Road Bike, Cycling Forums banner
1 - 20 of 30 Posts

·
Registered
Joined
·
5,524 Posts
Discussion Starter · #1 ·
It's well-documented here and in maintenance books that with chain "stretch" of 1% (1/8" over 12" span) it's time to replace a chain. The trick is the measurement.

The easy measurement is just a ruler alongside the chain. There are a couple tools, by Park or Rohloff, but I think it was Spoke Wrench who once said he calls that Park chain measurement tool the "Park chain selling tool." The reason is that the tool exaggerates the measured "stretch." This made an impression on me since SW has forgotten more about bike maintenance than I'll ever know. (Apologies to SW if I've mis-attributed the qoute.)

But I've been thinking...

The chain measurement tools tend to make their measurements with prongs that fit in between the chain links. The effect is that the measurement is made with the chain under tension. With chain "stretch" really being wear between pins and bushings, it seems that measuring with the chain under tension would force the loose pin-bushing couplings to their outside range of motion, thereby maximizing the stretch measurement. This seems like a reasonable thing to do since the chain is under tension when we pedal.

In contrast, the ruler measurement would often be done with no tension on the chain. The pins and bushings could be resting loosely anywhere in their over-sized gaps. So it seems that a ruler measurement with no tension on the chain would understate the extent of "stretch."

The Experiment...

So the experiment would be to mesure chain length with a ruler with the chain both relaxed and under tension. I did this with a well-worn chain , and it seems that tension on the chain increases the stretch measurement... but I can't tell how much... maybe between 1/32 and 1/64.... but my tired eyes can't tell for sure. This doesn't sound like much, but 1/32 would be a 25% error in the measurement, and 1/64 would be 12% error.

So, if anyone is interested, particularly someone with a caliper, I'd love to see if you could replicate and quantify this measurement... measuring relaxed chain length then chain length under tension. Or, if I'm just thinking of this all wrong, that would be interesting to hear too.

Thanks,
Mark
 

·
Banned
Joined
·
6,360 Posts
Aaarrghh...too much THINNNNNKIIIINNNNNGGGG!

Or you could follow the Close Enough Rule of Chain Replacment: When shifting gets sloppy, or you pull the chain off the big ring and can expose a whole tooth, or when it seems like a good idea, you go buy a new chain for 15 bucks.
 

·
Registered
Joined
·
5,524 Posts
Discussion Starter · #3 ·
Cory said:
Aaarrghh...too much THINNNNNKIIIINNNNNGGGG!
You're right, of course, as a practical matter. But this is SCIENCE. The pursuit of knowledge.

Well, ok, not really science, more like, well, ENGINEERING... ok, maybe not that either...

I know, more like QUESTIONING AN ASSUMPTION.... That's it... is a ruler along a relaxed chain really an accurate way to measure stretch.

In practice, I do use the Park tool because if nothing else it amounts to "when it seems like a good idea."
 

·
n00bsauce
Joined
·
13,507 Posts
Yes, a ruler is an accurate way

PdxMark said:
You're right, of course, as a practical matter. But this is SCIENCE. The pursuit of knowledge.

Well, ok, not really science, more like, well, ENGINEERING... ok, maybe not that either...

I know, more like QUESTIONING AN ASSUMPTION.... That's it... is a ruler along a relaxed chain really an accurate way to measure stretch.

In practice, I do use the Park tool because if nothing else it amounts to "when it seems like a good idea."
If your (and my) tired eyes can't see the difference (1/64-1/32) then there IS no difference!
 

·
Registered
Joined
·
5,558 Posts
I guess some of these chains are so expensive now (Campy especially) that people postpone replacing their chains. I still use relatively cheap SRAM chains and just replace 'em every 2000 miles or so. Never bothered to measure anything or think about it too much. If you wait until shifting goes south or you can pull the chain very far away from the c-ring it's already too late.
 

·
Registered
Joined
·
10,160 Posts
it's only 1/16"

Today, the recommendation (at least for Campy) is to change the chain when it is only .5% or 1/16" per foot longer.

Of course you always want to keep the chain tight during measurement. If you use a 12" machinist's scale it's easy to lay the chain on a bench. place the scale on the chain with the end of the scale aligned to the edge of a pin. Pull the chain tight and check the other end of the scale. The pin on the opposite end will be completely covered by the scale when new. When 1/2 the pin diameter is exposed, you've slightly exceeded the 1/16" per foot maximum elongation.

The early Park chain checkers weren't very accurate (I have one). They often show a new chain to have some wear and the dial produces a different scale reading depending on which way you turn the dial (the scale is only a decal applied to the dial).
 

