IEEE websites place cookies on your device to give you the best user experience. By using our websites, you agree to the placement of these cookies. To learn more, read our Privacy Policy.
Electrifying a bike can be electrifyingly easy
This simple electric-bike conversion involves just a few basic components: a brushless motor mounted under the chainstays [A]; a lithium-ion battery pack, electronic speed controller, and Arduino, housed in a normal trunk bag [B]; a Drok meter [C], two Hall-effect sensors [D] and a momentary rocker switch [E], all mounted to the handlebars.
In 2009, I wrote in these pages about my efforts to add an electric motor to a bicycle—a project that involved a considerable outlay of money and many hours in a machine shop behind a lathe. At the time, e-bikes were still rather exotic, at least in the United States, so it seemed worth the expense and effort to build my own.
Fast forward a dozen years. E-bike availability and ridership have exploded. Purpose-built electric bikes and e-bike conversions using hub-motors are now commonplace. One thing that hasn't really changed, though, is the expense. If anything, it's gone up. Consider the two e-bikes described in this month's Gizmo column, which each cost thousands of dollars.
That got me thinking, how hard would it be, and how much would it cost, to do a simple conversion now? Thanks to a few recent technical and commercial developments, I was able to come up with an e-bike conversion that cost me less than US $200 and yet functions impressively.
The approach I adopted is mechanically as simple as you can get. It uses an "outrunner" motor designed for electric skateboards. With such outrunners, the case of the motor rotates. Driving the bike with it merely requires mounting it to the frame in such a way that the case is in contact with the rear wheel.
I'd read about such friction-drive arrangements and had followed the work of a cyclist in Australia known as Kepler who had devised a clever articulated mount that allows the motor to retract from the wheel when not applying power. He'd been selling kits for his mount in small numbers. But Kepler's newest approach keeps the motor in constant contact with the wheel, which allows for regenerative braking. I thought: "What could be easier?"
So I bought an outrunner motor designed for electric skateboards for $88 and fashioned a mount for it out of a few pieces of scrap aluminum. The motor attaches to my bike under the chainstays, directly behind the bottom bracket. In that position, it's hardly visible. While it would have been nice to engineer some sort of spring-driven mechanism to control how forcefully the motor pushes against the tire, I kept things simple: I just deflated the tire somewhat, bolted the motor solidly in place, and reinflated the tire.
Most components are inexpensive and readily available. The only part I needed to fabricate was an aluminum mounting bracket for the motor [shown in outline]. James Provost
I next needed to decide on an electronic speed controller (ESC) to power the motor. For that I took advantage of the open-source VESC Project created by Benjamin Vedder. His work includes sophisticated hardware and associated firmware, as well as elaborate configuration software. Being an open-source design no doubt keeps the price of this hardware down: The controller I bought set me back a mere $85.
The third essential element of any e-bike is, of course, the battery. Here I took advantage of already owning an electric lawnmower, figuring that the 40-volt, 4-ampere-hour battery I had for it would serve well. All that was needed was a suitable adapter, which I bought for $18 from Terrafirma Technology, a company started by an engineer named Michael House who has gone into business supplying 3D-printed adapters to people who want to repurpose lithium-ion power-tool batteries.
On the handlebars are three controls: a momentary rocker switch operated by my right thumb and two Hall-effect sensors, which are switched when I squeeze the brake levers, to which I glued small neodymium magnets. These switches are wired to an Arduino Nano, which sends the appropriate signal to the speed controller.
The lack of noise, combined with the discreet positioning of the motor and other components, makes it hard for anyone to notice that I'm on an e-bike.
The most challenging part of this project was figuring out how to configure the ESC. Vedder's software contains "wizards" to help, but it's still complicated. Using it forced me to consider the difference between motor amps and battery amps and the virtues and drawbacks of controlling a motor by setting a target motor current or by changing the applied duty cycle.
I settled on controlling the motor current, which I adjust with the thumb-operated rocker switch on the handlebars. Squeezing the rear brake (which I'd normally rarely use on a road bike) applies a small amount of regenerative braking before the pads make contact–just as in my wife's Prius. Applying the front brake cuts power to the motor without triggering regenerative braking.
My first test ride convinced me that I needed some quantitative feedback about what the motor was doing. It would have been possible to have such information displayed on a smartphone, using a Bluetooth module plugged into the ESC. But I was in a hurry and simply mounted on the handlebars a Drok meter, which I had purchased for an earlier bicycle-based battery-charging project.
