This afternoon we carried on the recovery of the ITF:

• the last missing demodulation phase (B5_QD2_56MHz) was retuned, and once we were in dark fringe in a more stable way we fine-tuned the other ones; by doing that we noticed that a few random phases drifted of random amounts: the SIB2 ones were the most affected by far, with the RFC one drifted of more than 2 rad, making the relock of the RFC impossible at some point;
• going to dark fringe happened basically every time, with no unlock correlated to saturation; we looked into the switch to LOW_NOISE_1, which was troublesome for most of the shift: a couple of gains were retuned for the different 56 MHz amplitude, and we engaged the alignment loops (DIFFp, PR, and BS); while tuning the the phases we noticed that now the signals are quite poor, but we could close the loops; however, we had to decrease the gain for DIFFp_TY, as it started to build up the 7 Hz resonance;
• we tried once to engage the COMMp in LOW_NOISE_1; we noticed sometimes some very low frequency oscillations, so we left it off for the time being;
• the major offender was found in the change of locking filter for DARM; while this change could be done manually with no issues, it was not the case while doing it in the usual part of the lock; it is still not clear if this is due to the poor SNR of the 56 MHz, the general weak state of the ITF or because it is done during the huge transient where the B7/B8 powers go down by 30% just after reaching dark fringe; anyway, one attempt to move such change at the end of LOW_NOISE_1 (a bit of time after setting gains, engaging alignment loops and the boost for the SSFS) was successfull, to be confirmed of the next trials;
• engaging the Alpha subtraction did not kill the lock by itself, but it probably is wildly mistuned; moreover, since the RF sensor for MICH changed sign, probably also the coupling did as well, to be checked;
• we locked both OMCs a couple of times with no issue at all; during the first lock we noticed that OMC1 locked with a much higher power on B1s2_DC: this is due to the fact that changing the modulation amplitude of the 56 MHz sideband changes also its calibration in meters; we tuned the DARM offset accordingly (only for the RF part);
• from then on what we did was just retuning some parameters of the lock acquisition, and in the end we managed to get back to LOW_NOISE_3 once, and stay there for a few minutes; although, because of all of the above, and the fact that the working point of the ITF is yet to be tuned, we got no range or sensitivity; despite this, we feel relatively confident that the current path can be pursued.

We left the ITF in LOCKED_RECOMBINED.

Side notes:

• the 100 Hz noise seem to be present no matter what, and it is really noisy on DARM and PRCL (see Figures); also, the 50 Hz seemed higher than normal;
• the UGF of the SSFS is "rapidly" degrading over time in dark fringe, with losses of the order of a few kHz; also, we could observe high-gain oscillations even with nominally low UGFs;
• frame providers (moslty, but not only, FbmMainUsers) were quite weak for all the shift, with lots of "missing data" and an incredible number of dataDisplay crashing...
• thinlinc sessions were not robust either, with crashes of the IceWM DE, several freezes and one case of full death of a session.