Discontinuity in "slow out" signal

[cardavr]

I've been using linien 0.8.0 to lock a laser to a cavity using an externally generated PDH signal. The locking is working well, however the "slow out" port seems to have discontinuities that eventually cause the laser to lose lock. I've tried using both the "Analog 0" port and the "Fast Out 1" ports to provide the slow out control signal, but the behavior is identical. The discontinuities only appear to happen when the "control history" curve has a zero crossing, though it is not deterministic. I've copied a screenshot below showing the effect that we see. Yellow is the "slow out" signal that has the discontinuities, and you can see the impact it has on the green control history curve. In this case the glitches were not strong enough to cause the laser to unlock, but they eventually one will be strong enough. The "strength" parameter does affect this. If it's increased to say 10, the glitches pretty quickly become large enough to rail the output.

[bleykauf]

This should disappear if you use larger values for the strength, think >1000. It thought I added a note about this in the v0.8.0 release but it looks like it was not merged in the end...

Please report back if this fixed the issue.

[cardavr]

Unfortunately I'm unable to test if setting the slow output gain that high solves the discontinuity problem. We're using the slow feedback for thermal control which has bandwidth ~1 Hz. So setting the strength parameter to 1000 is not stable in our system. Strengths of 1-5 do seem perfect for steering the output with the appropriate gain, with the exception of the fast jumps that sporadically show up.

[bleykauf]

I see. Do you have the possibility to attenuate the signal from the slow output with a Minicircuits attenuator or similar? For documentation and future reference: are you using one of the two fast outputs (SMA) or analog out 0? cardavr ***@***.***> schrieb am Mi., 12. Juli 2023, 17:20:

…

Unfortunately I'm unable to test if setting the slow output gain that high solves the discontinuity problem. We're using the slow feedback for thermal control which has bandwidth ~1 Hz. So setting the strength parameter to 1000 is not stable in our system. Strengths of 1-5 do seem perfect for steering the output with the appropriate gain, with the exception of the fast jumps that sporadically show up. — Reply to this email directly, view it on GitHub <#347 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABO645MNZC55HLPW75WJW73XP26DHANCNFSM6AAAAAA2GK4CRM> . You are receiving this because you commented.Message ID: ***@***.***>

[cardavr]

I can attenuate the output, but this appears to not help significantly. It also has the tradeoff of limiting the overall tuning range, which is also undesirable.

For example, here is a screenshot showing the locking behavior with an 8 dB attenuator installed on the slow out port:

I currently have the slow out configured on the "Fast Out 1" SMA port, but I saw the same behavior from Analog Out 0.

Do you have any ideas what causes this?

[cardavr]

Here's one more image showing the behavior with the slow out port disconnected from the laser so that the catastrophic glitches don't cause the system to come unlocked: