Avoiding power idler by using Alfine hub gear

[StephenW]

Hello all

I have a Metabike, which I enjoy riding. It currently has 3*10 derailleur gears. The idler is quite noisy, because it only bends the chain through a small angle. I have discovered that it is possible to remove the idler and ride the bike like that. Unfortunately in some of the gears the chain rubs on the seat bracket, but in the gears where it does work, it is nice and smooth and quiet. It feels more efficient and less flexy, although that might just be psychological, as it makes no noise. (The less flexy part comes from the fact that the direction of the chain is more in line with the frame, thus causing less of a bending load on the frame).

I've been thinking that if I get a hub gear, I could run it without the power idler and still have plenty of gears. I would continue to use a return idler, but would mount it in such a way that it can be moved, so that it doubles as a chain tensioner. Perhaps the lower efficiency of the hub gear will be compensated for by removing the power idler? Also, using a hub gear means that it is possible to use only a largish chainring, reducing chain tension and frame flex compared to a small chainring.

The final part of my plan would be to put the chain inside some kind of conduit. Not the kind of chain tube which rubs on the chain, but a tube which is larger than the chain, and is firmly mounted to the frame. Like a chaincase on a roadster. This would keep both the chain and my legs/trousers clean. The chain would slap against the inside of the tube over bumps, but I don't mind this too much.

What do people think? Has someone done something like this before?

Thanks
Stephen

[hercule]

I've got a couple of upright bikes running 8 speed hub gears (2 Shimano, 1 Sturmey), and I quite like them - well, the Shimano ones anyway! The main downside for a recumbent would be the limited gear range, circa 305% on the 8 speed hubs and c. 400% on the Alfine 11 - much smaller than your current 3x10 set up, and you will obviously need the range on the Metabikes as you can't honk up hills! Plenty people have paired a hub gear with two or three chainrings up front, though that will result in some lateral running of the chain which may bring you back to your starting point...

Personally I've never had any issue with standard chain tubes and am delighted that they keep my chain much cleaner than on the uprights!

[UpWrong]

You'll lose more than you gain. A Rohloff would be more efficient than an Alfine but would probably make more noise than your power idler.

[StephenW]

Thanks for the replies.

Hercule: Yes the range of the 11 speed Alfine is a bit less than I had thought. Or rather, my current gear range is a bit more than I thought. I use all the gears I have.

UpWrong: I found an old thread on BROL, where someone from Bacchetta was suggesting losses of 1-2% for an idler. I'm not sure of the losses in the derailleur jockey wheels, but I guess it's not much. So from a strict efficiency point of view, I'm sure you are right. However, if removing the idler increased the stiffness through the drivetrain, this could be beneficial, especially for hill climbing. This is harder to quantify I think.

I had a bit of a read around. It seems there have been some recumbents without power idlers. People said that they were nice and smooth and quiet.

I do think that stiffness through the drivetrain is important. I don't believe that much energy is being converted into heat through flexing. (Although if it was, you probably wouldn't know, due to good thermal conductivity of the frame and good heat transfer through forced convection). I think that the problem with flex is rather that the feet have to make additional movements which are not contributing to the propulsion of the bike, and the flex also shortens the useful power stroke.

Anyway, I think the effect of chainring size on flex is important, and doesn't seem to get a lot of attention. Using the 22 tooth chainring, with 152 mm cranks, the bike feels noticeably noodly. (I am tall and have the boom extended out a long way). Perhaps there would be value in using one of the cassettes with a humongous 42 tooth sprocket, in order to give low gearing without requiring small chainrings. (Hub gear/hybrid gearing would also give low gears without small chainrings).

[OldBloke]

I would have thought the 'noodling' you experience StephenW is more to do with the ratio you are pushing than chain ring size. For example 22x11 shouldn't give anymore noodling than 42x21, while a 42x42 or 42x36 would give more noodling than 22x11.

OB

[squeaker]

I would have thought the 'noodling' you experience StephenW is more to do with the ratio you are pushing than chain ring size. For example 22x11 shouldn't give anymore noodling than 42x21, while a 42x42 or 42x36 would give more noodling than 22x11.

Except that with 22x11 you are doubling the chain tension compared with 42x21 (at the same cadence & power output), which IME is a significant contributor to boom flex...

[[XAP]Bob]

And when pushing with the drive side pedal you are countering the chain tension's tendency to twist the boom, when pushing the non drive side you are doubling the tendency to twist the boom.

[squeaker]

And when pushing with the drive side pedal you are countering the chain tension's tendency to twist the boom, when pushing the non drive side you are doubling the tendency to twist the boom.

