THC Servo Start

Hi, I had been using the corner lock with remapping for 3 days and it works nice.

I have changed the M67 by M68 because M67 makes a small stop (it is synchronized with moves).

Guys

Does anyone know if what is discussed in bug report #447 may have something to do with this? I know that even with all the work I've done to fix every corner hold problem I can find I am still getting "witness" marks where arcs and straight commands join. I was thinking about asking for help with how to optimize G64 blending but then I found the bug report.

John

how can I see the #447 bug report

John, I think you should upload a gcode sample file to that bug report with a photo of the resulting cut.

grijalvap wrote:

how can I see the #447 bug report

Pedro here:
github.com/LinuxCNC/linuxcnc/issues/447

Guys

I will try and get some photos of the marks posted in the next few days. The G code snippets will be a little harder, these parts are cut from an approximately 500kb file ( 12k lines of code). Does the community want to see the code where the marks occurred, at the start of the cut, at the start of the program?

John

John

Guys

I started looking into the bug #447 problem on plasma cutters and have found plenty of examples of "divots" at the junctions of some arc and straight sections. Now the bad news, these "divots" are also at places with strange small linear segments in the G code. So is this the problem reported in bug #447 or is it a post processor problem? I haven't had a chance to look over the original artwork yet and check for short segments but I will.


John

Tommy

This happens very fast! With halscope I would see the requested-vel go to zero and the back to full speed at the .ini acceleration settings.

I went back and checked the original drawing in Inkscape. The path is joined at every segment. The "divots" actually occur in a single segment. I have used Inkscape for a long time with plasma and am pretty good at producing cuttable files. When I import the file into SheetCAM something happens. SheetCAM breaks up the nice long path segments from Inkscape into shorter segments so each segment is an arc or a straight line, not the Bezier curve that Inkscape uses. The "divots" are at the junction of two of the SheetCAM generated arc segments. It gets even stranger because looking at the SheetCAM generated segments the "divots" are at the junction of two arc segments but the G code produced by the PP has tiny straight segments there. I am going to do some in depth testing of this problem over the next few days.

John

Guys

As promised here is my current kerf crossing component.

John