Shop Sabre (Win CNC based) post problems

[matthewsclark]

I recently purchased a ShopSabre CNC router with a rotary axis. I'm relatively new to CNC programming and G Code but comfortable in Fusion and Inventor.  I've been doing test cuts, right now dialed back to just a circle, to try and get the machine sorted and am having problems with Fusion's tool paths (there is a post for Shop Sabre on the Autodesk post page and shop Sabre provided me with an older one that they had modified as well, they both produce the same results here).

The machine came with Vectric as well. It outputs circles as 4 90 degree arcs while Fusion uses 2 180 degree arcs. Also their lead in and out strategies are different but I've tried to match them as closely as possible.

For some reason that I can't figure out, the Fusion toolpaths are making the circle not close, even with a very gentle lead in the circle starts out too large so when it comes around the circle and closes the cut there is a scallop outside of the circle diameter at the start of the cut.  The Vectric cuts are perfect.

As I understand it with I J coded arcs, it shouldn't matter if they are 90 or 180 because the start and end point is specified and the center is calculated, so I'm suspecting the lead in but that's about as far as my thinking has come.

I've attached the code from both programs (just a single finishing pass in a pocketed circle) and a photo, any ideas?

Thanks, Matt

(p.s. - error message says .tap is not a valid file extension for attachments haha, might want to add it - I made them .txt files)

[grigorov.rv]

Try not to use arcs, replace them with straight lines

[matthewsclark]

Could you explain this a bit more? I'm trying to understand the difference between the G-Code output from the two programs, hopefully so that I can fix / help develop the Fusion post because I'd much rather use it than the Vectric software. Since the 4 90 degree arcs cut fine, it doesn't seem like straight lines are necessary. Thanks in advance for the explanation.

[grigorov.rv]

Find in your postprocessor such a line (maximumCircularSweep = toRad(180);) And change to (90)

If it does not help, try replacing the arcs on the line.
Find the function (onCircular)

Comment on everything or delete and add this line: linearize(tolerance);

If it does not help, write, maybe I can help. I understand that the machine is 3 axial?

[matthewsclark]

I just realized I had tried to overlap the end of the cut so this fusion tool path is actually three arcs, but with the same result of the large overcut on lead in....

[matthewsclark]

I edited the post to create 90 degree arcs and ran a test, still has the lead in issue.  Before I start linearizing everything, I'd like to try and identify the source of the problem since the vectric code cuts great and uses 4 90 degree arcs still. Thanks.

[grigorov.rv]

G90
G0 Z0.8000
S14000
M3
G0 X3.6196 Y1.6804
G0 Z1.2
G0 Z0.25
G1 Z0.1 F25.
G1 Z-0.7
G1 X3.9196 Y2.2
G3 X4. Y2.5 I-0.5196 J0.3
G3 X1. I-1.5 J0. F30.
G3 X2.6249 Y1.0052 I1.5 J0.
G3 X3.9792 Y2.7488 I-0.1249 J1.4948
G3 X3.8502 Y3.0314 I-0.5917 J-0.0995 F25.
G1 X3.4681 Y3.494
G0 Z1.2
M5
G53 Z
X0Y0

Try running it

[matthewsclark]

I just modified the lead in in Fusion to eliminate the linear lead in distance before the lead in arc and made the lead in arc radius 1" and it fixed it, cut great. Still don't really know what exactly causes it though.

[grigorov.rv]

how is your progress?

[matthewsclark]

As far as I can tell so far the displaced tool paths have something to do with having an arc on a vertical lead in -  as long as I plunge straight down and have only a horizontal lead in arc, the cuts are as they should be. Don't know why - the WinCNC controller software and what kind of moves it supports I suppose??

Here is another example of displaced cuts. In the image I attached the roughing pass was larger than drawn and programmed on the right side of the cut - you can see the roughing tool marks and the thinner wall thickness. On the left it is in the correct location and the finish tool path hit it, the finish pass missed the cut on the right since it was displaced.

The finish pass is programmed like the test circle discussed earlier in the post - straight plunge, large lead in radius. The roughing operation was programmed using the 3D pocket clearing operation in Fusion, and while it does not have any vertical lead in radiuses, it does have a spiral plunge which may be causing the same errors that the leads are? - I'll test a simpler pocketing operation and post the results. 

Anyone else having any similar issues with a WinCNC based controller?

[Anonymous]

Hi, I am also running into a problem where my tap or nc file uses I and J coordinates and my shopsabre 3636 cannot read those coordinates. I also tried putting 90 degrees instead of 180 and that did not solve the problem. I have also removed the function (onCircular) to get an empty tap file. Any help would be appreciated. Thanks.

[grigorov.rv]

your problems in your postprocessors, they do not correctly display the data, I can later make postprocessors, send me control programs that work fine on your machines.

[Anonymous]

What exactly are control programs? I am new to cnc'ing and this is my first time using the shopsabre 3636.

[grigorov.rv]

this is the file that you run on the machine

[Anonymous]

The control program that I use is Shopsabre router control program 2.6.03.
It reads .tap, .nc, and .h files.
I looked into the PostP folder and found this file saved as a txt file. Is it what you are looking for?

[matthewsclark]

That is an old Vectric post that you attached, you can contact Shop Sabre for a Fusion 360 post or download one here - https://cam.autodesk.com/posts/

I'm having issues with accuracy with both posts as you can tell from this thread, but they will get you going. I will update this thread as I solve the problems I'm having.

[grigorov.rv]

try this postprocessor

[matthewsclark]

Hi Grigorov, is that just the win CNC post from Autodesk or did you modify it? If so, what did you change?

[grigorov.rv]

Have you checked it? all works correctly?

[matthewsclark]

No haven't tried it. I'm interested in understanding this process, not just trying things randomly. If you would like to share more info that would be great, if not then no worries, I'll figure it out.