Right way to tighten hub bearings
 

What I've been doing is tightening by finger until just at the point where I can't detect any play no matter what direction I yank or push, then hold in place with a cone wrench and tighten and reconfirm no play.

Is there a better way to do it?

If you have a QR axle, that may be too tight.
The axle compresses slightly when you tighten the QR. You have to allow for it.

The way I describe my bearing tightness is=
You know there's play there, but you can't feel it.

The only way to confirm the best bearing adjustment is after tightening lock nuts/rings and having the system in place. So with a hub that's with both sides of the axle's cones/locknuts tightened against each other and the wheel tightened in the frame. Then wiggle the rim/tire side to side to check for bearing play. If no play is felt you still don't know for sure if too much preload is present unless you've done a few adjustments from loose to just barely snug.


If this seems to be too much work then get someone else with better motivation to work on your bike. Andy

I try to get an adjustment such that with the wheel tightened in place I can't feel any play at the rim, yet when it's rotated it always stops at the same place.

One way we use to gage bearing adjustment was to watch the part's swing back and forth as it settles down from a spin. How far the wheel went back up the other side of the swing the back up the first side would suggest the bearing drag/preload. These days with contact seals and such this is less the case. Also the industry has moved to a mind set of disposable rolling elements in general.


Like other fine tuning aspects the techniques to do so are being lost. Andy

It's possible to adjust the cup-and-cone bearings on a quick-release hub with the axle under compression. This can be done using the frame with the wheel clamped on the *outside* of the dropout using the quick-release. See this section: Hub Overhaul and Adjustment: Cup and Cone Style | Park Tool
There is a similar approach on Sheldon Brown's site: http://sheldonbrown.com/cone-adjustment.html (scroll down to "Special tool for rapid cone adjustment under load").
This allows the bearings to be adjusted for zero play and zero pre-load, which is appropriate for ball bearings. When the quick-release is opened, the bearings will seem slightly loose because the axle expands to its original uncompressed length.
Solid-axle wheel bearings do not need to be adjusted under load since the axles are not compressed when the axle nuts are tightened.
Steve

Schwinn suggests a hint of end play when the cones and locks are tightened.

When the QR is open there should be a small amount in the bearings. There should be a small amount of preload after the QR is closed.

Bumping this
I do similar for the QR preload difference
however instead of clamping it to the dropout, i just keep a pair of old axle spacers around
one on each side of the axle so the QR can clamp down on them

That would mean it compresses stronger than the threads of the nuts, cones and axle resist...?

No.
It simply means the axle compresses or shortens.
It's just a hollow tube. The skewer is "pulling" the axle ends closer to each other via the frame & lock nuts. Like a vise.
When it compresses, the cones move closer together.
The bearing races in the hub don't.

My impression, is that the main 'compression' is in the thread interfaces of the cones and locknuts to the axle, not so much as deformation of the parts themselves.
the threads tolerances have quite a bit of play to them, you can feel it is there by hand before the locknuts vs cones interference takes up the slack.
clamping the QR biases that interference inwards; a tiny amount but enough that it's worth taking into account for bearing adjustment.

If you look at the Park site, the issue has nothing to do with the threads, but with a bowing, bending or curving of the axle as the quick release skewer is tightened. This has the effect of bringing the two cones very slightly together, thus forcing the balls more tightly into the hub races -- the hub, of course, being independent of the axle.

The role of the threads is in locking the cone and locknut together; in tightening the two together, the threads of the locknut "ride up" on the axle threads, thus creating an interference locking mechanism. It is the reason why the cone has to be held in one position while the locknut is tightening. If the cone is unchecked, the locknut will just keep turning on the threads along with the cone itself. At least, that is the way it has been explained to me.

As a side note, prior to my very first tour, as a very inexperienced bicycle mechanic, I fiddled with the cones on the rear wheel of my bike. Some 2,000km later, the cone and locknut wound together into the race of the hub and broke it neatly in half. I learned a lot from that experience about checking and double-checking pe-load and locknut tightness.