Setting the speed controller to use current control initially confused me when I tested it with the rear wheel in the air. With the power setting fixed, the wheel would spin increasingly fast and the amperage drawn from the battery would steadily increase. Online discussions convinced me, however, that this is the expected behavior when commanding a motor to maintain a fixed amperage in its windings (thus applying constant torque) while under little load
The handlebar-mounted controls connect with an Arduino, which interprets their settings and translates them into a PPM signal, which is applied to the electronic speed controller. The speed controller is connected to the motor and battery in the usual fashion, with an automotive-type blade fuse in line for safety.James Provost
This took some getting used to, but now I have the strategy down: I use the thumb switch to control how fast the bike accelerates and when I reach target speed, I tap my front brake, which returns the controller current setting to neutral. I can then goose it up from there as needed to maintain the desired speed.
The electric assist is virtually silent, because there's no gearing to make noise and because Vedder's electronics use quiet field-oriented control to power the motor. At speed I can't hear the motor at all. The lack of noise, combined with the discreet positioning of the motor and the fact that the battery and electronics are hidden inside a trunk bag, makes it hard for anyone to notice that I'm on an e-bike.
I took advantage of that stealthiness during an early test ride. While loitering along, experimenting with the controls, I was passed by a college student on a mountain bike who must have thought this gray-bearded old-timer on a four-decade-old 12-speed was huffing and puffing. So I poured on the juice and blew past him, thereafter maintaining Tour de France speed. As I raced ahead, I chuckled to myself. "OK boomer," indeed.
This article appears in the September 2021 print issue as "An Instant E-bike."
David Schneider is a senior editor at IEEE Spectrum. His beat focuses on computing, and he contributes frequently to Spectrum's Hands On column. He holds a bachelor's degree in geology from Yale, a master's in engineering from UC Berkeley, and a doctorate in geology from Columbia.
David Schneider’s suggestion to use friction-drive on the tires for e-bike conversion is interesting. If this method were efficient, standard bikes would be built this way, eliminating the inconvenient chain, chainring and freewheel. The power required to drive a bike is inversely proportional to the mechanical efficiency of the transmission. A chain can have an efficiency of 98.5%. I expect the figure for friction-drive to be considerably worse.
Schneider describes a conversion costing less than $200, but he omits the battery, which is by far the most expensive part. For a similar price, one could put a hub motor on the front wheel, install an e-bike controller and a throttle on the handlebars. This avoids the need to weatherproof and supply robust cabling to an Arduino.
A lithium battery will run $300 to $400. Adding up the prices, buying a complete electric bike for $1000 is a good deal.
If you want a bike for DYI electric bike parts, I have a 2010 Trek Valencia Ride+ for free, you pay shipping (I’m in Las Vegas). Has a 20” frame, but I’m assuming you would want to reformat the whole thing anyway. The lithium-ion battery pack is a 40.7V, 6.4Ah and died long ago, in fact was pretty worthless after 6 months riding. Bike details: https://archive.trekbikes.com/us/en/2010/trek/valenciaplus#/us/en/2010/trek/valenciaplus/details
The 350W motor had so much torque that spokes kept popping. Maybe why later models toned that down to 250W. The bike weighs over 55 pounds, and almost all distributed to the rear (may have contributed to popping spokes). But, was exhilarating to ride while it lasted.
One non-EE technicality. Everyone should be using their rear brake more often than the front! Many have recommended that. I learned the hard way once when I just used the front brake, flipped my bike, and broke my shoulder. However, I'm not recommending, without further study, that the front brake be used for regenerative braking.
It saves weight, energy, and space
The heart of any electric motor consists of a rotor that revolves around a stationary part, called a stator. The stator, traditionally made of iron, tends to be heavy. Stator iron accounts for about two-thirds of the weight of a conventional motor. To lighten the stator, some people proposed making it out of a printed circuit board.
Although the idea of replacing a hunk of iron with a lightweight, ultrathin, easy-to-make, long-lasting PCB was attractive from the outset, it didn’t gain widespread adoption in its earliest applications inside lawn equipment and wind turbines a little over a decade ago. Now, though, the PCB stator is getting a new lease on life. Expect it to save weight and thus energy in just about everything that uses electricity to impart motive force.
The layered components of an Infinitum Electric axial-flux motor are shown here, in exploded form.INFINITUM ELECTRIC
This saving of energy is critically important: Software may be eating the world, but electricity is increasingly what makes the world go round. Electric motors consume a little over half of the world’s electricity today. Some 800 million motors are now sold annually worldwide, according to the market research group Imarc, a number that has been increasing by 10 percent each year. Electric motors are making serious inroads into cars, trains, and aircraft, as well as industrial equipment and heating, ventilating, and air-conditioning systems. Transportation, construction, and HVAC together account for about 60 percent of all U.S. greenhouse gas emissions; more efficient electric motors will help cut emissions in these sectors.
