I've drawn up some magrid options assuming:
- toroidal coils
- torid magnets and virtual magnets covering equal areas of the sphere
- toroid minor radius set to cover 3/4 of the gap between torids
Note how much skinnier the 20 and 8 toroid forms are than their 12 and 6 sided duals.
Basic 6 toroid
8 toroid
12 toroid
Rhombic Dodecahedron base
20 toroid
30 toroid, colors added to make sense of the many parts.
The development of atomic power, though it could confer unimaginable blessings on mankind, is something that is dreaded by the owners of coal mines and oil wells. (Hazlitt)
What I want to do is to look up C. . . . I call him the Forgotten Man. (Sumner)
The development of atomic power, though it could confer unimaginable blessings on mankind, is something that is dreaded by the owners of coal mines and oil wells. (Hazlitt)
What I want to do is to look up C. . . . I call him the Forgotten Man. (Sumner)
The first two (6 and 8) are just complements of each other. They are both approximate cubeoctahedrons. In the "6", the 6 square faces of the cubeoctahectron are the real toroidal magnets and the 8 triangular faces are virtual. In the "8" it is the triangular faces that are real and the squares that are virtual. I am not sure the 8 is any better than the 6, but it could be.
I suspect the 12 and the 20 are the same situation for the icosidodecahedron.
Not for me, thanks. I did a little looking at openscad. I find the idea of a modeler-from-script counter-intuitive (but potentially interesting for serendipity). I admit my bias stems from my work with interactive modelers.
Both methods have their place, as there are somethings that are easier to do via scripting, while others work better to directly manipulate the polys "visually".
I can see that particularly for things like... screw threads! A more tedious chore to model I hope never to imagine. In addition to being a thankless task, modeling screw threads is or should be useless and unnecessary.