from several weeks, the maximum power stored in the FP-cavity was ~ 5kW.

today with Daniele, we finished to investigate the problem, and now the power inside the FP-cavity is back to ~50kW for 30% of laser amplifier ratio (~16W).

we optimized the signal received by the PDH photodiode by installing a large DET100 to collect more light.

if one installs a small photodiode (DET10) in the middle of the beam, the carrier signal when a FP-cavity crosses a resonance is larger because the photodiode "sees" only the part of the beam which is geometrically coupled to the cavity in its small active area, but :
1- once we will improve the geometrical coupling, the part of the incoming beam coupled to the FP-cavity will increase.
2- one need to work with a diffuser in front of this photodiode to precisely adapt the feedback loops gain : in that case the photodiode is sensitive to the whole input beam, whatever his active area size.
so, we decide to put a DET100 (which is given for 35ns rise time / 10MHz BW when connected on 50ohms).

see the scheme in attached file.
and a picture of the desktop with all the lock parameters :

the quality of the lock, seen on the reflected power signal is very good !
and the stability is only limited by the necessity to act on the laser Smaract motors to let the PZT in its working range.
dark blue : transmitted signal
green : PZT
pink : error signal
light blue : reflected signal

conclusion : it is not clear that the cavity Finesse have significantly increased during the last weeks, as we are roughly at the same level than before (47kW).
but as we precisely adapted the signal levels in the feedback scheme (PDH S/N ratio and Laselock parameters), the result is a more stable lock.