The reasons why someone would want a non-hub build…in spite of the complexity, noisy drive components, and extra cost…was discussed in Part-1. Here, in Part-2…you have decided that you want a non-hub drive system, and you want to know what components others have used with some record of success.
The two systems I’ll describe here are…first: a VERY high powered “one-speed” system that drives the left side of the rear wheel (leaving the stock right-side pedal-drive system just as you found it), and…second: a BB-drive, which is limited to a power level of around 2,200W /3-HP, but it gives the motor the use of the bikes gears, and that would provide a wider range of performance. Namely; a higher top-speed for the street, and still having great low-speed hill climbing for off-road (or places like San Francisco!).
If you are handy with tools, and want to give a non-hub build a try, I’d like to list some parts and components that have been frequently used, and some of the reasons why they are often found as the parts that are selected.
One of the first things you’re going to notice about non-hub drives is that there are frequently several stages to reduce the motor RPMs from their efficient range of around 3,300-RPMs, down to the rear wheel speed of around 330-RPMs (as a useful rule of thumb, a 26-inch wheel at 26-MPH is about 333-RPMs). This means that the amount of reduction from the motor to the rear wheel will need to be around 10:1 in order to achieve the best efficiency.
If we wanted to use a single chain from the motor to the rear wheel for a 10:1 reduction, what would that look like? and why don’t more guys do that?
It’s possible to make a motor sprocket that has a tooth-count as low as 6 teeth, but anything under 11 teeth will be very loud when running at 3,300-RPMs+. And at a 10:1 reduction, if the motor-sprocket is 11T, then the rear wheel sprocket would have to be at least 110T. This is the first problem, because it’s not difficult to find bicycle 3/32″ wide chain sprockets that are 60T, but anything more than that would have to be custom-made.
Even if you went to the expense of a ordering a custom 110T sprocket, that is still not ideal. Although an 11T motor-sprocket is much quieter than 10T (the 11T rule?), when using an 11T, the small number of teeth puts more stress on each tooth compared to a larger motor-sprocket. A 15T sprocket (roughly 7 teeth fully engaged with the chain instead of only 5 on an 11T sprocket), would need a 150T wheel sprocket for a 10:1 reduction. A 150T would be 24-inches in diameter!
The engineering gets a little easier if we go to the trouble and complexity of splitting the reduction into two stages by putting a jackshaft in-between the motor and the wheel. The first stage (primary) reduction will connect the motor to the jackshaft. And the secondary stage will connect the jackshaft to the wheel. The primary will be spinning much faster than the secondary, but it will also feel less torque, because the power that’s passing through it is spread across more RPMs. The secondary will feel more torque, but spin slower.
These characteristics will determine the appropriate parts to use in each stage. In the example below, a motor that is efficient at 3,300-RPM is driving a chainring attached to the left-side of the rear wheel (through a jackshaft), using an adapter that attaches chainrings to the 6-hole brake disc flange.
The Primary Reduction
I have seen both belts and chains used for the primary, and each choice has their benefits and drawbacks. Belts don’t need to be lubricated and they also run quieter, chains are somewhat less expensive, narrower, and it’s easier to make (or order) a custom sprocket (compared to custom-order pulleys).
HTD-5mm pitch is reasonably available from industrial suppliers, and the small teeth allow a small diameter motor-pulley to help achieve the most primary reduction. I have also seen the Gates Powergrip-GT2 5mm-pitch used. The one you should try might be driven more by which one provides a pulley with a width and tooth-count that achieves your design goals.
A primary belted reduction that provides even as little as 2:1 is a big improvement (of course 3:1 is better). The secondary chain reduction has its drive-sprocket speed cut in half, so chain noise is much less of an issue. If you are applying a lot of power to the primary belt, it may start skipping over the teeth, which will kill the belt fairly fast. The way to stop the belt-skipping is to either use a larger diameter motor-pulley to engage more teeth, or to widen the belt.
For lower-power systems, 15mm wide belts are widely available in many lengths, but 20mm and 25mm wide belts can also be found. The biggest issue is finding wider pulleys in the smaller tooth-counts. As an example of the power levels, a 20mm wide belt on a 36T-to-18T reduction (2:1) should be able to easily handle 3,000W / 4-HP. If you wanted a more-desireable 3:1 reduction, I would not recommend making the small pulley any smaller, you would be well-advised to look for a larger “big pulley”, with 54T:18T being a 3:1 reduction (if possible).
If you have a 2:1 reduction on the primary, the secondary chain only needs to be 5:1 in order to achieve the 10:1 total reduction this example is aiming for. If we use a 13T FW on the jackshaft (so when we are pedaling with the motor off, we aren’t back-driving the motor) it would require a 65T sprocket.
