Gear hub madness

Today I came across a good description of how the new SRAM G8 hub works, and I worked through the ratios in a spreadsheet. Its a strange hub and it took me a while to figure out what was going on, even when someone had already attempted to explain it. This got me thinking about another difficult nut to crack, the Alfine 11.

As far as I can determine, no one has done a teardown of an Alfine 11, counted the teeth, or posted any particular information about how its ratios are created.  In the past, I've thought it was quite similar in design to a Nexus 7, but now I've changed my mind.  However before that, lets go on a tour of gear ratios in hubs.

Gear hubs are made up of planetary gear sets: http://en.wikipedia.org/wiki/Epicyclic_gearing

Generic three speed:

The most basic hubs have one gear set, giving three speeds.  In the middle there is direct drive, then one ratio increased and one ratio decreased (the inverse of the increased ratio).

gear	1	2	3
ratio	0,750	1,000	1,333
ratio change		33 %	33 %
name	1/A	-	A
calculated ratio	0,750	1,000	1,333

SRAM P5:

This was extended to make 5 speeds such as the SRAM P5 by placing two gear sets into one module, so that one of two ratios could be used at a time, either up or down.

gear	1	2	3	4	5
ratio	0,633	0,781	1,000	1,281	1,579
ratio change		23 %	28 %	28 %	23 %
name	1/A	1/B	-	B	A
calculated ratio	0,633	0,781	1,000	1,281	1,579

SRAM S7:

This also worked for 7 speeds, such as the SRAM S7.

gear	1	2	3	4	5	6	7
ratio	0,574	0,677	0,809	1,000	1,236	1,476	1,742
ratio change		18 %	19 %	24 %	24 %	19 %	18 %
name	1/A	1/B	1/C	-	C	B	A
calculated ratio	0,574	0,678	0,809	1,000	1,236	1,476	1,742

SRAM i-Motion 9:

It even worked for a 9 speed, the i-Motion 9.  This hub was basically a failure on the market, apparently because it was too heavy, rough running, expensive, and poorly sealed against weather.  However it was apparently well constructed, and it did achieve 9 tight and evenly-spaced ratios while using only one planetary gear at a time, potentially giving high efficiency.  I think its a shame it didn't make it.

gear	1	2	3	4	5	6	7	8	9
ratio	0,542	0,621	0,727	0,853	1,000	1,172	1,375	1,611	1,844
ratio change		15 %	17 %	17 %	17 %	17 %	17 %	17 %	14 %
name	1/A	1/B	1/C	1/D	-	D	C	B	A
calculated ratio	0,542	0,621	0,727	0,853	1,000	1,172	1,375	1,611	1,844

Shimano Nexus 4:

But there are other fun things that might be done.  Consider a Nexus 4, which made four ratios from a set of three planetary gears, by only increasing ratios.  Leaving out the ability to decrease ratios simplified the hub.  This simplification, where each planetary gear is only used to either increase or decrease ratio but not both, is a major theme of all recent hubs.  There i-Motion 9 above might well have needed only 4 planetary gears to get 9 beautiful-looking ratios, but those 4 gears were too complicated.

gear	1	2	3	4
ratio	1,000	1,244	1,500	1,843
ratio change		24 %	21 %	23 %
name	-	A	B	C
calculated ratio	1,000	1,244	1,500	1,843

Shimano Nexus 5:

But all these hubs have fairly large steps between the gears.  Especially the gap between direct drive and the first ratio up or down was problematic.  While i-Motion 9 got this down to 17%, it was apparently unpleasant.  Even the 22% gap on Nexus 8 leads to some gear whine.  So Shimano got busy with dual-stage compounding its planetary gear sets.  Below is the Nexus 5, which used three planetary gears in two sets, and has no direct drive ratio at all, but is very smooth in my experience.

gear	1	2	3	4	5
ratio	0,750	1,001	1,159	1,335	1,545
ratio change		33 %	16 %	15 %	16 %
name	1/A	B/A	C/A	B	C
calculated ratio	0,750	1,001	1,159	1,335	1,545

Shimano Nexus 7:

The Nexus 7 is similar, but with four planetary gears arranged in two sets.  I've used two of these extensively and never been satisfied with either their smoothness or efficiency, but they seem to sell in vast quantities in Denmark.

gear	1	2	3	4	5	6	7
ratio	0,632	0,741	0,843	0,989	1,145	1,335	1,545
ratio change		17 %	14 %	17 %	16 %	17 %	16 %
name	1/A	1/B	C/A	C/B	D/B	C	D
calculated ratio	0,632	0,741	0,844	0,989	1,145	1,335	1,545

Shimano Nexus 8 / Shimano Alfine 8:

