How to Cheaply Increase the Top Speed of an Electric Bicycle

This instructable is about my investigation into increasing the maximum speed(and performance) of an electric bicycle with ‘standard’ hub DC motor and controller. Its in a diary-like format complete with all the setbacks and mistakes, because a lot were made along the way. I think pointing out what not to do is just as important as the final result.

I did a variety of online searches with key words to try and find any record of anyone else doing this, but nothing came up. That’s probably because its not a great idea™ for the correct way to design an ebike. If you just want to know the final setup, you can skip ahead to see that yes, a DC/DC boost converter can increase the top speed and performance of just about any small scale electric motor transport.

I’m an electrical engineer with a background in solar car and EVs, so I was comfortable in looking into this as a project. I was given a 26” wheel hub motor and 500W (36V?) electric bike kit by my father-in-law. He made a hasty purchase of it online, not realizing that it didn’t come with a battery(hence why it was so cheap, eh). With the extra cost and effort to make it work, he gave up and bought a premade bike, and donated the kit to me.

I bought a cheap bike ($25) to match the wheel size, as I needed to buy a bike anyway(my kids were just reaching bike riding age themselves). I had some aged 7.2AH 12V lead acid batteries lying around, so I made a quick setup using them. The controller itself wasn’t well marked, with an overly generic chinglish manual that just described in general terms how to set it up. The kit came with a charger that seemed to have voltages set to top off 36V of lead acid, so it seemed that 36V was the input rating.

The results with the setup were, to say the least, very disappointing. The added hub motor and batteries easily doubled the weight of the bike, and carrying it up even a few steps was now an Olympic event. The top speed was far lower than I could easily pedal to myself, and the extra weight of the kit meant I couldn’t travel as easily as before when the motor wasn’t driving it.

Putting on some monitoring to see what was happening, I could see that the battery voltage had some deep sag whenever the throttle opened up. It was down to near 20V for full throttle. It meant that the controller was well under its 500W rating, and rotation speed being proportional to voltage meant it was equally low. What I needed was a way to keep the input voltage to the controller at the maximum rating at all times, regardless of what the battery voltage was doing.

The answer was a DC/DC converter with settable constant voltage output. Obviously, putting one in adds another layer of failure and inefficiency, so I’m not suggesting that this is the right way to design an e-bike from scratch. The best way would be to start with a high voltage battery pack and controller that can handle the input. But there are a lot of low end ebikes already out there, and there may be other people interested in how to increase performance, even at the expense of efficiency.

I wasn’t in a big rush to make this, and I didn’t expect to permanently convert this bike to an ebike (I ride it because I want the exercise), so I was aiming to do this as cheaply as possible. Re-use in other projects was considered for all purchases. The DC/DC converter was the only part that ended up being specific for this, but I’m sure I’ll find another use for it. Buying them cheap off ebay meant a 6 week turnaround for every purchase, so with setbacks, this took over a year to do the design and testing.

A good hard look at the lead acids batteries led me to seeking out another source. They were just too heavy and old. I scored a slightly damaged 14kwh 28V NiMH pack, and this looked perfect for the project (and many other projects as well).

After browsing a few DC boosters I settled on a 1200W 20A 6V to 80V converter. 20A @ 28V was pretty close to 500W for the controller limit, so it looked like a good match. When it arrived, it worked out of the box with some simple testing, so I signed off the ebay purchase as ‘received’. Once I took it to work and put in on a kw load bank, problems appeared. It kept tripping up the current amp limit, and doing no voltage boosting(input=output) in an odd failed state. I took full notes, with pictures, and ‘proved’ to the supplier that the booster was actually faulty. No problem, they said, we’ll send you another one. For all my ebay purchases, I’ve had a 0% success rate on products being resent with no ebay ticket forcing them to do so. This was no exception. I waited 6 weeks for it to be delivered, then gave up and ordered another one.

Playing with the first converter again(while I waited for a replacement), if you coaxed it just right, and limited inrush current, I could get it to boost. So with it, a fuse, the battery pack and the wheel installed on the bike while it was free spinning upside-down, I did some tests. It was able to spin it at higher speeds, if I accelerated slowly so it didn’t ‘trip’ into voltage limit mode. Resetting it when it did required disconnecting the battery pack connectors. But then, the fuse blew, for no reason. I replaced it, and then plugged it back together again. Alas, the fault was in the DC/DC converter, not the ebike controller. So even with the bike controller off, the DC converter was pulling short circuit current. It fried the connector pin/sockets before the fuse blew. So now I had no booster, 2 blown fuses and damaged high current plugs. More ordering and multi-month delays before I could work on it again.