Despite the benefits of the PCB stator, people were slow to embrace the design because of a few misconceptions.
First, there was the mistaken belief that PCBs were good only for delicate applications. But in 2011, CORE Outdoor Power developed a leaf blower and a weed trimmer, both of which used a PCB stator and yet were rugged as well as quiet.
Second, there was a sense that PCB stators could be used only for low-power machines. But in 2012 Boulder Wind Power put a PCB stator in a 12-meter-diameter direct-drive generator for a wind turbine that output 3 megawatts of power and just over 2 million newton meters of torque. It was one of the smoothest-running high-power generators ever built.
Neither company endured. Boulder Wind Power ran out of funding before it could secure commercial contracts. CORE Outdoor Power couldn’t compete in a crowded market where there were cheaper options. Still, their pioneering achievements demonstrated the feasibility of PCB stators.
Axial-flux motors easily fit onto the axle of a car [top] and the fan drive of an HVAC system [bottom].INFINITUM ELECTRIC
Fast-forward to today. My company, Infinitum Electric, of Austin, Texas, has developed a PCB stator motor that fits a wide variety of purposes. Our motor generates as much power as a traditional alternating-current induction motor but has half the weight and size, makes a fraction of the noise, and emits at least 25 percent less carbon. It is now finding applications in HVAC, manufacturing, heavy industry, and electric vehicles. Here’s how it works.
The Infinitum Electric motor is what’s known as an axial-flux motor, a design in which the stator’s electromagnetic wiring stands parallel to a disk-shaped rotor containing permanent magnets. When alternating current flows through, it makes the rotor spin. The motor also has an air core—that is, there is no iron to mediate the magnetic flux and nothing in between the motor’s magnetic parts but thin air. Put all these things together and the result is an air-core axial-flux permanent-magnet motor.
In the past, attempts to build such a motor faced serious practical obstacles. A complex manufacturing process was needed to build the stator, the copper windings were bulky, and the coil support structure was intricate. As a result, the air gap was so wide that only a substantial magnet mass could create the necessary magnetic flux.
At Infinitum Electric, we did away with those copper windings and instead use photolithographic techniques to etch thin copper traces interleaved with epoxy-glass laminate, which insulates each coil from neighboring coils. Eliminating the iron core and minimizing copper together save 50 to 65 percent of the weight and 50 to 67 percent of the volume of the motor, when compared to an equivalent conventional iron-core motor. And conveniently, the copper and the laminate expand and contract similarly as the temperature rises and falls, avoiding stress that might otherwise slowly pull the components apart.
The absence of a stator core allows us to put two identical rotors facing each other on either side of the stator, with each rotor carrying powerful permanent magnets. This arrangement creates a constant magnetic flux. As in other axial-flux motors, that flux is parallel to the axis of rotation, rather than radial. Because the magnetic air gap is narrow, we need only a small magnet, which is why we can wring a lot of power from a given mass and volume.
Our motor generates as much power as a traditional AC induction motor but has half the weight and size, makes a fraction of the noise, and emits at least 25 percent less carbon.
What’s more, PCBs are manufactured by an automated process, which means they’re much more uniform and reliable than hand-wound machines. We made them even more reliable by simplifying their topology, which has to do with the motor’s phases.
An electrical phase is an alternating voltage that forms a sinusoidal wave that is shifted in time relative to the voltage in another phase. The various phases are synchronized so that the sum of the currents is always zero. When a multiphase voltage system is applied to a motor that has a separate winding for each phase, the circulation of several currents generates a magnetic field that rotates in space. The interaction of this rotating field and the field produced by rotor magnets is what turns the rotor.
Previous PCB stators mixed the copper traces from different phases in the same layer, which created the potential for short circuits. We instead have each layer carry only one electrical phase, and we minimize the number of connections between layers. That arrangement provides a continuous path for the electric current and reduces the risk of electrical failures.
Another advantage of the new layout is the freedom it gives designers to connect coils either in series or in parallel. Connecting the coils in series is appropriate for three-phase industrial applications and next-generation electric vehicles. Connecting in parallel is better for low-voltage applications, such as in an auxiliary EV motor.
Like other permanent-magnet motors, our axial-flux motor requires a variable frequency drive to smoothly start and accelerate the motor to the desired speed. The VFD also controls the speed and torque as required by the application.
