Modeling a hexagonal shaft that has been machined to a smooth circular point

I am trying to model a rather unusual cap-nut that was constructed by taking a hexagonal shaft, drilling a threaded hole into it, and machining the opposite end to a smooth, therefore circular point.

I came up with “an okay solution for now” by creating a bezier curve, applying the “screw” modifier to it, and converting that to a mesh, so I now have a mesh that looks rather like a Hershey’s Kiss (candy), which I then simply “dropped into place.”

But… I am a perfectionist. :ba: I’d really like to come up with a way that would create the geometry for real. The smooth cone shape would actually come down to the (hexagonal) edges of the shaft, and there would be “nothing to hide,” as well as not a whole lot of (if any…) manual mesh-stitching.


can you show pic of what you mean here
migth gives us better idea how to get it done!

happy 2.5

I don’t really have a good picture of it (except in US Patent #46111; see or

What’s been done is, as I described: a hexagonal shaft was put into a lathe and one end was machined to a rounded point. So, the profile of that portion is “a rounded cone,” but at the bottom of this round cone is a hexagonal shaft.

I could (and did…) model this by subdividing the hexagonal shaft and “stitching” another mesh made from a spun curve onto it … by hand. But I have been pondering what would be a better, cleaner, ideally curve-based way to do it, because there are a lot of similar machined shapes throughout this model. I’d like to find a good compromise between good-enough results and fast modeling.

I don’t really have a good picture of it (except in US Patent #46111; see or

Are you trying to say the image on that patent is the best example you have?
Why not just attach an image of the model you said you made ?

you want to do a complete steam governor ?

with high precision !

or can you show the piece you want to model if there is a pic of it !


Here’s one way (using 2.49, since that’s still better for this sort of precision modelling).

Start with the profile you want for the pointed end (possibly generated from a curve). I’ve kept mine simple:

Viewing from the end, spin it some multiple of 6 times; I’ve used 24, but if you are going to subsurf the final model you might even get away with just 12. As always after spinning, don’t forget to remove duplicate vertices.

From the same viewpoint, use the knife (exact, with Control-click to snap to vertices) to cut the hexagonal shaft.

(Continued in next post, since we seem now to be allowed only three images per message, instead of the old six …)

As I was saying, before I was so rudely interrupted …

Select the vertices outside the cut line,

and delete them. (The easiest way to select exactly the right ones is probably to start with just the centre vertex, keypad-+ to expand the selection to the cut line, then invert the selection.)

Select the new edge and extrude it along the axis.

Scale the extruded edge to 0 along the direction of extrusion (Z in my case), and there we are.

Best wishes,

Yes… I am modeling this (and a bunch of other) historic governors, which represented various significant advances in mechanical feedback mechanisms. And yes, I’ll post the WIP soon enough.

Incidentally, this patent was issued during a time when the US Patent Office still required patent models to be submitted with each new patent. This patent model still exists and is part of the permanent collection of the Museum of Science and Industry, Smithsonian Institution. The drawing that accompanies the patent application matches that model, and of course, the feature that I am discussing is at the very top.

The only difficulty in your technique, Matthew, is the treatment of the edge… a machined surface would not have the “arches” that are visible in your wireframe. In the actual milled piece, the bottom profile of the milled area is circular, not scalloped.

Still, the use of the knife-tool is an interesting idea.

The reason why I am contemplating the use of curves is that there are many examples of similar “irregular shapes” in this particular artifact, such as the supporting pins which extend at roughly a 45-degree angle upward from the vertical shaft to the weight-arms. There is no “weld” here. Just as the shaped hexagonal shaft progresses smoothly from a hexagonal shaft to a circular point, these supporting arms morph smoothly from a circular shaft to a flat coupling (with a hole drilled through it).

Usually, when modeling such things, I’m able to “cheat” what won’t be closely seen. But for this project, the geometry needs to be “right” if I can (in a reasonable amount of time = cost) make it so.

And yet the patent drawing shows exactly the same arch/scallop pattern at the edge. In fact, I cannot imagine how there can be a circular to hexagonal join without some such effect. You can move it around though: for example, if you make the circle fit the inside of the hexagon instead of the outside, you get a spandrel inside each vertex of the hexagon.

Best wishes,