I posted this on Thud's thread: http://www.ornl.gov/~webworks/cppr/y2001/rpt/120491.pdf
A similar design objective, but a radial example.
One more thing to throw into the constraints is cost. Super expensive efficient motors are already out there. We need cheap efficient motors. If you're going to DIY, then there are even more constraints in the availble materials and tools. There are going to be trade-offs in design, so anything Thud or anyone else is likely to come up with may not have super advanced fabrication techniques, but I think we could easily make something with more bang for the buck.
I'm always one for making maximum use of available materials. If I can buy something that's close and just modify it, that's easier than starting from scratch. This applies to the pieces as well as the whole.
The 'Sphere has morphed into one of the most open and dynamic idea incubators that I have ever encountered. Members come and go, but all add something along the way. Most of the techinical problems we face are quite solvable, but there just seems to be so many of them it's hard to keep up. I think we've had a large influence on the whole ebike sector either directly or indirectly. Ideas we were bouncing around a couple of years ago are starting to show up in the marketplace. Someday products might actually be built so they work and are reliable.