Servo Tuning Help

I disabled the drives internal PID so it is done externally. The drives are in velocity mode, that I am sure.

The problem is that the drives are not tuned to the motors either. If I run the drives with the PID enabled in them, and enable the drive, I hear them oscillate so I know they are not tuned either. I have no way of looking at the internal pid to see the signals, I can only go by ear. There is a rs485 and rs232 connection so I can remotely tune it, but that would be complicated to set up and I still would have to go by ear, so I can try with the front panel buttons and adjust a bit.

That is why I disabled them and tried to do it with Linuxcnc but I cant get them to stop seeking. They are always seeking too much.

I will upload the manual, There is alot of options......

Seems I cannot upload the manual, file exceeds admin's maximum file size.

Could you post a link to some place we could download the manual (such as the manufacturers web site)

If the velocity loop isn't tuned correctly getting a good tune of the position loop in Linuxcnc may be next to impossible.

The manual is located here. KRS Servo drive Manual

Pn146 page 80 is set to 0, which sets the PID external, from what I gather.

Then it becomes closed loop with linuxcnc and the PID control is via linuxcnc.

Am I wrong in this?

Wow, that is a fun manual to ponder on.
Make sure Pn002 is 1
I would agree Pn146 should be set to 0, But I would not have guessed it has anything to do with PID from the manual description.

I think you may still need to tune parameters Pn153-156

To attempt to tune them I would use Linuxcnc to monitor the velocity of the motor and if possible the velocity error (depending on how you generate the velocity command for the test)
Try to rig up a way to command a small square wave velocity command to the motor (60-200 rpm or so) for a second or 2 in each direction. Make these small enough so there is no danger of hurting your machine or crashing into an end stop. Then tune the parameters to get the best, smoothest, most accurate response to the above open loop input.

Thanks...... Got it done. You were right Todd. We had to change a boat load of parameters depending on what the signals were doing.

We got it pretty good now. We didn't realize that when we opened up the loop to linuxcnc, that the settings internal to the drive still applied. That's why we couldn't get linuxcnc dialed in.

On to getting feedback into the slave drive.... missing that somehow. Cant find any of the older hal files that didnt use trivkins or gantrykins.

Anyway onward we go....

i have the same chinese drives like you and the manual sucks..
Have you found how to tune these drives with linuxcnc ?
My problem is what settings must change in drives.
If you found any solution please help me..

Sorry to bring up an old thread, but I'm having exactly the same issue with one of these controllers. I'm using it as a spindle drive and the stability is dependent on the mass of the tool I'm using - and never good enough.

I'd agree that Pn153..Pn156 are the obvious culprits (also Pn115 and Pn116) but I've tried altering all of these across the full range of allowed values and none of them make the slightest difference. it's as if these parameters don't actually work and all the loop gains are permanently hard wired to values that don't quite work for my application.

Are there some other parameter values that have to be altered to get these to work? It's very frustrating as is since everything else is working as I want and the loop behaviour is not far off at all. in fact with smaller tools I'd get away with it if I really had to.