Swarf cutting fully mitered cabinet parts gives extreme tool postures

[licensKKGSY]

So swarf cutting is two contours mapped to drive the tool between the two in a smooth/accurate fashion, so I think I understand the fundamental limitation of what I'm going to ask for, but here is my problem;

I have modeled a sheet part with miters along the entire contour. For simplicity and to avoid axis limitations, I'm making a swarf toolpath for each miter, but when I do so the shorter top contour causes these unnecessary extreme (and I suspect slightly less accurate) tool postures at the start and end of the toolpath.

Thus far I haven't found a function within the Swarf dialogue which allows the tool to maintain the tool angle as it would be for two equal-length guiding curves. My workaround has been to generate various manufacturing models for the different miters, to obtain these equal-length guiding curves to drive my toolpaths, but the entire workflow is just messy. Either I'm simply unaware of the right function, but if not I suspect the request would be something like a "separate tangential extension" for top/bottom curves or "tangentially extend top curve to match bottom curve"? I know there is a new swarf coming, so perhaps this might be a moot point in a few months.

Edit: project file attached

[seth.madore]

Could you share a sample file here?
File > Export > Save to local folder, return to thread and attach the .f3d file in your reply

[licensKKGSY]

Here you go, project attached to original post. 

[seth.madore]

One solution would be to create a line at the top that extends to the boundary (or slightly beyond if you prefer) of the part. You can then select "Contour Pairs" in your Swarf toolpath, putting that into a Circular Pattern. However, this would not wrap around the chamfer, but I think that's what you're trying to avoid (with the extreme tool angle you mention):

 

 

[seth.madore]

---

@licensKKGSY wrote:

Here you go, project attached to original post. 

---

Take a look at my reply above and see if that works for you (we posted at roughly the same time, so you might have missed the notification)

[licensKKGSY]

Hi Seth, Thank you so much.

Yes I'm trying to avoid the wrapping-around-corner. 

The part in this case is arc shaped, so it doesn't lend itself nicely to patterns but yes, I have considered using a sketch as guiding curve. When I use "contour pair" selection, the guiding sketches must be created in a way where they won't be seen as a chain on their respective sketches, which is the case on the top sketch, where each curve extents beyond each other's end node, but not the case on the bottom sketch where they are seen as a chain by the selection even when I remove all coincident constraints from the sketch itself. I could extend each curve slightly, but by that time, the aforementioned solution of using two models with each their chamfer becomes easily as viable. 

I don't know if the addition of a "separate tangent extension" function to swarf has a broad enough application, but I would like for it to be considered.

[seth.madore]

If you press the ALT key and make your selection, you can select just the edge you want, not the whole contour

[DarthBane55]

If I can avoid 5-axis, specially swarf, I do, swarf always makes the finish a bit faceted (a lot in Fusion).  Anyways, see attached file, this is doable in 3-axis, using trace operation (which is found under 2-axis operations).  You don't get any weird moves on the ends because the tool axis is constant.  Your surface creates the illusion that it needs 5-axis, but it is in reality "planar" as far as tool axis goes.

[licensKKGSY]

Thank you Seth, that is good to know regardless of solutions!

DarthBane55 - Thank you so much. Using trace makes so much sense for these "simple" chamfers, so I will definitely add that to my toolbox, but technically the tool postures and the resulting chamfer on the part will not be correct for the curved sides, though the large radius on this part makes it look right. On a 90 degree arc you would easily see a constant tool orientation to be incorrect. 

[DarthBane55]

When I 1st read your answer, I agreed with you...  but, in your model, the trace does it perfectly.  Imagine your part being sliced by a plane, you still get an arc, but it is sliced, therefore it is flat.  Your part is like that, it's radiused up and down, but there is no curvature, it is flat.  Hard to explain, but trace is good in your sample part.

[licensKKGSY]

A bit off topic, but regarding the faceting you describe: besides setting the correct fine tolerance and multi-axis parameters from Fusion, you need to tune your machine control's smoothing function if you have that (which all 5-axis should IMO). In Fanuc it is either AICC2 or High Speed Smooth Tcp. It was night and day when got it working. swarf makes the difference very visible.

