Dan wrote:Niggle, you are (nearly) correct, the Rohloff Caliber 2 measures stretch at 0.075 and 0.1 mm PER LINK (according to the instructions), not 0.075% and 0.1% as I mistakenly thought. So I was out by a factor of 1/25.4=0.039
Taking 1 link to be 1 inch (25.4mm) of chain the Rohloff chain checker calibrations of 0.075mm and 0.1mm per link translate to 0.075/25.4x100=0.30% and 0.1/25.4x100=0.39%.
Let's get this straight once and for all (some hope
): We don't describe an inner link or an outer link as an inner or outer half-link, and the label on a new chain speaks of it having a hundred and something links, so I think it's safe to assume that the cycle industry universally regards one link as being that fraction of a chain measuring half an inch i.e. 12.7mm in length. So the Rohloff tool callibrations of 0.075mm and 0.1mm correspond with wear percentages of 0.6% and 0.8% respectively.
Direct measurement of my old centaur chain against the new KMC showed it had stretched by 5mm over the length of the whole chain. This gives 5/1370mm = 0.36%.
Interesting that this is less than the 0.39 according to the chain checker. Presumably 0.36% is the true figure (unless older chains are more elastic and have to be measured under tension) and the chain checker produces a reading of 0.39% because increased play in the rollers exaggerates the stretch, as Niggle pointed out in an earlier post.
Quite right and then some, with bells on, since the Rohloff chain checker actually indicated 0.8%, a massive
over-reading of the wear in this chain, due to the way this tool measures two doses of roller play/wear, in addition to the few doses of pin wear that are spanned in its very short length.
I once measured a brand new but very cheap and nasty chain with my Rohloff checker, according to which it was completely worn out already at 0.8% elongation, though when pulled out against a steel tape its 118 links measured precisely 59 inches. But this chain had very sloppy rollers. Q.E.D. Niggle
Before I lost it, I improved my Rohloff chain checker by tethering a wedge of metal to it, that I would insert into the link before the section of chain I was measuring. This wedge shoved the "zero" roller forward, in the same direction as the roller at the other end of the measuring section would be shoved by the checker. This took the spurious factor of roller play/wear completely out of the equation, so the tool now measured purely pin wear, like it was supposed to. After that its indications of wear became much more consistent with measurements of whole chains after removal.
In order to be sure that a new chain will run on the old cassette, I've found it's necessary to scrap the old one at about 0.5% actual wear. By 1% it's too late, some sprockets will jump. But if you're going by an un-improved chain checker that adds in two doses of roller play, it's possible that you have only 0.5% actual wear when the checker says 1%.
Anyway the chain has stretched enough to have worn my 11 tooth sprocket badly enough that a new chain jumps on it.
That's bad luck, given that you had less than 0.5% actual wear and a good illustration of the truism that YMMV (your mileage may vary)! Smaller sprockets will be more prone to wear and jumping from the same amount of use with a worn chain. But touring cyclists like me have very little use for the 11T top (or even the 13T next to it) that comes with the 34T we really want at the other end! So they're usually still in excellent nick regardless. When I leave a chain on too long, it's the middle-to-big ones that jump. I guess I should keep the pristine tiddlers from those cassettes and trade them with top gear mashers like you!