Dynomotion Forum

Hi Tom,
Iam having an issue with cutter comp on my mill. All tests are being done with KMCNC set to simulate. Attached is Gcode. It is a slot shape that is 2.191" wide by 3.191" long on an angle. See screenshot of FusionCAM. Also iam using the Fanuc Compsetting in KMCNC.
At line 22 i get a GCode Error window stating" Concave Corner With corner radius comp". This appears to be happening on the lead out.I have the Comp Radius in CAM set to .100"
For trouble shooting i set the Tool Diameter offset to .001" and still get same issue. If i set lead in and out to .5" radius in my cam software, then KMCNC will run Gcode with no error window.
I have ran similar code on Fanuc controls at work with no problem and with even smaller compensation moves. Also tested gcode with Mach4 software with no issue.

Currently using KMotion v4.34
Troy

Hi Troy,

Hmmm. The GCode does in fact have a Concave Corner of 4.1 degrees:


KMotionCNC currently allows gouging by 0.005 radians (2.86 degrees).

I suspect the problem is caused by the microscopic ~ 0.001 inch long segment which has rounding issues as the coordinates are only specified to 4 digits. The NIST code has a historical reference to 0.01 radian limit which would probably allow this.

I wonder what the limit for Mach4 or Fanuc are?

I suppose we could make the threshold an option.

Or ignore angles of "tiny" line segments. But that might be dangerous and I can't think of an easy way in general to determine what size is significant.

Hi Tom,
After your post i Generated toolpath again in Fusion and it picked a totally different spot on profile for Lead In/Out and KCNC ran fine. I also had same error on an outside profile yesterday. When i generated toolpath again today for this operation i got the same toolpath and same KCNC error as yesterday. So i selected a different starting point for Lead In/Out and KCNC ran fine with no errors at all.

Been digging around but have not found anything on the limit that Mach4 and Fanuc are using.

Hi Troy,

Yes the rounding error of +/- 0.00005 inches on the endpoints of that 0.001 inch long segment will change its direction by a few degrees pretty much at random.

Can you change the Fusion post processor to output another digit of precision? It sounds silly but rounding to 0.0001 on tiny segments can cause a smooth path to become all zig-zaggy nasty.