DIY non-hub builds, the why and how (part 1)

Common hub motor kits have a simple construction, which keeps the price low (especially with Chinese mass-production). The low price and easy installation makes them the big seller by a wide margin.  However, if you want a build which has unusually good efficiency (so a smaller battery can provide longer range), or…you want unusually good hill-climbing…a non-hub drive system just might be necessary to deliver your goals.

Be aware that a non-hub system will by definition be more complex, probably more expensive, and is likely to be noisier. And…the biggest problem is that a non-hub system will be either very expensive (like the EGO, and the M-Drive), or the more affordable drives have issues with longevity and component breakage (Cyclone and GNG). We’re not knocking Cyclone and GNG, they have no way of knowing if a customer is using their kits for light duty or racing…and non-hubs are often about reaching for exceptional performance.

Efficiency in a system design can help many aspects of your drive. An efficient off-road system is able to use the smallest and lightest battery pack, to improve the feel of the off-road experience. For a commuting road bike with a large battery, a more efficient system will not stress the batteries C-rate as much with occasional high peak amp-draws (lower amp-draws will lead to a longer battery life) and you will get more miles from the same size of pack whether it’s large or small.

As a general rule, using more volts to accomplish a specific performance goal will be more efficient than raising the amps. The biggest problem with designing a system around a higher voltage (if cost is no problem) is the limited selection of motors. The characteristics of concern about a motor are the shape, the size, heat-shedding ability, and the Kv. Thinner stator laminations using a higher-quality silicon steel will be more efficient due to fewer eddy-current losses, but designing and ordering custom motor laminations is a task that is beyond the average garage-builder.

Also of concern with higher volts, is the size and cost of the battery. If you substitute a 72V / 10-Ah battery for a 36V / 10-Ah battery, you have doubled the size and cost of the battery, plus you have to find a place on the bike frame to attach all that battery volume and weight.


Thud mid drive

Thud’s race bike (read the misadventures of Thud)


The recent introduction of the GNG BB-drive kit showcased a great E-bike motor. It can take 30A before it reaches any saturation. Saturation is where a certain size of coil has been fed the maximum amps it can convert into work (28mm wide stator for the GNG and the well-known 9C hub-motor family), any additional amps over 30A…and the heat slowly increases from turning watts into waste-heat. This motor can be run at 40A intermittently, but not continuously.

The GNG motor is small enough to be easily fitted into a common bike frame, and large enough in diameter to have a decent amount of copper mass and rotor leverage. As an inrunner, the hot coils of the stator are connected to the outer rim and the aluminum motor-shell, so…it passively sheds normal loads of heat fairly well without needing to add a fan.


Gng Drive

GNG Drive (read review)


It has been verified to run at 72V X 30A = 2,200W (3-HP), which is the maximum I would run this motor at continuously. Sadly, it currently available in only one Kv of 67 RPMs per volt applied. There are web-tutorials on stripping the wire out of a motor and re-wiring motors with thicker or thinner wire to change the Kv, but it’s such a pain to do it right.

This is one of the many reasons that we here also like the MAC, which…when used as a NON-hub…can take about 40% more watts (3,000-ish), and it’s readily available in 5 different Kv’s. If you want more than 3,000W (4-HP), we would advise that you consider the option of driving the left side of the wheel; using a large diameter sprocket attached to the 6-hole ISO disc brake flange, and using moped chain and tires.

ES builder “Thud” made several compact and robust 2-speed transmissions to help the efficiency of a high-watt system for several racers, but they are not cheap and the wait is long. There are pics available of builders using bicycle components at 4-HP+, but,… they simply don’t last long. Plus, a crash that is due to using light-duty parts that break at high speed is very dangerous. Here’s a pic of LFP-Lukes “Deathbike” at an electric bicycle race. Its true it can be pedaled (a requirement), but…there are NO lightweight components on it.



Live for Physics Bicycle of Doom


Since motors run at their most efficient around 3,000-4,000 RPMs, most industrial motors use the smallest possible adequate diameter, and make the stator/rotor longer in order to achieve an adequate amount of torque to prevent bogging down, resulting in a “can” shape (Cyclone uses these). To keep an E-bike drive configuration efficient and simple, the motor is often mounted somewhere in-line with the wheel, and that limits the width of an acceptable motor.

The GNG is 4-1/2 inches in diameter, and only 2-1/2 wide. The MAC is about 6 inches in diameter and about 5 inches wide. Both have more of a “pancake” shape. This is desirable for bicycles, because the pancake shape produces more torque per watt applied due to the added leverage of the longer stator-arms, and they do not require as much external RPM reduction to achieve 333 wheel-RPMs (26-MPH on a 26-inch wheel, a good calculation starting point).

