 1 controller operate. The blend filters are what is important not the loop gain (level). If you look at Sebastiens entry in the LLO log 13449 you will see that almost all of the extra seismic motion from the Tonga earth quake was between 20-150mHz. At these frequencies the springs are stiff and the platfrom wants to follow the ground so no (or very little) force is needed. When we are in low blend and the ground motion goes up by X100 it takes a lot of force to counter act the ground motion so the actuators saturate and the watch dogs trip. We (Sebastien) are in the process of writing a guradian module that will be able to switch the blends when a earthquake takes place, the tricky part is figuring out how to trigger it