·
Registered
Joined
·
2,601 Posts
I use the suggest Campy method of measuring between 6 links with a caliper. The distance may not exceed 132.60mm. I think originally it's 132mm so that would be about a 0.5% wear margin.

The below is something strange/interesting I read on Velonews a few days ago. It doesn't make sense if you ask me....unless their links are made of cheese metal.

"Wayne Stetina, Shimano's R&D manager, says, "If you remove the chain when it is only halfway worn out and flip it over," he says, "you will double your chain life." In other words, your chain will now be turned inside out. The other side of the rollers will now contact the gears, and the derailleurs will now be laterally bending the chain the opposite direction. Stetina says that Shimano engineers discovered this phenomenon quite by accident."
 

·
Registered
Joined
·
985 Posts
The Shimano "turn it over" is an interesting concept. I can visualize part of the logic where the wear is primarily creating an oval pin that is not at 90 degrees to the direction of the chain. I remember seeing some enlargements of pin wear that did show uneven wear. The problem in my feeble mind is that if I guess that the chain has been worn at 50%, turning it over will not double the life of the chain since I don't think the oval wear pattern will a) bring the chain back to no wear; and b) the pin that has been worn oval will now be wearing on a smaller surface so it may wear faster.
 

·
A wheelist
Joined
·
11,322 Posts
PdxMark said:
..........the ruler measurement would often be done with no tension on the chain. The pins and bushings could be resting loosely anywhere in their over-sized gaps. So it seems that a ruler measurement with no tension on the chain would understate the extent of "stretch."
I would never do that. I measure it with the derailer tension on the chain or with the chain hanging from a nail. Who's to know what the "correct" tension is but with consistent tension at least the measurements are consistent.

And 1/8" is way too much. 2x too much.
 

·
Registered
Joined
·
5,524 Posts
Discussion Starter · #10 ·
1/16".... Dohhhh... just when I think I know something...

As for that Shimano flip the chain trick... THAT seems weird... Kinda like...

Unidirectional chain stretch...

Pin/bushing couplings that are stretched if you measure left-to-right are no longer stretched if you measure right-to-left...
 

·
Registered
Joined
·
2,633 Posts
Where does brand new chain show up on the Park tool?

I would think that stretching would also work on a new chain, but I've never tried my tool on one. I have noticed that my cassettes last longer since I've replaced the chains more, but haven't tried to prove that it's cost effective. I've come to like changing them a little more often because I find old chains are harder to keep clean which adds to the wear on pulleys, cassettes, and chain rings.
 

·
Registered
Joined
·
5,524 Posts
Discussion Starter · #12 ·
rusa1586 said:
Where does brand new chain show up on the Park tool?
New chain shows up at zero stretch on the Park tool. It can even a bit hard to get the prongs into the chain.
 

·
Registered
Joined
·
408 Posts
PdxMark said:
As for that Shimano flip the chain trick... THAT seems weird... Kinda like...

Unidirectional chain stretch...

Pin/bushing couplings that are stretched if you measure left-to-right are no longer stretched if you measure right-to-left...
Shimano guy doesn't make sense. 'Stretch' is fore/aft wear on the pin/bushing. Inverting the chain won't change that, it merely reverses the top/bottom wear on the pin/bushing.
 

·
Registered
Joined
·
37 Posts
Bigger than the difference between a taught and a relaxed chain is the fact that a ruler measures a different thing than a chain checker. A ruler measures from pin center to pin center, while a ruler goes from roller surface to roller surface. The result is that the ruler measure doesn't include wear between the roller and bushing, only between the bushing and pin. There are two rotating surfaces inside a chain: between the pin and the inside of the bushing, and between the outside of the bushing and the roller. Wear between the pin and bushing changes the pitch of the chain (lengthens the distance between pin centers). Wear between the bushing and roller doesn't really change the pitch, but it does make the rollers rattle on the bushings, which means that when you stick the chain checker in, one roller goes to the front of the slop and the other goes to the back, and so the checker shows more distance and wear than the ruler.

Mathematically, you could think of it like this:

ruler measure = sum of pin<->bushing wear over 12"
chain checker measure = sum of pin<->bushing wear over 12" + bushing<->roller wear of 2 end pins

Sheldon Brown has photos of a heavily worn chain at http://sheldonbrown.com/chains.html. Look at the sloppy roller and worn bushing in the second photo. That's the wear isn't taken into account by a ruler.

I'm still trying to wrap my head around the Shimano reps statement. Yeah, it doesn't seem to make sense at first -- a lengthened chain is still a lengthened chain if you turn it around -- but it sounds like they got the result from experience, and that argues strongly. Why is another matter.
 

