Adjusting the bearings on a Mavic hub.

[Globalti]

Anybody who owns Mavic Ksyrium wheels and I expect several other models, will be familiar with the alloy ring on the LH side that has six holes drilled in it to take the plastic pegged adjustment tool. I actually removed mine from the rear fat axle last night to clean it all up and it's a beautiful little bit of engineering, the alloy ring has a deep thread, which has a groove half way through it that contains an O ring to create friction and prevent it from unscrewing.

Now I understand how bearings work and I know about side-loading a bearing being not a very good idea so I can't quite work out in my mind how these bearing ajusters work to take out play - do they in fact pre-load the inner race so as to take out any play resulting from wear?

 

[Yellow Saddle]

Good observation and good questions.

Most cartridge bearing hubs have no preload adjustment (I'm talking lateral preload here since radial pre-load is supposedly taken care of by the press-fit tightness) and rely purely on the manufacturing tolerances of the axle and hub and spacers and end caps to provide the correct neutral preload. Mass manufacturing and machining being what it is, cannot do that accurately and most cartridge hub bearings are never properly adjusted and we see evidence of it all the time - bearings fail prematurely. Most people think that is how it should be and live with it, but still go on and proclaim the virtue of cartridge bearing hubs which they incorrectly label as "sealed bearing" hubs.

An astute mechanic will know about this and perform little tricks to get the bearings to preload properly. For instance, if you install bearings in Tune hubs, you will notice that after you've driven the various bearings onto the axle, there is sometimes a bit of notchiness in the bearings. If you then just gently tap the axle from the opposite side to which it was driven in, the notchiness disappears, suggesting that the preload has changed.

Mavic has engineered their hubs so that you don't have to tap it from either side until you have just the right sweet spot. They have made it so you can dial in the sweet spot. It is a nice feature. They have also spec'd their Ksyrium wheels with nice big bearings which give you a reasonable bearing life. Stupid light wheels such as American Classic and Tune have a bearing life measured in Mayfly lifespans.

Cartridge bearings or to be specific with the ones used in wheels, are deep groove bearings and they are very sensitive to lateral loads. The slightest incorrect lateral pre-load destroys them quickly. I like Mavic's approach but I hate the plastic spanner. They strip too easily and although you may think that it is strong enough, the same spanner is used on MTB wheels where the adjuster cap is often frozen in place with dust.

 

[Globalti]

Good reply thanks, so help me with the bearing overload bit: if you side-load the inner race you begin to force the rolling balls to carry the load on their shoulders, so to speak, rather than vertically through their centre axis? In other words if you look at a ball bearing and divide it up like a clock face, the greater the side load the more the bottom point of load transfer moves up from 6 o'clock towards 3 or 9, meaning that the load places a shearing force on the ball. This forces the ball down the wall of the groove subjecting it to a scrubbing action as it spins and no lubricant will be able to remain stuck to the ball or the groove in those circumstances.

So the Mavic "sweet spot" adjustment can only be a means of delaying the demise of a bearing that's worn and the more play you adjust out, the more you hasten the demise of the bearing?

 

[Yellow Saddle]

Sketch 1: A deep groove bearing (cartridge bearing) in perfect alignment with the ball perfectly inside the channel created by the groove in both races. The ball does not scrub on the sides and runs in a well-sized groove that does not constrict the ball. The ball in the sketch is supposed to be perfectly round, not egg-shaped.
Sketch 3. A deep groove bearing in poor alignment due to either a side force from cornering on your bike or, poor alignment of the bearing in the hub. The ball is constricted in a narrowed channel and scrubs on the sides of the channel that's narrowed by the relative movement of the two races. The squiggly thing in the middle is the distorted ball.
Sketch 2: An angular contact bearing as in cup-and-cone bearings found in Shimano and Campag bearings. The bearing can tolerate both a lateral and radial force because of the race's 45 degree orientation and shallow groove.

Back to your question.

Bearing overload and bearing side load is not one and the same. In an overload situation with the two races properly positioned, the ball is squashed radially and becomes shaped like a rugby ball. Because ball and races are only case-hardened, the hard shell cracks and starts to peel off. Overloaded bearings crack like an orange sheds its peel.

Constricted bearings fail differently, they start to gall and the evidence is smear marks in the steel, but not cracking like in overload. When the balls are constricted, they are in shear like you say and the ball is forced into an S-shape. These shapes I describe are of course not literal but only intended to indicate the way it distorts.

Most bicycle hubs display scenario 3 above. Sometimes temporarily as you corner, sometimes all the time because of poor alignment because of poor part tolerance manufacture or, poor installation. Mavic's way of improving (not fixing) this is to allow you to make a neutral adjustment of the lateral preload (which is actually zero preload in the neutral position).

When the bearing is pressed into it's seat in the hub, it is also preloaded radially. The hub squeezes the bearing. Manufacturing tolerances determine the amount of radial preload, which also have an effect on bearing life. In my experience, bearing life is not so much a function of bearing size, but of preload accuracy. This can be seen with Hope hubs which have many, large bearings but a short service life. American classic wheels on the other hand, have tiny bearings but they last as long as a Hope hub bearings, if not longer.

The reason most hub manufacturers use cartridge bearings is because it is the cheapest way to make a hub unless you have huge market share. These hubs use off-the-shelf bearing whereas cup-and-cone hubs require in-house bearing factories. The best wheels will have cup-and-cone bearings. On a top-end Campag wheel you can easily expect 100 000 kms without ever opening the bearing. Re-greasing is via a grease port.

True bearing wear is absolutely minimal. A bearing in a bicycle wheel that is properly adjusted and runs in uncontaminated grease, doesn't wear out.

