Motorized Bicycle
Project Background

Some time after I attempted to build a Segway, midway though 2012 in my sophomore year of high school, I realized that a motorized bike was a much more feasible project for my level of knowledge at the time than a working Segway. In addition, I felt that the required materials would be far less expensive. Using spare parts that I already had from other projects, some store-bought materials, a few motor scooter motors, and large lead acid batteries, I was able to convert a bike into a mostly-working motorized vehicle. Operating it was fairly dangerous when I first finished building it, but I have made many improvements to its design and have more planned for the future. I plan to incorporate a microcontroller to pulse-width-modulate the motor-power signal with MOSFETs to decrease the initial current spike when accelerating and to allow the continuous adjustment of speed while riding. I will likely pick this project back up when I once again have somewhere to temporarily store and work on such a large device.

Detailed Description

I began by purchasing a few DC motors online and drawing up a few sketches of ideal the framing necessary to mount those motors to a bike’s frame. I then picked up a cheap used bike from a nearby garage sale and bought some weldable steel angle iron and 3/8-inch nuts and bolts from Sears Hardware down the street. Using the weldable steel and a few nuts and bolts I formed the frame that I had drawn, then I drilled holes into it near the top in the general location where it would interface rigidly with the bike. The motors were mounted to my steel frame using many of the remaining fasteners that I had purchased, and I experimented a bit to figure out exactly where the frame would need to be attached to the bike to accommodate for the primary sprocket’s position and the available space on the bike frame that overlapped with the motor frame.

My goal with these projects has usually been to get to the final product without spending a ton of money whenever possible, so I picked up scraps to use whenever I could find them. I found a large, thick aluminum sheet which I cut and used to interface the frame to the bike by effectively increasing the overlapping area that they shared, then I proceeded with setting up the sprockets so that the motor could drive the bike chain. I realized later on that the tooth size on the motor scooter sprockets and chain really was too small to be driving the loads that the bike would be subjected to. Sometimes the frame would flex, once I had the bike working, and the chain would jump over a few teeth, which slowly degraded my primary sprocket.

My original design had included a 4:1 gear ratio between each motor and the chain-drive sprocket. I eventually adjusted that to be a 5:1 ratio to reduce the torque experienced by the primary sprocket, which I expected to be the root cause of the chain skipping over teeth. The pedals of the bike were in the way of my motor frame, and in my design were unnecessary as the motors were to replace manpower, so I removed them, leaving just the axle with three of the bike’s original sprockets from its 21-speed drive system. To interface the sprocket driven by the motor with the sprockets driving the bike wheel, I used an aluminum hub from a motor scooter wheel as a spacer and bolted them together as concentrically as I could manage by eye using a few 3/8-inch fasteners.

The plan that I started with for combining the motors, sprockets and chain had many flaws. I positioned the three 280-watt motors in a triangle surrounding the larger sprocket expecting that the chain would contact the large sprocket at the three sides of the triangle and believing that that contact would be enough to keep them connected. At the time I was not familiar with the rules regarding how much a chain should wrap around a gear to have good, consistent contact, and to prevent the chain from jumping off or skipping teeth. In a perfectly rigid model this would have worked, but it was far from rigid, and I quickly learned that the chain would not stay in contact with the internal sprocket. I tried a few tricks to force them into contact that, looking back on them today, were extremely poor engineering solutions.


There was plenty of aluminum available from the motor scooters that I had gutted for parts which I used as a base on which to mount three ball bearings behind the chain in order to put pressure on it in the radial direction, towards the central sprocket, essentially forcing it into contact with the sprocket teeth. I made these fine-tune-able so I would be able to adjust their distances from the chain, not too far so they would do their job and keep the chain in contact with the sprocket, but not too close so they wouldn’t grind together. There are many reasons why I wouldn’t do this today, and I’m glad to say this was removed from the current design. That did work slightly, enough to film one of the videos I have of the motors propelling the wheels, but when loaded with a human body the interface still fell far short of the required torque transmission without severely damaging components.


