Index Home About Blog
From: (Jobst Brandt)
Subject: Re: Magnets and hub bearings
Date: 20 Sep 1999 15:56:45 GMT

Mike Lackey writes:

> I remember reading long ago that one should not use a magnet to
> remove ball bearings from a front/rear hub "lest ye be sorry".
> That's about all the explanation the author gave, and I've wondered
> ever since what.

The balls become magnetized and will forever after collect metal
debris that can be found in bearings, debris that is better left to
settle in quiet corners of the bearing.  The balls become garbage
collectors and self destruct.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: Magnets and hub bearings
Date: 21 Sep 1999 17:25:35 GMT

Sheldon Brown writes:

> Theoretically There's A Chance Of Magnetizing The Balls, Which Could
> Make Them Attract Ferromagnetic Crud, But Does Anybody Really Still
> Try To Reuse Old Bearing Balls When Repacking Hubs?  This Seems Like
> A Bad Idea To Me, Partly 'Cause They're Pretty Cheap, And Partly
> 'Cause, How Can You Ever Know That You've Got Them All Clean?

That perception should be taken with a bag of salt, considering that
Sheldon sells ball bearings.  If the bearing balls are shiny, then
they are as good as new, the shine being the first thing that is gone
if the balls are dimensionally incorrect.  Another consideration is
that new balls are not always on hand when a bearing needs attention.
Unless the hardened crust of the ball is chipping (spalling), it is
probably good for another tour of duty.

> There May Be A Big Lougie Stuck To One Of Them That You Can't See
> Because You Never Actually Saw That Side Of The Ball.

Oh oh!  Watch out for the bogeyman lurking behind what you don't know.
In fact, if you magnetize them, there will be something stuck to them
but not necessarily when you assemble the bearing.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: Bearing adjustment tolerances
Date: 16 Nov 1999 23:02:03 GMT

Kirby Krieger writes:

> Back before the Internet, I and my biking buddies would spend no few
> moments perfecting the adjustment of the bearing surfaces on our
> bikes.  We would make sure that wheels reversed direction four times
> before settling with the heaviest point at the bottom.  We would
> pack our pedals carefully and be sure they had an effortless spin.
> My question is: was this silly?

Yes it was silly because bearings don't carry loads well without
preload.  In the unloaded condition that you were adjusting, the fully
free ball bearing just barely makes contact.  Under load the balls
under the axle sink into the race elastically, one at a time as they
pass.  This is why you'll notice auto mechanics always adjust front
wheel bearings with preload.  This must be done on differential gear
bearings.  The most important place on a bicycle for preload is the BB
spindle of the conventional cup and cone type.  The spindle should
turn with perceptible drag when rotated by thumb and forefinger...
about 4-5 oz-in torque.

> Today's expensive wheels seem as often as not to have "sealed" and
> un-adjustable bearings -- and don't seem to pass the four reversals
> test.  My Speedplay X-2 pedals are sealed with O-rings which
> prohibit freely spinning the pedal.  What's up here?

They require less assembly time in production.  Other than that they
are a disservice to the repair shop and customer.  They don't last as
long but then most bearings don't fail from load but rather by water
intrusion.  So called sealed bearings do that as well or better than
cup and cone bearings.  O-ring seals are the poorest form of contact
seals for shafts.  O-rings are best used for static seals.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: How Bearings are Made
Date: 22 Nov 1999 23:05:20 GMT

Roger (who?) writes:

> Did you know you should only ever use ball bearings from the same
> batch in one side of a race?  They're not exactly the same size
> between batches.  Never simply replace that naughty one that bounced
> into the corner of the garage - replace the other 10 (or whatever)
> too!

You are making this up.  The tolerance between bearing balls is so
small as to be below a small fraction of the elastic compliance of the
steel bearing.  Besides, the races of bicycle bearings are so rough
that a tight bearing feels lumpy.  In high precision bearings used on
computer disk storage devices, preload causes a smooth viscous drag.
Even for these bearings the balls are not identical but are made to a
prescribed tolerance.  I don't believe I understand what you mean by
the same batch.  Each bearing is not made in the same finishing
process as the others in a shipment of balls.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: How Bearings are Made
Date: 2 Dec 1999 17:04:23 GMT

David L. Johnson writes:

>> Suppose you had a conventional deep groove ball bearing with 10
>> balls.  You might be able to replace the cage with 4 additional
>> balls to make a 14 ball full complement bearing.

> Well, my experience with most components is that, once you chuck the
> cage you can only fit in one extra ball.  Maybe two in a headset.
> But, no matter.

