Trajectory Planner using Ruckig Lib

I have created a new test file from the same dxf file. This one uses the lcnc plasmac PP from sheetcam. Then delete the plasmac specific bits.

This IS WORKING. I am recoding a run with supporting log as I post this. The difference seems to be that circles are done as two arcs. Where as the previous file had circles as a single arc. This was correctky dealt with by the standard tp but was not being handled by the scurve.

Sorry Grotius but I am still not convinced by the zero arc argument. It does not gell with what I am seeing in the gcode or in the way it is interpreted both by linuxcnc gcode plot or other gcode viewers.

Cheers - J.

@Joco,

Sorry Grotius but I am still not convinced by the zero arc argument.
If there is a arc or line with lenght=0. Previously the function returned a "nan", after a source code edit it now returns a "0".
Then as a result no program freezing anymore. It just passes this zero lenght segment.

The cavaliercontours version that cam uses indeed put's out 2 arc segments to create a full circle.
That should not be a problem for the scurve.

Note : If you see a gcode mismatch, please provide the gcode line and or a picture. Then i can figur it out.

@Joco,

You suggest a full circle is the problem for scurve?

Indeed. Stupid..

This is ok, outputs ~157mm.
struct sc_pnt start={0,0,0};
struct sc_pnt way={50,50,0};
struct sc_pnt end={100,0,0};
double l= arc_lenght_c(start,way,end);
printf("arc lenght : %f \n",l);

Then this is fault : It outputs nan, but is a circle.
struct sc_pnt start={0,0,0};
struct sc_pnt way={100,0,0};
struct sc_pnt end={0,0,0};
double l= arc_lenght_c(start,way,end);
printf("arc lenght : %f \n",l);

Ok will fix this.
 

Okay this test run ran for circa 7mins before stopping early.  Attached is video of the run.  It shows the interesting mix of motion. Sharp corners can still ber quite hard while the transitions between arcs can have a relatively long pause.

The program stalls before completion. See log details on

Log file trapped for this run.
 
 
 

New test file based off the sheetcam plasmac PP.  This has circles made of two arcs. Where as the previous test file would have single arc circles.  This was still handled by the standard tp
 

Cheers - J
 

I  don't have the visibility you do of the internals so I am operating from external observation.

The line which was causing the issue is a circle defined by a single arc with the same start and end. Which would seem to be the definition of a circle if represented by a single arc.

This second file I ran used two arcs to define a circle each being 180degs are turn.  This caused the planner no issues.  However the previous definition of a circle did cause it an issue. It could not hanlde that definition.  Yet the standard planner does.

As I said, all based on obersvation of what is feed in and the apprarent behaviour based on those inputs.

Cheers - J.

Note : If you see a gcode mismatch, please provide the gcode line and or a picture. Then i can figur it out.

That was why I gave the link to the gcode viewer for stepping through the gcode file and being able to see the shape rendered and the movemments.  But not problem to provide the pics.  I am pretty sure I quoted the gcode line in question in an earlier post. But no matter.  Happy to add whatever info makes things easier.

Cheers - J.