G-code bug

Moderators: TomKerekes, dynomotion

GabrielR922
Posts: 79
Joined: Thu Aug 13, 2020 10:07 pm

Re: G-code bug

Post by GabrielR922 » Thu Jun 03, 2021 2:43 pm

Hi Tom.

Again with this problem, with the same configs as before and facet angle of 0.8.

Moving to a facet angle of 0.9 solved the problem.

Thanks
Attachments
tp08.PNG
miguel.txt
(16.82 KiB) Downloaded 40 times

User avatar
TomKerekes
Posts: 2529
Joined: Mon Dec 04, 2017 1:49 am

Re: G-code bug

Post by TomKerekes » Thu Jun 03, 2021 4:22 pm

Hi Gabriel,

Sorry for the delay. I think we have a fix. That Corner Rounding Patch introduced a bug. The determination of whether a segment was completely rounded and eliminated or not had a bug. The inconsistancy resulted in taking the square root of a negative number. The bug was somewhat random as it occurred when the corner rounding portion coincidentally matched the original segment length. So changing most anything would likely shuffle things and avoid the problem.

Thanks for the precise example.

The Corner Rounding patch has been updated: GCodeInterpreter.dll and source if needed TrajectoryPlanner.cpp

Please let us know if you have further issues.

Regarding your earlier question on what facet angle to use: We don't have a simple answer. A smaller value results in in more segments needing to be sent to and processed by KFLOP. For example a facet angle of 0.5 results in 180 segments for a single 90 degree angle. KFLOP can process ~2000 segments/second. Generally we try to keep the value > 1.
Regards,

Tom Kerekes
Dynomotion, Inc.

GabrielR922
Posts: 79
Joined: Thu Aug 13, 2020 10:07 pm

Re: G-code bug

Post by GabrielR922 » Thu Jun 03, 2021 5:25 pm

Hi Tom.


Will use facet angle of 1 degree. Any future problem I will let you know.

Thanks for the solution and the help.

Post Reply