This discussion isn't getting anywhere because these are not the same
type of bearing.  Cartridge (radial, deep groove) bearings are
eccentrically loaded, using the largest balls practical, hence 7 or 9.
These are inserted between inner and outer race with the inner race
off center, filling about 180 degrees of the outer race, after which
the inner race is centered and the balls spread uniformly by means of
a cage.  The cage is essential to the bearing assembly as is an odd
number of balls, because with eccentric loading, the last ball must be
inserted in the center of the complement.  By this method the inner
race rests against one side of the outer race, leaving just enough
space to insert a ball on the opposite side.  Balls can be inserted
only until the outer race is half full (ball centers on the diameter)
and still allow the inner race to move to center.

Exceptions are slot loaded bearings and split outer race bearings.
Both of these use full complement (ball to ball contact) balls for
special purposes where the disadvantage of the slot or split race are

The bearings of which we speak in bicycles are primarily cup and cone
bearings in which the cage serves merely as a convenience for
assembly.  It has no function other than that.  Cheap bearings, that
we needn't consider, often use cages to reduce the number of balls,
cages being cheaper than balls.

The contact of one ball on another is not the reason for cages
although this is a problem with needle or roller bearings that have a
long contact line.  They also skew if run without a cage.  Connecting
rod roller bearings present a special case where inertial forces are a
main concern with roller to roller forces becoming substantial at high
speed without a cage.  None of this applies to bicycle bearings.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: How Bearings are Made
Date: 29 Nov 1999 22:25:04 GMT

Jim Adney writes:

>> Won't argue with most of this, but can you provide more information
>> on: Conventional wisdom suggests an angular contact ball bearing
>> (which a cup and cone bearing is a form) can be operated preloaded
>> under normal operating temperature. The easiest way to ruin a
>> precision angular contact bearing is to run it unloaded or even
>> lightly loaded. The problem is magnified in cageless or full
>> complement bearings.

> I think the point is that just because there is no preload, this does
> not mean that there is no load.

>> The ROT on preload is to use the maximum amout of preload that does
>> not add an intolerable amount of additional drag. Certainly you can
>> have too much.  Bicycle bearings (even Campagnolo, I would suspect)
>> are not precision enough, otherwise they would benefit from
>> preload.

> Putting preload into a bearing forces the races and balls to undergo
> greater continuous elastic deformation.  The more preload, the more
> deformation.  But greater elastic deformation causes fatigue failure
> sooner.  There is also some increased drag, but this can be small in
> a quality bearing (better than usually found in bicycles.)  Preload
> is also used to stiffen up assemblies where precise positioning is
> important as in lathe headstocks and grinding machine spindles.

BB bearings require preload to survive reasonable service.  Pitted
spindles of the standard old angular contact type experience severe
"ball drop" if used without preload.  This is the elastic drop of the
spindle between balls because there is no contact area when such a
bearing is adjusted to have no preload but also no clearance.  The
Spindle makes point contact at this juncture when not loaded but sinks
into the balls, or more accurately the one ball, beneath the spindle
elastically, leaving the upper balls with clearance.  Ball drop is the
wavy contour a spindle makes when rolling over successive balls
without preload.  It causes higher stress than suitable preload.

GM cars used angular contact ball bearings in place of tapered rollers
for a short while, years ago, with many failures because mechanics did
not follow the torque requirement for preload and assumed that tapered
roller preload (light) was sufficient.  It is not and similarly on
bicycles, pitted BB spindles are largely a result of no preload.

>> The vast majority of automobiles in the world are now made with
>> angular contact ball bearing based hubs. The only exception is in
>> North America.  These hubs operate with preload, and quite often
>> have an maintenance interval greater than the life of the car.

> The cars I'm familiar with use either angular contact tapered
> bearings with clearance, or radial contact bearing assemblies which
> are not adjustable and I assume to be assembled by the bearing
> manufacturer with some minimal clearance. I am not well versed in
> modern cars, however.

Tapered rollers do not run on clearance but rather with light preload
because they are so much more stiff in relation to balls.  The bearing
is first tightened and the nut backed off until the washer can be
moved stiffly with a screw driver.  This is not without preload.  The
reason angular contact balls are most often not adjustable is for the
same reason that we are talking about it here.  The users cannot grasp
the importance of preload and mis-adjust the bearing for early failure
because they hope to reduce friction.

>> The standards for balls are set by AFBMA not ABEC, and a grade
>> represents the deviation in diameter and spherical form in
>> microinches as an earlier post mentioned.  And you can certainly
>> get a grade 3 or grade 5 steel ball.  Let me know if you need the
>> name of a supplier.  A lot (or batch) that is grade '3' for example
>> will have the diameter or each ball in the lot within 6 microinches
>> of every other one in the lot.

> I would be interested in learning where I could buy grades better than
> 25.  Do you know what quantities are required?

