The Breakdown Lane
Mission has even bigger problems. Like MotoCzysz, its bike completed one of the qualifying laps and broke down in the other–but the team has no idea why. It’s the night before the big race, the one that counts; the bike is busted, and all Mission really knows is what its rider Tom Montano can describe. The bike was feeling really good–fast, even, he says. He was passing other riders left and right, and then the machine just gave out. “All I can compare it to,” Montano says, “is when a gas bike starts lugging and then binds up.”
Hearing this, Jon Wagner, Mission’s CTO, gets on his hands and knees and opens the bike’s power plant. It sits low in the bike’s frame just forward of the swing arm. “I’m getting a sinking feeling that we’ve got a jenky motor,” Wagner says. Placing the two probes of a digital multimeter to the motor’s guts, he takes three measurements of internal resistance: .018 ohms for the first and .021 ohms for the second two. The measurements are consistent with a short in one of the three windings. “We may have to take this thing apart and relacquer the coils,” he concludes.
Wagner has found the failure, but that doesn’t explain why the motor quit in the first place. Mission was counting on its custom software to give it an edge, but forget stunts like “Segway mode”–the Mission bike didn’t even have brains enough to shunt current away from an overheating motor. Even worse, when data-acquisition tech Ray Shan downloads the race log from the bike, he finds that Mission would have been better off if it hadn’t used a race computer at all. “We completed 31 percent of the track before we broke down,” says Shan, in disbelief, “but we used 40 percent of our total power.” Even if the motor hadn’t blown, the bike would have run out of juice before the end of the qualifier.
It’s Seth LaForge, Mission’s lead software engineer, formerly of Google, who starts to connect the dots. What if the software loaded on the ride computer was not updated to account for the larger sprocket that was swapped onto the back wheel before the race? Then the bike would be running faster than its speedometer would indicate–and this would explain why Montano reported passing other riders left and right.
To test LaForge’s hypothesis, Shan recalculates the bike’s speed by extrapolating from the tachometer data. Since electric bikes generally don’t have gearboxes, the relationship between rotor speed and actual speed is fixed. The revised speed calculations indicate that the bike was topping 100 miles per hour for the first seven miles of the course–an energy-guzzling pace, for sure. But why didn’t the bike just run out of charge before the finish, like the MotoCzysz bike? Why did it break down instead? The answer comes when Shan superimposes the corrected speed data onto a motor efficiency map. “One hundred miles per hour is right at the edge of the chart,” says LaForge, gasping a little when he sees the graph. The bike was redlining the entire way, dumping energy in the form of heat. A faulty setting in the motor control software was feeding the motor too much electricity. The bike just cooked itself.
LaForge would be a hero, except it’s his code that didn’t account for the larger gear in the first place. Garbage in, catastrophic motor failure out. The team works all night to replace the motor.