Simply repositioning the motors was all that I needed to do to increase the amount of chain wrapped around each of the sprockets, so that’s what I ended up doing next. The new chain path was a major improvement as it then entirely prevented the chain from jumping once I replaced the old damaged sprocket.

All that remained was to mount the 18 Amp-hour 12-volt batteries that I had purchased to be the bike’s power supply to the back end of the bike frame. As a temporary solution, I wrapped ratchet-straps around the bike frame and both batteries and suspended them from the top structural bar of the bike. Using this setup, I was able to ride the bike without the chain skipping and without any electronic failures and achieved a maximum speed of 26 miles per hour on flat ground. My only method of speed control available was a third break lever that I installed on the right handle bar which was set to pull a cable attached to a momentary switch that would trigger all 3 relays.

The relays were set up in parallel with each other in order to share the loads from the three motors and prevent them from welding themselves shut from excessive inrush current, however ultimately that is what happened. I made the mistake of creating a dead-man switch that controlled only the low-power signal sent through the relays. There was no real high-load switch wired in with the motors, so when a single relay was welded closed, it greatly damaged one of the motors and destroyed that relay. The relay was welded shut, stuck in the on-state due to high inrush current while I was testing the limits of the vehicle.

When starting from a complete stop, the momentary current through the motors supplying the torque necessary for acceleration peaks at tens to hundreds of amps. I had a fairly rudimentary knowledge of circuits, current, voltage and power at the time and failed to recognize that the system would behave that way under load. In order to shut off the motor that was stuck on, which at the time was locked in place, preventing it from spinning, I had to quickly pull apart the connection between its wires and the batteries. I have since added a real kill switch in case I have that problem again.

For a few years I set this project aside, but I took another look at it during my junior year of college when I had a larger workspace. At the time I was very much interested in redesigning the motor apparatus to operate with six motors instead of only three which posed a few new design problems. After the conclusion of my sophomore year, I was finally armed with knowledge of engineering 3D modeling and was excited to test out what I had learned, so I modeled each component of the motor apparatus with my desired dimensions, then I brought them together into an assembly. It wasn’t until the end of 2018, following my graduation, that I taught myself to animate the mechanisms that I had I made within Creo which required a major rework of the assembly and its motion constraints, but the building blocks were all in good shape.

I wanted the bike to be able to reach speeds upwards of 37 miles per hour which, I calculated, would be achievable with six motors running electrically in parallel. The main changes that I made to achieve this included the addition of three new 280-watt motors for a total of six, a welded frame instead of one held together with 3/8-inch bolts, and a heavy-duty size-420 chain with a ½-inch pitch. The chain was essentially a high-strength bike chain meant to stay on one track, and not designed to switch sprockets. The final design contained six 1/8-inch-thick 11-tooth sprockets attached to the motor heads as well as one 44-tooth primary sprocket centered within the motors.

During my original attempt to build the electric bike, I realized how weak the motor scooter chain that I was using to drive it really was. The tooth size was far too small for my application and could not transmit large amounts of torque without slipping and damaging the teeth. This also was largely influenced by the rigidity of the system which was not perfect, evident in the fact that the frame would flex towards the large sprocket whenever the motors abruptly started. It was because of this that I decided to weld the frame and switch to a chain designed to handle much larger loads.

While visiting one of the engineering machine shop’s labs and using their band saw to cut a small piece of metal for my arm brace project, I began talking to one of the TAs that worked there. After I described the projects that I was working on for a bit he offered to waterjet-cut the sprockets that I had designed in Creo for my bike. A few weeks later I had them in hand and began to grind chamfers and rounds into its sharp edges, as well as a bit of clearance that I hadn’t accounted for in my designs. I then tumbled them for a while in my metal tumbler to remove any burrs that I had missed and polish the surfaces.

The larger 44-tooth sprocket was easy to find on eBay for a few dollars, so I purchased it and welded it to the bike sprocket assembly. I included a few spacers in between the welded sprockets to provide the spacing that I calculated in Creo would be necessary to align the sprocket with the motors.

