Has anyone looked at the source code to verify that? I don’t believe that’s the case.
Can you attach a G code program and your INI file that is demonstrating that behavior? My guess is you have lines of G code that are commanding such small movements that theoretically multiple lines of G code should be evaluated in a single polling loop of the servo thread. If a move is supposed to take less than 1ms (I’m guessing that’s your servo thread period), it will still take 1ms because it only evaluates one command per loop. We have encountered this limitation on our machines, but there are work arounds to improve performance. Generally, you don’t need movements that small so the G code output can be changed to issue larger segments. You could also try using G64 to eliminate segments that are very tiny. If you have a computer capable of it, you could even try lowering your polling period. Again, looking at your G code and INI could help verify this theory. > On Jul 7, 2023, at 1:17 AM, Marius <[email protected]> wrote: > > >> On 2023/07/06 22:54, John Allwine wrote: >> The latest software on the Pocket NC is using the same trajectory planner as >> what is in LinuxCNC. >> >> I haven’t followed the whole thread. What seems to be wrong with the >> trajectory planner when doing 4 or 5-axis? > > The trajectory planner only has one line look ahead for any rotary axis. So > this means that trajectory planning is not really happening and performance > is very poor when trying to achieve faster moves. > > > >>>> On Jul 6, 2023, at 12:34 PM, Eric Keller <[email protected]> wrote: >>> >>> The Pocket NC runs Machinekit. I'm told machinekit has a better TP, >>> but I haven't seen any evidence of that. >>> >>>> On Thu, Jul 6, 2023 at 12:27 PM Roland Jollivet >>>> <[email protected]> wrote: >>>> >>>> The small 4-axis CNC machine,'Pocket Nc' uses LinuxCNC as far as I can >>>> tell, and it seems to run just fine. Or does it? >>>> >>>> Roland >>>> >>>> >>>> >>>>> On Tue, 4 Jul 2023 at 20:16, andy pugh <[email protected]> wrote: >>>>> >>>>>> On Tue, 4 Jul 2023 at 18:34, Marius <[email protected]> wrote: >>>>>> The problem seems to be that our trajectory planner lack the ability to >>>>>> look ahead for any rotary axis. This means that a coordinated move that >>>>>> includes a rotary move will be dysfunctional at best. >>>>> It's not as bad as that. >>>>> >>>>> There is a 9-axis blending planner in Tormach, but it's not trivial to >>>>> merge it with Linuxcnc due to divergence in the codebase. >>>>> >>>>> If we could get hold of the patches that introduced it to Tormach then >>>>> we could look at integrating it, but doing it just from comparing the >>>>> code sounds hard. >>>>> >>>>> -- >>>>> atp >>>>> "A motorcycle is a bicycle with a pandemonium attachment and is >>>>> designed for the especial use of mechanical geniuses, daredevils and >>>>> lunatics." >>>>> — George Fitch, Atlanta Constitution Newspaper, 1912 >>>>> >>>>> >>>>> _______________________________________________ >>>>> Emc-users mailing list >>>>> [email protected] >>>>> https://lists.sourceforge.net/lists/listinfo/emc-users >>>>> >>>> _______________________________________________ >>>> Emc-users mailing list >>>> [email protected] >>>> https://lists.sourceforge.net/lists/listinfo/emc-users >>> >>> _______________________________________________ >>> Emc-users mailing list >>> [email protected] >>> https://lists.sourceforge.net/lists/listinfo/emc-users >> >> _______________________________________________ >> Emc-users mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/emc-users > > > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
