UCCNC Turn Problems

[andrej]

Hi, hope one of the devs can read this.

I've been playing with latest test release on a lathe and have some issues. Here is what I've found in last couple of weeks:

Using LinuxCNC PP in Fusion360 as it appears to give best results, tried also Fanuc PP, Mach3 PP and Grbl Turn PP.

1. The acrs are not displayed correctly in UCCNC and the toolpath is wrong. See attached screenshots. In UCCNC the toolpath has sharp edges where arcs meet, no matter if radius acrs are enabled or not. In ncviewer for example the toolpath is nice and smooth.
Gcode for the operation with IK arcs:

Code: Select all

N39 M1
N40 M0 (CHANGE TO T4 ON REAR TOOL POST)
N41 T4 M6 G43 H4
N43 G90 G94 G18
N44 G54
N45 G97 S1150 M3
N46 G0 X22. Z50.4
N47 G0 Z-3.456
N48 X13.543
N49 G1 X10.418 Z-4.704 F500.
N50 G2 X8.939 Z-6.044 I1.248 K-1.563
N51 G3 X6.971 Z-8.486 I-4.87 K0.544 F100.
N52 G2 X5.924 Z-13.33 I3.711 K-2.851
N53 G3 X11.395 Z-22.955 I-27.901 K-13.131
N54 X10.472 Z-29.104 I-16.323 K-1.868
N55 G1 X14.472 F500.
N56 Z-19.904
N57 G0 Z-3.456
N58 X13.543
N59 G1 X10.418 Z-4.704 F500.
N60 G2 X8.939 Z-6.044 I1.248 K-1.563
N61 G3 X6.971 Z-8.486 I-4.87 K0.544 F100.
N62 G2 X5.924 Z-13.33 I3.711 K-2.851
N63 G3 X11.395 Z-22.955 I-27.901 K-13.131
N64 X10.472 Z-29.104 I-16.323 K-1.868
N65 G1 X14.599 F500.
N66 G0 X22.
N67 Z50.4

Gcode for the operation with radius arcs:

Code: Select all

N41 M1
N42 M0 (CHANGE TO T4 ON REAR TOOL POST)
N43 T4 M6 G43 H4
N45 G90 G94 G18
N46 G54
N47 G97 S1150 M3
N48 G0 X22. Z50.4
N49 G0 Z-3.456
N50 X13.543
N51 G1 X10.418 Z-4.704 F500.
N52 G2 X8.939 Z-6.044 R2.
N53 G3 X6.971 Z-8.486 R4.9 F100.
N54 G2 X5.924 Z-13.33 R4.68
N55 G3 X11.395 Z-22.955 R30.837
N56 X10.472 Z-29.104 R16.429
N57 G1 X14.472 F500.
N58 Z-19.904
N59 G0 Z-3.456
N60 X13.543
N61 G1 X10.418 Z-4.704 F500.
N62 G2 X8.939 Z-6.044 R2.
N63 G3 X6.971 Z-8.486 R4.9 F100.
N64 G2 X5.924 Z-13.33 R4.68
N65 G3 X11.395 Z-22.955 R30.837
N66 X10.472 Z-29.104 R16.429
N67 G1 X14.599 F500.
N68 G0 X22.
N69 Z50.4

2. Sometimes the UCCNC control software gets stuck after finishing whole cycle. Entering for example G1 Z0 F500 to bring the tool back to part zero does nothing, on the screen the DTG for Z axis changes to the distance it should travel, but there is no actual move. But G0 (rapid) works. Restart of UCCNC is needed for it to work properly again.

3. Constant surface speed is a bit erratic I think. Although I am not sure how it is supposed to work exactly. But appears to work sometimes, sometimes not as the spindle RPM do not change in either control software or audibly. I can try to replicate it when I have time and make a video if necessary.

Any ideas what could be wrong? Will there be an update soon perhaps?

I will update this thread if I find other issues when I get the lathe running again after I fix some mechanical issues.

