This article was first published in 2005.
Last week we introduced you to the design of our DIY electric bike
(Building an Electric Bike, Part 1). We showed you how you could add electric power to a normal bike and
the simple control system that allows both performance and regenerative braking
modes. (If you’re coming to this article first, it won’t make much sense without
starting at the beginning of the series!) This week we build the bike, starting
with the motor.
In this design, the motor cops a caning. But the approach works very, very
well. Cheap, too. Basically, what we do is to use an excellent quality,
nominally 12V motor - and often run it at 24 volts!
Aaaagghhhhhh! I can hear all the design engineers sputtering. You may get
theoretically four times the power, but what about the heat? What about the
bearings? What about....? Three points. The first is that data-logging the current
flow to the motor shows that typically it’s drawing less than it would under
high loads on 12V. Secondly, when it is drawing very high currents, it’s
usually only for a short time. And the third point? We dramatically improve the
The motor that’s used in the bike is from a Honda electric power steering
system. More and more small cars are running electric power steering and the
motors are beautifully made, high torque designs. As said previously, the motor
shown here was sourced from the tip (the complete electric steering rack was a
warranty cast-off from a Honda dealer - no, I’ve never been able to find what’s
supposed to be wrong with it) but similar motors can be obtained from Japanese
importing wreckers. When looking for a motor, aim for an ‘orphan’: a steering
rack imported with perhaps the front half of a car - the engine in the half-cut
suitable for local models but the electric steering rack an unknown.
Heavy duty windscreen wiper motors and hydraulic brake boosters are
alternative sources of motors - again, you don’t want to request a specific part
but instead see if you can find an unwanted one. Look for heavy gauge wiring
going to the motor - the thicker it is, the better
In most cases, the output shaft of these motors uses a worm that drives a
spur gear. What we do is cut away part of the output gearbox casting so that the
worm itself can be pressed against the tyre tread. (In fact we replace the worm
with a roller but we’ll come back to that in a moment.) So in addition to
looking for a good quality 12V motor, also assess the shape of its output
housing. It’s even better if the shaft is supported at both ends with roller
bearings, but in some cases only plain bearings (eg bronze bushes) are used.
If you do score an electric power steering motor, there’s another advantage -
many have an electric clutch built into the assembly. Disconnect power and the
output shaft turns freely; apply power and the shaft is locked to the motor.
This function gives better motor-off performance as the bike wheel is spinning
just the roller and a short section of shaft, rather than driving the whole
Lifting the Roller
In the case of the bike shown here, the motor assembly has a built-in
electric clutch. With the clutch wired in parallel with the motor, switching off
the motor enables the output roller to freewheel. However, most motors won’t
have this function and it may be advantageous to add a mechanism that allows the
rider to operate a lever to pull the roller up off the tyre. As the motor
assembly can be moved upwards against the spring tension, this is easily
Either by unbolting an extension housing or by pulling the armature out of
the motor, you should be able to gain full access to the worm. Originally, I
created the roller by forcing heavy-duty rubber hose over the worm. This was
quiet and initially effective, but once the hose-roller got hot, the rubber was
(I should add at this point that the prototype was
bicycle torture: the hills around here have very steep gradients of up to 17 per
cent and when I started this project, I hadn’t ridden a bike frequently for a
very long time. So the motor was required to do a lot of work...)
On the flat, the rubber hose roller may last very well - and it’s certainly
the easiest way of turning the worm into a roller that can bear against the tyre
tread. However, in this case, a metal roller had to be placed over the worm.
While initially I thought it would be a requirement that the roller be knurled
(ie have ‘grip’ marks machined into it), the friction between even a smooth
roller and the tyre is OK - especially on flat roads.
How you turn the worm into a metal roller is up to your ingenuity - I ended
up using a drilled-out 10mm deep-style socket surrounded by a roller salvaged
from the output shaft of a stepper motor. The roller was slipped over the top of
the socket, being held in place with a grubscrew.
However, when I realised I just loved riding this electric bike, I decided to
remove my (relatively) bodged-up roller and have a small machine shop turn-up a
new shaft, complete with knurled roller. This cost AUD$140.
Lots of home-brew electric assist bikes use wheelchair or automotive
windscreen wiper motors, complete with the worm drive gearbox. Typically, the
output gear drives the chain. However, a downside of this approach is that
regenerative braking cannot occur - very few worms will drive backwards.
The relationship between the ‘natural’ speed of the motor, the available
motor power, and the diameters of the roller and the tyre will determine the
on-the-flat cruising speed.
There are far too many unknowns to even make a guess as to what the gearing
outcome will be - testing is the only way. With the bike shown here, the
diameter of the roller on the output shaft is about 25mm and the tyre is
nominally a 26-inch design. This gives an on-the-flat cruise speed of about 18
km/h - or 20 km/h with just very gentle pedalling. This speed works really well
on the road - not too fast to attract attention but quick enough to get places.
(After all, 20 km/h averaged for an hour gets you 20 kays away!)
If the on-the-flat cruising speed is too slow, increase the diameter of the
roller. If it is too fast (or more likely, your hill-climbing ability is really
bad), decrease the diameter of the roller. Changes in roller diameter are easily
tested by using different diameter rubber hoses - even if the roller life is
short, it will give you a guide to the outcome.
Note that mountain bike knobbly tyres should be replaced with smooth highway
tyres - they’re available for wide mountain bike rims but you might have to
scout around a few shops to find them.
If you want to get very tricky, it’s possible to replace the cylindrical
roller with one shaped like a cone. If the motor and cone-shaped output roller
are moved laterally across the tyre, the gear ratio of the drive system can be
Motor Mounting and Cooling
The motor’s output roller needs to be held rigidly at right-angles to the
tyre tread. Further, it needs to be pushed down into the tyre so that the tyre
and roller have a good frictional grip on one another. It also helps if existing
holes in the bicycle frame are used.
In this case, 40mm square, 3mm wall aluminium tube was used to form a
T-shaped frame that held the motor in place in the right position. The motor is
attached to the frame using long heavy-duty plastic hose-clamps and is
preventing from swivelling by 50 x 3mm aluminium angle. (It’s easier to look at
the pictures than read a description!)
The roller pressure on the tyre is maintained by the use of an extension
piece of smaller aluminium tube that goes to the underside of the seat, a
threaded bolt and a stiff compression spring.
Lots of testing showed that the motor could get really hot after long, very
steep hills. (In these conditions the bike speed is down to 10 km/h or so and
I’m pedalling in 1st or 2nd gear of 21 speeds.) To stop the motor getting
so hot, a cooling air scoop was then added. It’s made from a 75mm stormwater
pipe plastic elbow that was cut, heated and bent to form the required shape.
The scoop dropped moving motor temps a lot but if the bike was stopped after
climbing a long hill (or descending it, where regen braking loads the motor),
the external casing rose in temp to 55 - 60 degrees C. To stop this temp
increase, a PC cooling fan was integrated into the scoop. It runs off just one
of the 12V batteries (its current load is tiny) and is triggered by a
thermostatic switch mounted on the motor. The fan comes on at 57 degrees C and
switches back off at 48 degrees C. The switch was salvaged from a
On hot days the fan operates when the bike is moving, but in most cases it
switches on only when the bike has come to stop. In this situation it runs for
10 or so minutes.
Next week: the batteries, and riding the bike!