Rotational linear Hall Effect Sensor

Jose,

I would recommend you take a look at TMAG5170 or TMAG5273.  These devices have digital interfaces, and have integrated the CORDIC angle algorithm to calculate the resulting angle from the magnetic inputs to 0.25 degree resolution.  Tighter resolution can be achieved if using the XY,YZ,orZX outputs directly in the microcontroller.  The overall accuracy may depend on the quality of the magnet rotation, as mechanical errors will influence the amplitude, offset, phase, and distortion of the inputs.  

For some guidance about this function please consider these resources:

https://www.ti.com/lit/pdf/sbaa463

https://www.ti.com/lit/pdf/sbaa503

https://www.ti.com/lit/slya036

https://training.ti.com/angle-detection?context=1139747-1139746-1137749-1139635-1148102

https://www.ti.com/lit/zip/sbar012

The magnet selection will partly depend on the application and accessibility to mount the magnet on the motor shaft.  One approach could use a diametric cylinder magnet mounted on the end of the motor shaft.  This would be easiest accomplished by fixing a cup to the end of the shaft that will snugly seat the magnet.  You would then fix the magnet using an epoxy.  

If the end of the shaft is needed to connect to the load, then you could use a diametric ring magnet slid onto the shaft.  You will want to ensure that the magnet does not slip on the shaft during rotation, so it would similarly be necessary to either epoxy the magnet or use a clamp to hold the magnet in place.  

In either case, the rotating magnet will produce field components which are naturally 90 degrees out of phase that can be used to calculate the magnet angle.  However, you will find that in most cases, these inputs to the sensor are not equal in magnitude, and therefore would result in potentially large angle errors.  This can be corrected for using the Sensitivity Gain adjustment register settings of either TMAG5170 or TMAG5273.

I would expect that a diametrically magnetized toroid may work equally well, but may present some mounting challenges.

The ideal measurement case for any magnet will produce a pure sine/cosine relationship between the two components being measured.  This is most easily observed with the magnet on end of shaft and the sensor placed on the axis of rotation.

For a magnet supplier, you might consider the list of vendors at the bottom of this thread:

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/355909/faq-i-want-to-use-hall-effect-sensors---where-do-i-start?tisearch=e2e-sitesearch&keymatch=faq%25252525252525253Atrue%252525252525252520hall%252525252525252520effect#

Magnet cost will be the predominant factor in many solutions.  This may vary largely based on the size of the magnet and the quality and tolerances required.  Smaller magnets matching the vendors standard portfolio will be the most cost effective.  However, small magnets will produce less magnetic field which may result in reduced resolution.  

You can check availability of inventory on TI.com here:

https://www.ti.com/product/TMAG5170#order-quality

https://www.ti.com/product/TMAG5273#order-quality

Thanks,

Scott