during the last days we changed the FP-cavity - RF reference locking system to force the right bucket:
in addition to the 500MHz beating which is the signal input to the PID, we added a 33MHz beating signal which is the trigger signal of the PID.
then, the lock is triggered only when this trigger signal is close to 0.

as the peak input signal is Ap = 0.5V, the beating signal is Ap*sin(phi)
one needs a phi determination below 2pi/15 which correspond to a particular 500MHz bucket (500MHz/33MHz = 15).

so, Ap*sin(+/- 2pi/30) = +/- 0.2*Ap = +/- 0.1V
we should be able to put a trigger intervalle value between -0.1V and +0.1V to always fall into the same bucket.
we did a test with -0.05V et +0.05V and we always fall into the same bucket => we will have to test the intervalle -0.1V / +0.1V to check if it is ok or not.

as the crossing is around 0V, there is 2 possible locking trigger phase position with Pi intervalle.
for the moment, there is no way to get rid of this problem. one can only let the phase drift until it reaches the right phase to start the lock.

if one tries to detect the maximum or minimum value to fall into the same bucket, the voltage level is very sensitive :
Ap*sin(pi/2 +/- pi/30) =  0.995 * Ap => it has to be sensitive at 0.5% !

we changed also some locking condition on the 2 Laselock PID regulators : see the attached picture.
search and relock : reset mode with 0V for both => when the 2 PID are losing the lock, they search a new lock starting with the same 0V value on the PZT (it can prevent they are out of range in some cases)
with these condition, it seems the lock is quicker to come back.

we can check the phase bucket with the remote oscilloscope at the IP address : 192.168.229.21 (see attached picture)
blue : trigger signal @ 10Hz (will be 50Hz in next ThomX update)
red : reference 33MHz signal, phase linked with the 500MHz of the ring
yellow : laser 33MHz signal (need to have always the same phase relation with the reference 33MHz when locked to RF reference)
 

the rms jitter with the RF reference is still ~15ps => to be improved...