By simply adding a few teeth to the 36T primary pulley, the secondary reduction can use an affordable and available 60T chainring for $40 (no custom-ordered parts). A 60T-to-13T secondary chain reduction is 4.6:1, which is pretty good, especially since we can still use affordable and available parts. If a 4.6:1 secondary is close to ideal, then we only need a 2.2:1 minimum reduction on the primary to be as good as it can get. Using an 18T motor-pulley, the big pulley (on the jackshaft) needs to be a 40T (or, larger would be better if you can find it affordably).
I have seen 73T chainrings at recumbent websites, but they are expensive ($190?…Yikes!)
It’s better to start out a new design with too much reduction, leading to better-than-necessary hill-climbing. Then, you can swap-in the small pulley for a larger one to increase top-speed and also improve the drive tooth-count (which reduces belt-skipping). Or, you can swap the 13T FW at the jackshaft for something larger to increase the top-speed. The best strength ENO FW has a 16T, 18T and 20T version, and the ACS-Crossfire can be found with tooth-counts from 13T-22T. Either one of the smaller drive-units are a cheaper swap during experimentation compared to swapping-in the large pulley/chainring.
If you are open to considering a chain for the primary drive (from motor to the jackshaft), the most popular choices have been #25 chain and #219 Karting chain. Both have a 1/4-inch distance between the link-pins (called the pitch), which makes their links half the size of common 1/2-inch-pitch bicycle chain.
The 2 in #25/#219 means 2/8ths, and #410/#40 industrial chain means it is 4/8ths pitch. #40 sprockets are too thick for a bicycle chain to fit on it, but they have the same tooth profile, so a machinist can thin cut them thinner to fit 3/32″-wide bicycle chain, 1/8th-inch wide BMX chain, or heavy-duty 3/16″ width #415H chain.
Both #25 and #219 chain is readily available, if you can find sprockets in a tooth-count that matches your goals. The benefit of chain for the primary is a lower cost, its narrower, and chain has the ability to have a much bigger reduction in the primary without any tooth-skipping, because the drive-sprocket can be smaller than a drive-pulley.
If the motor-sprocket is at least 11T (for lower noise), the large sprocket can be easily found as an 85T in both #25 and #219, so an 8:1 reduction can be had with no fear of tooth-skipping. In this example, the secondary with a 13T FW would only need a 1.5:1 reduction using a 20T chainring. This provides a wide range of available sprocket tooth-count options, so no custom-made sprockets would be necessary.
Most Kart #219 11T/12T motor-sprockets have a tapered bore, and are so case-hardened, it is not advised to drill them out. The B-style has a cylindrical 3/4-inch bore (0.750″), so when using them you will need a bushing to fit the 0.500-inch / 12mm jackshaft to the 3/4″ bore. The flange style may suit others.
A more realistic example is a secondary reduction using a 60T bicycle (3/32″ wide) chainring on the wheel that’s driven by 13T FW (4.6:1) on the jackshaft. Then the primary would use #25/#219 chain with 37T:17T sprockets for a 2.2:1 primary reduction. 2.2 X 4.6 = the 10:1 reduction we need. All using affordable and available parts.
Having the largest of the 4 sprockets on the wheel (the 60T, 3/32″ chain), means the other sprockets can be reasonably smaller, cheaper, easier to fit on the frame, and do not cause ground clearance issues…
Due to the lower 1600-RPMs of the jackshaft (when driven by a 3,300-RPM motor and a roughly 2:1 belted reduction), it may be possible to use common 3/32″ bicycle chain on the secondary (or the stronger 1/8″ BMX chain), but since the 1/2-inch pitch bicycle chain links are twice as large as #25/#219, the 11T minimum drive sprocket would be twice as large as the smaller link chain types, and this can sometimes make achieving adequate reduction difficult.
If we use #25 or #219 chain as the secondary, we again have the benefit of having a fairly quiet 11T minimum drive sprocket, and it is twice as small as an 11T in a 3/32″ bicycle chain sprocket (13T/15T will be even quieter than 11T if that fits your design).
Bear in mind, that we are still considering a system that drives the left side of the rear wheel. Several web-sites sell a spider that allows a chainring to be attached to the disc-brake flange on the left side of a rear-wheel (some wheels do not have a rear disc brake option, but rear wheels with a disc flange are easy to find).
20-inch wheels help RPM reduction
All of the above figures were calculated using a 26-inch wheel as the example. If additional reduction is needed, another option is to use a 24-inch or a 20-inch wheel on the rear. As previously stated, a 26-inch wheel at 26-MPH is only doing 333-RPMs, but… a 20-inch wheel at 26-MPH is doing 437-RPMs (it takes a lot less reduction to get the motor-RPMs down to 437 compared to 333-wheel RPMs). 3,300-RPMs down to 437 is only a 7.6:1 reduction, much easier than 10:1!