But thats not the only way to do multi-stage compounding.  The Nexus/Alfine 8 is a lot like the old Nexus 4 with a new planetary gear enabling the four original ratios to be used twice.  This leaves a direct drive gear in place which is good, but creates an odd situation where the least efficient gear (4) is right next to the most efficient (5), and switching between the "high" and "low" gears can be problematic.  Still, this seems to be a robust design.

gear	1	2	3	4	5	6	7	8
ratio	0,527	0,644	0,748	0,851	1,000	1,223	1,419	1,615
ratio change		22 %	16 %	14 %	18 %	22 %	16 %	14 %
name	1/A	B/A	C/A	D/A	-	B	C	D
calculated ratio	0,527	0,645	0,748	0,851	1,000	1,223	1,419	1,615

Rohloff:

And then there is the monsterous 14 speed Rohloff, that uses 5 planetary gears in three sets.  Gears 3 and 5 actually use triple-stage compounding, which is hard to notice in my experience.  In some ways this combines concepts which are found on both Nexus 7 and Nexus 8, sort of a combination of those designs.

gear	1	2	3	4	5	6	7	8	9	10	11	12	13	14
ratio	0,279	0,316	0,360	0,409	0,464	0,528	0,600	0,682	0,774	0,881	1,000	1,135	1,292	1,467
ratio change		13 %	14 %	14 %	13 %	14 %	14 %	14 %	13 %	14 %	14 %	14 %	14 %	14 %
name	1/B * 1/A	1/C * 1/A	D/B * 1/A	1/A	E/C * 1/A	D/A	E/A	1/B	1/C	D/B	-	E/C	D	E
calculated ratio	0,279	0,317	0,360	0,409	0,464	0,528	0,600	0,682	0,774	0,881	1,000	1,135	1,292	1,467

SRAM G8:

Here is the SRAM G8 in all its confusing glory.  Five planetary gears in two sets, sort of arranged like a backwards Nexus/Alfine 8.  No direct drive, and 6 of the 8 ratios are two-stage compounded.  http://sheldonbrown.com/sram-g8.html

gear	1	2	3	4	5	6	7	8
ratio	0,609	0,710	0,803	0,903	1,054	1,204	1,355	1,581
ratio change		17 %	13 %	12 %	17 %	14 %	13 %	17 %
name	1/B	1/C	E/A	E/B	E/C	D/A	D/B	D/C		1/A	D	E
calculated ratio	0,609	0,710	0,803	0,904	1,054	1,204	1,356	1,580		0,5411	2,2258	1,4839

Shimano Alfine 11:

Then finally we get to the Alfine 11.  I spent some time on this.  There are lots of ways to make a good approximation of the published ratios, but the way I am most satisfied with uses four planetary gears arranged in three sets.  The main difference between this and Nexus/Alfine 8 is that one gear (number 7) is able to combine with gears 8 and 9 to create ratios 10 and 11.  Then all those resulting 5 ratios are re-used in positions 2-6.  This very nicely explains the 29.2% gap between gears 1 and 2... its same ratio as gear 7 has.  This also has no planetary gears which are used to both increase and decrease ratios, which appears to be very much out of style, it explains the alternating 13% and 14% ratio changes, and it explains how the 11 speed is able to match the weight of the 8 speed.  Essentially, the same number and strength of gears needed to be used.

There is no direct drive ratio, which is instead replaced by a triple-compound approximation, I imagine to achieve mechanical simplicity.  The only difference between 2-6 and 7-11 is whether or not the "low range gear" is active.

gear	1	2	3	4	5	6	7	8	9	10	11
ratio	0,527	0,681	0,770	0,878	0,995	1,134	1,292	1,462	1,667	1,888	2,153
ratio change		29,2 %	13,1 %	14,0 %	13,3 %	14,0 %	13,9 %	13,2 %	14,0 %	13,3 %	14,0 %
name	1/A	B/A	C/A	D/A	(B*C)/A	(B*D)/A	B	C	D	B*C	B*D
calculated ratio	0,527	0,681	0,770	0,879	0,995	1,135	1,292	1,462	1,667	1,889	2,154

Further reading, here is the grand daddy of complicated bike gear hubs: http://sheldonbrown.com/elanratios.html

Nasty slush

Oslo just hasn't been a proper winter city this year.  After three weeks frozen, its been going back and forth between snow and rain.  Quite a mess on the way to the daycare, but the conditions were nearly perfect during the commute from the daycare to work.  On most days I chose to use the Nihola with less-impressive gearing and the Schwalbe Snow Stud, to benefit the commute part of the daily journey.

Harmless looking, but difficult.  That rut took some effort, but the hill was climbed.
Slope less than 10%, one kid, perhaps 6kg on the rear.

