Joint 0 on limit switch

In your config's main hal file, can you check for a line something like this line:

net x-vel-fb => pid.x.command-deriv

It should be changed to:
net x-vel-fb => pid.x.feedback-deriv

The other axes are similar.

This is a bug that was fixed awhile ago but has not been released yet.
It makes good servo tuning impossible.

Chris M

Whooha

net x-vel-fb => pid.x.feedback-deriv

That has completely freaked out my motors.

Ok, For testing I'm using some ex-printer motors. 15 volt, 4 amp with 500 cpr encoders. They are quite light weight motors but very good quality ones (Cost me a whole $10 from a computer wrecker. I bought 12 of them)

What do you suggest I use to start with for the motor tuning.

With net x-vel-fb => pid.x.command-deriv and P=1 I=0.05 D=0.01

it was quite good.

I will let others counsel you on PID tuning (it's kind of an art).

Though it might help to know your analog max scaling ( A PNCconf term for PWM scale)

It's scaled to 10 by default which is voltage.
And I think you are using velocity mode Copley amps that have there own feedback ?
In other words the encoder or a tach is connected to the Copley amp?
and Linuxcnc sees it's own encoder or a copy of the amps encoder?

Chris M

net x-vel-fb => pid.x.feedback-deriv

That has completely freaked out my motors.

This is the correct connection, if it freaks out your motors it
is because you have too much D term
(you may not need any if the drives have tachometer feedback)