An Awesome Electric Bike


Converting a motorcycle to electric is always a favorite project of ours, and [Peter]’s build is up there with the rest of them.

The bike is a 2002 ZX6E he bought from a salvage shop. It had been parted out over the years and for $250 this very light aluminum frame made the for the perfect electric conversion frame. After learning MIG welding from his brother, [Peter] cut up a few plates and built a motor mount for his new 4.2 kW power plant.

The controller is a 300 amp IGBT he found on eBay, with an extraordinarily sturdy looking circuit built into an ammo box. The motor from the bike was replaced with 16 60Ah LiFe cells providing 52 volts. [Peter] also built his own battery management system using a Cypress PSoC 3 microcontroller and a beautiful custom PCB.

It’s still a long way from being finished, but already [Peter] has a great looking bike and an awesome weekend project on his hands.

16 thoughts on “An Awesome Electric Bike

  1. 4.2kW is way too wimpy to keep up with traffic, highway speeds are completely out of the picture, to be practical, it needs more power (8-10kW will give plenty of acceleration and still have enough to move fast)… also, for that kind of power, the 300A controller is rather overkill…

    1. One question right here: The 300A Controller might be Overkill. I’ve had to choose an controller for a different project the other day and the major nonsense narrative was, that you cant overrate a controller. You always want some overhead for spikes or a later upgrade. Can someone shine a bit of light into this corner and tell me: How to choose the right Amperage Controller?

      1. this controller is so high powered because he wants room for improvements. As long as you can handle current spikes from inductance and you have a good amount of wiggle room above peak current it should be fine. You should also derate igbts and fets, its standard to limit maximum voltage/current in your transistors to 50-60% of their rating

        1. Yeah, after looking at this thing, the 4.2kW motor is only 5.6 horsepower. I can’t imagine that’s enough to go very fast on a chassis like this. Heck, the brushless outrunner motor on my electric longboard is almost 3kW. Also, the 300A IGBT will likely need to be upgraded if he does get a more powerful motor.

      1. In the UK (seems the blog is also a 16 year old can ride a moped, they *roughly* define this as a motorcycle with a speed of upto 28mph which puts this firmly into the restricted 50cc zone. The piaggio zip 50 is just a standard vehicle fitting this limitation, 2.5Kw at shaft, just about all the so called “provisional” or “learner legal” motorcycles in the UK are around this mark too.

        4.2 will be better than the restricted 50. Restricted 50 copes fine around the city as it is.

        A non modified 80cc scooter engine apparently kicks out about 6 horsepower. 4.2Kw is approx 5.6 horsepower. So is around the lower end of the 80cc engine region.

  2. The motorcontroller won’t actually survive 300A for sustained periods, even though it says “continuous” in the datasheet. There’s a reason it also says “absolute maximum”, and the 300A only applies if you can maintain a junction temperature of 25C. Good luck with that… You’d usually want at least a safety factor of 1.5-2 on those ratings.

    Also, keep in mind that the motor current is always at least as large, and most of the time much large than the battery current, so you don’t actually need to draw 15kW from the battery before the motor current, and therefor the current through the IGBT, would exceed 300A. At low speeds, the motor voltage would be much smaller than the battery voltage, while the current would be proportionally larger (disregarding losses in the half-bridge). It’s basically a buck converter with a mechanical output instead of an electrical one.

    IGBTs aren’t exactly the most efficient choice for this type of application; at 150A, it would dissipate about 450W (while the IGBT is conducting, the losses in the diode somewhat smaller, about 300W). At the same time, a could of paralleled MOSFETs could have an total Rdson of 3 milliohms or less, resulting in power losses of about 70W. The main advantage of using an IGBT in such a controller is the fact that they are very tough suckers, and nearly indestructible as long as you don’t royally screw up the gate drive. I think he made the right choice; build something very sturdy but slightly inefficient first, you can always replace the motor controller with a more efficient design later on, when the rest of the build is done.

    However, I would really recommend actively controlling the motor current, to protect the motor, batteries, and IGBT, and also because current (and, therefor, torque) control drastically improves the throttle response, compared to a straight throttle to PWM conversion, which is used now, if I understand correctly. Measuring the motor current doesn’t have to be very difficult or expensive; the easiest way would be a hall effect current sensor that you can feed one of the cables through, these would cost somewhere between 25 and 50 euros, or alternatively you could measure the voltage drop across one of the motor cables; the one permanently connected to the battery (instead of the cable going to the half-bridge) would be easiest, because it doesn’t have a high frequency common mode voltage.

    1. I agree that the controller needs some more attention. I did something very similar about 6 years ago trying to build a motor controller for an electric car conversion. My controller looked nearly identical to his. I’ve learned a LOT since then, including what I did wrong back then, and I did a LOT wrong back then.

      He’s on the right track using the dual IGBT module for the body diode on the upper unit. What’s missing is bi-polar gate drive, desaturation protection, and current control.

      Bipolar gate drive can be achieved with a small DC-DC converter of an appropriate rating. 1W minimum should do it. The gate should be switched to -5v when turning off to ensure complete turn-off at high currents. This becomes an issue when the gate capacitance increases (proportional to Vce). It may be okay with just 52v supply though. He should calculate the time constant of his gate drive circuit and compare it to the switching period. Switching losses can cause the IGBT to heat up and fail very quickly under load. Also, gate voltages need to be closer to 20v for a good turn-on. Search for Powerex App Note BG2B for lots of good info on IGBT gate drive.

      Desaturation (overcurrent) protection is an absolute must. A desaturation event needs to be corrected (by turning off the gate) within 10uS. IGBTs are surprisingly easy to blow, and it’s expensive to replace them. This cannot be done in software or with hall-effect sensors. Hall effect sensor response time plus propagation delays will not work. I suggest using an HCPL-316 driver for this as it’s built in to the chip and is very effective. I’ve had very good luck with these devices.

      Current control can be done in his microcontroller with hall effect sensors. Ideally, the throttle would control motor current (and thus motor torque), not duty cycle (voltage). Speed control on the throttle instead of torque control is very difficult to handle when driving, and good throttle response is even more important on a motorbike. I’d suggest a PI controller with current setpoint (from throttle) as input and duty-cycle as output. Alternatively, an SG3524 or similar could be used and get rid of the microcontroller all together.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.