After today's maintenance we found nearly impossible to lock, even the initial PRITF configuration was difficult to get and to keep; we found a very big line in SSFS signals exactly at 10 kHz (see Figure 1), which seemed suspicious from the start. We contacted SSFS experts, and Alain found out that such line appeared on Sat Jun 3rd at around 19:56 UTC, and began to grow (see Figure 2), with some additional noise at very low frequency as well. This was in between locks, when FmodErr was being tuned, and he suspected our tuning of the demodulation frequency (which I thought had been disabled for the time being) as a culprit. Indeed there was a big change in the demodulation frequency in correspondence of the appearance of the line, and the same happened today (see Figure 3); also, the 56 MHz frequency was mistuned by 2 Hz.

Given the ITF behaviour of the last days, one could assume that the system recovered in some way only when the mechanical tuning of FmodErr was triggered, and then progressively degrade again; this is far from being a valid hypothesis, but the trend of the MC Z position right after the LNFS jumps suggests something similar. After this we decided to remove the LNFS frequency tuning from the FmodErr procedure, and Benoit changed the requested demodulation frequencies used by the demodulations mezzanines of the daqboxes to make them consistent to what FmodErr has changed.

A reload of all the *_dbox_bench configurations was needed, and this made all the demodulation phases to be retuned:

• the RFC phases have been tuned as first thing; the RFC lock was immediately recovered but maybe some fine tuning of the demodulation phase is still needed, to be better checked tomorrow;
• the phases for all the longitudinal signals have been correctly recovered;
• I spent the rest of evening trying to tune the phases for the QPD signals, in order to restore the good alignment configuration; the B2 QD2 signals, used in PRITF @ MICH_SET = 0.7, were recovered and the path towards dark fringe is working reasonably well, and we went back to DC_READOUT; as we were trying to recover the other QPD signals, with some difficulty due to some strange sidebands behaviour and some signals being poorly sensitive on both quadratures, we had to quit to leave time to some noise injections.

The last phases we tuned and the switches that will need to be uncommented are in ITF_LOCK, lines 2415-2440.