I had a 20A circuit breaker(on the battery side) that also acted as a main power switch. I had repaired the connectors, and beefed up the wiring in general with correct terminations(instead of some basic splicing). The battery, dc converter and controller were in a rack bag on the back of the bike, which I secured crudely with a tie down strap. Loose cables were taped to the bike frame. It looked ready to ride. A quick trip with super safe voltage settings up and down the street was a success. It drew the neighbor kids like the pied piper back to my driveway to get a better look at it. I set it up to try and do a max torque test by measuring the voltage dip while driving it into a solid wall. I was looking at the voltmeter so I didn’t notice that the wheel motor axles had slipped and spun a bit(about half a turn). I thought I just hadn’t tightened the nuts enough, so I gave them an extra hard crank.

With the pressure of a gaggle of kids shouting for me to go, and the joy of initial success, I launched down my driveway again. I wasn’t going too fast, but it was fast enough to hit the driveway/curb join with a bit of a bump. This bump was enough to jump the front axle out of the forks, bringing the bike to a dead stop and causing me to rush headfirst towards the pavement. The motor, free of the forks, continued to spin and twist up the power and control wires in a coil until they snapped. The multi-kg battery pack, loosely held in with a single tie-down strap, launched to maximum height before coming down on me and the crumpled mess of a bike. The children cheered! There was only the odd lingering parent that thought coming to my aid might be in order. Show's over kids, go home.

I assessed what went wrong, and thought about repairs. It didn’t appear that the twisting of the wires did any internal damage to the wheel hub. It was just the external cabling that broke. In future weeks, this was soldered back together.

I looked at the bike forks, they were now giant Vs. Some googling later, I discovered some new terms – torque arms and axle spreading. It also revealed that one of the locking torque washers(part of the axle) was actually installed backwards, and this is unfixable without chopping connectors(assembled wrong in the factory). The general rule seems to be that under 500W, torque washers will do, but above that, torque arms are required. I only had one functional torque washer, so I was well under the required amount. With no torque strain relief, the square edged axle spins in the forks, and widens up the space so there is then no resistance to free spinning from then on. Attempts to re-bend the forks were just damaging the bike frame. Torque arms, at $30+shipping a piece, would be a big expense and delay for a temporary project. The cheap and nasty solution was to buy a single 9mm spanner and cable tie it to the frame. I already had another one from a separate set, so with one of them on each side, I reassembled the wheel to the forks, wedged on the spanners, and cabled tied them on. The end result seemed robust enough to trust my life to it again.

With my 3rd major attempt, I made everything more secure. Initially, I used 2 multimeters setup to measure voltage and current. I didn’t get off the driveway without blowing the fuse in the current measuring multimeter. I still wanted to do measurements, so I scalped a 20A DC load power meter from another project, and changed the wiring so it could be put in line. Initial tests were all successful, but I couldn’t go for more than a minute without the DC converter going into a weird state again, and no longer providing power. I guessed this was probably due to overheating, seeing as it was housed deep inside the rack bag with no air flow surrounded by other components that could get warm. Some more changes in wiring layout, and I had the converter in a plastic box on the bike frame so that the FET heatsinks got some airflow during movement. This assumption was correct, and the DC converter rarely tripped in the same way again.

It was looking good to say design was now complete, and I was ready for testing. I pulled out my ancient stand alone GPS and taped it to the handle bars. I trust it for the fundamental tasks of movement measurement, and it is more expendable than using a cell phone GPS app if I crash again.

I had to move the DC power meter to a better location so I could read it while still vaguely looking where I was going. A much better solution would be actual logging of data, so I don’t need to look at all. If I was doing serious research into this I would have done that, but I could get by with static readings committed to memory for now.

A method for measuring repeatable results was still eluding me though. The current and speed readings were all over the place. While I could generally demonstrate that increased voltage made the acceleration and top speed higher, I couldn’t say by how much. Working out just how high I could safely push the voltage was a dilemma too – if the only feedback I got that I had raised it too high was a broken motor/controller, then that’s hard to plan testing around.

In the end I opted to set the converter across a spectrum of outputs from 36V to 55V and do repeated tests on the same stretch of road. It had a slight slope to it(not visible to the eye), and I took readings at the same 2 places going both down the hill, and back up the hill. The overall run extended quite far past these points, so there was a lot of distance to reach an equilibrium before taking a measurement.

