Axis jerk, servo tuning and capturing data

[GabrielR922]

Hi Tom.

One more feedback.

Here goes the output for 4m/min. (Very bumpy, i can't hear any noise.)

4.5m/min and 5m/min goes very smooth.
Also for 29m/min.

[TomKerekes]

Hi Gabriel,

I had the Y axis following error set to 10.000, lowering it to 100 solved the problem. (my system is tuned for around 50 counts).

Did you mean raising it from 10 to 100?

Check the attached photo. i need to lower the Gain to raise D. what can i use?

I don't understand the question.

is ok to use for example a low-pass filter of 300 ?

I often use a 2nd order low pass filter of 500Hz Q=1.4 in the last Filter #2 and clear #0 and #1.

[GabrielR922]

Hi Tom.

Did you mean raising it from 10 to 100?

No. I mean lowering from 10000 to 100.
With a max following error of 10 my machine won't even move. And will trigger a following error in the console.

I often use a 2nd order low pass filter of 500Hz Q=1.4 in the last Filter #2 and clear #0 and #1.

Did you find any anormal motion in the above graphs?

I can post a photo of the part with a lot of bumps if you need.

Thanks!!

[TomKerekes]

No. I mean lowering from 10000 to 100.

I don't see how that would make any difference.

Did you find any anormal motion in the above graphs?

I see an oscillation at ~ 17Hz

[GabrielR922]

I don't see how that would make any difference.

Yeah, me too Tom, i can further test to see if it happens again, but that was the solution.

I see an oscillation at ~ 17Hz

So, I should use a pole-zero filter to increase gain in this region ?

Do you have any suggestion to what I can do?

I already tried several options and can't find a way to solve it.

Thanks

[TomKerekes]

I already tried several options and can't find a way to solve it.

What have you tried and what were the results?

You might do some math to determine what is at that frequency. Your system may have a resonance at 17Hz. You might hit the axis with a rubber hammer and check if it oscillates at ~17Hz. You might also do this while capturing a Plot of size zero. And it may have an unbalanced lead screw or something that excites this when rotating 17 revs/sec. What speed is the Motor or leadscrew rotating at the problem speed of 4m/min?

I doubt if a filter would help.

[GabrielR922]

Hi Tom,

As my knowledge in this is limited, I tried different axis tuning, modifying PID to different values while maintaining error small. And all tunings resulted in marks on the finished part. (tried different tools and holders to verify also, with different speeds I got better results. even faster speeds.)

The axis has a resolution of 15880counts/inch, so ~625 counts/mm, 4000mm/min gives around 42000counts/sec. It also uses a rotating ballnut design, so it removes the unbalanced lead screw option. to capture a plot of size zero, is the same that done before capturing the following error ? Or I need to use a different setup ?

Sorry for the lots of questions and problems, this system gone through a lot of problems, a lot of backslash in this axis, all have been rebuilt and adjusted for perfect performance now. the tuning of the axis are very good and the trajectory planner is working beautiful.

Also KmotionCNC is working very fine and doing a great jog. The only thing missing is a "console" in kmotioncnc itself, that could print 2,3 lines of errors or reports about the machine.

Thanks!

[TomKerekes]

What speed is the Motor or ballnut rotating at the problem speed of 4m/min?

to capture a plot of size zero, is the same that done before capturing the following error ? Or I need to use a different setup ?

Same method. Just set size to zero.

[GabrielR922]

Hi Tom!

The axis pitch is 40mm, 4000 / 40 = 100 rpm. It also has 2 pulleys, 2.5 to 1, 100 * 2,5 = 250rpm.

Is there a way to take away this resonance ?

Thanks!

[TomKerekes]

Nothing seems to match the frequency. Unless its some multiple. 250RPM / 60 x 4 = 16.7Hz. You might also try double or half speed.

I doubt if the servo can correct for that at 17Hz. As a rule of thumb to correct for errors of a frequency the servo bandwidth needs to be like 10X which would be 170Hz in your case. I doubt if you can achieve that.

Like I suggested you might run experiments and see if you can identify the cause and if it really is a resonance and not servo instability. Servo instability is not usually triggered at a particular speed. You might reduce gains a lot to make a very stable (yet not very accurate) system. If the oscillation is still there then it is not likely servo tuning.

A resonance is normally caused by a mass (actually 2) and something acting like a spring. Your belt might be a suspect.