During the break we decided to increase both the TCS powers by 100 mW, in order to recover the best working point for finesse asymmetry.
We realized soon that the new lensing compensation was not compatible with the stability of locking/alignment, because an oscillation of PR_tx at 3.2 Hz appeared. It didn't disappear even reducing the gain by a factor 20; there is clearly a cross-talk with some locking loop, the same problem observed many times for PR_ty. This kind of effect was not there during the commissioning time, when the locking scheme was different (PRCL on B2).
The NI TCS power has been reduced a couple of time. Now is lower then this morning, while the WI is higher. It seems that PR_tx is stable, and the finesse asymmetry will stabilize in a better position.
At this point it will not be possible to recover by TCS a further reduction of mirror temperature.