Measuring current turned out to be misleading, so I used the inbuilt power consumption of the DC load meter as the instantaneous power input, and the GPS nav page for the speed reading. Both had a refresh rate of about 2Hz. At low voltages, readings were quite steady state at the measurement locations. At high voltages, there were times when things weren’t quite at equilibrium, so a one off reading didn’t truly represent what was happening. For each run, I would memorize the speed and voltage at the 2 points, then write them into a log book at the turn around point, and do the same on the return journey. I’d then adjust the voltage, and repeat. I would regularly recharge the battery and re-do the same voltage setting on a different day to try and get an average sampling of results across different conditions. Headwinds (on some days) seemed to influence results far more than I imagined.

The main data is in the graph. It shows 4 lines of the readings, which is the samples of 2 locations in the 2 different directions. Up to 45V, the bike had generally reached top speed/power on the run(about 1km total). Above 45V, initial conditions and weather was still influencing the performance, so it felt like higher speeds might be possible if the road was longer/flatter. You can see that, in general, the higher the bus voltage the higher speed at the same location. In the raw data, I swapped between amps and watts for more representative readings. An X value meant I couldnt sample, for various reasons.

I calculated that up to 45V should be completely safe. Assuming the system was meant for 3 lead acid 12V batteries, you get (3 X 13.8V + 10% tolerance = 45V). Beyond that, I could only speculate what the limits might be before a failure of some kind. I put a hand on the motor and controller between tests, and they were never too hot, but these were also fairly short testing cycles.

Other factors came into play that kept my maximum test voltage at 55V. The DC converter was only rated for 20A. On one test, the motor power was reading over 1000W. This was pulling 60A through the DC converter to make this happen, 3 times its nominal rating. At this level, even for a fairly short test, it tripped out from over-temperature again, so I took it as a hint to back off a little or suffer another hardware failure. The circuit breaker I bought was supposedly 20A, but it only tripped once during the entire testing, even though it was regularly pulling 40A for extended durations. It was another 60A episode that finally tripped it.

I was also concerned at the ability to actually pull 60A with this setup. Everything was rated for less than this – breaker, converter, power meter, connectors and battery pack. I don’t know how much current the NiMH batteries were actually rated to survive, but current that high was definitely shortening their lives. Without serious $$ of my own, I was never going to get a battery pack like this again, so I didnt want to kill it.

Also, 55V allowed you to get speeds that were approaching hazardous for bicycle safety. I was actually breaking neighbourhood speed limits when I decided to really go for it full throttle. Cars and pedestrians were not expecting a bike to be going that fast, and the existing bike brakes were underpowered to stop me in time when things came out in front. I decided that 52V would be my maximum if I was make this a permanent setup.

One added advantage of using a DC converter is that I now had a settable current limit capability built into the power train. I experimented with leaving the voltage high, but reducing the current until it started to affect performance. The converter efficiency of constant voltage vs constant current left me feeling constant voltage was much better for overall conversion efficiency. I didn’t specifically measure the relative power/speeds, but the general sluggishness of performance had me backing off on the throttle until it hit the constant voltage threshold again, and then adjusting within this setting to get the maximum zippiness. So in terms of riding discipline, this was a good method of forcing me to pedal hard during the high power starting torque conditions, and only using the motor to boost me up to ‘faster than I can pedal’ speeds.

20A breaker (note: took 60A to pop)

20A DC load power meter (was able to operate and measure 60A(some specs say 99A), but was always under the nominal 2kW limit)

1200W 20A DC Converter Boost Step-up Power Supply Module IN 8-60V OUT 12-83V

28V 14kWh NiMH Battery pack

‘500W’ 36V DC ebike motor and controller.

2 torque arms(9mm spanners) + cable ties

26” wheel bike

Cables and connectors (capable of handling high amps, 50A+)

Portable GPS (testing only)

For just $20, you could convert any existing ebike to use a constant voltage DC/DC converter to increase overall top speed. The approximate 15% energy loss from conversion may be worth it if you have spare battery capacity for your average journey, and you value the increased top speed and performance more.

The downsides are that this method will kill your battery pack even faster(if you’re already having aged battery voltage sagging issues). Also, some clever packaging would be required to permanently mount the dc/dc converter so that it still got the required airflow but was protected from rain.

The long term effects of repeated high power use of the converter, and potential damage to EV components from higher than rated voltages, was not investigated.

Key points:

Add torque arms if not already installed.

Add a circuit breaker to act as a switch so you can 'reset' the DC/DC converter if it goes into over-temperature mode.

Dont make the bike go faster than the brakes ability to stop it.