I turned up the gains of ASC-MICH pit and yaw, and gave them each a different lowpass, and was able to get our BNS range up a few Mpcs. I also switched to using a SOFT_P cutoff in both CSOFT and DSOFT with a corner at 10 Hz, rather than 14Hz. These 2 things combined give us about 110 Mpcs.
However, the ASC seemed more delicate in this situation. I'm not sure if it was Keita in the LVEA for DIFF VCO tests (likely not) or a small amount more ground motion, but the IFO wobbled at 0.5 Hz a few times, and then eventually had a ring-up of this frequency and lost lock.
In the attached plot, it seems like INP1, SOFT and the DC centering loops all saw this ring up at roughly the same time, starting a little more than 200 seconds before the lockloss. It's hard to say if it's a pitch or a yaw problem. INP1, DC3 and DC4 see the instability most stronly in yaw, while DC1 and DC2 see it more in pitch. Other loops such as SRC2, DHARD, and CHARD all see the ring-up, but they mostly start later, after it's pretty clear in other degrees of freedom.
(Squeezing was on during this lock)
I've put these changes into the guardian, and they've worked for at least 2 locks so far. So far, the 0.5 Hz instability hasn't come back, although that is mostly luck and low ground motion, more than anything deliberate.
I took a spectrum with no lines (no Cal, no ADS), and we got up to 117 Mpc!
BruCo report available here:
https://ldas-jobs.ligo.caltech.edu/~gabriele.vajente/bruco_lho_1236049000/
During the Friday commissioning meeting several of us were wondering how H1 with 30 W and 2 dB sqz have the same shot noise as L1 with 40 W and 3 dB sqz. I believe that the L1 curve shown in H1_DARM_FOM has low squeezing. The shot noise at 1 kHz is shown to be ~4e-20 m/rtHz. The attached plot shows a typical L1 noise curve with 40 W and ~3 dB sqz. The shot noise at 1 kHz is closer to ~3e-20 m/rtHz as one would expect from the factor ~sqrt(30/40)*10^-0.05=0.77. The L1 40 W 0dB sqz spectrum is also shown.
