[speedydumpster]

All,
Thank you for your suggestions and help. I was finally able to "tune" my servos in torque mode. I have attached some pictures of my tuning for both my X and Y axis. Problem: I machine a rectangle at 50 ipm - the rectangle comes out correctly. I even did this under load by taking a .05" DOC pass using a .75" cutter. The rectangle cut just fine at 50 IPM. If I turn the feedrate up to 100 ipm then the result on the mill looks like a parallelogram (about 15 degrees off normal). You can see where the Y axis starts moving before the X gets all the way to the end of the rectangle. I did a stress test going back and forth in the X axis to see if I was loosing steps, and I didn't see any indications of loosing steps on my dial indicator. I would like to be able to run at 150 ipm on this mill. I have 2000 ppr encoders. Any suggestions?

Also, if anyone has any tuning tips I'm willing to learn. This is my first time tuning a system, and when I was reading I heard that you can get it to be +- one encoder count. Right now I'm at +-12 counts.

Thank you for your help!
Sincerely,
Dave

[TomKerekes]

Hi Dave,

I suspect what is happening is that your system (Motors, Amps, or Supply) is just not capable of the feedrate you are commanding. If a trajectory is commanded faster than what the system can move then the motors fall way behind the trajectory while going as fast as they can. Eventually when the trajectory slows down then the motors catch up to where they are supposed to be.

Normally this would cause a following error fault but it looks like you have the Max Following Error set at 1 billion.

You should test high speed moves using the KMotion Step Response Screen. The Plots you posted were of very tiny moves - 4000 counts. Your system has very high resolution - 160,000 counts/inch, so this is only 0.025 inches which is not enough distance to accelerate up to any speed.

Try much larger moves and you should see the problem clearly. You can plot Velocity to easily see what speeds you are getting up to.

Your Max Jerk setting is quite low. The numeric value is almost the same as the Max Acceleration which means it would take almost a full second to apply full acceleration. A more reasonable value might be 50X the acceleration which would apply full force in 1/50th of a second.

You will probably see that for large fast moves that the output (green plot) is saturated at max output of 10V yet the motor continues moving at a slower speed. There may just be a setting in the Amp limiting the speed.

Your Bode plots and small moves look quite good. I wouldn't expect following errors of +/- 1 count. 10 counts with 160,000 counts/inch resolution is only 60 microinches.

HTH

[speedydumpster]

Tom,
As always, you have been very helpful and I appreciate it.
I have a few questions that may help solve these issues.
1. What is the maximum output supposed to be? I am running in analog mode, not in counts, so basically i want the voltage to go from -9V to +9V (that way I don't hit the 10V limit, clip, and go unstable). So what is the Output supposed to be under the Max Limits section?
2. The error - is 200 usually a good value, or do you usually set this to a much smaller number?
Thanks again, I'm really glad that you support your product so well. Sorry I'm so dense, but I'm learning. [ATTACH][/ATTACH]
Regards,
Dave

[TomKerekes]

Hi Dave,

Regarding the Max Output of +/- 9V I'm not sure that really helps stability. That just means you would clip even sooner. Since in your case the Output is commanding torque you would want the limit such that you might limit the torque/current/forces in the system to safe levels. The idea is that if under normal operating conditions the system never needs to output beyond a certain level then you can set the max to that level so that if anything goes wrong the output will not go to a level higher than this.

Regarding Max Error of 200: that should be a good value. Under normal conditions the error should always be less than that amount so the limit will have no effect. Only in abnormal conditions like hitting something or commanding a speed higher than possible would the error grow to bigger values, and in those cases limiting the error to a low value of 200 rather than letting it grow to a huge value might prevent a violent correction attempt. With a Max Following Error of 200 the system will disable as soon as there is an error of 200 so the Max Error will never come into play.

So as you can see in your big move plot. The output maxes out at 2000 (~+10V) and the velocity maxes out at ~ 370,000 counts/sec (138ips) which is much lower than the trajectory of 500,000 counts/sec which is why you see huge errors.

Regards
TK