This was especially nasty, hard to drive on even when level, and it packed into the tread.
Spinning solved nothing here.  Having two kids in the box didn't make it easier.

It turns out that really dense snow, stuff which is transitioning into slush and ice via rain, is really a challenge on a hill.  It drags on the tires, and digging might never reach down to anything better.  It seems that bikes have an advantage here due to better weight distribution and more tire per unit of mass.

Parking lot slush.  I should have taken a picture of the ice that followed.

On level ground, slushy snowy crap can be much deeper and rutted without causing too much trouble.  But remember to rock the trike to "push" past the worst parts.

Looks like snow on top, but it transitions to dense slush.

It doesn't take a lot of dense snow underlain by slush and ice to cause a problem on a climb.  The stuff I encountered was really heavy, so just getting the tires through it was a challenge, but if the rear tire dug much it would just come to some sort of dense semi-slush ice.  On level ground it went well enough.

My son eagerly does a bit of cleanup.  This gate is kind of irritating in difficult conditions.
Slope is as much as 15% through here.

Picking up my son after a daytime snow flurry.  Clean snow isn't a problem.

This period has nicely illustrated to me that I am dependent on someone to do at least a halfassed job of snowplowing on the steeper sections of the climb to my son's daycare.

Nexus 8 SG-8C20

When I broke the SRAM P5 that came with our blue Nihola, we spent considerable time trying to find the right replacement.  It would have been a Nuvinci N360 with a roller brake, except the shop in England selling N360's wasn't really interested in ordering one with a roller brake mount, and bike shop employees in Denmark seemed to have little time for "strange" hub gears.  Basically people were not interested in being paid to do anything they didn't normally do; perhaps the margins were too small for them to care.

The Shimano Nexus 8 with foot brake would be another good option, except in Denmark (and some other countries) it was only possible to order the old SG-8C20 version.  It seemed that Shimano refused to sell the SG-8C31 until the old model was sold out (on a per-country basis), and no one could order one for me, nor could I find any for sale online.

I also considered the SG-8R36 with a roller brake, but it was more expensive and we wanted to stick with a coaster brake (foot brake) unless we had a really good reason to change.

So we ordered the old model, the SG-8C20.  It had probably been sitting on a warehouse shelf for years.  I installed it myself.  Right away I had some issues because the "cassette joint" wanted to occupy a bit of space which the chainstay was using.  I installed some extra washers and bent the cassette joint a little to get the shifting in order, but the whole thing was always a pain to fiddle with.

The hub itself was functional, but made me nervous that it was headed for premature death.  From day #1 it made various clicks and had various shifting issues between gears 4 & 5.  Gear 5 (direct drive) seemed like it might slip at any time (but it never actually slipped, except after shifting from 4th).  Adjustment of the shifter could address the various issues but always at a cost somewhere else.  At some point, the clicking and humming alone was enough for me to know what gear my wife was pedaling in, if I rode along behind her.  Still, it was working.

Happily it seems there are ways to address these problems.

The Nihola company had started using Nexus 8 hubs (I presume SG-8R31), so when visiting Copenhagen I checked with them to see how they handled the cassette joint.  Turns out they had some different non-turn washers that I didn't realize existed: 5L and 5R.  They sold me a set of them for a reasonable cost, and one source of irritation dissappeared.

On my next visit to Denmark I took the Nexus 8 into Byman Cykler and had them put some new grease in it.  Now this was the magic the hub needed.  Most of the noises and shift problems vanished and haven't come back (in over 4 months).  So, moral of the story: hubs like fresh grease.

There is one lingering irritation with this hub, however.  It works its axle nuts loose over time, and then pulls itself crooked in the frame.  I guess I'm putting too much torque on it.  Always carry a 15mm wrench!

So now I can make a bit of a verdict on the Nexus 8 with coaster brake as a hub for a Nihola.  Its pretty good.  A bit rough in 3rd and especially 4th gear, in fact its generally a rough hub.  Buts its strong: ours is set up with a 24t sprocket on the hub and 38t on the crank, well beyond any ratio any hub company endorses, and I've put my entire weight on it countless times, plus added arm strength.  I've gotten small blisters from pulling on the handles to force the pedals around.  (I do these terrible things in 1st gear exclusively, which is probably the strongest gear available.)

The Nexus 8 is a strange hub.  Its not symmetric (i.e. the top ratios are not mirror images of the bottom ratios) and the least efficient gears are in the middle of the range (3 & 4) rather than the ends.  However this works out well enough on a Nihola.  For any real hill, I just drop to 1st gear and pedal.  On any reasonably flat ground, I can stick to gears 5-8, which are mostly pleasant.  I never make any strenuous effort in 3rd or 4th gears.