The old frame had a large number of holes in it due to previous attempts at mounting the motors. I took it apart, cut its components to their new shapes following the dimensions given by my Creo model, and filled every hole using my arc welder. After grinding the filled holes down to be flush with the surface, I laid them out on the concrete and welded the three sections into a much more rigid frame than the original had been. I marked out the hole pattern for each of the motors, again using my Creo model as a reference for dimensions, then I drilled them all out with a hand drill.


Using the drill press that I had recently acquired, I drilled three half-inch holes through the top of the motor frame to match the half-inch bolts that would soon interface the bike with the motor apparatus. This ensured that the holes were drilled normal to the back face of the frame, which would have been very difficult to achieve using a hand drill. I then drilled matching holes in the aluminum plate on the bike and re-drilled the holes for the fasteners holding the aluminum plate to the bike because it needed to be repositioned to accept the modified motor apparatus.

I was able to get most components to mesh nicely once everything was mounted but wanted to implement a tensioner for the heavy chain to pull out slack and prevent it from slipping off of the sprockets. I attempted to do so with some bearings and a spare 11-tooth sprocket, however there is barely any room to add an additional mechanism, so it was really crammed in and could potentially interfere with the rest of the mechanism. There is no room at the top of the apparatus despite that section containing the longest stretch of open chain simply because the chain passes between the three half-inch bolts at that location, causing the chain’s path to be predetermined along the top. On either side of the primary sprocket there is a small stretch of open chain but not much metal available on the frame to which a tensioning mechanism could be attached. My first attempt to add the tensioner to one of these sides was subpar, so I am designing a better attachment method that I will implement in the near future.

With the addition of the new sprockets, the welded frame and half-inch bolts replacing 3/8-inch bolts, the entire system is far more rigid and well-equipped to handle the 1680 watts of continuous output power that the motors can deliver. I have two main prospects for this project in the future. I don’t think that it is safe to ride in its current state at all both because the thin steel bike frame is flimsy and because of the many holes that I have added to it. I either plan to replace the bike with a heavier, much more rigid and high-strength bike, and using as few structurally compromising fasteners as possible, or welding my own frame which is the far more likely, once I acquire a tig welder and a better workspace. The welds that I can achieve with my arc welder suffice for now but are extremely messy and hard to control, so I wish to try my hand at a more professional form of welding and become more skilled at welding in general.

In addition to the issues with the frame, the current setup for supplying power to the motors is less than ideal. Using an on-off switch to toggle relays into two states is a poor and dangerous way to control motors. Using what I’ve learned in my work with Arduino, I know that I can easily write a program to pulse-width-modulate an output signal to high-current-capacity MOSFETs which can power the motors as a much better alternative. I will implement a potentiometer within the bike handle that will increase in resistance as the handle is twisted, as most handle-controlled electric vehicles use, or I will use some type of encoder or rotary resolver. The voltage across the potentiometer will be read as an analog input into the Arduino which will interpret the voltage strength and convert its magnitude into corresponding pulse widths to send to the motor MOSFETs, allowing me to effectively control power sent to the motors and drastically decreasing inrush current.

I have yet to decide what method I will use to prevent static from frying my MOSFETS, as they essentially run the whole system and their failure would cripple the machine. Preventing the MOSFETs from failing closed and thus preventing them from forcing the bike to run at full power with no easy way to turn it off would be preferable. I expect that a combination of traditional diodes and Zener diodes as well as resistors would be preferable, but I’m also sure that circuitry meant for this exact purpose has been designed before and is available for purchase online. I will be looking into that before I move ahead with designing my circuit.

The PWM frequency of the power signal is also limited by the Arduino’s hardware which may force me to look for different microcontrollers. I will need to either look up specifications of the motors or test the response of the motors myself using an oscilloscope and choose a pulse-frequency tailored to that response. If the resulting frequency is within what the Arduino can produce, I will most likely run with Arduino because of my experience with the Arduino IDE, Arduino’s pin interfaces, and its limitations. I have a lot of work ahead, and much to learn about circuitry before I will consider the vehicle safe to ride, but this project has already taught me a fair amount about CAD, chains and sprockets, wiring and part-fabrication through trial and error and by pointing me towards useful topics to research.

  • LinkedIn
  • YouTube
  • Facebook