We want faster PID performance!

We’ve been enjoying using the Moku:Go in our lab. We recently purchased the Laser Lock Box instrument, and it’s really nice.

However, we were surprised to learn that the PID performance* of the Moku:Go is limited to ~20 kHz. Even if we upgrade to the Lab or the Pro, the performance is limited to 100 kHz or 120 kHz respectively. Increasing the frequency would be critical for many laser-locking applications.

Is it possible to increase the frequency at which the PID controller can provide a low-phase-lag response?



*The “performance” of the PID is limited by the phase lag at higher frequencies. The frequencies listed above are the frequencies where the phase lag is specified to be less than 30 degrees in the specification sheets for the Moku units.

Hi Dan, sorry for the slow reply but thanks for the feedback! You’re absolutely right, the limiting factor for the PID performance is the latency of the instrument as a whole. For the Go, the latency is mostly in the decimation chain that reduces the sample rate to something manageable for the Go while maintaining good noise performance. For the Lab and (in particular) Pro, it’s limited by the latency of the communication busses to the ADCs and DACs themselves, which prioritize throughput over latency.

So: We’re looking at ways to improve latency in the Go at the cost of noise performance, I can’t make any promises but we are actively looking in to it. For the higher-end products, this same effort may bring some incremental improvements but it’s mostly hardware-limited.


Thanks for your reply and for the information! I appreciate the insight. Hopefully you can make some improvements to this in future products. Getting PID feedback up to a few hundred kHz would make a big difference in our lab. If you can reach ~1 MHz, then the Moku devices would be directly competitive with the best analog servos (Vescent D2-125, Newport LB1005).

Useful note, thanks Dan! We’ll keep you posted.

I think it would be very difficult for a digital solution to truly compete with an analogue controller for fast laser locking. Even with an extremely fast Toptica FALC (15 ns delay) we got a phase crossover of just about ~3MHz. And we had to put the controller right on top of the laser to keep the conduction delays small. But it would be really great as secondary control loop to compensate and monitor medium/slow perturbations!
We got a Moku:Go in the hope of using them for fiber-noise cancellation, but 20kHz bandwidth is just a bit too slow for that. It would be great to get it up to around 50kHz maybe, that would make the Go quite versatile already!

I agree that digital is unlikely to beat analogue in terms of pure performance. But, the best analogue laser locking electronics (to my knowledge) have a phase crossover of about 10 MHz, which seems attainable using fast FPGAs. For example, using the Linien software on a Red Pitaya, I recorded a phase crossover of about 4 MHz. And the Red Pitaya is only $400. So, I can totally envision a future where an affordable and flexible FPGA-based product could offer several MHz level locking PID performance that would directly compete with the best commercial analogue lock-boxes currently available.

Thanks TiSapph, Dan,

Yeah the FPGA is typically capable of single-digit-MHz crossovers if you try nice and hard, but it is typically delays elsewhere in the system that limit you. Paradoxically some cheaper devices can have better phase crossover performance because if you’re dealing with low data rates, you don’t need to deal so much with deep buffers, error detection/correction etc like you do with high speed converters. We’re definitely looking at options here, we’d love to at least be able to push our Laser Lock Box “fast” path up in to the MHz and if we can do it for that instrument, we can imrpove the standalone PID too.