Schaublin 125-CNC retrofit.

That is how most Chinese engineers seem to draw ground shielded cables, not sure why, but they do it. So if you see that on Chinamart schematics, it's usually the outer woven ground shield jacket, then attached to ground. Sometimes they bundle a jacket on each twisted pair, sometimes not. Depends on what day of the week and which supplier had cables. Solar panel stuff is really wild at times !!!

That is the nice thing about standards, there are so many to choose between

I am thinking it is best to remove LinuxCNC from the loop, replacing it with a 1/2V PSU and switch,  and first tuning the driver, and only add the LinuxCNC position loop and tune that using Tommy's excellent servo tuning guide.
I find the instructions in the drive manual difficult to understand or follow.So if anyone has already set up a JMC servo with analog (Speed control) input, I would love to hear and suggestions you have.
Mark

Hi. always managed to get acceptable results without fiddling with the drive velocity loop. You should be able to get it fairly close by only tuning the linuxcnc pos loop.

Keep in mind that i havnt used JMC servos yet, but i have used a number of chinese brands.

But if you do want to exclude the linixcnc loop, i would do it like this....

In your hal file, connect the position cmd to the position feedback so lcnc thinks it os always in position and ignores the encoder feedback.

Then in your PID loop, make p=0 and give a value to ff1. You will still be able to see the encoder velocity and position in halshow and halscope

Now you will be able to jog with the full load but with no feedback for the position loop.

If you look at halshow encoder velocity while jogging, you can tune ff1 untill the reported velocity from the encoder matches the lcnc jogging speed. That should be what the final ff1 value should be.

Wrong direction.....change the encoder scale sign and/or the output scale sign. It womt run away because there is no feedback and no p.

To tune the drive, now set the acceleration in the ini file to near infinity....99999999 will do. If you now jog, it will seem like a step chamge signal to the drive and you can tune it by looking at the encoder velocity in halscope.

it could be that the 0-10 v needs to be reversed if that's the case, the pulse train timing coming in 180* out of sync will certainly cause motion faults. I have the t shirt for that one. IIRC you can change the encoder direction in the configuration utility pnconf

Since you inspired me to start my own Linux CNC mill project i feel obligated offer you some help if you want to. While i have no specific Linux CNC experience i do have quite allot of motion control tuning experience on several industrial motion platforms including designing complete drive concepts from start to finish. Next to that i have a master degree in systems & control theory. If you want we can schedule a Skype call and maybe i can help you understand the manual and explain you the basics of the analog tuning concepts. The only thing that i want in return is that you tell me what info is new to you and possibly this forum then I will write it down so that everybody can use this.

Thanks guys. I tried reversing the encoder but didn't include it in the video, as it gave similar behaviour in both directions.

Henk, thanks for the settings needed to bypass the LCNC loop. That is great idea.

Wout, thanks a lot for your generous offer off your time. I will try to set things up with Henks instructions, and come back to you to set up a call thanks.
Mark

reversing the encoder - not sure how you did that but keep in mind there are two things reading it and you can have it backwards on either.
Since you motors jog properly by the motor drive, the wiring is fine between them.

When you jog (with the drive controls) towards the chuck in Z, does the feedback count down in linuxcnc? It should.
You can set that by negating the feedback scale.

Then you need to check if the linuxcnc jog buttons move the axis is the right direction.
you can negate the scale to change that.

The target is having the linuxcnc positive jog button to move the axis in the positive
direction and the feedback to count in the positive direction.

Otherwise PID tuning is useless.

There is a openloop test in pncconf to help with this (if you used pncconf)

Chris

in that case, try turning up/down the maximum velocity in linux CNC config, see if it is a issue of acceleration versus actual