DIY belt drive for bikes

[bobc]

Thought I'd open a new thread on this: a few years ago I built up a road bike with a gates "carbondrive" toothed belt + homemade (laser cut) pulleys and a shimano alfine 11 IGH. It has been a great bike, I've commuted with it, given it zero or minimal maintenance & it continues (I'm guessing at 8 to 10k miles down the road) to just work.
Couple of years ago I thought I'd make a new one, specifically to address the issue of the outrageous cost of the gates belt (which was £76 for the belt alone). I got an industrial 14mm HTD belt cut to 10mm wide & put it on an old steel frame with a Sturmey 8spd. The belt was clearly not strong enough so the bike has been gathering dust since & I've carried on using the old bike.
Anyway, back to present, I started tinkering again..... I bought a rohloff IGH (oh yeah....) a year or so back and got a carbon reinforced 14mm timing belt off ebay last week. I made the toothed belt pulley for the rohloff by sandwiching their 13tooth sprocket in stainless lasered laminations.
The belt was £40 delivered but it's 20mm wide so I hope I'll get 2 out of it.
I'll put pictures up here as and when things get made; Here's the rohloff + sprocket

Speedhub + belt pulley

[greyingbeard]

a car fan belt might stretch too much, can you find a timing belt the right length ?
Or use a "too short" fan belt, pre-stretch it with weights hanging off. Fan belts can be had for a fiver.

[bobc]

Yeah you can get timing belts in a wide range of sizes. Timing belts are preferred for this job compared to vee- belts (fanbelts) because they are more efficient & don't slip when imperfectly tensioned.
"Gates carbondrive" is a well established supplier into bike manufacturers using specialised timing belts.

[mig]

i really admire projects such as this with the ideas & skills involved. i'd love to try this sort of set up on a fixed gear bike just to see how 'solid' it feels and how smoothly it would run.

[Brucey]

functional V belts have been around for over a century. They have never been used on bicycles because they are inefficient. In order for them not to slip when dry, they need a huge preload on them, and they can still slip when wet.

The idea that they are 'very stretchy' is correct in that they would stretch far more than a toothed belt or a chain (and thus be inefficient), but is otherwise not a helpful idea; like toothed belts they have fabric reinforcement that is intended not to stretch permanently over time.

Toothed belts need less preload in order not to slip but a skinny belt will be fighting preload tension if it is not to slip under load. Even then it may feel a bit stretchy in use, and may also tend to knock the hub and BB bearings out. A belt that is stiffer in tension (so can run with less preload), but still flexible in bending, is a tough ask; twice as wide is twice as stiff and has twice the losses under low load. If the internal reinforcement is made twice as thick, the belt starts to be rather hysteretic in bending and again therefore more lossy.

In the fullness of time we may see belts that are reinforced with CNTs and then we may be going somewhere. In the meantime there are compromises....

cheers

[bobc]

Just a bit more info on my "industrial drive" experience.
First of all, "doing the math" for a bike transmission application always ends up being way too hard- the belt manufacturers all publish nomographs and loading data but it never extends down to the rpm range, custom belt width and low power requirement of a pedal drive. You can try to extrapolate but it has to be done several ways, and belt strength vs width will not be linear in the small width range.
And the americans insist on using imperial units.......
So I'd figured that a "megadyne" (sounds like a movie villain) 14mm pitch belt 10mm wide should be strong enough (just - I was assuming they had already applied a factor of safety in their recommendations....) and built up a bike.
Because of availability issues I used a 1288mm long belt and, together with the SA8 IGH gearing -up proclivities I ended up with a smallish chainring, pitch radius, equivalent to a 44tooth on a 1/2" chain. That's a pitch radius of 89mm. let's say I'm 115kg and I have 170mm cranks, when I stand on the pedals that's 220kg (2155N) tension in the top belt run.
The bike was absolutely fine until I stood on the pedals. There was no failure but there was clearly an elastic stretch, you could feel it and you could see it in that on the visible "outside" of the top run of the belt, "tiger stripes" became apparent at the knobble pitch. And the belt unshipped at that time, presumably because the stretch on top resulted in slack underneath, or maybe made the stretched part climb out of the chainring notches. Who knows.
The belt with carbon tensile members is advertised as being over twice as strong as the fibreglass one, so I think it will be up to snuff.

