Sure Tom,
Rule # 1 is that the preferred set of magnets is usually the cheapest.
Rule # 2 is that the more magnets, the more flux you get, and usually the most voltage.
So I went choosing between different arrangements of magnets to find the best coverage, and kept an eye on the dollars because I'm cheap. Below are the three contenders for which I kept screen-shots. The first is an arc-segment. That turned out to be expensive, and the dimensions have to be "perfect" for the right amount of flux. The second is an 8-magnet arrangment. I could fit magnets 3/4" thick as long as I only had 8 per ring. But there are big pie-shaped gaps between each magnet. So I tried to make smaller magnets work, and found that I could make the best coverage with 1/2" magnets. This had high flux, but the cogging was going to be pretty bad because 12 magnets line up really well with 36 teeth...
Then I hit upon the idea of using some 1"x1"x1/2" and 1"x1/2"x1/2" mags in alternation. By doing that 3-D drawing I wrapped my brain around how I was going to machine such a thing, and that's when I decided to go with it. I could flip the pattern of the middle ring backwards, and its cogging attraction "clockwise" would counter-act one of the other ring's cogging "counter-clockwise". This plan also turned out to be pretty cheap, like 250$ or something. I spent a little more on screws and the epoxy was already paid for. The blank of steel for the rotor is yet another junk-yard discovery. Got an extra 18" of it, so I can make 6 more if I want!!
To mill the flats, I started with the four rows 1 inch wide for the 1"x1"x1/2" magnets. Then I clamped it in the vise at the odd 37 degree angle I needed for the next big flats, measured very carefully. Once that flat was done, it became a process of repeated iterations of clamping against the previous 37 degree flat to mill the next one at 90 degrees to it... At these angles I could not mill straight across, either. I had to mill each 1" flat on the ends, then come back later and mill the central 1" flats at the opposite 37 degrees. With all of those done, I only had the 1/2" wide flats to finish, which followed the same procedure, but turned at yet a different cock-eyed angle. Time consuming!
Have to talk the boss into getting a 5th axis drive for the CNC....
I guess there are a lot of details left out because I do a bunch of this stuff either on paper or in spreadsheets scattered about my computer hard drive that don't even make sense to me a few months later.
I took some pictures of the machining process so maybe if I can find them it would help. Here are the FEMM drawings.
ARC SEGMENTS1" X 3/4" MAGNETS1" X 1/2" MAGNETS + 1/2" X 1/2" MAGNETS