bobc wrote:I'll make up a jig with 3 sensors at 60degrees mechanical on the outside & see what the signals look like; but I do fear the signals will be noisy. (60degrees signals is a single signal inversion away feom being 120degrees signals, but far more attractive mechanically! Or I could just fit the middle sensor upside down??) Originally I thought the halls could be right next to each other - getting the 120degree phasing from a single magnet... but this thread has shown the error of that assumption!
You can glue the same number of magnets to the outside of the can as there are inside and use those for position sensor, you can use any sort of little magnet as long as each side only has one pole. I am pretty sure someone already mentioned that in this thread, I am just to lazy to try and find it. I think the flexible magnets (like you would put on a fridge) have south and north poles on the same side, so they won't work. Use a compass to make sure you alternate positions. To figure out the mechanical spacing between sensors, use the following logic/math.
each pole pair is 360electrical degrees
If you have 20 magnet poles in the rotor, that gives you 360degrees electrical is 18degrees mechanical. or 1degree electrical is 0.05degrees mechanical.
That means that 120degrees electrical is 6degrees mechanical. So as long as you put your sensors multiple of 6degrees apart from each other, you will get some form of 120 degree commutation, There is a little bit more to it to make sure that they are all in the proper phase, but if you keep them all equally spaced, it should work. The 60deree separation that you proposed should work fine, but something that comes out of the above calculation is the precision that you have to have on your sensor placement. If you are out by 2 mechanical degrees, you have just moved your timing out by 40electrical degrees, so when you built your sensor mount, you need to ensure that the sensors are accurately placed relative to each other.