Pedersen VP2000 810M retrofit, severe servo hunting

Z movement is achieved by lifting the bed rather than the spindle via a central ballscrew, and as far as I am aware there is no counterbalancing.

I tried a few different BIAS value (0.001-100, as well as negative values) without seeing any change in the oscillation. At values above 10 the machine jerks when motion is enabled, which is to be expected I guess.

Further, increased Integral values decreases the period of oscillation but seems to be necessary for a low f-error. I have a tune (something like P=10, D=0.1, FF1=0.01, FF2=0.001) which produces very little hunting (more like 0.002mm rather than 0.015) but has a completely unsatisfactory f-error while moving.

Lastly what I found a bit odd is that the hunting shows up as instant on the DRO, it basically jumps 0.015 up and down repeatedly without any inbetween.

andypugh wrote:

What is the physical size of a single encoder count?
You might need to add a little deadband to the PID.

The slides are graduated to 0.004mm but it's hardware interpolated to 0.001mm. The latter number is the deadband set on X and Y and it works well there. Raising it leads to more severe hunting on those axes.