router spindle PWM control (open loop)

I have a hard time keeping track of all these variables, I really need to cleanup my hal

I some following interogations, I've setup a lincurve comp to offset my 6000-18000 input to a 0-18000 output which works fine, but the PWM response seems very un linear. For 0-5v out from the DAC seems to correspond to the 6000-6400rpm range and than everything above that requested speed (so 6400rpm+) results a variation from 5 to 10v (out of the DAC.
I had the same issue before adding the lincurve. any idea what might cause this?

unrelated to my spindle setup, but is there a way to setup the scale for the step gen as an expression? I have a belt setup resulting in 256000/78steps/mm or
328,205128...etc and I feel this can only cause dimensional inaccuracy

For linearity issues I would take lincurve out of the data path and just check
straight linearity.

The behaviour you show may be more reflective of how lincurve's table
parameters are setup than your PWM--> Analog circuits linearity.

If the PWM --> Analog circuit is isolated, a lot of non -linearity
is caused by asymmetrical OPTO switching times. You may be
able to improve this by lowering the PWM frequency. (if too low
you will get more ripple)