Many of the racing E-bikes used an extra-wide 20-inch wheel with a 16-inch moped tire (mopeds and motorcycles measure tires at the rim, bicycles measure the outer tread). The fat, heavy-duty, flat-resistant, and higher-speed moped tires ended up with an approximately 22-inch diameter tire OD. (with a 22-inch diameter, speed will be roughly 15-RPMs per 1-MPH)
There are dozens of models that will work, but a frequently-listed Moped tire for hot-rod builds is the Pirelli-ML75, 16 X 2.50-inch, which fits well in a 1.85-2.125-inch wide BMX rim. They are rated for a continuous 62 MPH (100 Km/H). They are heavier than bicycle tires, but…surprisingly, they are also cheaper than high-performance bicycle tires. Coupled with 16-inch motorcycle tubes, they are unlikely to ever get a flat.
For very high-power builds that have too much power to run through the bicycle components on the right side: many are driving the left side of the rear wheel (like liveforphysics/Luke’s death-bike). The final drive chain to the left side of rear wheel for the most powerful builds has been #428 chain. This is a light chain for a motorcycle, but it is much stronger than most E-bikes would ever need. A chain that is about halfway in strength between the common 3/32″-1/8″ bicycle chain and #428, is heavy-duty #415 BMX / moped chain. The difference between these three chain options is the width of the sprockets, and the thickness of the chains side plates, pins, and links. #415 cannot be used through a derailleur (it’s too wide), so it is found on single-speeds.
A light and narrow 9-speed chain is sized so that nine gears can be squeezed into the axle space that normally holds 7 gears, so 9-speed chain has no place in a non-hub build, since it is only made for a 200W human. Also, thicker sprockets are quieter, if that is of interest to you.
Here is a pic of a handlebar stem used as a jackshaft which is attached to the seat-tube. Both the seat-tube and bearing have thin cylinder-shims added to ensure a snug fit.
I have had good luck buying case-hardened shafts from McMaster-Carr. They have affordable shafts already on the shelf, cut in one-inch increments, and they will custom-cut any shaft they sell for a small additional fee. I have received them in a mailed padded envelope in less than a week.
I recommend either a 1/2-inch diameter (0.500″), or a 12mm diameter shaft (12mm is 0.472″). Both will allow small tooth-counts on your drive-sprockets. Even if a larger diameter shaft allows a small #25/#219 sprocket at our 11T minimum, the smaller shaft would leave more “meat” on the sprocket collar, which holds the set-screws.
Staton-inc has affordable FW-to-shaft adapters for 1/2-inch shafts, and…for a 12mm bore, they will sell an adapter that only has a small pilot hole. This allows the customer to bore the adapter to any custom size they need.
For bearings, VXB is the “go to” supplier that is always my first stop when searching. You must sort out which pulleys and sprockets you want to use with a shaft before you order anything. After finding a perfect pulley-set for you, it might not come available in the shaft diameter you need. But…keep in mind, most pulley and sprocket suppliers will provide a pilot-bored unit if asked, so the customer can custom-drill it to their desired final shaft diameter (or drop it off at a local machine shop).
For a jackshaft mount, there are dozens of ways to do that, and it would depend more on what your personal skills and tool-set can accomplish. If you want to experiment with an aluminum boxed jackshaft, you can buy square U-channel from onlinemetals.co, and possibly use flanged bearings. McMaster-Carr has shaft clamps that can be used to clamp a bearing to an aluminum channel.
A pillow-block can be used to attach a jackshaft onto a motors plate-mount. Specify one with an ID that is the same as the OD of your shaft bearings, or bore a shaft-clamp pillow-block to fit your bearings. One example would be a shaft pillow-block with a 30mm ID, and VXB bearings that are 30mm OD and 12mm ID, for a 12mm jackshaft.
At some point, some of the mounts for the motor or the jackshaft will have long oval slots, so the mount can slide to tighten the chain/belt. In order to avoid needing to use two wrenches while trying to also hold the object in tension, I have used T-slot nuts and also Riv-nuts. I do not normally collapse the rivet-nut in they way that it’s intended (although I sometimes do that just to make them shorter). I drill a hole in aluminum plate and simply epoxy a steel riv-nut in the hole with the flange on the opposite side from the bolt. This leaves steel threads in aluminum without needing to use a small tap to cut threads.
If the riv-nut is too long, I have made a fat aluminum washer to set under the flange, so the threaded end is flush with the other side. Another option is to use a shallow steel “Thumb nut” that is epoxied into a hole.
A non-hub that drives the rear wheel directly is a “one-speed” system. Due to the broad and fairly flat torque band of an electric motor, a one-speed is often enough (especially if the top-speed is fairly low). If you want both high top-speed for the street, and an optional low bike speed that allows the motor to stay in its high-efficiency motor-RPM range (for off-road steep climbs)…you can get a lot of bang for your buck if you give the motor some gears.