[DarthBane55]

Yes I know about that, but thanks!  In the case of your part, the slope is so gradual that it probably would be no problem.  But on some parts with more drastic curvature and directional changes, I've done side by side comparison with code from Fusion and code from Siemens NX, and it's day and night, with the same tweaks and tolerances applied.  Not sure what it is, but NX is much better on swarf.  I'm thinking that the new swarf coming up will be much better than current one though (because ModuleWorks).

[licensKKGSY]

If I understand your way of defining the toolaxis you create a slicing sketch on the midplane between the two non-curved edges and define that as your z. I fail to see how this can be correct (though one might accept the deviation in less extreme cases). The stock simulation shows this.

What could be done, if multi-axis was added to trace, was to define your tool axis towards a point at the center of the arc.

[licensKKGSY]

Yes, had the same experience coming from Alphacam, which in so many other ways are less refined on 5-axis stuff, which is why I only became aware of Fanuc HSSTCP after I started using Fusions, but (I suspect) the finer codes provided by alphacam just masked the real issue of not having the control tuned correctly.

[DarthBane55]

I am only referring to your sample part.  See picture below, I think the deviation would work in your case.  Of course the smaller the radius and the larger the radius angle, the worse it gets.

[DarthBane55]

---

@licensKKGSY wrote:

Yes, had the same experience coming from Alphacam, which in so many other ways are less refined on 5-axis stuff, which is why I only became aware of Fanuc HSSTCP after I started using Fusions, but (I suspect) the finer codes provided by alphacam just masked the real issue of not having the control tuned correctly.

---

Are you talking about G121 or G131 here, or something completely different?

[licensKKGSY]

I think we understand eachother regarding the trace toolpath. 

I think g121/131 has to do with axis transformation - I'm talking about high speed functions (fanuc's own term). In this case the ones affecting tool posture and therfore smoothness when swarf cutting. 

In fanuc there are a few options for high speed machining, where AI contour control (G5. 1) at its highest levels can apply look ahead interpolation/nurbs spline to all 5 axis with some parameters defining tolerance etc.

High speed smooth tcp (g43. 4p3) is the more refined and combines tool posture control (rotation axis moves along a rulled surfaces of specified deviation from linear) between programmet points) with the nurbs interpolation at both the axis rotation point AND the tooltip. 

[DarthBane55]

G121/G131 is the smoothing, or high speed function.  A more advanced (recent) version of G5.1 if you will.  It controls how accurately it will reach each point.  For example, G121R10 will reach each point of the toolpath very accurately, great for 2d contour maybe, but for 5-axis it is really not great, because the machine gets jerky as it reaches each of the hundreds of coordinates.  As you understand it well I think, you want the toolpath to sort of flow thru the average of the points smoothly when you feed it hundreds of points per seconds.  That makes it smooth, and because you have so many points, it won't stray far from wanted point anyway.  This would be achieved with G121R1.  So that is for smoothness.

We use G43.4 for 5-axis simultaneous, but I don't know about the G43.4P3, I will now try to find out exactly what it is and if we have it!  G43.4 is nothing for smoothing the path, but I assume the P3 parameter is...  G43.4 "welds" the coordinate system to the part, so when the part moves, the coordinate system moves with it.  At that point, Z is not pointing upwards anymore, it follows the part as it moves.

Anyways, I think you know all this already, but I did not know about G43.4P3, which I will now research, thanks!

[licensKKGSY]

Huh, never come across g121/131 in my reading on hsm function, but then again fanuc always seems to have a handfull of functions covering something very similar but still different.

G5.1 Q1 Rx (x being between 1-10) does exacle what you describe, for 2/3 linear axis that is.I searched and only found a fanuc whitepaper mentioning g121- it's about circular interpolation speed control on 30i A controller. My knowledge is based on 30i/31i B. 

[Laurens-3DTechDraw]

With what you describe here I would probably go the trace route myself. But for other parts, or if you do want it to be a 5-axis solution, there is often an easy way to create the guide curves you want with a ruled surface. Create a quick and dirty video for you;