I took bike.3 out for another test, this time giving it 50 volts. As soon as I powered it up, the voltmeter stopped working; I think it's died. Not a big problem, those little votmeters only cost about £1.50. IThe bike went very nicely at level I (50%), then I switched it up to level II and ... sput.
The motor pretty much stopped working, it would growl a bit, but no real power. So I pedalled home, hoping that I hadn't broken something important.
I have an e-bike tester, it's a nice little gadget, cost about $10. It'll test motor windings, motor Halls, controller power output, controller Hall control and throttle (which for some reason it calls "steering"). So I disconnected the controller from the motor, and tested the motor. Two of the three windings worked, the third (the yellow one) didn't. That's looking bad. But before I thought about opening up the motor, I thought I'd check the connections to the motor, and when I unpeeled the insulating tape, the yellow connection looked very ropey. I'd used small plastic connectors, probably rated at 6 amps, and I think that with the new higher current, they just weren't up to the job. So I cut away the bad bit, and made a new connection, this time using 30 amp connectors. I connected up the bike, and bingo! It works.
I'm going to London again tomorrow. I'm planning to go round the M25 and down the M11, I'll park at Forest Gate, near the cemetary (I like parking near cemetaries, there's usually good parking there), and then I'll take the bike in to the City from there. I'm hoping to go all the way in, because I want to have another go at London Invasion. I've tried three times so far, but now I have a very different theory of where the final is.