Axis short stop (very short) about every 15-20 sec. when cutting circle.

Thanks PCW for testing it on real hardware!

So, it looks like the g-code causes the problems.That would be great, because that means that at least i have no problem with my hardware.

Unfortunately i dont get HAL Scope to work. Will try that again.

I think your little bit of code has uncovered a bug in the new trajectory planner.

If I run the code a little faster I am getting the pauses as well.
If I disable the new trajectory planner the code runs smoothly with no pauses.
(to disable it and revert to the old planner can be done with an ini file setting.)
linuxcnc.org/docs/2.7/html/config/ini-config.html#_traj_section

You guys are awesome. I always get help pretty quickly here. THANKS!
I was again stressed since yesterday, because i had so many problems and i thought there is the next one...

Nice to hear that my "litte bit of code " has uncovered a bug
I am not any more in the workshop right now, so on tomorrow i will disable the new trajectory planner, but i hope that is correct:
ARC_BLEND_ENABLE = 0 (1 is default)
ARC_BLEND_FALLBACK_ENABLE = 0 (0 is default)
I hope i am right that this are the only commands i have to look for or write in the ini if its not there.

Now i am very courious if my accuracy will be better when cutting again the circle without the new trajectory planner. Lets see, i will test it on tomorrow

I just tried with arc blend enable off and I get the same velocity drops

So if its a TP bug it affects the old TP also

The gcode does look a bit insane however...

I'm having trouble putting my finger precisely on offending lines of g-code that are triggering the bug. It seems to require a sequence of lines to trigger it. If I pair it down to less than about 5 lines of code the pause I am zooming in on goes away. But I think I have it narrowed down to what triggers it. If there is a helical arc followed by three or more regular arcs, there will be a pause between the first two regular arcs.

Here is a simplified g-code file that reproduces the bug.

New TP enabled


New TP disabled with ini setting ARC_BLEND_ENABLE = 0

is it possible that just the switch between helical and plain arcs needs to slowdown to avoid a Z axis accel limit?

Old TP around line 73 in gcode:



Not clear whether this is a TP bug or just bad gcode

Those two tiny velocity blips with the old TP are not the same as the full stops showing up in the new TP. (Nor are they in the same location in the G-code.)

I can not get the repeating predictable stops with the old TP as I can with the new one. I can get small slow downs but they look more like the old line length limitation issue that lead to the creation of the new TP.

PCW wrote:

Old TP around line 73 in gcode:

Not clear whether this is a TP bug or just bad gcode

Near lIne 73 (actually line 72) looks like a short line segment slow down in the old TP. (it isn't a full stop like I'm seeing in the new TP.)