hall effect sensors for magnet position

Question

this may be more of a geometry question than an EE one, but i'll ask it here anyway. I'm trying to build a weather vane, and am looking at using several linear hall effect sensors (such as the allegro 1324) laid out around a shaft with a free-spinning magnet on it.

how many sensors would i need? this is what i'm looking at doing now (they're the sot23w package)

i could get away with 3 laid out on the corners of an equilateral triangle, but... can i get away with half of the schematic above, and just look at the effect from the north pole vs the south pole?

Answer 1

Your magnet has a south pole and a north pole. A noth pole will produce the opposite signal in a Hall sensor compared to a south pole and so, if you have two Hall sensors at 90 degrees, you can infer any angle from the two readings you acquire.

You have to watch out for increased noise though but this should not be a problem with reasonably sensitive devices. Check the data sheet. Worst case scenario for noise is this: -

The magnet poles are furthest away from both sensor active areas. Any rotation in either direction makes the magnetic field stronger on one sensor and if these sensors are calibrated and fairly accurate you can use the sensor that has the strongest reading to determine angle.

Having said that I'd probably go for three sensors at 120 degrees.

Answer 2

Check out the MagAlpha devices. Only one sensor will do the whole job and the precision will be much higher. You just need a ring magnet that is magnetized diametrically.