Tonight we were able, through trial & error, to tune the dither alignment offsets for the test masses in dark fringe.  Paolo found a set of offsets that provided a stable locking point, and we had several locks at dark fringe that lasted for 5+ minutes.  Our record lock was more than 15 minutes long.

Diego automated the transition to dark fringe in metatron, and throughout the night we tuned the gains to have good UGFs as the alignment stabilized in dark fringe.

First plot attached is our longest lock.

Second plot attached is the darkest lock, which lasted for about ten minutes.  There are two minutes of unperturbed data started at 22:32 UTC.

The third plot compares DARM between yesterday and today.  The main difference is the SSFS gain, which Diego increased by 50% in the dark fringe state.  This seemed to reduce the noise in DARM between 10 and 100 Hz.  We can probably push the gain higher; also we should see if this noise can be reduced by running FModerr more regularly.

The fourth plot is a fine resolution FFT of the correction signals sent to the mirrors. For the end mirrors, the rms actuation is around 0.01V above 1Hz. At high frequency it is dominated by the UGF monitor line, which can easily be reduced by 10x.

The fifth plot attached is a zoom of the high-frequency region in the SSFS fast channel of the B4 6MHz signal.  The most prominent line (at 15625Hz) is a well-known CRT display refresh rate.  We don't see any evidence for lines associated with test mass solid body modes, but we didn't look very hard.

We haven't seen any sign of thermal effects in the time-series data from tonight; we couldn't pick out any consistent trends in the DC powers, but maybe there's something to learn from a deeper analysis.