scurve trajectory planner

Hello! I have intersting problem. I use solidcam iMachining procedures for milling. With ordinary TP I have some strange deaccelerations and my servos don't like it. Then I check your tp, it try to stop every g-code segment. I can't understand where is problem. Help me, please. White screenshot - new tp, black - ordinary tp. All tests on pictures are in simulation.

Hi Evgen,

If you try to do roughing G64 P1 Q0 or even G64 P2 Q0
The velocity will be more constant. And the path tends to be G3 continious.

My conclusion for the file:
From what i see is that the file set a velmax for a lot off indiviual G1 line segments. The planner will react to this.
Some off the line - line segments transitions in the file are very sharp, non smooth. Tiny clothoid fillets, will slow down the
motion.

Hello! I try P1. The movement becomes smoother, but still srange. And also srange deaccelerations in traditional tp at moment when movement is smooth and feedrate doesn't change. same programs an non linuxcnc machines works fine. I have such problem on much 3 many years ago. The reason was CPU usage.

Are you sure that is not just an artifact of running a non-realtime (SIM) installation?

Yes,I'm sure. Firstly I got this problem on real machine with MESA card.

Can you post the gcode?
 

Hi,

When i run Posix non-realtime or Posix realtime scurve planner, i both get the same results as Evgen is telling us.

The hal pins that are flippering, blinking :

• motion.feed-mm-per-second
• motion.feed-mm-per-minute
• motion.feed-inches-per-second
• motion.feed-inches-per-minute
• motion.feed-upm

I didn't look any further.

What is using the hal pin :  motion.feed-mm-per-second for example?
I personally never used it, that's maybe i didn't noticed this behaviour.

@Pcw, link to gcode he posted before:
forum.linuxcnc.org/media/kunena/attachme...i3DRough_target2.ngc