S. Dwyer, J. Kissel, K. Kawabe After installation of the infrastructure (LHO aLOG 16655), and copying of LLO's filters (and adjusting for the specific frequency of 9.7305 [Hz]; LHO aLOG 16658), we tried damping the H1 SUS ETMY's highest vertical mode (a.k.a. "bounce" mode). For the first attempt, the IFO was locked only using ALS diff, with Sheila in the driver's seat. At the time, the mode had been rung up to ~7e-12 (DARM) [m/rtHz] @ 9.7 [Hz]. We had tried a few configurations of the filter bank, and only adjust the gain. We'd found how to ring *up* the mode with a positive gain, with the +60 [deg] (FM2) and bp9.73 (FM4) filters engaged -- then flipped the gain sign (i.e. flipped the phase 180 [deg]), and immediately could see reduction. We had the gain as high as -64, using ~50% of the DAC range, after which took about ~10 [min] for the mode to cool down to the ALS DIFF noise floor of ~1e-12 [m/rtHz] @ 9.7 [Hz]. After a lock loss of two, we were able to get as high the IFO guardian state "RESONANCE," with CARM controlled using digitally normalized RELFAIR9, and DARM has been transitioned to AS45 Q. At this point we saw the mode was still quite run up, so we again turned on the DARM DAMP V filter -- same filter combo, and we could see just as quick a reduction with a gain of -64. This time however, we were using much less of the DAC range, so I went up to a gain of -100, and the mode was quickly damped to the RESONANCE noise floor of ~1e-13 [m/rtHz] @ 9.7 [Hz] within a minute or three. With these two victories, I'm reasonably confident that this will be our ticket to future bounce-free success. Design strings: FM1 "+60dg" zpk([0],[2.16667+i*12.8182;2.16667-i*12.8182],1,"n")gain(0.0523988) FM2 "-60dg" zpk([0],[1.21667+i*7.1979;1.21667-i*7.1979],1,"n")gain(0.0911865) FM4 "bp9.73" butter("BandPass", 4, 9.3, 10.4)gain(120, "dB") with all filters set to an input switching of "zero history" and output switching of "immediately." Attached is a bode plot of the final good set of filters together, FM2 and FM4. Note that these filters were loaded individually from each bank, Keita did *not* load the the whole foton file.
Why local damping does not work:
Before the DARM bounce damping was implemented, I started playing with the BOSEM damping at the top stage, and concluded that it will not work even though the bottom stage bounce mode is clearly visible in the top BOSEM.
The reason for this is that the feedback only sees the top to top transfer fuction.
This TF at the bottom bounce resonant frequency (9.7305 something something Hz) is not that different from that at off-the-resonance proximity (e.g. 9.7Hz). I confirmed this by various things like injecting band limited white noise, injecting sine wave tuned to the resonance as good as possible (9.7305 something level), injecting sine wave at proximity frequency, feeding back with a band pass filter and turning up the gain until it does something.
This means that, since the coupling from the top to the bottom is small, the top mass starts oscillating at proximity before the feed back can do something significant to the bottom motion.
The decay time (1/e) for the bottom bounce mode, measured by the top BOSEMs, was measured to be 13000 to 14000 seconds (Q of 4E5 or so).
I was able to reduce this to 8000 to 9000 seconds by top mass local feedback, which is useless.
Why DARM to top mass damping works:
The feedback loop sees the top to the bottom TF, which has a very sharp peak (as in Q of 4E5 sharp) at the resonance. Therefore you can touch the resonance without touching anything else.
Note that DARM to top mass bounce damping affects the calibration, so there will be a calibration hole at 9.7Hz if the damping filter is on.
When the bounce motion measured by DARM was on the order of 10^-11m, we needed to use the full range of DAC to damp the motion using DARM bounce damping path.