Indeed, and with a spotlight on in the dark the effects are all too obvious

[[XAP]Bob]

Yes, it happens a bit in other vehicles, but we tend to have the power stroke in line with the chain tension, rather than perpendicular (as on an up wrong) as well as having a light mounted there to make it more obvious. It's also not supported in as many directions as a conventional BB position (down tube, seat tube, chain stays)

So there can be a feeling of the boom swaying from side to side. Not sure I'd ever call it 'noodle' on mine....
I can watch the ICE boom sway, but it's not a huge deflection...
Can't see it on the raptobike, but the hamster bars don't put the boom in easy view anyway...

[squeaker]

Think the Rapto is significantly stiffer than ICE (did some sums a while back: my Grasshopper is about 2x as stiff as ICE).

[[XAP]Bob]

Think the Rapto is significantly stiffer than ICE (did some sums a while back: my Grasshopper is about 2x as stiff as ICE).

I suspect it is as well, the boom is significantly larger diameter.

BUT the ICE boom is back compatible a long way, and I don't see the flex as a significant issue.

[Brucey]

re the noodly feeling; as others have commented the absolute chain tension will affect this. So if you are worried by it, it is better to use larger sprockets and chainrings to achieve any given gear ratio. There is another factor which is that, at speed, the (small) change in speed with each pedal stroke is smaller than at low speed; this means that the peak torque at the crank is liable to be higher than when using a lower gear at lower speed.

Finally the machine won't be going in a straight line, not quite; there will be some weaving as the weight of your legs is moved and some more if you are reacting any load through the steering controls. So when you see the headlight beam move from side to side there are a number of contributions to this, as well as real flex in the boom. The 'lever' is huge, too, so a tiny movement may make the headlight beam move a long way.

Re idler efficiency vs long chain runs: I don't like chain tubes or idlers either, but they are the lesser of several evils in many cases. A long unsupported derailleur chain run will sag (top and bottom) into a catenary curve whenever there is no power being transmitted, and there are losses and a spongy feeling associated with pulling the chain run taut again with every pedal stroke. The sagging is dependent on the length of the span and the weight of the chain, so is likely to vary (to a first approximation) with the square of the span; i.e. doubling the span may make the sagging about four times worse. Tests to assess losses this way are difficult, since only a pulsing power source will generate losses of this kind. Most transmission tests are carried out using continuous drive electric motors, which are not subject to the same effects at all. [ BTW There is a real dilemma in choosing the power level used for such tests; peak, average, or something inbetween?]

I built a recumbent machine with a 700C rear wheel a couple of years ago and I wrestled with similar design considerations. I wanted to use simple parts, have a wide gear range, and have at least a passing nod to efficiency. The machine I built was intended to be used on the road for normal riding, and perhaps occasionally for racing. In the end I mounted a SA 3s hub as a mid-drive in the frame. The conversion was simple; a 3/32" output sprocket was bolted to the steel flange of the 3s hub. This allows a 1/8" chain (which can be rejoined and extended etc easily) to be used at the front and a 9s chain to a cassette on the rear.

Both chains are carefully (luckily...) fractionally less than 1-1/2 normal chains long, so buy three chains of each type, get two complete sets of chains for the bike. The rear chain needs one extra quicklink to make it work. There are no chain tubes. The chain runs are not so long that there is excessive sag, so a normal rear mech will work OK. The front chain is almost in line with the rear chain too, which means that the reaction load onto the bearings in the mid-drive (in the middle gear) is very small, which should make for good efficiency.

The idea is that for touring, a large input sprocket is fitted to the mid-drive, giving low gears for hills etc. Normal riding on the flat is in the high gear of the 3s hub, which is a bit lossy. For racing, a smaller input sprocket is fitted to the mid-drive, which means that the (efficient) direct drive ratio gives useful gears that allow you to ride in the ~25mph range, with a high range that comes into play for forays above 30mph and a low range that isn't really low enough for climbing steep hills. It seems to work OK. The most noticeable thing is that the transmission is (by comparison with the usual racket from idlers, tubes etc) virtually silent. The single part that makes most noise is the rear mech, and that is far enough away that you don't hear it.

So with an output sprocket of ~22T and a chainring of 44T, a 'touring' input sprocket of 24T gives virtual chainring sizes of 30.25, 40.3T, 53.8T. Obviously a smaller chainring would get those gears lower if required, if crawler ratios are required. For racing a 16T input sprocket gives virtual chainring sizes of 45.4T, 60.5T, 80.7T. In the direct drive gear I hope the losses are comparable with an idler, and in the other gears the losses are likely to be two or three percent worse than that.

