Simply put:
Spalling is due to particles breaking free from one surface and being attached to onto another. It usually occurs when a softer material is in contact with a harder one under high pressure. The softer material breaks down.
Brinelling is due to impact of a harder part into a softer one such as a hard ball bearing into a softer race. This creates indentations in the softer part.
Fretting can be caused by a repetitious rocking motion usually between hardened parts and attacks both surfaces causing localized damage in the area of contact.
Cheap case hardened loosely adjusted headsets usually fail due to brinelling! The depth of the 60+Rc case hardened surface can be as little as 0.005" (0.13mm) and the steel below it very soft. The same thing can happen to any insufficiently hardened bearing surface. I'm going to dig out some damaged Campy and other make headset parts and have the hardness tested.
Fretting is a very feasible failure mode, that's why I mentioned that I was intrigued by Jobst's hypothesis. Parts can fail due to both brinelling AND fretting. The indentations caused initially by brinelling can force the ball bearings to remain in the same position and then they start the process of fretting (remember the elephant).
Look at the indentations in a headset race under magnification. Most of the time you'll see that the bottom of the "craters" in the races are usually shiny indicating that the ball bearings are still hard enough and smooth enough to create a burnished surface, ergo brinelling.
In a fretted condition both the ball bearings and the bearing races will tend to have a rough surface finish because fretting attacks both surfaces.
The point about less Campy headset failures on large frames is interesting. I don't recall seeing many damaged headsets on larger frames, same thing with the smallest sizes (maybe due to rider weight).
Something to consider, larger size frames tend to have steeper head tube angles. This would tend to lessen radial loading on the headset bearings. Also, long top tubes would tend to put less weight on the headset.
Axial loading is parallel to the plane of the steering tube: think of an axle. The forces are going up into the lower headset race.
Radial loading is perpendicular to the plane of the steering tube: think rays of the sun or spokes of a wheel. The forces are going to the outside of the bearing races (in this case to the front and rear of the bearing races).
BTW, Jobst Brandt rides huge frames!
http://www.sheldonbrown.com/
Chas. Colerich Oakland, CA USA
Tom Sanders wrote:
>
>
> We used to have some trouble with Deisel's cam shifts years ago in the auto
> industry. We were losing bits of metal until failure due to a less than
> ideal lubricant. This was termed "Spalliing" , if I remember correctly. Is
> this the same as Fretting, or have we yet another condition here?
>
> Tom Sanders
>
> Lansing, MI USA