My original RV120 motor had hall sensors that did not function. The signals were disturbed by stray fields from the coils due to their poor postion and wiring. Therefore i mounted new sensors on the outside of the motor in this thread:
Now after two winters of riding the sensors started giving bad signals. It turns out there are two issues:
1) The sensor magnets are badly corroded, see cracked surface treatment and corroded neodym in this pic
2) One of the sensors magnets fell off. I used 14 magnets to represent the 14 motor poles. When one falls off or stops working then that's really bad as the latching sensors stay in the same state until they get switched by an opposite polarity magnet. Commutation is then totally wrong.
I decided to mill seats for a complete magnet ring instead. Four magnets then represent each motor pole. If one magnet falls off then it will still work.
The corrosion is harder to solve as we have tons of salt on the road here in winter. I'll spray the sensor magnets with an extra layer of Zinc spray to further protect them.
One thing i noted when changing the side cover was that the motor magnets seemed to be starting corroding also. That means the end of my motor eventually
I built a new sensor body equipped with ss411 bipolar halls as the latching sensors are no longer needed with a full magnet ring.
Bi-polar, non-latching sensors are supposed to give better efficiency as latching sensors normally have a larger hysteresis; latching sensors will then switch a bit later and cause the motor to lose some of the possible torque and efficiency.
I checked the data sheets for the respective sensors i've used again to see how they compare:
SS411 sensor (often used&recommended here):
Sensor is quite stable over temperature
Sensor typically switches in 0.15 microseconds
SS460P latching sensor which i used for the first external sensor body:
I realize that the SS411 hall sensors have a larger hysteresis than the 460P sensor (60 vs 30G) but typically switches 10x faster (0.15 vs 1.5).
Max switching distance for the sensor magnets i use can be calculated at KJ magnetics site:
The magnets will switch the sensors reliably at a distance of around 7mm (a little more than a 1/4")
I think overall that the SS411 hall sensor is a better choice for commutation sensor than the 460P (in theory) but i guess that the unavoidable misalignment of the sensor in the sensor body contributes more to losses than the switching time difference..