Since 2018 I have been working on an ebike project where I convert a regular Costco-brand bicycle to electric. I started with a prototype using a proof-of-concept electrical system, and over the years have added upgrades such as a new battery pack. The following is a gallery of the progress.
Starting with the electromechanical side of things, the power plant is a 6374 size, 194KV brush-less motor from Turnigy. This kind of motor is really meant for large-propped model airplanes and probably quad-copters too, but myself and others have repurposed them into motors for locomotion, such as in my electric skateboard project. The motor has a max voltage rating of 12S (50.4V), max loading of 70A, but interestingly, a lower-than expected power rating of 2250W.
The motor was mounted to the rear wheel of the bike. The brake rotor was replaced with a water-jet cut sprocket (#25 ANSI chain), and a much smaller sprocket was used on the motor. The smaller sprocket was not adequately sized for the application (not enough teeth engaged). An aluminum bracket holds the motor to the frame where the brake pads were originally.
Side view of the motor mount. About 70% of braking is done by the front brake, and regenerative braking is available on the motor side, so it’s not a stretch to get rid of the rear brake. The exposed motor windings are not great; a shield would go a long way to protecting the motor from dirty water. The only other issue is a small misalignment of the motor relative to the driven sprocket. This results in torsion on the singular motor shaft bearing which eventually caused premature wear.
This was the first prototype of the e-bike. The battery is two 5Ah 3S packs in series, with discharge rating of 20C. The ESC is a no-name brand 1/10 scale model ESC for RC trucks. The throttle was provided through an RC remote control for RC trucks. This setup served to validate the system but was obviously not very usable since you had to have the controller in one hand. The ESC was also extremely loud as it utilized a BLDC-type control scheme. Liberal amounts of tape held everything together.
The next iteration saw an upgrade of the ESC and enclosure. The ESC I used was the VESC (Vedder ESC), an open-source project by Benjamin Vedder. This controller is commonly used in electric skateboards as well as ebikes and was one that I had lying around. With a firmware upgrade the VESC supports FOC (field oriented control) mode which almost eliminates all noise from the electrical switching of the motor and provides higher torque. Cogging was also no longer an issue at lower speeds. The throttle was changed to a thumb-throttle which required a soldered connection on the VESC. No BMS on this battery pack.
Side view of the bike at this point in time. The project box was made out of high-density polyethylene. While the battery provided lots of power, it had a limited capacity, and more importantly, no BMS.
I decided to upgrade the battery pack by creating a new one out of space 18650 cells. I harvested these cells over many years from new-old-stock laptop battery packs from eBay. With the shift away from cylindrical cells towards pouch-style cells in modern PC laptops, I believe that many laptop repair services have found themselves with plenty of inventory for 18650 packs and nobody to sell them to.
The cells were assembled into an inexpensive carrier case and spot welded together. 10S4P battery pack. The spot welder used was the kWeld, which works by essentially shorting another lipo battery across the metal strip you are using. Frankly, it was very difficult to use and required optimal conditions to create even a half decent weld. The system is sensitive to the battery that you use for welding (I used a 3S 10Ah 30C pack I believe), as well as temperature and the wiring harness used to provide power to the unit. In my case, at cold temperatures the system would fail to weld because of an over-current condition. As soon as I could get one weld though, the battery and wires would heat up, increasing resistance and decreasing weld power. Eventually, the battery would heat up and I would be forced to stop welding. Warm 30C lipos scare me.
Of course this time I also added a BMS to the unit. A standard, no-name, 10S BMS with balance charging. I think the current limit is set to 15A by default, but don’t quote me on that. With the cells that I am using I don’t want to push them harder than 2C, so with the 310Wh battery this equates to about 600W of power. I would later set a current limit on the VESC to limit battery output.
Assembling the battery pack. I tried to fit the VESC inside the battery case with the BMS but could not quite get the dimensions to line up. The red PCB you see is one of my high-power LED switches that switches power to the VESC. Why that functionality couldn’t be part of the VESC is baffling to me. Beggars not choosers though.
Does not fit in the case. The charging plug is shown here, a 2S “waterproof” port from eBay is wired to the BMS.
The entire battery pack was later protected with a thin layer of foam and plenty of shrink-wrap. The switch was placed in the unit with the BMS while the VESC stayed outside the case.
Since the VESC could not fit inside the battery case I bought an inexpensive project box specifically for this application and placed it on the opposite side of the bike frame. In the bottom you can also see a three-pin connector for the throttle potentiometer (or hall effect, I’m not sure).
How everything fits with the battery and VESC.
Now with the case on the VESC side of the electronics.
At the end of the day the bike performs well and probably costs a fraction of the cost it would have taken to build with a kit. The max range without any assist is roughly 10km, and about 30km with moderate pedal assist. Top speed is around 15km/hr without assist and almost 30km/h with assist. Not too shabby for something cobbled together mostly using parts lying around!