TinyG post - unexpected X-axis movement

[jtellier466]

I'm new to Fusion360, as well as Gcode  🙂  I've got a 3-axis CNC router driven by a TinyG controller.

I'm in the testing/bring-up phase of building this thing; I took a test model (STL), imported into F360 and converted the mesh to a component/body.  Then defined a 2D Contour to cut a "notch" along the Y-axis top of the stock, like this:Y-Axis Notch Test

I set the Passes/Tolerance parameter to 0.1mm, and the Maximum Stepdown to 2mm.  I end up with Gcode that F360 simulates "plausibly", with the toolpath resembling this:Simulation

Now, if I were to do this simple operation via hand-written Gcode, it might look something like this:

(MANUALNOTCH)
(T1 D=4.762 CR=0. - ZMIN=-6. - FLAT END MILL)
G0 G90 G94 G17
G21
M5
M9
G55
M3 S15000
G0 X53.0 Y-10.0 Z10.0
G1 Z-2.0 F300.
G1 Y85. F250.
G1 X55.0 F250.
G1 Y-10.0 F250.
G1 X53.0 Z-4.0 F300.
G1 Y85. F250.
G1 X55.0 F250.
G1 Y-10.0 F250.
G1 Z20. F300.
M30

------------------

But what TinyG Post produces is way more complicated; but what started me down this path is the ERROR 202's that I've been seeing when I try to run it on the controller.  So I started looking at why there are such "fine" steps comprising the cut.   Well, what I see is that not only the Y-axis is being finely stepped, the X-axis isn't being held constant!   An example segment of the gcode after the lead-in completes looks like this (with my comments included):

(...previous G1....)  ... Y53.646 Z-4.    <--- stepping down to 2nd depth, ok
Y53.663        
Y76.638 F250.   <--- correct feed rate for cut
X52.374 Y77.187 <--- plausible Y values (but I would claim Unnecessary stepping)
X52.349 Y77.393 <--- but note how "X" is not held constant!  This is supposed to be
X52.329 Y77.496  <--- a perpendicular cut on the Y-axis!
X52.305 Y77.599
X52.265 Y77.736
X52.215 Y77.873
X52.156 Y78.01
X52.087 Y78.147
X52.005 Y78.285
X51.892 Y78.446
X51.828 Y78.526
X51.758 Y78.607
X51.639 Y78.728
X51.52 Y78.833
X51.365 Y78.951
X51.211 Y79.05
X51.074 Y79.125
X50.937 Y79.189
X50.801 Y79.243
X50.664 Y79.287
X50.527 Y79.322
X50.39 Y79.349
X50.254 Y79.368
X50.117 Y79.378
Y79.382
X55.693
X55.709 Y79.363
X55.586 Y79.345
X55.449 Y79.317
X55.312 Y79.28
X55.176 Y79.234
X55.039 Y79.179
X54.902 Y79.113
X54.766 Y79.036
X54.629 Y78.947
X54.492 Y78.843
X54.48 Y78.833
X54.34 Y78.708
X54.2 Y78.559
X54.09 Y78.422
X53.995 Y78.285
X53.945 Y78.203
X53.881 Y78.087
X53.825 Y77.97
X53.777 Y77.853
X53.735 Y77.736
X53.695 Y77.599
X53.664 Y77.462
X53.641 Y77.324
X53.626 Y77.187
X53.619 Y76.638
Y0.362    <<<<<<<<---- BIG jump back to front of stock            
X53.626 Y-0.187
X53.641 Y-0.324
X53.664 Y-0.462
X53.695 Y-0.599

-----------

SO, I'm confused.  Is there any way (similar to Maximum Stepdown) to force longer traversals of the tool on the path?  I would seem to require that to reduce the "202" errors? Not sure.

And why the X-axis behavior?  Could this possibly be some kind of "compensation" being gen'd intentionally?   

BTW, I am sure that the original model "notch" is in fact perpendicular to "X" 🙂

I am testing this "import" stl process because I have some models that were done using a different tool, and I need to CAM them!

