[aaron, shruti]

Shruti has observed a 144 kHz oscillation on the PDH error signal from our cavity-with-cantilever. The oscillation was railing the PDH signal, making it impossible to maintain lock for more than a few seconds. I came in to troubleshoot.

1. Observed a flaky lock with PDH signal railing
   • initial settings on the LB box were LFGL = 90 dB, G = 6.76, PI corner = Int
2. Adjusted gain on the LB box to G=5, then back to 6.76 after observing no change
3. The PDH signal was being split with a BNC T junction after the lowpass, sending one end to the LB box channel A input (50 Ohm) and the other to the oscilloscope (1 MOhm). I removed the T junction and sent the PDH signal directly to the LB box, and instead sent the LB box's 'Error Monitor' output to the oscilloscope (still 1 MOhm). I observed no significant change
4. I noticed that the current drive output of the LB box was being split with two T-junctions, each terminated with 50 Ohm, resulting in the LB box's output driving 25 Ohm. I removed the unnecessary T junction, so the output is now sent to a 20 dB attenuator, followed by a T-junction with one end terminated in 50 Ohm and the other sent to the current drive mod in.
   • The cavity can now remain locked for about 10s
   • After reducing the gain to G=5.70 and tweaking the laser temperature, the PDH error signal is not quite railing (oscillating at 142 kHz with 500 mV pkpk)
5. I realized I should have changed the input impedance for the oscilloscope channel monitoring the PDH error signal in step 3. 'Error Monitor' output wants to drive 50 Ohm.
   • After adding adding a T-junction to 50 Ohms between the error mon output and the 1 MOhm scope, I observed 300 mV pkpk oscillations at 142 kHz and stable locks for up to a minute.
6. I removed some extraneous BNC T-junctions (were just open ended), and disconnected aux out from the back of the LB box. No change in the 142 kHz oscillation.
7. I next checked the LB box offsets by
   • turning off the servo and terminating both inputs with 50 Ohm.
   • Tuned the input offset to null the LB box error monitor point (viewed on Moku scope) with the servo off.
   • Also tuned the sweep offset to null the LB box output with the servo off, though this should not matter for the in-loop behavior    
   • No change for the 142 kHz oscillation

That's all for me this morning. I think the oscillation is sufficiently low we could try using the DAQ to feed back to temperature as we were before. It would be useful for diagnostic purposes to maintain a more extended lock, and I'm finding I need to tune temperature anytime I reacquire lock. Maybe we're just always sitting close to the edge of the current control loop.