I'm working on my second eBike project with a VESC. Both have used the Sempu T9 pedal sensor. This second one, however, has a noise problem. For the longest time, I thought it was a connector issue. After swapping out cables and sensors many times, I realized that's not it.
The T9 has "wireless" (inductive) power. The torque signal is 1.5V at rest, 3.0V full-scale. The supply voltage is 8-12V. (I'm using a 5-9V booster and a 9V battery for testing.) This works fine when I initially wire it up with a front wheel motor. But when I zip tie the cables to the frame, it gets buggy. The torque signal often dips close to 0V. The problem is clearly when the sensor cable is cinched to the frame. The cadence sensor, however, continues to function. (I doubt it uses the inductive power.)
If, however, I disconnect the motor, the sensor is fine. I can manipulate the connectors and the cable extensively without ever seeing a drop in torque signal. From my measurements, none of the motor legs has any direct connection to the fork. However, when even one of the motor legs is connected to the VESC, I see a >1.5V AC potential between the VESC ground and the fork. (And that's when the problems happen.) It seems odd that having only one leg of the motor could cause this problem.
I've ordered some shielding for cables. And ferrite beads. I'm considering a ground strap from the controller to the frame.
Any suggestions?