The foot brake is strong enough to slide the rear tire at will, though Oslo hills might cook the hub.  More than once I have been unable to touch the hub's shell after a good downhill.  After that happened a few times, I decided to have a V-brake installed in addition to the foot brake, so now that Nihola has two rear brakes.  Overall I am very pleased with the foot brake.  I'm convinced foot brakes are the best rear brakes for Niholas, except possibly in deep snow, ice, or mud.

But is the Nexus 8 a worthy upgrade over a SRAM P5?  For any moderately serious cyclist, I would say yes.  The ratios are closer, which makes it easier to go fast without knee injuries.  The lowest ratio is good and low.  The hub itself is stronger.  It shifts easier.  Its easier to handle a freezing shift cable.  There is no "clickbox" sticking out in the way of pannier bags.  On the other hand, the P5 is a lot smoother in operation, its likely a bit more efficient, and it doesn't take tools to disconnect its shift cable.

Anyway, Nexus 8 is a good Nihola hub.

The right winter tire

So we have two Niholas.  They've got different gear hubs and winter tires on them, and I swap back and forth as appropriate.  (We basically only take out both trikes at the same time in the summer.)  Since I can jump back and forth, I've got rather more insight into winter tires on Niholas than most.  Allow me to share.

The fancy trike has a Nokian/Suomi Extreme 294 (26x2.1), and the basic one has a Schwalbe Snow Stud (26x1.9).  While both are studded winter tires, the tires are in different categories.  Which is to say, they can be compared, but the differences outweigh the similarities.  The Extreme has way more studs and has more aggressive tread, which pays off on climbs, and on pavement it sounds like a wheel covered in velcro (especially at low pressures).  The Snow Stud is mild mannered on pavement, making hardly any noise of note, wasting less energy, and doubtless lasting longer, while also being a lot cheaper.  Of course it doesn't climb as well.

And to be clear, the Nokian/Suomi Extreme does climb.  It digs and it scratches; pure ice can possibly be climbed up to a 15% slope with something like 20kg in front 10kg in back, and low pressure.  Softer stuff can be climbed beyond that, however this can require patience, low gearing, and stamina.  The Snow Stud gives and spins up much sooner in almost any snowy, slushy or icy surface.  The worst surface I have experienced for the Snow Stud is one that is icy and slightly lumpy, it just completely defeats the tire.

So here is the difficult part about transporting children as part of a cycle commute in an Oslo winter: each tire is sub-optimal, one could say annoying, when its out of its element.  I haul a kid up a fairly steep and marginally-maintained hill on my way to work, and I also go the rest of the way into the office with less load, on better paths, on the same Nihola.  The tire that can conquer the hill (the Extreme) proceeds to sap my energy and destroy itself on km's of pavement.  The tire (Snow Stud) that is pleasant on the pavement can be difficult to scale the slopes with, in many conditions.  Happily, rear tire inflation can be adjusted to significantly widen the range of situations a given tire can perform acceptably.  Airing up and down is kind of annoying, but its a lot less work than cycling km's with an inefficient tire.  (So: want traction?  Air down.)

For a more normal Nihola usage scenario, which would be a relatively short tour of schools, stores and public transportation, either tire would vastly outperform a standard slick summer tire (in snow or on ice).  Also, with some basic steps like airing down and carrying some kilos on the rear rack, even the Snow Stud will likely provide more traction that the standard gearing of a Nihola can require.  (With the exception of smooth ice.)

A few more nuggets.  Its pretty easy to lock up the rear tire of a Nihola while going down hill on ice.  If not addressed, this can easily lead to the trike not pointing in the same direction it is moving, which is not necessarily cause for alarm, but should probably be corrected.  A rear tire with more studs will make sliding harder and recovery easier.  (I would advise using front and rear brakes on icy hills.)

I have purchased a pair of studded front tires (20x1.6 Schwalbe Marathon Winter), but not installed them.  The regular Schwalbe Marathon Plus tires work pretty well on some pretty slippery stuff, and unlike on a bike, sliding the front tires is not necessarily cause for alarm.  Studs would probably help, but also that adds rolling resistance and noise.  I have decided that the conditions are not currently bad enough to bother swapping the tires.

I would like to try a Schwalbe 26x2.0 Marathon Winter in back.  Looks like a great choice for ice, or even pavement, but its harder to say if it can dig well enough in snow.  One of the problems I've had with the Snow Stud is that it won't dig sufficiently in some types of dense slushy snow.  Sheets of ice may or may not be a primary concern, and the Snow Stud is priced fairly low, so it could be that the humble Snow Stud remains the best trike tire for reasonably flat places with reasonably mild winters.

Update a month later:

Still running the Snow Stud, and its pretty harmless on pavement.  Really not a bad tire for insecure weather.

A Schwalbe Snow Stud is happy even with no snow in sight.