PID tuning of Rexroth Ecodrive

He guys,

I have set up my cnc machine with Rexroth Indramat Ecodrive units with +/-10 V inputs.
I have connected the drives with a shielded cable to the 7i33TA which is mounted in an aluminum case .
The shielding of the cable is connected one time with the aluminum case which is connected to ground and on the other side its connected with the drive case which is also connected to ground.

The Ecodrive unit is an closed loop drive by itself with an emulated encoder output which Im using to get the encoder position back to linuxcnc. I set the encoder count per turn output of the Ecodrive to 6000 lines per turn as its possible to use any value for the encoder emulation output.

Now im planing on how to get these drives to work fine for me .
I tried to use the drives in velocity mode but with bad results as i think.

My settings are:

but i cant decrease the following error of this axis any better :
I made a run from 0 to 50mm with the max velocity of 4000mm/min



I think that the error is too large at the start and end of the motion but FF2 does not affect the hight of the spikes showing in the picture... theres also a ripple on the ferror line with a hight of 0,5m the whole time the motor is at position

What other way can I ty to make this dives work fine ? torque mode ?
The ecodrives can handle different input modes as velocity , torque and positioning control

Should i try to set the drive to torque mode ??

Velocity mode should be easier to tune than torque mode

Is the velocity loop of the drive tuned? Thats the first step...

Also any velocity command input filtering should be turned off

@BigJohnT
Okay i also tried with shlieding on just one side but that didnt change anything ...

@PCW
The velocity lopp of the drive should be okay as the Ecodrive seem to use their own parameters for each fitting motor ..

Velocity loop tuning is load dependent so you must tune the drives velocity loop
(with final mechanics)

So I tried optimizing the pid loop fo a fw hous now and cant gt better results than this :





When the motor isnt in motion theres always an ferror of about 1m visible on the scope does anybody have an idea where this can come from?

The spikes while de/acceleration also dont change with my modifikations

Is that following error in millimeters? So that's 2 milli-millimeters or 2 microns per division. And a micron really is not much at all.
That is not bad, or is it?

A couple of spikes at the beginning and ending of the move don't surprise me either; the LinuxCNC trajectory generator does infinite jerk (jerk=derivative of acceleration). Real motors do not do infinite jerk, the connected mechanics are not infinitely stiff and the drive electronics are not infinitely fast.

Your Igain seems to be zero, so that might explain the residual following error. If the ferror stays at 1 micron at rest I would not care too much about adding Igain.

Dear DaBit,

I now added some I gain and the ferror while motor stands still is now significant better
I have now a ferror while no motion of about 200μ which I think is aceptable ?!

But the spikes at de/acceleration wont get any smaller

My actual settings are:



Will it be better to set a higher Encoder rate on the Ecorive unit ? Actually I set it to 6000 lines per turn ..
I also set the drive to 4000 rpm per 10V input signal... its possible to set this value to max 8000 rpm per 10V

Is this a point where I can optimize things ?