If you think you can gain anything by more precise bearing balls, you
are in the realm of, not trifles, but trifling trifles.  Consider ball
drop first.  It is not trivial.  In reality most wheels run with
preload because they are QR wheels, in which tightening the skewer
preloads the bearing.  This is likely because nearly every well
maintained wheel I have had in hand had no perceptible bearing
clearance, an amount that the skewer shortens the axle when tightened.
This can usually be seen by letting a front wheel coast to a stop.  It
homes between ball irregularities instead of swinging like a pendulum.
With the chain off, rear wheels demonstrate the same effect.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: Loose balls vs. retainer?
Date: Mon, 22 May 1995 21:13:46 GMT

Gary Helfrich writes:

> I agree the the 11 ball cages do not separate the rolling elements.  Is 
> there any proof that the 11 ball design has a superior lifespan?  This 
> would not be the first time that a mechanically inferior design was used 
> in the bike industry. 

Ball bearing failure occurs primarily through subsurface fatigue that
results in spalling, the flaking of the ball surface.  Especially at
the speeds of bicycle use, surface speeds and frictional wear are
insignificant.  The two modes of failure in bicycles are spalling and
contamination with dirt and water.  Neither of these are affected by
balls making contact with one another.

> Please post the bearing number and manufacturer who produces an 
> automotive bearing where the tapered rollers are in contact with each 
> other.  I think that you plucked this one from air. 

Among the machines on which I have worked, several had steel cages
that did not separate the rollers of a tapered roller bearing.  I
neither have access to these bearings now nor am I not going to run
down to the auto repair shop in order to assist you in proving that
your claim of the importance of bearing cages was valid.

> If the simple act of adding balls to a bearing would increase the 
> capacity with no other ill effects, this practice would be commonplace.  

Balls in cartridge bearings are eccentrically loaded by moving the
inner race off center so the balls can be introduced.  This means that
about one more ball than can fit in a half circumference can be
inserted.  For this reason, most such bearings have little more than a
half complement.  

Full complement bearings are either slot loaded or have the outer race
fractured on the diameter and are recombined with a press fit ring.
The fractured ring method is the only one that works for high speed it
diminishes fatigue life due to the structural discontinuity.  With the
low ball density of eccentric loading, cages are necessary to keep the
balls spaced.  Only in full complement bearings does oil film spacing
keep the balls separated if the bearing.

Only angular contact bearings (cup and cone) lend themselves to a full
complement of bearing balls to gain full benefit of additional

> The bearings in my milling machine are quite massive compared to those 
> found in small two stroke engines (110 mm bore) and have a very heavy 
> preload.  When rebuilding the spindle, Pope, ( the rebuilding company) 
> was concerned that the correct retainer material was used in the rebuilt 
> spindle. These are very expensive separable angular contact bearings.  If 
> eliminating the retainer would increase performance, I suspect this 
> option would be offered,  (It is not)

With surface speeds encountered in such bearings, ball to ball contact
causes hot spots and oscillations can occur.  In this case, precision
overrules other considerations.  Headstock bearings get hot just
running under no load and have little in common with bicycle bearings
that run at about 1/100 the speed and precision.

> Even non Conrad type bearing construction uses retainers, though 
> this style uses a filling groove for the maximum number of balls.  
> In this case the retainer is an extra step in the construction of the 
> bearing which adds to the expense.

So?  It is evident that a badminton racket is not a tennis racket and
that different demands require different solutions.  I think we are
pretty far afield of bicycle crank spindle bearings with 9 or 11 balls.

> What is the reference for the statement that the rolling elements in 
> preloaded bearings will not contact in the absence of a cage? 

"Rolling Bearing Analysis" Tedric Harris, Wylie Interscience.

Jobst Brandt      <> 

From: (Jobst Brandt)
Subject: Re: FAQ Question: Jobst on "Sealed" Bearings
Date: 19 Nov 1998 17:05:08 GMT

Sheldon Brown writes:

>>> Of course all this is my theory since I have never seen anything
>>> on bearing testing or theory.

>> I think you should read "Rolling Bearing Analysis" by Tedric Harris.

> Gee, I dunno... based on your previous recommendation of this book I
> checked it out.  Although Newton has an unusually large and fine
> library, they didn't have this, nor did any of the other public
> libraries in the Eastern Massachusetts system.  My library
> eventually got it in on interlibrary loan from Hartford.

This is the work recognized by tribologists as the source for
lubrication and stress analysis for all types of ball and roller
bearings.  I find it useful for understanding the problems of rolling
element bearings.

> I found very little in this book that was of any use in bicycle
> maintenance applications.  It had nothing whatsoever on cup-and-cone
> bearings.  It seemed to be mainly aimed at engineers designing new
> equipment, with information about types of bearings needed to handle
> various types of loads, very heavy on formulae and graphs which have
> no bicycle application.

