I wanted to invoke thermal rollback on my Bafang powered Fighter so I could wind up the amps a bit.
So I went to the trouble of removing the wheel and stripping the motor to get to the windings. No problems so far. However, when it came time to get the thermocouple wires through the axle with the phase and hall wire bundle still in there, it just wasn't going to happen. The only way I was going to get the extra wires through would be to completely de solder the phase and hall wires, withdraw the bundle, and start again. Decided this was going to take just too much time and effort so reluctantly, I re assembled the motor and re fitted the wheel.
However, it wasn't a total loss and seeing the stator removed from the motor with just the axle attached to it got me thinking.
As far as thermal paths go, the motor windings are effectively in direct contact with motor shaft. With this in mind, it seemed fair to summate that temperature changes at the windings will also be seen at the motor axle. So the next question was just how reactive would the changes in temperature be at shaft in comparison to changes at the windings. Naturally there would be a significant DeltT between the windings and the shaft but this would be easy to compensate for.
So time for some testing.
A 10K thermistor was wired to the CA and inserted into the end of the hollow motor axle. It only goes in about 5mm but is held in place with thermal putty to make sure it has good thermal contact with the inside of the shaft.
Needed a starting point to set the CA to so I went with temperatures I typically see on the motor casing when measuring with a infrared thermometer. I know from previous experience including motor failures that 55 degC is a safe max temperature with anything over 60 degC being dangerous territory. with this in mind, rollback was set to 50 degC and cutout to 60 degC as a starting point.
So off for a ride at one of my local torture tracks to get some data.
First thing I noticed was that the temp at the motor shaft was surprisingly reactive to changes in motor load. Hard hill climbs would see temps rise quickly with coasting down the other side quickly seeing temps drop off again. My expectations were that the heat sink effect of the stator and motor shaft would make temperature reactions very slow but this certainly wasn't the case.
Every few minutes I would stop and measure the motor casing with my infrared thermometer and discovered that the two temperatures tracked quite closely but with a DeltT of around 8K (shaft temperature showing approx 8 degC lower then the motor case).
So with the data collected, l set the CA up for 42 degC rollback and 52 degC cutout.
Still plenty of testing to do but a rollback temperature setting of 42 degC certainly gave me plenty of scope to push the motor hard including 3000W peaks and some serious hill climbs. Pushing hard to finally see the rollback coming into play would see temps stabilise under the enforced power reduction setting.
So in summary, obviously having the thermistor directly on the motor windings is the correct and ideal way of measuring temperature. However if fitting an internal thermistor is not practical to carryout, reasonable motor protection through thermal rollback looks like it can be achieved with this very simple installation method.
Still needs plenty of testing but so far, the results are quite promising.