today, we had a strange effect :
both the laser and the FP cavity were locked on the RF frequency (500.10045 MHz), and we were producing X-rays but with a slowing (several seconds) fluctuation exactly as the electrons-photons phase was drifting and the fluctuation can goes to 0 Xrays produced even at the "right" phase.
but the ring people says there is no phase fluctuation in their measurements.
and I looked the baseline of the 500MHz beating (which is the synchronisation signal with electrons) and I didn't see any fluctuation either.

we tried 2 other ring frequencies : 500.10030 MHz and 500.1 MHz.
there were still X-rays production fluctuations but with a faster rate.
when we went back to 500.10045 MHz, we found back the same X-rays production fluctions at a slow rate.

to remove this effect, I had to add an integrator gain (I = 1e-6) in the Laselock in the FPC/RF loop.
this integrator was not used (I = 0) in the previous stable X-rays productions....

it means that we have now a slow phase fluctuation.
we have to check if we see these fluctuations in the 500MHz beating signal.
loosing the Xrays means we moved by the packet length dt ~ 50ps which is equivalent to 9° @ 500MHz of fluctuation.
with A= +/-1V signal, it would be equivalent to a drift of ~ A*(2pi*dt/T) ~ 157mV !!!
it is strange we didn't see it => to be checked !!!

conclusion : now, we have to put this I=1e-6 in the FPC/RF loop to get a stable Xray production.
but the locking acquisition works better with I=0.
so we need to remove it before starting to lock and then, when the lock is stable (but the X rays are poor), we need to put I=1e-6 => the Xray flux inscreases dramatically.