Who said anything about bicycles.  The subject was bearings and their
loads.  The title should also give you an idea that it is about
bearings, not bicycles.  It has a slew of subjects on angular contact
ball (aka cup and cone) bearings that show how this affects spin of
the balls and how they scuff.  You appear to have given the book a
biased review.

> There was also information that would be invaluable to someone
> trying to design a new bearing from scratch...not something that
> Nick nor I am likely to do in the near future.

Hold it!  The reference was for people who claim to or want to know
about the workings of rolling element bearings, not how to maintain
a bicycle.

> There were a very few nuggets scattered here and there explaining
> wear mechanisms, from which the chief lesson is to keep the bearing
> clean and well lubricated, and not to exceed its design
> load/speed/temperature.

Well?  Isn't that what the original subject was?

> All in all, I'd have to say that, from a bicycle mechanic's point of
> view, this very intimidating book is well over 99% noise.

Mechanical engineering is intimidating to many people.  For some it
cannot be reduced to practical terms and will always remain obscure.
Harris' book is a classic in the industry and is a cornerstone for
those working with bearings.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: spoke-spoke notching
Date: 4 May 2000 22:47:30 GMT

Terry Morse writes:

> Now how do we apply it to hub adjustment? I currently adjust the
> cones until there is barely any perceptible play, figuring that the
> QR will take up the slack. I follow this up with a slow-spin test on
> the bike to make sure the bearings are not binding.

I choose to adjust just loose enough to be felt as clearnace and small
enough that it goes away when the QR is closed.  To hell with
perfection, they last long enough for me.  On the other hand, BB
spindles are easily adjusted with preload because that can be felt
with fingertip rotation of the spindle.  Besides, the BB bearign is
heavily loaded and needs all the help it can get.  It should turn so
that using the tips of thumb and forefinger, the spindle can be turned
as if it were lubricated with stiff (lumpy) molasses.  Really good
bearings would turn as with smooth molasses.

> Now that I've read your explanation of preloading, I'm afraid my
> method may not produce any preload at all. Would it be better to
> tighten until there is noticeable binding (when the QR is
> tightened), then back off slightly?

I think this can be overdone, but it's good to understand what is
going on.  I can see it now, wheel torque micro merasurements and
special cleaners and oils... or even wax.  All vegetable water based
elixers are next.  Keep it simple.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: Rolf Vector Comp Front Wheel Problem
Date: 9 Jul 2000 02:55:12 GMT

anonymous writes:

> I purchased a set of Rolf Vector Comp wheels in February of this
> year.  A few weeks after I began riding on them, the front hub began
> making a terrible crackling noise when I was out of the saddle.  The
> bearings were fine, but I suspect that something between the fork
> drop outs and hub spindle was fretting.  I also have a set of Vector
> Pros that I have had for a couple of years now. Those wheels do not
> have this problem.

The noise as you describe is typical for water intrusion into ball
bearings.  The only thing that helps is to clean the bearing balls and
races as soon as you can and add 30W oil.  I don't know what sort of
bearings these wheels use but if they are replaceable cartridges, that
is what you need to do.  Sealed bearings are not "sealed" against
water intrusion but are merely sealed to prevent airflow in electric
motors where rotation acts as a vacuum cleaner and fills bearings with
dust.  Phil Wood did not understand this when he started the "sealed"
bearing fad.  Only labyrinth seals will protect a wheel bearing over
the long run.  They also have no drag.

Jobst Brandt      <>

From: (Jobst Brandt)
Subject: Re: Scratches on Outer Hub Races
Date: 13 Sep 2000 21:35:25 GMT

Alex Thomson writes:

> After getting a flat which ended up needing a new tire, I noticed
> that my front hub made twanging and scratching noises when spun. If
> you really listened it felt like there was some sand inside the hub.
> I took it apart and virtually no sand was visible. What I did notice
> was that half of the inside surface of both of the outer races was
> severely scratched. The inside races were fine, as was most of the
> track on the outer races. It's such a weird occurrence that I'm
> wondering whether it was a defect in manufacturing.  Has anyone
> heard of this happening before? Also, the hub was a really cheap one
> (AceraX) and it only had 3000km on it so nothing really should be
> going wrong yet.

Cracking clicking noises from ball bearing hubs come mainly from water
intrusion that cases rust.  Even a small amount of rust that barely
colors the oil/grease brown can cause great snapping noises.  Check it
out.  Scratches is an odd term for ball bearing races.  Clean them out
and put new balls and chassis grease in there.  The bicycle shop has
bearings in bulk and grease is best found in auto parts stores.

Jobst Brandt      <>

Index Home About Blog