The true power rating of a motor is how many watts you can apply while driving a specific load, and the motors core temperature (its ability to shed heat faster than it comes in) does not exceed 200F / 93C at a maximum peak. Normal operations should be from 100F-140F, 38C-60C. If your motor is running cooler than that, you may have used a motor that was larger, heavier, and more expensive than necessary for the job you are giving it (not a problem, just pointing it out).

If you are running hotter than than 140F, then you are probably converting too many of your batteries potential watts into heat instead of work. Either the motor is too small, or you are running it at too slow an RPM, and the controller is throwing high amps at it to get its RPMs up (and beating up on your battery to accomplish that).

If your motors temperature regularly exceeds 200F, several bad things can happen. The cheap assembly-line solder can melt just enough in one thin spot to break a vital connection, leading to the “Walk Of Shame” back home. The weakest of the three hall sensors will die abruptly (…WOS). The clear insulating epoxy on the copper wires will turn dark brown and degrade…eventually falling apart and allowing the copper wire windings in the coils to short (more WOS). Since shorted windings cost more in time and hassle to fix than they are worth, most hot-rodders just replace the motor with a new one, and the old one becomes “wall art”.




Concerning heat in controllers, a FET is a Field Effect Transistor, or an “on/off switch” for the motor coils. Common motors have three phases (groups of electromagnetic coils) spread out evenly in a pattern (A, B, C, A, B, C, etc). This layout uses the minimum number of phase groups that provide reasonably smooth power (I have seen 7-phase motors). Sooo…controllers add FETs in groups of three to affordably raise the amount of amps a controller can provide without overheating.

The vendor Lyen sells and modifies controllers, and currently his best sellers are the 36V-100V 6-FET, and the 36V-72V 12-FET (both are highly programmable). The 6-FET is physically smaller, so it’s easier to find a place to mount on an E-bike, but the longer 12-FET can flow twice as many amps without overheating. I have frequently seen the 6-FET set with a limit of 30A, and the 12-FET to 45A, with good performance results and no overheating. More amps than 45A, and you should seriously consider moped driveline components.

Both 6-FET and 12-FET controllers can be had with a direct-plug in port for the CycleAnalyst E-bike computer. This allows the builder to easily change the max amp-limits. Start out with too few amps while applying your normal loads, and slowly raise the amps until heat becomes an issue (your hills may be steeper than my hills). There is only a $30 difference between the 6-FET and the 12-FET, so the 12-FET is an experimenters favorite, because then you would only need a temp probe on the motor to monitor the system because the controller would always be running cooler than the motor. For E-bike racing, an expensive 18-FET/24-FET can be sourced.

Adam (ES member “Itchynackers“) won his class in the Pikes Peak hill climb on July14/2012, and “PaulD” won the Arizona death race and the Socal Grange race in 2011. Lets compare their builds…


Itchynackers Pikes Peak 100V bicycle with a 9C rear hub and an 18-FET controller

Itchynackers Pikes Peak 111V bicycle with a 9C rear hub and an 18-FET controller (read  about Itchy’s victory)


Adam used a Direct-Drive (DD) rear hub-motor. Probably the worst choice for efficiency, and yet he won his class. So…how did he keep the amp-heat within safe limits? He water-proofed the inside of the motor to prevent shorts, because he decided to cut large ventilation holes in the side-covers to let the heat out (just like a cars alternator). Plus, he trained for a year to be able to pedal along with the motor to take as much load off of the motor as possible.

Also, one strategy he used to get the maximum power from the motor (while the amps were being limited to the systems heat-shedding ability), was to use a 111V system, but bear in mind the race was short enough he didn’t need a lot of Amp-hours (Ah) of battery volume.

A geared hub would have been slightly more efficient on flatter ground because a geared- motor spins 5 times faster than the wheel, but…the only available geared hubs have a lower copper mass and don’t shed heat well compared to the large diameter DD hubs. Other than the 111V battery, Adam chose a very affordable and easy-to-build system.

PaulD used a reasonably efficient design. On a practical level, this meant that…he didn’t make as much heat in the first place, so shedding heat was less of a concern (an affordable 12-FET is all he needed for a controller). Since PaulD wanted a non-hub, and the amount of power he wanted to use was way beyond the bicycle chain and sprockets ability to survive (much less perform well), so his motor drove the left side of the rear wheel.

He used a right-side belted 2:1 reduction from the motor to a frame-welded jackshaft, and on the left side he used #219 Kart chain to the rear wheel. The right-side bicycle pedal-drive is stock.