 

[Globalti]

This explains why the bearings in my old Hope hubs wear out so fast, despite there being two 6001 bearings in the hub and two inside the freehub.

A ball bearing in a rigid race must by definition be a compromise, designed to tolerate a small amount of mis-alignment. In a perfect engineering environment such as a fully rigid shaft running in a rigid housing a roller bearing is almost the perfect solution and wouldn't even need much lubrication. Similarly a taper roller bearing offers an excellent B solution where there are side-loads (such as in a car wheel hub) inside a very solid housing. Unfortunately taper and plain roller bearings are bulkier and heavier than angular contact bearings. I know a little about this because I used to own and service a Land Rover, which has a complex transmission system containing many cartridge ball bearings, poorly protected from moisture and dirt and needing to tolerate the movements of massive supension travel and backlash due to the complexity and length of the drivetrain. Amusingly the Land Rover doctine on transmission noise is: "Continue to operate the vehicle until the noise becomes intolerable, at which point consider an overhaul"!

Thinking on ahead, I guess a plain journal bearing must be the ultimate solution but it requires a guaranteed film of pressurised lubricant, possible in the controlled, well-serviced, cool and extremely rigid environment of, for example, a steam mill engine, many of which are still running smoothly 150 years after they were built. In internal combustion engines these bearings usually only fail if the lubricant becomes contaminated or the parts are subjected to extreme pressure such as when the engine is flogged in high gear.

 

[Yellow Saddle]

This explains why the bearings in my old Hope hubs wear out so fast, despite there being two 6001 bearings in the hub and two inside the freehub.

A ball bearing in a rigid race must by definition be a compromise, designed to tolerate a small amount of mis-alignment. In a perfect engineering environment such as a fully rigid shaft running in a rigid housing a roller bearing is almost the perfect solution and wouldn't even need much lubrication. Similarly a taper roller bearing offers an excellent B solution where there are side-loads (such as in a car wheel hub) inside a very solid housing. Unfortunately taper and plain roller bearings are bulkier and heavier than angular contact bearings. I know a little about this because I used to own and service a Land Rover, which has a complex transmission system containing many cartridge ball bearings, poorly protected from moisture and dirt and needing to tolerate the movements of massive supension travel and backlash due to the complexity and length of the drivetrain. Amusingly the Land Rover doctine on transmission noise is: "Continue to operate the vehicle until the noise becomes intolerable, at which point consider an overhaul"!

Thinking on ahead, I guess a plain journal bearing must be the ultimate solution but it requires a guaranteed film of pressurised lubricant, possible in the controlled, well-serviced, cool and extremely rigid environment of, for example, a steam mill engine, many of which are still running smoothly 150 years after they were built. In internal combustion engines these bearings usually only fail if the lubricant becomes contaminated or the parts are subjected to extreme pressure such as when the engine is flogged in high gear.

Newer Hope hubs have five bearings in the rear and still.....
A plain journal bearing would still require axial preload of sorts and that's looking for trouble in a place where a 45degree bearing would do the trick.
Roller bearings also have more complex lubrication requirements and are best run in oil baths. Some years ago Shimano designed a Dura Ace Octalink BB with roller bearings running in grease and that was a disaster. Balls are better at recycling the grease than rollers, which tend to push it aside and then run dry.
My friends all have Land Rovers and it is comical to see them justifying their choice. Their cars spent many hours on trestles outside my workshop or in their natural habitat - on a flatbed tow-truck.
Ship propeller shafts are another good example of journal bearings in action. These shafts weigh several tons, yet float perfectly on a layer of pressurized, filtered oil.

 

[migrantwing]

Interesting....

 

[Tim Hall]

Yes, fascinating stuff. @Yellow Saddle , are the sideways (axial?) loads on a cartridge bearing different on a trike, than they are on a bike? Logic suggests this might be the case, as a cornering trike keeps its wheels perpendicular to the road surface (unless cornered very enthusiastically), whilst on a bike the wheels lean.

 

[Yellow Saddle]

Yes, fascinating stuff. @Yellow Saddle , are the sideways (axial?) loads on a cartridge bearing different on a trike, than they are on a bike? Logic suggests this might be the case, as a cornering trike keeps its wheels perpendicular to the road surface (unless cornered very enthusiastically), whilst on a bike the wheels lean.

Yes, axial loads. Although a two-wheeler doesn't corner perpendicularly and one's intuition is therefore that it doesn't load axially, it actually does. For instance, if you sprint (even a little) and move the bike from side to side, you load the wheel axially as well as radially. When you get on and stand on only one pedal, the wheels are loaded axially. In fact, every time the frame flexes, the wheels are loaded axially, and that's every pedal stroke.

The axial loads would be hugely different on a trike and although I have zero experience with trike wheels, I would dare to guess that trike bearings wouldn't last as long as their two-wheeled counterparts, if they are standard deep-groove cartridge bearings.

Companies like SKF do make ACB (Angular Contact Bearing) equivalents for some of the 6000-series bearings used in bicycle hubs, but those bearings require a preload mechanism like that on the Mavic and on the suspension wishbone of a Santa Cruz Tallboy or Blur frame. I once had to convert a set of Zipp front wheels for use on a racing wheelchair where the wheels have to be angled. The bearings didn't last at all, so I substituted them with a pair of 7000-series SKFs and concocted a very time-consuming preload mechanism on my lathe. As far as I know, they are still in service, two years on.

Racing wheelchairs must not topple when leaning, hence the angled wheels.

It is therefore not as if the bicycle industry doesn't have a choice because cartridge bearings are only available in deep groove, it is then the added cost of a screw-on "cone" with locking mechanism. And, don't forget, people who buy Zipp wheels weight things on little electronic cocaine scales. They'll pick up on the preload mechanism's grams and vote with their dollars.