In my experience carbon doesn't stretch - it snaps.....

[bobc]

Lasered a torque arm in 3mm stainless: together on hub in this pic.
I bought one of those clips sturmey archer do for brake torque arms, which I'll fit on the chainstay (if it can do front brake torque, it can do this

lasered torque arm and brake disc

[Brucey]

looks nice that, but a bit stressy at the front? I guess you have done the sums and know what kind of reaction torque you will be dealing with? I guess with a Rohloff the reaction torque in low gears is the main concern; the loads will be a lot less in the high gears because of the spread of gear ratios.

For the rear hub reaction arm I quite fancy a braze-on with a captive fastener (could be a wing nut) that passes through a (sort of keyhole-) shaped slot in the reaction arm. If the arm is long enough (like yours) it ought to be easy enough to fit it since not only the wheel will move but also the arm will flex sideways without too much effort. Once in the slot, the fastener can't come out or be lost (because it is captive anyway).

cheers

[bobc]

20mm wide belt has arrived. I need to slice this down the middle to hopefully end up with 2 10mm wide belts. The belt seems rather stiff so I'm not confident that a Stanley knife blade will do the job.. First though I have to make a wall mounted pulley behind the lathe so I can run it and try slitting methods. I suspect I will end up using one of those fine circular saw blades in a dremel.

[bobc]

Brucey makes a good point, I've got the gear to analyse this properly & I should use it.....
So what's the axle torque? 2 approaches:
1) what can I supply? 100kg on 170mm cranks is 100 x 9.81 x 0.17 = 167Nm
gear ratio 60/18 so rear sprocket torque 167 x 18 / 60 = 50Nm
wheel torque = that / gear ratio = 50/0.279 = 179Nm (bottom)
= 50/1.467 = 34.1Nm (Top)
The difference between these torques comes from the axle & that's 167 - 50 = 117Nm in bottom
and 50 - 34.1 = 16Nm in the other direction in top. There's 7x as much axle reaction torque in bottom gear c.f. top
2) what gets me up my steepest climb? Let's say that's 15%, that's a force of 100 x 9.81 x 0.15 = 147N
Wheel radius is 0.325m so wheel torque needed is 147 x 0.325 = 48Nm
Axle torque in bottom gear in this case is 48 x (1 - 0.279) = 34.5Nm
Interesting dilemma; do I stress it for what it will see (34.5Nm) or what I can give it (117Nm). Hmmmmm
The item in the picture above WILL get me up that hill, but would yield if I placed the front wheel of the bike against a house side & then stood on the pedals. I did a "beefed up" design that should be OK in the latter case, and should work on the bike, the picture below shows a FEA stress analysis of the beefier design, red areas are overstressed. I think I'll end up using this design

too much stress........

[Brucey]

I'd err on the side of caution too; the peak torque is often double the average required to climb a given gradient, and any time you decide to get a shift on (or when starting off) the torque can approach the maximum for a few turns of the pedals at least...

cheers

[bobc]

FEA "stress" output. Note the deformation has been exaggerated by the software by a factor of about 12. I've cut out some unused steel...

similar levels of stress

[bobc]

Slicing the belt down the middle: pulley/sprocket in 3jaw + idler rawlbolted to wall above. Started cut with a dremel(lookalike) & continued it with a Stanley knife blade (blade tended to wander without a pretty good groove to run in)

slicing toothed belt

[bobc]

cut's not exactly in the middle (cutting blade moved under pressure

belt slicing 2

I'll start using the 8mm wide part & that should be a handy stress test...

[greyingbeard]

I love a bit of diy...

great stuff, keep it coming