The loss of transmitted power of the RFC occurs together with an increase of reflected power, so it seems that it is indeed some kind of misalignment taking place (also, the IMC power stays constant during the drift of the RFC). In the past it occurred sometimes that a misparked reflected beam from the PR warmed up the SIB1 so inducing some misalignment of the RFC injection, but this time I am not sure it is the case (see PR tx/ty signals).
What is really wierd is the fact that at some point the loop (?) drives completely away the IP of the mode cleaner mirror (>1mm), so causing a very big thermal correction (almost -2V) on the ML to hold the resonance. This is definetely not expected, especially since Paolo Ruggi implemented stricter limits to the search for RFC resonance.

Tonight I paused the INJ node, then brought the MC inverted pendulum to its last "normal" position (around -2600), then reset the thermal correction to the nominal value (+0.15V) and tried to scan for the RFC resonance. The peak was very low, at a maximum of less than 3V on INJ_RFC_TRA_DC (used to be >6V). I set the threshold to trig the lock at 2.5 and then 2.1 in the INJ_main.ini file of metathron and reloaded the configuration to check whether the lock works fine, but unfortunately, despite the presence of a clear error signal, the loop to lock the RFC does not seem to engage (second plot).

We will investigate tomorrow.

I leave the INJ node paused, RFC loop not engaged.

In fact the RFC misalignment was due the already seen thermal effect related to a PR wrong angular position. In this period there is no periodic alignment of PR ( the last one occurred on Dec 18th), and the local controls are not able to guarantee enough long term accuracy. The frost of these days probably accerated the drift.