This afternoon we carried on the recovery of the ITF:

• the West arm was realigned and relocked;
• we performed the tuning of FmodErr: since we did not do that for a very long time, the misplacement of MC_Z was quite big; this led to the first occurrence of the problem described in #42159 and to the re-discussion about how to deal with the master laser's thermal correction when FmodErr is engaged; probably the best thing will be to hasten the re-deployment of the old calibrated FmodErr, hopefully next week;
• after that we began to recover the ITF and to study the saturation issues arised in the lock acquisition since the increase of the input power:
  • we reverted all the changes done for the SDB1 intervention without stumbling into any issue;
  • we did a standard lock acquisition and we noticed that, near and at dark fringe, the B2 photodiode is the one saturating the most, mostly due to the 16 MHz, coherently with what Julia and Yuta recently reported and the thresholds reported by DET people (B2_PD2_sample above 0.4 mW, see Figure 1);
  • we investigated by hand reducing the modulation amplitude for the 8 MHz sideband during the MICH offset reduction, and at MICH_SET = 0.3 we are above the threshold, and at 0.5 we are already close enough; we reduced the modulation amplitude from 15 dBm to 9 dBm for the whole lock acquisition (but resetting it in DOWN, since it is also used for FmodErr) and things got better, however in dark fringe there was still saturation of such sidebands, so we reduced it further down to 6 dBm; both PRCL and the PR alignment loops were adjusted in gain but they did not seem to suffer from the reduced signal;
  • at this point it was noticeable that the 6 MHz sideband was still causing trouble on B2, so we tried to reduce it even more at MICH_SET = 0.2, from 9 dBm to 6 dBm; despite the very bad SNR, B7 and B8 were still usable, so we left it as default since it was definitely improving things;
  • in Figure 2 there is the comparison at MICH_SET = 0.1 between July 2nd and tonight, where the [6, 8, 56] MHz modulation amplitudes were respectively [9, 15, 7] and [5, 5, 7];
  • at this point we managed a few times to reach dark fringe and, eventually, LOW_NOISE_1; statistics is not high, but the locks seem quite promising;
  • at a first glance, the alignment loops seem to work fine, maybe some fine tuning of their working point for 25 W is needed, since we didn't do that before the Commissioning Break;
  • we tried once to lock OMC1: as expected the temperature scan took a long time, given the several interventions and the long shutdown; starting from the usual temperature of 22.69 °C, we could find a good amount of power on B1s2 at 22.745 °C but the scan went on, then at 22.86 °C another was found, but the temperature reset happened just after; at the second attempt I manually ramped up to 22.745 °C but then the OMC locked on a 1st order mode and the DARM hand-off failed (see Figure 3); in the end I reverted manually the temperature to 22.69 °C and we gave up for the night;
  • Figure 4 is Yuta's GUI showing the overview of the ITF when we're locked in LOW_NOISE_1, with all the powers nicely scaled also visually.

We left the ITF in LOCKED_RECOMBINED with the FmodErr tuning disabled.

Side notes:

• the West arm is usually quite shaky after an unlock, and sometimes the 7 Hz shows up a bit in DARM, but it does not build up as before;
• looking at the PD signals, we could correlate the flashes we see on the B7/B8 powers when near dark fringe with similar flashes on SIB2_B2_PD2_RF_{min,max};
• the LNFS bug reported in #42015 is still there: for "low" modulation amplitudes, the LNFS somewhat misses the correct integer, and selects the closest one: for this reason, the 6 MHz sideband is reduced to nominally 6 dBm, which become 5 dBm, so the (hardcoded) sanity check is done against 5 dBm; the same happens to the 8 MHz sideband, but there is no check (yet);
• the change of the 8 MHz amplitude is done in the old way with the external script, as the current PyLNFS100 package doesn't have the appropriate cm handler; however, this is trivial to implement and it will be done tomorrow; looking at a few sequences, I did not find any issue with the commands already present in the DOWN state, while the other ones happen in different places altogether;
• all the "8 MHz" gains were rescaled by hand, but in the ITF_LOCK.ini file there are the old values, the values for 9 dBm and the current ones for 5 dBm; if we think to stick to this, I will setup something similar to what we already have for the 56 MHz gains;
• since the further reduction of the 6 MHz amplitude requires a bigger increment of the DARM gain, and given the different places they must be done into, such increment at MICH_SET = 0.2 uses twice the ramp w.r.t. the other gains/setpoints, so the asynchronicity between ramps is done on purpose;
• the 100 Hz line and its harmonics are HUGE, always polluting PRCL (see Figure 5) and sometimes DARM, but they're not particularly impacting on the corrections (see Figure 6); however, a number of unlocks are due to this PRCL weakness;
• most of the unlocks when trying to go to dark fringe after the reduction of the modulation amplitudes (after 18:50 UTC onwards) are due to the fact that the transfer function between MICH before and at dark fringe (B1p_SUM_NORM and SSFS_B4_56MHz_Q respectively) changed sign for unknown reasons.