Every so often, a topic gets going in an email exchange and has enough "meat" in it to share as interesting with others. Greg Priddy, cyclist, soundman, and the other voice on The Outspoken Cyclist had a question about the "wear" on his Shimano chain and started a "conversation" with Brian. I give you, in its entirety, that discussion - enjoy!
The chain was elongated by 1/2 link over the entire length, as compared to a new chain. That equates to 0.75% on my chain gauge, which is an earlier version of this: http://www.parktool.com/product/chain-wear-indicator-CC-3-2. Mine measures 0.75 and 1.0. The newest model measures 0.5 and 0.75. Maybe I need one of these, which gives a true measurement, rather than just a g0/no-go: http://www.parktool.com/product/chain-checker-cc-2.
I will bring the old chain to Diane on Thursday.
I grossly underestimated my miles on that bike for this year. It was actually 2,392, plus the roller miles from the winter.
The old chain was also an Ultegra, but it was a CN-6600, vs. the new one which is a CN-6701, hence the difference in appearance.
I examined the cable-end in the right shifter, and it looks fine, but I'll keep that in mind as something to check from time to time.
Thanks for the sage advice.
(From Brian to Greg)
A full half-a-link over the chain's length is roughly 0.45%, correct?
I'm estimating the number of links in your chain to be 112, and to
have the lengths differ by one-half link (change) over 112 links
(total)... or 0.25" (change) over 56" (total)...
That works out to be around 0.021" over 12", or a bit less than 1/32"
over a foot.
According to the old rule(s) about when to replace a chain - '1/16"
(0.063") over a foot' seems familiar - you still have some life left.
Relative to my experience with more modern drive-trains, this wear
seems reasonable for the distance you say you've ridden. Although I
would have expected/hoped it to go further before having chatter
issues, I'm now far less surprised about the lifespan it seems to have
had. Given that the old chain is apparently a 6600, it now seems
possible that it might have have been the original chain as well, and
possibly already had a few hundred miles as part of the original
demo-group. This too would help to explain the wear and need for
As far as tools go, as I tried to say earlier, almost any
length-change checking tool is okay - once you become familiar with
its idiosyncrasies and calibration (or lack of it). Like any
measuring tool, inaccuracies can be used to your advantage once you
are aware of them. These chain tools are notoriously inaccurate, but
most still do their jobs perfectly as long as the customer isn't
looking for numbers to trust blindly.
A common argument between old-timer frame-builders and the newbies is
about surface-plates. Everyone wants perfection, and the newbies seem
to think that a perfect surface plate will help them build a frame
perfectly straight. The first step to getting there (there being
APPROACHING perfection) of course is to understand that no surface is
perfectly flat, and the next step is to understand that the more
familiar you are with a surface's imperfections, the better you
account for them. Indeed a skilled builder can use a badly warped
surface (or string for that matter) by simply understanding how,
where, and how much the surface is warped. As usual, I digress.
My point is: ignore the numbers on your chain-measuring tools as
meaning something. Start by putting a tool you like on a new chain.
Feel it, see it, try it over and over again, until you trust how a new
chain should fit in the tool. Now you have a datum to compare to
later on when the chain wears and begins acting poorly. Also try some
worn chains, some still performing properly and some not, and get an
idea about the point at which a Shimano 10-speed chain will measure in
your preferred tool just when it might begin under-performing. Over
time (and chains) you'll know just when to expect it's time to change
them, based on performance, distance, conditions, and of course the
tool's reading (regardless of accuracy in the actual number).
Oddly, I have never used a Park chain-wear tool, although the ones I
remember from years past were most similar to the CC-2. The CC-3.2 is
superior-but-similar-in-principle to the ProLink tool, which is
TERRIBLE because its calibration is so easy to knock out (think
Schrodinger's Cat paradox) and so difficult to correct. I doubt the
CC-3.2 has this problem, but it does appear that it requires a certain
"feel", and probably is a bit too much of a go-no-go gauge. Although
I like the old gauges I have, I still trust my steel rule. That all
said... if the chain has significant mileage on it, and doesn't seem
to be performing properly... it gets replaced, regardless of what any
gauge or rule says.
(From Greg to Brian)
It is quite a topic.
My chain is only 105 links (I recounted the old one just to be sure), which makes the elongation 0.47%, but still not enough that I would expect engagement issues.
I understand what Brian is saying about knowing the imperfections of a tool or a fixture. However, it seems like there should be a more accurate way to measure chains than the options we have now. Maybe it's because the purpose-built tools are taking a sample over an insufficient length of chain? I wonder if a 12" digital caliper would do any better?
Of course, even if we can measure accurately, we still have to correlate that measurement to a real world decision on whether or no the chain needs replacing.
I do know that the first ride on a fresh chain is a beautiful thing. I could actually hear all the debris I was riding over on the bike lane because the drivetrain was so quiet.
Here's an interesting article (which is probably old hat to Brian, or at least the concepts are).
The thing I found interesting is that most of the tools available include both roller-bushing wear and roller-pin wear, while only roller-pin wear is said to be significant.
That would mean that these tools are routinely overstating wear, which would explain why my Park CC-3 claimed 0.75%. Perhaps the conventional wisdom has been adjusted over time to recommend replacement at 0.75% as measured by commonly used tools, which equates to roughly 0.5% actual wear.
The Shimano tool seems to be one of the only ones that measures from the "same" side of the rollers, thus not adding roller-bushing wear.
After seeing a picture of it, I can definitely see how the Prolink tool with its big arc shape could easily be bent out of calibration.