The ALS PLL now locks stably for several minutes by using a modified version of the servo filter.
The bandwidth is about 12.5 kHz and the phase at the UGF is 127 degrees.

Earlier the loop was stable with only a very small gain of the common mode servo board (-47dB). Because of that, the PLL range was insufficient and the lock would brake after few seconds.

To better redistribute the gain, Keita made a small filter with gain of 1 at DC, a pole a 4Hz and a zero a 4kHz inside of a small Pomona box (see attached schematic). This provides a gain of -40dB above 4kHz and so it relieves the attenuation by the common mode board.
Today we connected this filter between the output of the phase-frequency discriminator and the input of the common mode board.
After a bit of tinkering with the common mode board gains and switches, Keita found a stable set of parameters.

Input 1: ENABLED
Input 1 Polarity: ENABLED
Com. Comp: ENABLED
COM. EXC.: ENABLED
Fast: ENABLED
Gain IN1: 0
Gain Fast: -6

The lock now breaks only when the laser temperature drifts out of the PZT range (+/- 15V).
The slow control will have to make up for that. Work is in progress.