Sheila and Jeff have been looking at the idea of trying to damp some of the triples' bounce and roll modes (mostly the bounce modes), since we seem to be seeing some of them in DARM. The alog thread with notes and candidate IDs for some of the modes starts with alog 77182. While we may not end up wanting to actually damp these modes during Observing (especially if it broadens them), we would at least like to know which modes are associated with a suspension, since other modes could be associated with environmental factors (like HVAC fans).
After chatting with Sheila this morning, I interleaved some attempts at actuating on some of those modes with other work that was going on during this morning's commissioning time. The initial idea was to use a length signal (eg DARM) that we already have access to (via the LSC output matrix) and try to actuate on a triple suspension that is not used for length locking. I ran into some hiccups, but then at least tried something. I think we'll need some model changes in order to try much more.
The challenges:
- The LSC output matrix does not have rows (at least on the medm screen) for actuating on either PR3 or SR3. That makes sense, since we never envisioned using them for length locking. But, it does mean that we can't use the LSC output matrix to send signals to those suspensions.
- The LSC output matrix does not have the ability to send DARM to any optic other than the quads. Again, this makes sense because we didn't envision using it that way, but it again means that even for testing with PR2 or SR2 we can't use DARM as the length signal for damping the bounce / roll modes.
The successes:
- I think I've got filters in place in PR3 to try damping the 28.2 Hz bounce mode, once we have an error signal for it. These filters (a bandpass, as well as a +60 deg and a -60 deg filter) are in the M2 drivealign matrix of filter banks, in the L2L, L2P, and L2Y filter banks. I chose PR3 and this mode since it seemed nice and loud, but also less susceptible to the confusion of which mode is from which optic that may be happenning in the 27 Hz modes. But, after I put the filters in, I realized that I didn't have a way to send a length signal there. All the filters are loaded, but everything in the PR3 suspension is reverted to what was in the Observe.snap SDF file (which is basically no signals going anywhere).
- I think I've got filters in place in PR2 to try damping the 40.9 Hz roll mode. This is the only roll mode (or, peak that could be a triple roll mode) that seems to be prominant in DARM. It's not clear which optic it is from, but since it's louder in PRCL than in SRCL, I started with PR2 (and didn't get to SR2). There didn't seem to be any obvious PR2 bounce modes in Jeff's plots in alog 77184, which is part of why I started with the roll mode.
- Again these filters are a bandpass (this time around 40.9 Hz, with a magnitude of 1 in the pass band), and a +60 deg and a -60 deg filter (at 40.9 Hz). They are in the M2 drivealign L2L, L2P, and L2Y filter banks.
- I first checked that the gain of the M2 L2L, L2P, L2Y outputs were zero, and that there was no way I could actuate in length on PR2's M3 or M1 stages (the switch after the M3 LOCK_L bank is set to OFF, and the gain in the M1 LOCK_L bank is zero or one of the switches is set to off.).
- I then set the LSC output matrix to have a non-zero value in the PRCL -> PR2 element. I started with something very small, but ended up with a 1 there.
- I made sure that the M3 ISCINF_L filter bank was letting signal through, and also that the M3 LOCK_L filter bank was letting signal through, and that the M2 LOCK_L filter bank was letting signal through, and I turned off all frequency dependent filters in those banks. The gains on all of these filter banks was 1.
- I had on my bandpass in the M2 L2L filter bank, and no other filters (so no phase filters), and slowly increased the output gain until I started seeing signal going through to the M2 coil actuators.
- With a gain of +30, I could ring up the 40.9 Hz mode (this was a few to a few tens of counts on PR2's M2 coil master outs). The mode seemed to be growing, as I'd expect if I had correctly identified it (albeit with a lot of luck on my first try).
- With a gain of -30 I could ring it down again. -100 was still okay, but a gain of -200 caused the mode to start growing again.
- When I was increasing the mode, I saw it grow in both PRCL and DARM outs. When I was decreasing it, I could make it pretty much disappear in PRCL_OUT, but it never really got smaller than the ambient level in DARM.
- At this point it was nearing the end of the commissioning window, so I left all my new filters in place (but off and unused), and otherwise reverted everything in PR2 (PR3 had already been reverted).
Some thoughts:
- Sheila pointed out that, since I was using a PRCL signal to actuate on a PRC optic, it's possible that the mode is not on PR2, but that I was just cancelling it out on PR2. So, today's test isn't necessarily definitive that I've identified a mode on PR2.
- PeterF asked the interesting question of how much these modes are showing up in DARM, relative to their expected levels from thermal fluctuations. This is going to require some math to figure out.
- In order to test more things, the next step is probably to create pathways for PR2, PR3, SR2 and SR3 to be actuated by DARM. It's likely to be challenging to use the infrastructure we have right now to do much more.