Skip to main content
Submitted by ErikT on Wed, 04/06/2011 - 08:26

Hello, I have a bit of a problem tuning.

I have a machine that has a 45 kW servomotor that runs through a 80:1 reduction - end result is it has a heck of a lot of torque, because eventually it will be pulling on hot steel. Right now though I'm just trying to get it to behave properly while empty.

It has a peak torque limit of 9.99 volts and a constant torque limit of 5.0 volts. At 5 volts it is yanking pretty hard, much more than I need to get the mass attached to the motor to move without a part in place.

It will eventually be a dual loop system, but at the moment I am only using the motor encoder. This has a configurable resolution, currently 6000 pulses per revolution (24,000 quadrature) which works out to be almost exactly the same as the linear encoder I will be using with dual loop mode. (I am aware the motor encoder resolution should be increased later, but that would only compound my problem now.)

I'm walking through my tuning process and I seem to be able to get the machine to respond well enough to the step response I am throwing at it (considering the mass it has to move) - KD is about 150 before increasing further results in no change.

Unfortunately the KP and KI parameters prove to be a lot more difficult. KP only allows steps of 0.125. Setting it to 0.250 results in a nice response but I feel as if I could get it better, unfortunately 0.125 and 0.375 are too far away from the sweet spot and increase overshoot a fair bit.

KI is even worse. I need it - accurate position tracking during moves is a must - but if I set it to [b]anything[/b] it causes the machine to rock back and forth around the target position until it finally decides to stop in the right position. Obviously, KI is too high... but 0.001 is even too much!

Any ideas? This is extremely frustrating!

Comments 2

Galil_AndyH on 04/06/2011 - 15:13

This was handled offline. The problem is related to the fact that the amplifier gain is too high. Increasing TM, lowering the amplifier gain and lowering IL will provide a better response.

ErikT on 04/07/2011 - 08:49

Increasing TM did help with the system as it stood yesterday - I was able to improve tuning somewhat.

However, there was another more important issue I discovered - the drives had a 25% torque limit set for both peak and average as opposed to the 200% and 100% appropriate for my motors!

Essentially, I was running the system with a motor that was 75% to 87.5% too small and the PID loop's output was trying to compensate.

I am now getting very satisfactory results with TM back at the default value of 1000. Without changing my linear encoders, I will probably have to double my motor's rotary encoder resolution output with [i]may[/i] result in requiring tweaks to TM later for best response, but I am in a much better mood now that I've discovered the actual source of the problem!

Thanks to DJ at support for the ideas.