Rich,

can you also post the screen shots of test motion graph for each axis.  this will help to pin point the issue lot quicker. 

also, if the PID tuning is not tight enough, the final position will not be close to commanded, or it may take more time to get to the commanded.

Regards,
Rufi
VSI

Here are the motion graph screens with a few PID modifications. I will also post the PID settings screenshots. We have also re-calibrated the Z-axis which changed the "steps per" count, I will post those settings as well. Thanks for looking at this.




Rich

More settings screens...

A few tweaks on the motor tuning as well.

your test motion obviously shows the PID tuning or drive response is not effective at all.  the blue and red lines should be as close as possible, ideally should overlap.   there could be a number of issues:
- PID tuning parameters need to be updated
- Servo Drive Gain setting is too low
- Servo drive and motor not capable of handling the load or velocity settings.

check your drives, they should deliver maximum velocity or maximum torque at +-10volts.

there is a section of pid tuning in the dspmc user manual.  please check that out for some tuning steps.

we can also login to your system to help you tune the pid parameters.  please email to support at vitalsystem dot com

thanks
rufi

To add to what Rufi has said your following error is so large that it is allowing the system to function when it should not. The greater the following error the farther from the true path you get before it faults out.

With 80k encoders on my mill I run a following error of 1k or less.

Based on my experience tuning my machines your "I" value is way too high. I normally end up with a value below 10 and on some systems it could be zero.

HTH
Mike

Thanks for the direction.We have tweaked the PID and achieved smoother movements and the machine is positioning closer but it is still not quite right. As you can see, we sent the x-axis 1 inch and it didn't quite make it there. The pulse count for one inch should be 5079.99999228, that is what the "Cmd Pos" reads but the "encoder" and "feedback" are still off. We have been able to get it to position within 1 thousandth of an inch but not consistently. The very next move may be off by 20 thousandths or it may also be within 1 thousandth.

We are running linear encoders, rack and pinion drive servo motors (Dunker w/15:1 gearbox) with Advanced Motion Control 16A20AC drives and DSPMC interface.



Thanks,
Rich

Hi Rich,
Your graph looks better but you still have a long way to go. Increase the D. I would guess it has to go way up. IIRC the D term is normally larger than the P by quite a bit. The quick tutorial says start with only the P term and keep going up until you have instability. I.E. The servo wont stay still and goes "crazy". don't let it do this for any length of time or damage may occur. Once you get to the point of instability, ad in the "D" term. IIRR D is normally 32 times the P term. There are of course no hard and fast rules on PID tuning but this should get you in the ball park. Once you have the P and the D where the system is running well with minimal following error, then add in small amounts of I. The small amounts of I allow for the system to get to the final position. Too much I and it can take quite a while for the system to get to the final position.

Hope this helps,
Mike

hi Rich,

you should test with short moves like quarter inch or less and at 2 to 3 times target accel and velocity.  this way at regular speeds you get even better results.

regards,
Rufi

I am working with Rich on this issue.  I have followed the manual regarding the setup of the PID settings.  With the last posted settings I can get about 0.001" of accuracy on the initial move.  When I begin adding D to the setup it doesn't seem to change anything at all other than I seem to have to increase my max following error to allow any move at all.  Is there any other settings I can change that I may be missing?  Basically if I zero everything out and tell the machine to move to 5" it will be 5.001" on the first move.  If I send it to 10" and then back to 5" the results are 4.9895" to 5.0195".  I have started over and followed the manuals procedures multiple times.

Joe