·
Chili hed & old bike fixr
Joined
·
1,126 Posts
ahaile said:
Bigger than the difference between a taught and a relaxed chain is the fact that a ruler measures a different thing than a chain checker. A ruler measures from pin center to pin center, while a ruler goes from roller surface to roller surface. The result is that the ruler measure doesn't include wear between the roller and bushing, only between the bushing and pin. There are two rotating surfaces inside a chain: between the pin and the inside of the bushing, and between the outside of the bushing and the roller. Wear between the pin and bushing changes the pitch of the chain (lengthens the distance between pin centers). Wear between the bushing and roller doesn't really change the pitch, but it does make the rollers rattle on the bushings, which means that when you stick the chain checker in, one roller goes to the front of the slop and the other goes to the back, and so the checker shows more distance and wear than the ruler.

Mathematically, you could think of it like this:

ruler measure = sum of pin<->bushing wear over 12"
chain checker measure = sum of pin<->bushing wear over 12" + bushing<->roller wear of 2 end pins

Sheldon Brown has photos of a heavily worn chain at http://sheldonbrown.com/chains.html. Look at the sloppy roller and worn bushing in the second photo. That's the wear isn't taken into account by a ruler.

I'm still trying to wrap my head around the Shimano reps statement. Yeah, it doesn't seem to make sense at first -- a lengthened chain is still a lengthened chain if you turn it around -- but it sounds like they got the result from experience, and that argues strongly. Why is another matter.
Bigger than the difference between a taught chain( what if the chain is stupid and taut rather than taught?
Today's chains are known as bushingless( which means without bushings) chains!
 

·
Registered
Joined
·
37 Posts
curlybike said:
Bigger than the difference between a taught chain( what if the chain is stupid and taut rather than taught?
Heh :).

curlybike said:
Today's chains are known as bushingless( which means without bushings) chains!
Yeah, you're right, I should have put "bushing" in quotes (like Sheldon does on his page). The same holds true though -- "bushingless" chains still have a bushing-like thing (or whatever you want to call it), it's just pressed from the sideplate. It still wears on the inside and the outside, and a ruler only measures the inside wear (bushing-like-part-of-sideplate<->pin) and not the outside wear (bushing-like-part-of-sideplate<->roller).
 

·
NeoRetroGrouch
Joined
·
6,491 Posts
I wonder if those who....

remove their chain with a removable link for washing are, unknowingly, getting more life from the chain. Odds are that half the time you would put it on one direction and half the time the other.

TF
 

·
The Edge
Joined
·
205 Posts
I feel like I've been getting more life because I'm able to clean out the gunk better when I remove the chain and soak it in an agitating mineral spirit bath. I read somewhere that chain wear in our application is caused more by contaminates getting inbetween contact surfaces and not by metal to metal contact. I'm fairly sure this is true because the o-rings on a piston experience far more distance and metal to metal contact than our chain pins and "bushings", but don't wear nearly as fast.
 

·
Chili hed & old bike fixr
Joined
·
1,126 Posts
ahaile said:
Heh :).



Yeah, you're right, I should have put "bushing" in quotes (like Sheldon does on his page). The same holds true though -- "bushingless" chains still have a bushing-like thing (or whatever you want to call it), it's just pressed from the sideplate. It still wears on the inside and the outside, and a ruler only measures the inside wear (bushing-like-part-of-sideplate<->pin) and not the outside wear (bushing-like-part-of-sideplate<->roller).
I might be mistaken, but I think that the roller part is free to revolve and if you were to disassemble an old chain and compare the o.d. of the rollers to that of a new one, you would find the wear on the o.d. to be of no consequence. It would help to use a micrometer to see the minute wear. I guess this will have to be a project for me to do. Then we will all learn.
 

·
Chili hed & old bike fixr
Joined
·
1,126 Posts
continuation of prior thoughts

ahaile said:
Heh :).



Yeah, you're right, I should have put "bushing" in quotes (like Sheldon does on his page). The same holds true though -- "bushingless" chains still have a bushing-like thing (or whatever you want to call it), it's just pressed from the sideplate. It still wears on the inside and the outside, and a ruler only measures the inside wear (bushing-like-part-of-sideplate<->pin) and not the outside wear (bushing-like-part-of-sideplate<->roller).
After looking again at Sheldons site about chain wear, I agree with most of what he sez. I am not too sure that the wear of the actual roller, inside and out, is anything but minor. The ability to spin greatly reduces the wear in any one area and the other parts do not have that advantage. After saying all that, it really works best to measure the chain on the bike with the derailleur spring tension on the chain. The Chain should be on the big ring to get more tension and make it easier to measure. I always use the edge of the pin because that is easier to see than the center of the pin. I have old eyes and they can see the edge much more sharply.
 
1 - 20 of 30 Posts
Top