Thanks for any thoughts you may have!

Jim

[HughesTooling]

You can see in the simulation in Fusion there are arcs at each end of each pass, in your code there are no G02/03 codes are generated so you get all the short segment moves. I'm not sure why you're not getting arcs in the code but seem to remember TinyG had a problem with arcs so it might be disabled in the post.

Looks like you might have used a pocket in the picture, you could get a lot closer to the manual programing if you used 2d contour and add a tangential extension to clear the part. If you want to simplify even more disable leads.

If you need more help Export your design as an f3d and attach. Don't use the share links, they're a waste of time.

Mark 

---

[jtellier466]

Hi Mark! Thanks for the reply! (I see now that my other 2 attempts to post to the forum finally showed up! HAHA!  I'll see if I can remove them as redundant).

The gcode fragment I posted (pun?) earlier Was from a 2DContour, not a Pocket.  I'm not 100% sure what you mean by "tangential extension" tho.   I did specify Contour/Geometry/Additional Offset = 10mm (w/ Tool Outside Boundary); is that what you refer to or something else?

Anyway, I tried your suggestion of disabling Leads; I also selected a specific "Entry Position" near

the front of the stock, so as to avoid plunging down into it to start the cut.  Simulation seems to look good.  The Gcode also seems much more "direct", in that there is a definite Y-axis traversal across the stock, e.g.-

Z-6. F100.  <--- 3rd depth at 2mm stepdown
X52.381 Y0.362 F250. <--- starts at front
Y76.638  <--- full Y-axis feed across stock, no intervening X-axis weirdness 🙂
X52.374 Y77.187 <--- moving laterally to continue cut, etc...
X52.349 Y77.393
X52.329 Y77.496
X52.305 Y77.599
X52.265 Y77.736
X52.215 Y77.873
X52.156 Y78.01
X52.087 Y78.147
X52.005 Y78.285
X51.892 Y78.446
X51.828 Y78.526

=======

So, I will try this on the machine later today; but I'm optimistic! 🙂

Thanks much for the assist!

Jim

[HughesTooling]

It's a lot easier to help if you export your f3d design with setup and ops then attach here. At least then I can see exactly what you are trying to do.

Mark

[jtellier466]

OK, my bad.... I forgot to do that earlier... here 'tis....

[HughesTooling]

Take a look at the attached using a 2d contour, 3d is really not suited to something this simple. Talking about simple why are you importing a mesh for something you could model in Fusion quicker that the time you spent importing? If all you are going to try and do with fusion is import meshes you're going to make work and get bad results if there are any curved parts.

Have you looked through this thread on New to CAM?

Mark

[jtellier466]

Hi Mark,

   I've been going thru the "newbie" links as much as I can; I may be able to re-create some of the (existing STL) parts I've got in F360 once I get over the learning hump; but I kind thought "if a process exists, it probably ought to be usable", so I went down the "import" path.

Your comments and intuition are most likely based on experience, so I'll value that.

BTW, I didn't see any attachment as you mentioned last.

I'm looking at getting more up to speed with direct F360 modeling over the next few days, so I'll hopefully have better luck with the CAM.  Thanks!

Jim

[HughesTooling]

Hi Jim

Sorry about the missing attachment, should be attached this time. For some jobs STL might be OK, something like a 3d logo or relief. But for engineering\prismatic parts a brep model will work a lot better, for example none of the drill and hole ops will work on a mesh and picking boundaries off a mesh will be a problem.

Have you looked through the How To and tutorials in the help, a good place to start is sketching like this one.

Mark

[jtellier466 · Accepted solution]

Hi Mark!

   Your example of doing the contour set me straight on a few things!  First, when I was selecting edges to define contour, I had selected the "short" x-axis edges at the bottom of the notch (along with the "long" y-axis edges of same.  Didn't realize that wasn't necessary, so I guess that introduced the "dancing" step behavior.  LOL!   

  Thanks for your pointer to the "sketch" stuff... very helpful! Lots to learn, but I'm at least moving Forward now!   Thanks for your help! 

Jim