I was sending from Tapatalk, which won't work properly on my phone. I try and post a message and it stalls. Whether you go back or forwards, you can't see what's happened.
I'm not sure that our German friends have got it right. I discovered today that there are two elements to the gear-changing decision: Torque and speed. Both can be measured by the controller CPU using the current and hall-sensor signal.
If you force it into low gear, then stop it and lift the wheel off the ground, it always starts in high gear, but if your weight is on the bike, it starts in low gear.
There cannot be any sort of solenoid inside because there's no wire thick enough to power one. That's if there's any spare wire at all. I'm pretty sure that the white wire isn't connected. If the gear-changing were mechanical, there would have to be torque on the motor to change gear, but the changing happens while it's free-wheeling or stationary. Another point is how the controller would know to cut the power while it's changing. I can't see how it could be triggered. the changing has to be initiated by the controller - as I said by two simple conditions - speed and current (or manual switch). Also, to change gear from the switch would need a solenoid in the motor, which i don't believe it has.
Everything, to me, points to the controller making the motor reverse direction to change gear despite the patents banded about. I think they're red herrings. This method is simple, logical, and doesn't require any modification to the controller other than software. all it needs is a double ratchet/clutch in the motor.
The mystery will soon e solved when I get my next ones, which will give me the chance to open one.