Spindle PID control

Hi,

I have a retrofit lathe with Mesa 5i25+7i76 and Omron 1000ppr ABZ spindle encoder. Spindle speed is controlled with 0-10v to VFD, but the speeds are wrong. Voltages coming from 7i76 are correct, so I think there's some non-linearity in the VFD control. At lower speeds the actual speed is too high and on high rpm range actual speeds are lower.

For few days I've read about the PID control and different threads about it, but so far I haven't really understood how it should work.

The configs are attached. I read somewhere that for a start I should change the FF0 until the speed is sort-of-correct and then follow with I and P. But changing these values doesn't change anything for me. Currently the configs are as they came from config wizard, no my changes present.

How should I proceed with debugging and fixing it? It looks to me like the PID "process" should be attached and processing "stuff", but I'm not getting the results...

Note these three lines in your hal file: (reorganized for clarity)

net spindle-output <= pid.s.output
net spindle-vel-cmd-rpm-abs <= motion.spindle-speed-out-abs
net spindle-vel-cmd-rpm-abs => hm2_5i25.0.7i76.0.0.spinout

These mean that the analog output is set by the spindle speed from motion
(motion.spindle-speed-out-abs), not from the PID output

To have the PID output set the analog voltage, you would need to change
net spindle-vel-cmd-rpm-abs => hm2_5i25.0.7i76.0.0.spinout
to
net pid.s.output => hm2_5i25.0.7i76.0.0.spinout

Allright, I think the logic is dawning now.

There's also defined:
net spindle-output <= pid.s.output

So I could just as well do this, right?
net spindle-output => hm2_5i25.0.7i76.0.0.spinout

Just setting variables as the arrow is pointing...

Yes, you are correct, my example would not work as you
cannot have a pin (like pid.s.output) used as a net

net syntax is:

net net-name pin pin pin pin ...

Note that the arrows are just for looks (they dont do anything)

Well, I tried the config changes out, but something is still wrong and I have no idea how to approach this. Currently config states:

net spindle-output => hm2_5i25.0.7i76.0.0.spinout

But the feedback loop behaves strangely. Whenever I set only P value, the spindle speed is changed to static value, either higher or lower from the commanded one. I think that is expected.

However, if I set something for the I value, the speed is ramped up to the max (in case of positive I) or min (negative I), it doesn't settle to the commanded value. Speed of change depends on the I value. I tried very small I value to see if the rate of change somehow messes with the "settlement", but nope. I will go to max or min, no matter the I value.

D value didn't seem to do anything. Or I didn't know which combination of the PID to try.

I poked around a little with halmeter and got these values for different pid.s values


That was for S200. pid.s.feedback seems to be always 0, is that correct?

lincurve could be an option indeed, but PID would be much awesomer

Thanks, that makes sense.

To see if that really made sense, here's the spindle speed part from my current config

So, to connect the abs spindle speed to the pid feedback I'd need to do this:
Or should I use the non-lowpassed abs value for feedback?

I got this working with lincurve, that is nice.

I tried many-many combinations to get the PID control loop working, but to no avail. Just couldn't get it stable. Method I tried was to increase the P until output started oscillating (happened at around P=4) and the started increasing I to get rid of the oscillation. But increasing I never got rid of the oscillation, and at some higher levels it started to increase it.

Was this method wrong somehow? How should it actually be done and what are the approx numbers the P, I and D? 0.1, 1, 10, 100, 1000? I have no idea.