[funding33]

Hello,

I'm using kflop with servo drives which takes STEP/DIR position input. The Kflop is set for STEP/DIR output but this seems to invalidate any PID loop parameters.
Whatever is set in P,I,D - output allways seen 0 in step response screen and no effect.

From Step response it looks like there is constant delay from output to actual position. Attached is proof how 2x changed velocity produces constant 2x changed position error.
I was willing to tune it out - but P,I,D just does not work.

What is the correct way to go in such situation?

[TomKerekes]

Hi,

In Step/Dir mode the Controller only commands how to move. The manner the motor actually moves entirely depends on the Drive. KFLOP only monitors how the motor moves. Thousands of counts of following error seems very large. What kind of system do you have? Where is the encoder? What is the resolution? How is the axis configured?

If you want to have KFLOP make corrections to what the Drive is doing use Closed Loop Step Direction mode. See here.

[funding33]

Thanks for tip. Tried "CL Step" output but it does limit pulse frequency to preset kflop max velocity and this solutin has not helped as seen in new screens. The input/output is scaled to real world mircometers with 1 encoder count= 2.22um
The gantry has 40mm pitch leadscrew plus 20:36 timing belt reducer from motor. So 40mm==1.8 motor rev = 18000 encoder counts. Mechanics very rigid - retrofit from hi end machinery.
The drives in question is chinese AASD-20A and motor 90ST-M04025 with 2500 line encoder. Estimated lag is 16ms. Is it realistic for such size drives?

Would it be a way for kflop to put Output as pulse frequency? This would make drive to work as it would be in speed mode and so kflop could be tuned to close position loop?

[TomKerekes]

Hi,

The plots show the corrections being limited to +/- 200 counts so with thousands of counts of error it isn’t doing much. I think this is caused by the huge Max Limit Output of 1e22 which is outside the expected valid range and is being rejected and ignored. Change to something more reasonable like 1e5. See the Console Screen for error messages.

But before this I would investigate why the Drives are doing such a poor job of Following properly. I would expect them to be orders of magnitude better.

[funding33]

Hi,
Yes, the MaxLimit did the trick. Now was able to reach +/-50um error which is +/-22 encoder counts. Could that be considered normal/typical? To me it looks reasonable and I tended to accept this.
Surprisingly biggest effect and help gave Feed Forward values of V and A. How can this be explained? Does it tell something where and what to investigate in Drive parameters?

[TomKerekes]

Hi,

I don’t recall you telling us what kind of system this is, but the motion is quite aggressive. 900mm/s with 5G acceleration? So 50um of error might be acceptable.

Looks like more A FF might be needed as there is lag during Acceleration and lead during Deceleration.

I gain may be low as there is a small but persistent error after the move.

I’m not familiar with your drives. Possibly high D gain without high P gain may cause a lag proportional to velocity.