First, did you save all the orignal settings in PR and CA, to restore them later after experimenting if necessary? (if not, you should do this now).
Next, have you tried connecting throttle directly to PR without the CA connected? This will eliminate all the CA settings, so that if it still has the same problem you can then tune the PR until it works correctly, and *then* put the CA back in the loop to verify operation.
If the PR doesn't have the PAS sensor directly connected to it, you don't want to tell it it has one; it should be set so that it does not have one.
If the CA has the PAS sensor connected to it, then *it* will read the sensor and provide a throttle signal to the PR when it is used.
If you want to use the PAS sensor on the PR, you have to direclty connect it to the PR, and not to the CA, and then configure the PR to use it as you wish.
For a torque-controlling system like the PR, it is going to operate normally only when loaded. Unloaded, a tiny amoutn of throttle may give full wheel speed, and if there are any speed or other limits in the system, it may then cease motor output, rapidly drop in speed and then reapply what looks like full power (but isn't), and keep doing this.
To do a real test of an FOC controller or any ocnytroller modulating torque (current) instead of speed (voltage) you have to test it while riding, or with a load applied (like a dyno).
For a geared hubmotor with a freewheeling clutch like the Ezee, it will make the normal motor noise and loosely spin the wheel (from friction in the clutch) backwards if it is being driven in reverse.
If sometimes the motor spins forwards and sometimes backwards, then my first guess is the ERPM**** of the motor is too high for the PR to keep up with the sensor outputs. Or if the hall sensors are not being read correctly for any other reason, so it is falling back to sensorless, and is not always detecing the correct direction of hte motor when commanded to spin.
****some geared hubmotors ahve a high pole count, so that times the reduction means that the ocntroller has to do a lot more work in a lot less time, and a lot of controllers just can't keep up with that. If it's in a small wheel at high speed, it can make a system unusable....or offground for a test with a torque-based controller that spins up fast welll past intended max speed with just a little throttle input.
