Analyzing The Wire Thickness Error
Here's the
easy way to look at why the no load speed went up...
20 AWG wire has a resistance of 0.0333 ohm/m
22 AWG wire has a resistance of 0.0530 ohm/m
...now the turns are the same length no matter what thickness you use, so you can do a simple multiplication :
0.0333 ohm/m * 18 turns * length = 0.5994
0.0530 ohm/m * 27 turns * length = 1.4310
...taking the ratio we get:
1.4310 / 0.5994 =
2.387
...no we multiply the stock no load speed by this and get:
No load speed 36V 3000 * 2.387 = 7162 rpm
...reducing to match the 24 volts I'm actually using:
7162 rpm * 24/36 = 4774 rpm
...but 27 turns of wire actually produces more magnetic force so:
18 turns / 27 turns = 0.6667 or 1/
1.5
4774 rpm * 18/27 = 3182 rpm
...which just so happens to match the bike. It's interesting that the Triple wind resistance change (due to picking the wrong gauge wire) added a speed factor of 2.387 and took away a factor of 1.5. So a change in thickness has a non-linear relationship to the no load speed. (probably because a cross section is "pie r squared")
-----------------------------------
The overall resistance of the motor was reduced because of the Triple wind design, but the magnetic flux density remains constant because it's tied to voltage and no matter how much you divide the coil up into separate winds
the DENSITY of the magnetic flux is bound by the thickness-to-voltage relationship. It's the density that went wrong... the total magnetic force goes up because of the reduced resistance overall... but the density does not and that's what actually creates the backemf and sets the limit of the no load speed.
Just keep the wire gauge constant... it makes things easier...