All you need for the CNC is the path - not the geometry. (in fact, the geometry confuses the issue as far as creating CNC path)
Considering manufacturing tolerance and clerance between conveyed part - why need perfect (can't manufacture perfect parts anyhow).
Apparently the solid sweep cut in SolidWorks is severly limited
I beg to disagree, at least for the case of a fixed-pitch helix. Sweeping a 3D solid will leave a helical cut with a perfectly consistent 2D profile. The challenge for Inventor users is to work out what that 2D profile looks like. But I agree, too, that perfection in the model is not necessary in the case of feed screws. Nobody would dream of trying to produce one by NC directly from the model!
I'm afraid sweeping of 2D profile - no matter how shaped and positioned in relation to 3D path curvature - will not produce results identical to sweeping 3D solid.
This is definitely not a simple problem. Here is one cross-section, cut on a plane parallel to the helical axis.
This shape was created with the patterned cylindrical cut feature mentioned above. The shape appears to be approximated by a series of elliptical arcs, each with a slightly longer major axis (vertical in this image) and a constant minor axis (equal to the diameter of the small cylinder--the dashed circle in the image). The center mark of each elliptical section is shown, and they appear to be laid out in a parabolic shape, or maybe elliptical, or even sinusoidal.
Not an easy shape to sketch in Inventor, anyway.
I think you're on to something here: the locus (of the points) is probably sinusoidal (projection of the helix's path?). If you draw a circle at each locus, and construct a spline that's tangent to those circles' outer boundary, I think you'll have a very good approximation.
I'll try it and see what transpires.
In the mean time, here's an improved version of the IPT I posted originally. Now fully parametric (you can change the cutting depth, diameter, pitch, etc). See the parameters in the file.
The latest part you posted is quite interesting. How did you come up with that approach? The logical progression isn't quite clear to me, although the approximation is impressively good.
I figured you have to take the 3D path into account somehow, and the easiest is just a silhouette of that 3D path/volume.
There's a 3rd, more terse approach, see attached.
Sadly, I'm working on a 4th method...
Start with some of our most frequented solutions to get help installing your software.