A short path for magnetic flux is made possible by sandwiching rotors [gray] with magnets [red and blue] around a thinly printed circuit board stator [green].
However, the air-core design gives the motor exceptionally low impedance (typically just 5 to 7 percent as much as in a conventional iron-core motor), because air cannot contain as much magnetic energy as iron can. There is thus very little magnetic energy available to smooth out the variations in the voltage supplied to the motor by the VFD. To remedy this deficiency, we added another element: an integrated variable frequency drive that is fine-tuned to operate with a low-impedance motor. Our VFD uses high-efficiency silicon carbide MOSFETs, which reduce losses and contribute to overall system efficiency.
The VFD also monitors performance, and the results can be reported via the cloud, if the user wishes. The motor’s software can also be updated in this fashion. Such remote monitoring offers a variety of ways to conserve energy, manage performance, and predict when maintenance may be needed.
The thinness of the PCB furnishes a high surface-to-volume ratio, which makes for more efficient cooling, thus allowing us to push two to three times as much current for a given amount of copper. The cooling can be done by blowing air over fins on the outside of the motor and across the electronic compartments.
Removing the iron core eliminates loss due to the cyclical magnetization and demagnetization of the iron, while also avoiding energy-wasting eddy currents in the metal. Our air-core motor can thus operate at a high efficiency over loads ranging from 25 percent to 100 percent of the rated power. Skipping the iron core also means the magnets on the rotors face a constant reluctance and a constant magnetic field as the rotor turns. This arrangement eliminates eddy-current losses in the magnets and the rotors, which therefore can be made of standard, unlaminated, low-carbon steel plates.
In a typical electric motor, both the stator and rotor are made of ferromagnetic materials. Once electric currents are applied and rotating magnetic fields are established, these fields produce two forces: one that produces useful torque and causes the rotor to rotate, and another that pulls the rotor radially towards the stator. This radial force does nothing useful, and it aggravates noise and vibration, because the slots in the stator—needed to accommodate the copper coils—generate pulses.
Here’s why that happens: A magnetic flux produces a force that at first points in the same direction the rotor is moving in; then, as the rotor turns, the alignment of the rotor poles changes in relation to the stator slots until the force points in the opposite direction. This alternating force produces torque ripples, which can cause metal fatigue in the motor and in the machinery that it is driving.
But there is no such alternating magnetic force in the Infinitum motor. This advantage, together with other efficiencies, is why its noise averages about 5 decibels lower than that of conventional motors. That may not seem like much of a reduction, but this component of motor noise tends to be at a particularly bothersome pitch.
The motor design is based on a printed circuit board [top], the thinness of which allows for a package that is far more compact than an equivalent motor based on a conventional iron core [bottom].INFINITUM ELECTRIC
By combining the lightness of an air-core motor with the high torque density of an axial-flux machine, the Infinitum motor is well suited for building ventilation and HVAC systems. That’s particularly useful now that the pandemic has put a priority on purifying indoor air. Heat pumps, which heat and cool in one system, are another application in which the motor can save energy, ease installation, and reduce noise. According to recent tests conducted by the U.S. General Services Administration and the U.S. Department of Energy, Infinitum Electric motors could save up to US $8 million annually if deployed in the GSA’s HVAC plants.
Electric vehicles are another big market for this new motor. EVs are projected to make up 31 percent of the global fleet by 2050, according to the U.S. Energy Information Administration.
Our company is working with a leading automotive supplier to develop an oil-cooled motor for a long-range hybrid vehicle. Oil cooling works much more efficiently in our design than in a conventional motor because the coolant can be readily applied to the entire surface of the PCB. With oil cooling, we have realized a threefold increase in power density over our own air-cooled motor, bringing the power density into the 8 kilowatt/kilogram to 12 kW/kg range. That makes the oil-cooled version suitable for use in electric aviation, another promising market.
We’re also working with companies that specialize in material handling, such as forklifts, conveyor systems, and the mixing equipment that’s used to make food and beverages. Caterpillar Venture Capital has invested in Infinitum Electric to develop a new line of alternators that are one-third the size and weight of existing models and quieter and more efficient as well. The alternator market is estimated at $17 billion a year and growing.
We estimate that if every motor in the world were replaced by an Infinitum Electric motor, it would reduce carbon emissions by 860 million tonnes per year. That’s the equivalent of eliminating the emissions from 200 million cars annually. As motors become ubiquitous, even small improvements in efficiency have the power to make a big difference for our planet today and over the next century.
This article appears in the April 2022 print issue as “This Axial-Flux Motor With a PCB Stator Is Ripe for an Electrified World.”