The mid-drive sits in a 'cradle' which has bolted-in 'dropouts' which can be respaced. This means that the mid-drive can be substituted for almost any hub (from ~110mm up to about 140mm OLN) if needs be, including a derailleur one, although this would be noisier. Due to bad planning, the mid-drive is slightly higher than is optimal (a last minute change in design allows a longer rear mech in the front drive if needs be) so a large diameter IGH mid-drive would clash with one of the frame tubes, but if necessary this could be resolved with a little more cutting and welding; the main spar of the machine has two large diameter steel tubes, one above the other, so is plenty stiff enough to cut about if needs be.

Of course the simplest thing would be to use a pair of input sprockets on the SA hub (it is an older one with a longer driver which will accept two without modification), and a rear mech to shift between them. With a ~32T chainring, and the same input sprockets, this could give six virtual chainring sizes from ~21T to ~60T with lots of efficient ratios in the midrange.

So I don't think this transmission is perfect by any means but it is built from readily available parts, is simple, quiet, strong, and has a wide enough range for most folk. It is tolerably efficient and only a little heavier than normal. It is of course incredibly versatile and flexible in nature; if needs be it can even be reconfigured to have, uh, an idler in the mid-drive mountings.....

cheers

[[XAP]Bob]

Finally the machine won't be going in a straight line, not quite; there will be some weaving as the weight of your legs is moved and some more if you are reacting any load through the steering controls. So when you see the headlight beam move from side to side there are a number of contributions to this, as well as real flex in the boom. The 'lever' is huge, too, so a tiny movement may make the headlight beam move a long way.

Of course on a trike the steering response is somewhat different, but I agree that the light does make a very sensitive measuring device (it's used in many experiments for that reason, as well as in early long distance telegraph machines.
On the trike however I can see my feet, and the chain, and I can see the boom flex to the right as I push with my left foot and pull with my right, and straighten on the opposite phase.
It never flexes left - because the chain tension is always pulling to the right.



As for mid drives of the sort you built - I agree that they make an awful lot of sense where they can be fitted. A BWR hub might even be convinced to sit between ones knees on a low racer - though there would be a significant thrust vector into the bearings (how would that compare with a hub in a 'normal' frame though?).

[Brucey]

re the thrust vectors into the hub bearings; these are (net, I think) less bad in a mid drive than in a rear hub. In the middle gear, with similar chain tensions in and out, and the chains well aligned, the losses might be comparable with a good idler, but with the benefit that the chain runs into the mid-drive sag less, and don't necessarily need guide tubes in the same way etc

It occurs to me that the amount of chain wrap around an idler affects the way it works; there is a fundamental difference depending on whether the chain wrap is an exact number of half links long, or 1/4" different to that.

cheers

[UpWrong]

My Stratus XP doesn't have a power idler and the transmission feels quite efficient when climbing. If I remember correctly, the Metabikes idler is a smooth untoothed one, but with little deflection? Do Terracyle make a double idler for it (toothed power, smooth return)? That might improve it although noise is not necessarily a function of inefficiency. Pat Franz of terracycle has written about how the noise can change with the number of teeth enagaged by the chain in the idler. For a smaller deflection it seems you need a larger diameter idler in order to keep 3 teeth engaged. Found this quote of his:

"If you analyze the path of a chain around an object, you'll find that since the chain is made up of links, the path isn't at all like something continuous, (rope, for instance), would be. The chain only bends at each pin. This means the paths of the pins and the path of the chain tension line between the pins may or may not be the same. Effectively, there are distinct numbers of contact points between the chain and the idler. When the chain angle is low, the number of contact points alternates between 1 and 2 as the links move past (at ridiculously low angles, it alternates between 0 and 1, of course). As the angle increases, it goes to alternating between 2 and 3 points. When it gets big enough, there are always at least 3. Three points define an arc, and once you get 3 or more points of contact, the chain flows smoothly because the contact points are on an arc that matches the idler.

At less than 3 points, the path of the chain doesn't match the idler. When you zoom in and look at it, the tension line going from pin centerpoint to pin centerpoint goes up and down as each link makes contact and leaves. This is what produces the throbbing sound from an idler. It doesn't matter whether it's a cog or a groove, since chains only bend at the links. As long as the centerpoints of at least 3 pins aren't on the arc of the idler, the tension line won't move along the arc of the idler, and there will be throbbing. A soft idler will smush and deaden the effect, but energy is lost and not recovered. The smaller the idler, the larger the angle required to get enough contact. Here, cogged idlers have a slight advantage, since they contact the chain slightly further out, but the difference isn't that much.

This effect is why, paradoxically, straighter chainlines are noisier. It is amazing to listen to an idler as you change the chain angle. As you transition from 1/2 points of contact to 2/3, there is a big drop in noise. As you transition to 3 or more, things get very quiet, assuming the tooth shape is right. On some bikes/trikes, the chainlines are right at the critical angles, and you can hear the difference as you shift."