aLIGO LHO Logbook

H1 SUS

jenne.driggers@LIGO.ORG - posted 16:49, Wednesday 04 December 2019 - last comment - 15:03, Monday 09 December 2019(53690)

Violin mode guardian now resets reference level for checking ring-ups

[Keita, Jenne]

Keita wondered if we could make things such that operators don't have to take the IFO out of Observe to damp violin modes. The restriction right now is that our violin damping guardian is written such that it will overwrite any gain changes you try to make, when it is in its nominal state. So, in order to change damping settings, you must take the violin damping guardian out of its nominal state, which then pops the IFO out of Observe.

A situation where we see this is that if the guardian determines that a mode is more than an order of magnitude larger than the value it had when we first enter the nominal state, it will write a zero to the damping gain forever more. We fixed this case (we think) by having a copy of the mode amplitude that is reset to the new (higher) value if the guardian needs to set the gain to zero. The idea was that the guardian would then leave the gain alone, and a human could try damping the mode. So far, that doesn't actually work, in that the guardian then thinks the mode is okay for damping, and starts to damp the mode. So, humans still can't enter different gain values (including to set the gain to zero temporarily in order to allow a filter change). But, modes that have historically been left undamped even though they're not actually growing are now being damped. So, hopefully we'll have more of our violin modes damped than have been damped in the past.

More logic needs to go into the guardian however, in order to allow humans to enter gain values.

Comments related to this report

jenne.driggers@LIGO.ORG - 15:03, Monday 09 December 2019 (53776)

Link

[Jenne, TJ, Camilla, Patrick]

We've made some more modifications to the violin damping guardian, this time successfully making it so that a person can change the damping gain values while we are still in the nominal Damping_on_DC state, so that we can stay in Observe while damping violing modes.

Note from the weekend that these 1kHz modes on ETMY that are rung up don't seem to have been made worse by the new logic from last week, but this change will enable operators to work on trying to damp them without taking us out of observing.

The logic from last week was modifying the reference amplitude, and then checking against this updated value to see if the mode was ringing up. If a mode was ringing up according to the most recent reference value, the damping gain would be set to zero. But, since the reference value kept getting updated, there was a risk that modes would never get turned off when they needed to be. So, now the guardian will update the reference values twice, but then will stop updating the reference value and so will leave the damping gain at zero. This allows some modes to continue getting damping, in the case where they are a bit higher after lock than when we initially acquire, but for which their damping settings do work well. But, if they have damping settings that do not work well they should have their gains set to zero.

The new logic that was added today was to allow the guardian to determine if a human is trying to change the damping gain.

If the current damping gain is different from the last value that the guardian thinks it wrote, then it assumes a human is working on the damping, and the guardian will not do anything to that violin mode. It will not check to see if it is ringing up, it will not turn the gain to zero. So, if a person is working on damping a mode, then they must be watching that mode. In practice, this is already what our operators are doing anyway.
If at any time the gain value returns to the value that the guardian last set, then the guardian will take back control of that mode. There is a notification (that blinks annoyingly - I'll try to follow TJ's suggestion on how to make it not blink) that is also written to the log file of what that last guardian value was. So, if you put the gain back to that value, the guardian then goes back to monitoring the mode and adjusting the gain. By doing this, you can give control back to guardian while staying in Observe.
Since all of the reference values are reset in the main part of the DAMPING_ON_DC guardian state, another way to give control back to guardian would be to toggle over to DAMPING_ON_SIMPLE, and then back to DAMPING_ON_DC. However, this will take the interferometer out of Observing, so is not preferred.

So far, things seem to be working; TJ, Camilla, and Patrick are working on damping the EY modes that are high right now, and we're in Observing!