The easiest way to give the motor some gears is to somehow have the motor drive the bikes stock rear-wheel gears. Right now, the most common way to do that is a Bottom-Bracket (BB) drive. With this method, a motor drives one of the two chainrings on the crankset, and the second chainring drives the rear wheel.The weakest link in a BB-drive is the freewheel that is built into the crankset.
Many of the most appropriate motors for a BB-drive reach saturation around 30A. You can get more power by raising the amps above that, but…the higher your amps go above 30A, the more waste heat it produces, instead of your battery watts getting converted to actual work. The GNG motor has been run at 72V X 30A =2160W (a hair under 3-HP), which is about the limit for the freewheels available. It’s possible to run more power than this through a BB-drive, but then the FWs will wear out very fast.
The White-Industries ENO flanged freewheel is the absolute strongest FW you can buy, and it’s strong when used applying power in a straight line. but it has a single bearing that is not designed to handle the twisting loads that are applied by a BB-drive that has two chainrings on one FW. The flanged ACS-Crossfire is the best FW available right now for the freewheeling chainrings of a BB-drive.
IGH, or external gear cluster?
I have always liked Internally Geared Hubs (IGHs). The Rohloff has very strong gears, but it’s expensive. The Nexus 3-speed that is made for commercial pedi-cab trike use is probably the IGH that is the next strongest, but it’s still fairly affordable. Even them, I have read about high-powered electric builds snapping their internal gear-teeth. Some controllers have the ability to program a different phase-amp scale to the motor. This allows you to have full motor-power at the top-RPMs, but less power at the lower RPMs. By definition, it removes the ability to “pop a wheelie”, but…it might be just enough to allow a slick-looking single chainline to a 3-speed IGH with no external derailleur.
For the fast and easy BB-drive hot rod, use a common 7-speed gear cluster with a 15T minimum as the small cog, and it will take the most abuse that is thrown at it. If you are putting too much power to it, the chain will simply skip, although…you may occasionally break off a tooth. In a worst-case scenario, and you damage the chain and/or the gear cluster…a new chain and cluster is easy to find off the shelf and very affordable to buy. It is possible to find a gear cluster with an 11T in order to improve top-speed, but the fewer teeth you have, the more load you are applying to each tooth. If you use an 11T on a high-powered system, it may work, but plan on replacing them often.
Since the extra-strong #415 (or slightly stronger #415H) chain will not work with a derailleur, you might consider a 7-speed 3/32″ wide bike chain by Wippermann, which I am told has a good record for taking high-power (although I have no personal experience with them).
Let’s wrap it up
I’ll add more pics and more info when something presents itself as relevant. I want to emphasize that these parts that are listed are not the only way to do a non-hub build (I have left out #35, 6mm, and 8mm chain…nothing wrong with them if they work for you) . I just wanted to highlight existing parts and methods so interested builders could get an idea of the parts that can make a difficult issue a little easier to resolve.
There is a place for factory plug and play E-bikes, and we will continue to review them to help you understand the benefits and drawbacks of the available models. However, adding a kit to your favorite frame is growing in popularity, so we will cautiously bring you information on kits when we feel enough data is available to compare them fairly.
This last year we noticed a lot of manufacturers adding a BB-drive to their line-up, so it gives us comfort to know that large companies with professional engineers have come to the same conclusion that the real-world has shown us. Namely, that…when a law in some country limits the power of an E-bike…it really helps the performance when you can give the motor at least 3 gears to choose from.
We like technical info in order to understand something as well as it can be understood, but we also like the proverb: “One test is worth a hundred theories“.
This last year has also seen a boom in E-bike racing, and when we were watching that evolve, we quickly came to see that a NON-HUB system has a lot of performance benefits. The affordable non-hub solutions use off-the-shelf bicycle components, which are lasting fairly well up to around 3-HP / 2,200W. Above that and they begin wearing out too fast and they occasionally break.
For a drive system that will use 48V X 30A = 1,500W /2-HP or more…consider driving the left-side of the rear wheel for a long lasting reliable hot-rod (for “off road” use only?…*wink).
You may have noticed that electricbike.com has added a STORE page to our website. Over time, we have contacted various manufacturers to ask if they could provide a part for E-bikes that is very similar to their existing line. Some companies have responded, but others must see a huge demand before carrying a new part in their catalog. Someone needs to “prime the pump”, and occasionally it will be us.
One example is EB.com’s Eric, who wanted a fat bike for riding in the sand near the ocean, but existing frames didn’t have space for a mid drive, especially one big enough to help a heavy fat-bike through the soft sand. He had one frame custom-made, and then a dozen…and now a wholesaler has ordered 20. We don’t know what we’ll do next, but…it might be interesting to stop by here once in a while.
Written by Ron/Spinningmagnets, March 2013