Last night something happened on the beat between the PSL beam and the auxiliary NPRO beam, that spoiled the quality of the beating I had before. As a result the PLL has become unable to lock the two lasers.
The amplitude of the beat at the spectrum analyzer has gone down to -40 dBm from -10 that it was earlier. The frequency has also become more unstable so that now it can be seen writhing within tens of KHz.
Meanwhile the power of the single beams at the PLL photodiode hasn't changed, suggesting that the alignment of the two beam didn't change much.
Changes in the efficiency of the beating between the two beams are not unusual. Although that typically affects only the amplitude of the beat and wouldn't explain why also its frequency has become unstable. Tuning the alignment of the PLL optics usually brings the amplitude back, but it was uneffective today.
It looks like something changed in either one of the two beams. In particular the frequency of one of the two lasers has become less stable.
Another strange thing that I've been observing is that the amplitude of the beat goes down (several dBm) as the beat frequency is pushed below 50 MHz. Under 10 MHz it even gets to about -60 dBm.
I noticed the change yesterday evening at about 6pm, while I was taking measurements of the PLL open loop tranfer function and everything was fine. I don't know whether it is just a coincidence or it is somehow related to this, but Jenne and Sanjit had then just rebooted the frame builder.
Problem found. Inspecting with Koji we found that there was a broken SMA-to-BNC connector in the BNC cable from the photodiode.