Each cone-locknut pair is fixed on the threaded axle; there is no play if the parts are tightened against each other. Actually, the tiny bit of axle between the cone and lock-nut is under tension and stretched a tiny bit... it is this stretch that keeps them locked together.
The axle doesn't (shouldn't!) bend. Under compressive loading, it actually becomes slightly shorter, so the cones become closer to each other. Bending ("bowing") is undesirable in a compressed elastic material, because that can lead to failure of the part.
Steve

Putting aside the quick release compression issue and back to how to get the basic adjustment right consistently: I used to do it by repeated trial and error. I even had to get someone to lend a third hand to keep a wrench on the other side of the hub so I wasn't turning the axle when I was dealing with the cone and locknut I had adjusted. I don't any more. Now, I simply finger tighten the cone then I finger tighten the locknut onto the cone and then put my wrenches on them. Here's the trick... instead of trying to hold the cone while torquing the locknut down on the cone, I hold the locknut while torquing the cone up into locknut. That almost always gives me the correct result without the trial and error "Kentucky windage" problem of how much to back off the cone before coming down on it with the locknut.

In summary, finger tighten (only up to first resistance) both cone and the locknut, then with wrenches on each, back the cone into the locknut to lock the adjustment. Very precise and consistent results.

The way just about every pro wrench I have worked with does hub bearing adjustments is to tighten each side's cone and lock nut then "counter" tighten or loosen as needed with a fit into the frame/fork with QR clamping to check. Repeat till right. After you've done a few hundred it doesn't take too much time or repeats to get it right.


Counter tightening/loosening is when you take two cone wrenches and placing them on the cones loosen one cone WRt the other (if the initial adjustment is tight) or two lock nut wrenches on them and tighten WRT each other if a sloppy adjustment.


http://www.jastein.com/PDF/Hub%20Vise%20Web.pdf


Stein makes a tool which allows the axle to be clamped against it by the QR, the tool (a plate of steel) held in a vice and the free end of the axle can be accessed with a cone nd lock nut set of wrenches. The ae is under compression/bowing as the free end has the QR bearing down on the tip of the axle (not the lock nut). While this works well it can take longer then counter tightening/loosening. Especially iv your bench vice is in the back room or at the other end of the shop. Andy

I would be more interested in what Campagnolo and Shimano suggest. As they actually made Hubs.

I would agree with what @Andrew A Stewart said.

I had one of those Stein tools, but tossed it in favor of the Hozan axle vise. Procedure is the same, but the vise holds the axle securely while you go through the exercise.

No problem, thanks for the reply

While the adjusting process and goal might be the same for an axle vice (campy and Park also make them) The Stein also loads the axle with the compressive/bowing force that the QR gives. An axle vice doesn't. This is a major difference in actual method of adjustment. Andy

I understand that. I use the QR skewer through the axle and the axle vise in the same way it's done with the Stein tool. The Hozan vise is a unique design as axle vises go, and makes this possible.
http://i66.tinypic.com/2akgx35.jpg

With the Stein tool I found it difficult to differentiate between bearing knock and the axle moving in the hole in the plate with the QR open. With the axle firmly mounted in the axle vise, this becomes a non issue.
not a knock against Jim Stein. I have a number of his tools and like them a lot.

Dan- Yes I have had a Hozan axle vice (and I know it's thread pitch sensitive to a degree. I like that the two halves have a number on then to index their fit). It's a great device to hold the axle from turning while working the cone or locknut. I admit that I never also loaded a QR skewer with the Hozan vice. Good point.


I will say that after my comments on the Stein tool I find that the fine tuning of a hub bearing adjustment is not a black and white thing. The QR securing force is a variable that can change from one setting (as in a tool) and another setting (as in the actual frame). Additionally I believe that sometimes the drop out material, alignment can effect the clamping pressures and axle compression/bowing.


I just find it easiest to confirm my axle bearing adjustments with the same QR, frame/fork securing as will actually be used in riding. Anything else is a step away from the real situation.


BTW, one use of the Hozan vice that I use to do a lot was tightening the fixed cone on English (Raleigh usually) front axles where the fixed cone was run up against a ridge on the axle and the other cone was held by the outer axle nut. Andy (who's walking down memory lane)

Yes. Holding one end of the axle in a vice does makes it much easier. Imo a tiny bit of play at the rim, once mounted in the frame, is fine and prevents over tightening that is otherwise hard to gauge because of the large wheel diameter.

The only problem is that a little play at the rim means the bearings have no preload and wear quicker.


Wheel Bearing adjustment by Jobst Brandt

So called "pre load" means the bearings or races are deflecting. This will ruin bearing in the long run. Take your time and with the wheel mounted and the quick release tight, you should be able to feel and ever so slight movement of the wheel at that rim several places around the rim.

I have always wondered where this nonsense of preload ever come from.