Here's the motor as it comes apart.

Here's the multi-sized winding.

Here is what I was originally thinking of to use the quarter inch by quarter inch mags in one of these motors.

After drawing it & having it on screen to play with, I am thinking that the 1.25 X .562 X .125 mags might give a better field. They would cover the stator segments quite well. If those who know could give any pointers on this...

I don't believe the 1/8 x 1/16" will do much at all. The 1/16" is too thin. The 1/8" will have a lot of losses and are hard to get together to amount to much surface area. If you decide to go with the tiny magnets, leave the factory wires and reconnect them as in my diagram in the link I posted above. I have a feeling you will need all the turns you can get. They can be changed later if need be.

The 1.25 X .562 X .125 will have a very large effective airgap. It may help with the cogging, but the output will suffer. And the magnets are SO wide it may cause cogging... I'm not sure about that one.

The 1/4 x 1/4 should be OK if all 9 are used.

I don't think there is a common flat magnet much better suited to this than a 3/4 x 3/4" and 5/16 or 3/8" thick.

I only have a few in various stages of completion, or lack there of. Pure 'Jerry Rig' works best on most with factory wiring. Delta output is less. Star output is lowest for 12V because of the resistance (do not be fooled by the high open voltages).
An amp at 12V is easy with the factory wire and fan blade, given some Good Wind.

If I ever get around to rewinding one, I'm going to try 130 turns per coil (tooth), test it, and connect it Jerry Rig or Star to suit my needs.  A single degaussing coil from a TV or 15" monitor would be plenty of #20 or #21 wire.

Do not get too carried away with 'better neos'.  It can cause cogging in these.  My best guess is the flux leaks to the other coil laminations after the intended lamination is past saturation.
http://www.fieldlines.com/story/2006/8/23/144336/344

Having access to good machines, it could be worth looking for a motor like this...
It is almost the same size, but far thicker, and with much heavier wire (2.5 ohms per coil by the meter).  1/6 HP 115VAC 6.2A.
It may be a decent cantidate for the 1.25 X .562 X .125 magnets, standing up and stacked a few deep. It is from a vent fan of some kind. Semi common.
Stuck rotor with 4 curved magnets, but it shows some of the good stuff.
http://www.otherpower.com/images/scimages/2050/Stuck.jpg

Now that things are slowing down a bit....

As I am back to work now, I also have time to tinker. One of the lakewood fan motors went to work with me today. First the armature was turned down to fit 4 neos with a .015" airgap. The mags are now glued to the armature, and will be poured in fiberglass resin.

The shunts were completely removed from the stator. I used a grinder with a scotch brite disc to skim them down, and then a small punch to knock the remaining copper piece out of the slot.

There is a bit of cogging with the 4 magnets, although I wouldn't expect any output in low windspeeds and the cogging isn't severe enough to keep things from breaking loose at a moderate windspeed where this small gen could actually create power. It can be turned by hand, although it certainly doesn't free-spin. The image below is a video clip 263kb of the motor being spun with the armature in place. It is still on the original bushings.

The next thing happening is plasti-dip to recoat the stator. The rotor will be poured in fiberglass resin. Providing the resin sets overnight, it will go to work tomorrow to be turned down. Bearings will be coming soon.

The armature of the lakewood is now poured. I had an idea which I wasn't too sure of, so I didn't post until it proved out. Here's how it goes... First, the magnets were superglued in place. A cheap plastic food container ($2 at Kroger) was drilled to allow the shaft to go through. I hot glued the face of the armature to the bottom of the container to keep the fiberglass from getting on the shaft.

I was primarily concerned about the hot glue giving loose & letting fiberglass pour down the shaft. It leaked very slightly, but not enough to make it to the shaft.

Aside from wasting a little fiberglass due to the large diameter, it worked flawlessly.

The cured shaft went in to work this morning, and got turned down. The gap from magnet face to stator tooth is .02" There is .01" fiberglass over the top of the magnets. It looks kinda like a new age roller skate wheel.

So much for the plasti-dip. It seems to peel & split too easily. My first wire wrap shorted to the stator, so strips of good old electrical tape are now working well. The first test coil on the lakewood is 225 wraps of 25 gauge. 2.5 ohm resistance. Using a craftsman drill to drive it (rated at 1200 rpm) it is showing 7 volts. There's enough current to throw a spark when shorted. Am I thinking right that with 4 mags, 2 poles in series producing current, I would have to run 550 wraps per pole to get the 12v cutin below 600 rpm?

It is starting to look like all six poles would be required to produce any amperage, and with a .5 inch thick armature & stator and the wide stator segments this isn't going to produce much of anything.

225 turns of 25 gauge wire per tooth checks around 16-18vac per pair at just over 1000 rpm and throws quite a spark when shorted out. Rewinding this motor is not a fun task with my fat fingers. The tape over the coils is to hold the leads in place while I'm winding. One more to go.

The completed lakewood box fan mini mill. 2.8 amps short circuit. 12v cutin around 500 rpm. Very low budget.

225 turns per tooth 25 awg salvaged wire. 4 neos. Wired 4 lead star. Two rectifiers mounted inside the can, 2 leads coming out. It is definitely not a powerhouse mill, although it is small, lightweight, and can actually generate a couple amps of useable power for very little invested.