The obvious question, and I certainly hope the answer is, "not as fast as it will when I put the new gear on". Still, it'd be nice to know, so yesterday I took my GPS and set out to find out.

I tried for a bit of control over conditions. I found a reasonably straight section of bikepath, about 330m long, which didn't go up or down more than a meter or two (or three, the GPS could never make up its mind). I pumped up the tires, topped off the battery, measured the temperature (a cool 5 C), and scowled at the wind. I did flying starts, and used the GPS to measure average speed over the course while not pedalling. To try and adjust for the wind and slope, I averaged runs going in both directions, and also tried it both with and without the lights on, to see if that made any difference. To add one more measurement, I took the battery voltage after every run. As far as aerodynamics, I was wearing bulky street clothes and sat upright. So how did it do?

The average of the averages, over all the round trip runs, is 25.0 km/hr, std 0.4, with a max of 25.3 km/hr. If I do statistics against voltage and lights, I model a top speed of 25.6 +- 0.3 km/hr at 26.0 V with no lights, losing 0.4 km/hr for turning the lights on. At 24.0 V, this drops to 24.3 +- 0.5 km/hr. Lights do make a difference, as does the state of charge.

I discovered one more disturbing thing, though. At the end of the session, as I was starting what was to have been the last run, the motor cut out. I can't tell you what I was thinking, because it was more or less unprintable. This is the kind of behavior that caused me to buy a new motor in the first place. The motor was cool, so it's not heat. Some debugging showed that rebooting fixed the problem, and the motor could be reliably started at no / low load, but would reliably fail at high load. Now we're getting somewhere. I managed to get a voltage reading under load (I'm starting to want one of these or these), which showed that, though it might read > 24 V no load, the battery could drop to 22 V or below. I saw 21 V before the motor cut out. What I suspect is happening is that the controller sees the low voltage, and cuts out, to protect a battery pack from damage due to over discharging. In my case the voltage drop is due to too high an internal resistance. Another way to say it is that the batteries were effectively fully discharged, after only 6 km of actual use!

We'll see if it goes away with a fully charged battery, but I suspect that the motor is fine (who knows maybe the previous one was too), and the culprit is the old batteries. The solution is to baby these ones and start looking to replace them.