80mm X 100mm motor, 72V, belted primary reduction with a #219 chain secondary.

80mm X 100mm motor, 72V, belted primary reduction with a #219 chain secondary.


Since he felt he could achieve his goal with only 72V, and chose a low-Kv motor (Kv is the number of RPMs per volt that a motor provides), and he also used a smaller 24-inch wheel, to keep the motors RPMs up in its best efficiency range (normally 3,000-4,000 RPMs). Remember, a popular 26-inch wheel is turning only 333-RPMs when the bike is traveling 26-MPH, which is about 1/10th the RPMs needed at the motor to be as efficient as possible.

This higher efficiency compared to Adams DD-hub, was a clear performance benefit (PaulD easily won with almost no pedaling or fear of component overheat), but,… this is not a cheap or easy plug-and-play system. It required making custom parts and having a shop alter existing parts.

Neither of these two systems gave the motor the use of any gears, so the motor was driving a “one speed”. Since a motor can provide max torque from one-RPM and up, their broad torque curve often doesn’t require any gears to achieve their goals (a transmission of some type). Adam and PaulD didn’t really need the extra weight, complexity, and cost of a transmission because they were in a race, and they were only in the lower half of the RPM range during the first few seconds off the starting line. I point this out because these two didn’t have any stoplights, subjecting their bike to the frequent stop-n-go of traffic that a commuter E-bike would suffer.


Thuds DIY 2-speed transmission driving a parallel right-side drive so the rear disc brake can still be used.

Thuds DIY 2-speed transmission driving a parallel right-side drive so the rear disc brake on the left can still be used.


I was looking at a data-log from a Castle Creations controller on a low-power/high efficiency system of mine. This system uses a small RC motor the size of a large coffee cup (63mm diameter, 2-1/2 inches) and 26:1 reduction to the rear wheel. I was experimenting with adding capacitors to reduce voltage ripple, but the important data points to report are the amp draw readings. For the few seconds while I am accelerating it pulled 60A, and as soon as I begin maintaining a stable top speed, the amp-draw quickly lowers to 10A continuous. 36V x 10A = 360W is very low powered and efficient for long range on flat ground, but it was no race bike, or steep hill-climber.

If I took this bike onto a steep hill, and IF the power of the system was too weak to maintain the motor in the top 10% of its RPMs, amp-draw would quickly rise and stay at a high enough level that it might overheat, and it would definitely drain the battery at least 3 times faster. If I somehow gave this motor at least 3 speeds, the bike would still need to slow down on the steep uphill by shifting into a lower gear, but…the motor RPMs would stay high. Motor/controller heat, and battery range, would be hurt much less if we keep the RPMs up.

I am a fan of motor ventilation holes, and also “active cooling” air-fans. But, adding more cooling should be a secondary desire, after first raising system efficiency to reduce us converting the battery watts into waste-heat in the first place. A well-designed system will last a long time, but batteries will have to be occasionally replaced when they wear out, so…an efficient system saves on battery costs over the long run.


Matt (recumpence) now has optional fans for his high-quality Astro motors.

Matt (recumpence) now has optional fans for his high-quality Astro motors. Read about Matt’s 50mph trike build


So…if these engineering principles are so well-known, why aren’t more factory systems being seen that are configured to make sense? The reason is that large corporations follow the mass market, and most customers are very price sensitive. If we look at the average task that the common E-bike or E-kit gets used for, we can see that the hub-motors efficiency is, well…”good enough“…and low price is a bigger factor in actually getting the public to buy.

E-bikes and kits are much more popular in Europe than in North America, but bear in mind that gasoline in the Netherlands, Italy, Denmark, and Germany is around $8/gallon! (1700Eu/liter). This is also the reason behind the lightweight and expensive 250W BB-drives from Panasonic, Bosch, Cleanmobile, etc. If a country is limited to 250W for their street commuters, buyers are willing to pay extra for a quality long-lasting system that uses the efficiency of giving the tiny motor the use of the bikes gears.

In part-2, I have listed the most common parts that are being used now, for garage-builders to make a non-hub system.

Written by Ron/Spinningmagnets and’s Eric, February 2013

…many thanks to Luke/liveforphysics, Matt/recumpence, Todd/Thud, Adam/Itchynackers, and PaulD for their generous support and encouragement.

Grew up in Los Angeles California, US Navy submarine mechanic from 1977-81/SanDiego. Hydraulic mechanic in the 1980's/Los Angeles. Heavy equipment operator in the 1990's/traveled to various locations. Dump truck driver in the 2000's/SW Utah. Currently a water plant operator since 2010/NW Kansas


Leave a Reply