PID tuning by Graph

I have found a way to understand PID tuning that is easy for me to understand and is working well.
While running a program that moves an axis back and forth and watching the f-error graph I experimented with the effect of adjusting each setting until I could see what each one did to the f-error.

Knowing that made it easy to know what setting to adjust to reduce the f-error

The image below shows what effect adjusting some of the settings has on the shape of the f-error graph.

I hope this is helpful to someone.

hmm this would be good info to add to the manual as well as the pics from here:
wiki.linuxcnc.org/cgi-bin/wiki.pl?PWM_Servo_Amplifiers

Or something similar.

What ya think John?
PID tuning still seems like black magic ....

Chris M

Yah, I was really frustrated by it when i was following the text instructions.

It doesn't seem like a black art at all now.

Maybe the text meant the same to some people as the graphic I posted.

Using this method, I could tune an axis on my machine in 10 minutes maximum. The second axis in 2 minutes since the graph was already set up..