It doesn't sound pll-related if you get clunks when idle/standing still.
You can try and record this. If the clunks are from commanded position disruptions
you should see it as disruptions in position feedback.
You can record long sequences to try to catch it. Use the "sampler" hal component.
Possibly add a ddt component to take the derivative of the position feedback.
It will show some spike/peak when a clunk happens, if there is something with the position.
Plot the data, and evaluate.