[cncdrive]

Hi,

Your code is errorous, with UCCNC it shows up, because it supports arcs with different starting and end radius, however it is considered code error, because the RS274NGC standard does not allow that.
Your code does not even load in Mach3 for example, it gives an arc start and end differs error and you can't even run your g-code file.
So, something is wrong in your CAM settings in my opinion.

[andrej]

I quieckly tried Mach3 Turn and GRBL Turn again, same issue with arcs....
All PPs I tried are official from Fusion 360 library so nothing custom made.
Can you recommend any wokring PP if you tried any? Or will you write one specific for UCCNC at some point?

Mach3 PP code from Fusion 360 for example.

Code: Select all

N37 M0 (CHANGE TO T4 ON REAR TOOL POST)
N38 T0404
N39 G90 G94 G18
N40 G54
N41 G97 S1150 M3
N42 G0 X22. Z50.4
N43 G0 Z-3.456
N44 X13.543
N45 G1 X10.418 Z-4.704 F500.
N46 G2 X8.939 Z-6.044 I1.248 K-1.563
N47 G3 X6.971 Z-8.486 I-4.87 K0.544 F100.
N48 G2 X5.924 Z-13.33 I3.711 K-2.851
N49 G3 X11.395 Z-22.955 I-27.901 K-13.131
N50 X10.472 Z-29.104 I-16.323 K-1.868
N51 G1 X14.472 F500.
N52 Z-19.904
N53 G0 Z-3.456
N54 X13.543
N55 G1 X10.418 Z-4.704 F500.
N56 G2 X8.939 Z-6.044 I1.248 K-1.563
N57 G3 X6.971 Z-8.486 I-4.87 K0.544 F100.
N58 G2 X5.924 Z-13.33 I3.711 K-2.851
N59 G3 X11.395 Z-22.955 I-27.901 K-13.131
N60 X10.472 Z-29.104 I-16.323 K-1.868
N61 G1 X14.599 F500.
N62 G0 X22.
N63 Z50.4

[andrej]

I am puzzled, see attached screenshot. Same gcode (grbl turn PP, F360). on top old (stable) version 1.2113, below latest version 1.2116.

Any PP in F360 I try always outputs incorrect arcs in latest SW version. So where is the problem, UCCNC or F360?

[cncdrive]

Your file loads the same at me in both 1.2113 and 1.2116.
I'm attaching 2 pictures. The picture with the tangential knife circle enabled is the 1.2116 version.

[andrej]

Yes the code generated with mach3 or Linux cnc behaves the same in both versions. But for some reason code gerated by grbl turn postprocessor looks OK (as on the screenshot in previous post) in the last stable release. No idea why... All other code generated by mach3, fanuc, etc looks the same in both versions.
Can't understand why fusion would generate incompatible code. Been using uccnc on a cnc router for many years without issues. But with lathe it just behaves weirdly IMHO.
Not sure if more people use it, but if you want me to test anything that might help let me know. Simple stuff works fine, but multi arc lines do not, atleast not for me.
Will test some other objects designed in Fusion and generate lathe code for it and see if the problem is still there.

[cncdrive]

What about asking the Fusion360 developers?

[needleworks]

I may be wrong but I don't think this is a Fusion 360 problem. I draw in Solidworks and use Cambam to generate code, I am also having problems with the generated code.

[andrej]

Yes, something weird is going on. I tried to open the gcode in LinuxCNC and it is nice and smooth with tangent arcs.

In UCCNC the arcs are not tangent at all, and the toolpath shown is also present on the finished part.

Also looking closely at the toolpath I see that the roughing toolpath is closer to the part final surface than the finishing one despite set to leave 0.15mm stock. No matter if finishing operation is done with the same tool or not.

Seems weird when other control software can interpret the toolpaths OK, but UCCNC not

I asked in Fusion CAM group, looking for answers, but no reasonable explanation yet.

[cncdrive]

Mach3 and Mach4 also not interpret your g-codes correctly, I have tested them.
UCCNC interprets the arcs as they are programmed, so I don't know what is different in LinuxCNC.