IoW - and new cogs

Simon Brooke said:

The experimental method I propose is as follows:

(i) clamp sprocket in the vice

(ii) drape new chain over sprocket

(iii) Attach shock-cord to the pulling side of the chain to simulate
load

If my hypothesis is correct the link on the second engaged tooth away
from the load will be slack and capable of being moved radially of the
order of half a millimetre. Would you agree this proposed experiment
would be conclusive? If not, what experimental method would you
propose?

I'd suggest a few amendments:

- If you can get the chain arranged with a (very) strong pull in the
pulling direction, and a weak pull (jockey wheel spring) in the slack
direction, I'd be happier.

- You would need to do your test at every tooth "engaged" with the chain
- and then rotate the entire thing on by, say, 1/10th of a tooth
spacing, repeat, and keep going ten times. This would check if the
"wear point(s)" are moving.

- your test presupposes (I think) that slack link <==> no wear and thus
tight link <==> wear. However, I can almost see a case where this may
not be so: the link could be tight because it is under very much more
tension than "normal" because either or both of the two adjacent links
are pulling it, due to badly worn teeth engaging with the adjacent links
- but the link itself isn't engaging at all with the teeth.

R.

Tony Raven said:

It says nothing at all about wear unless you can show that the load-time
product for the chain is much greater for a worn cog than an unworn one.
And I can't see any mechanism for that to happen from what you have
described. Indeed if the load-time product is greater then the work
done for the same movement of the wheel would go up proportionately and
you would find yourself riding a very sluggish and hard work bike.

Just found what is an explanation that makes sense on Sheldon's site.
His view is that with a new chain on a worn cog, the link rolls up the
worn tooth under load (too worn and it rolls off the top of the tooth
causing the traditional skipping). That rolling causes a loaded
rotation of the roller about the pin which generates wear in the roller
and pin and accelerates chain wear. This does not happen on a new cog
where the rolling up does not occur.

--
Tony

"A facility for quotation covers the absence of original thought" Lord
Peter Wimsey (Dorothy L. Sayers)

Tony Raven said:

Just found what is an explanation that makes sense on Sheldon's site.
His view is that with a new chain on a worn cog, the link rolls up the
worn tooth under load (too worn and it rolls off the top of the tooth
causing the traditional skipping). That rolling causes a loaded
rotation of the roller about the pin which generates wear in the
roller and pin and accelerates chain wear. This does not happen on a
new cog where the rolling up does not occur.

I read the same thing last night (and felt very silly for not re-reading
the article earlier).

How much of a factor do you think this specific rolling is to wear? Does
it "greatly" accelerate wear, as Simon claims old cogs do to new chain?
And what about bushingless chains? (That section of the article refers to
bushings).

Personally, I doubt much riding up is going on with semi-worn cogs or that
it causes the majority of chain wear anyway.

~PB

Pete Biggs said:

How much of a factor do you think this specific rolling is to wear? Does
it "greatly" accelerate wear, as Simon claims old cogs do to new chain?
And what about bushingless chains? (That section of the article refers to
bushings).

The answer is I don't know. My first hurdle was to find a plausible
mechanism by which it could occur before arguing about its magnitude.

--
Tony

"A facility for quotation covers the absence of original thought" Lord
Peter Wimsey (Dorothy L. Sayers)

Tony Raven said:

It still wouldn't make any difference because a link is now
only engaged for a fraction of the time it would be on an
unworn sprocket. So unless, as someone else pointed out,
wear is superlinear with load, it will average out at x times
the load for 1/x times the time = the same wear.

Peter Biggs said:

What difference does it make to total wear if x load is
concentrated on one link at a time instead of x/6 load
applied to six links at a time?

If there's no relative motion between adjacent links, there can be no wear.
As a chainring (assuming a round one) has a constant radius, the only
relative motion between adjacent links occurs when the chain meets the
chainring (or sprocket) and when it leaves it. Between these two points,
the chain maintains a fixed curvature - each link maintains a fixed angular
displacement from the next. The important wear can only occur in the first
link meeting the chainring (and leaving the sprocket), so the tension
change across that link is crucial.

James Thomson

Tony Raven said:

Quoted message said:

How much of a factor do you think this specific rolling is to wear?
Does it "greatly" accelerate wear, as Simon claims old cogs do to
new chain? And what about bushingless chains? (That section of the
article refers to bushings).

The answer is I don't know. My first hurdle was to find a plausible
mechanism by which it could occur before arguing about its magnitude.

Thanks Tony. I know what you mean :-)

~PB

in message <[email hidden]>, James

Thomson (') said:

Tony Raven said:

It still wouldn't make any difference because a link is now
only engaged for a fraction of the time it would be on an
unworn sprocket. So unless, as someone else pointed out,
wear is superlinear with load, it will average out at x times
the load for 1/x times the time = the same wear.

Peter Biggs said:

What difference does it make to total wear if x load is
concentrated on one link at a time instead of x/6 load
applied to six links at a time?

If there's no relative motion between adjacent links, there can be no
wear.

That's over-simplistic. It would be true in the case of a platonic cycle
chain link, but a real link is made of metal which has finite surface
hardness and measurable surface irregularity. Furthermore, in the real
world, between link and sprocket there's a certain amount of road grit
which also has a certain degree of hardness. In crushing it must damage
the surface of the roller, and if the load is increased by an order of
magnitude all of these interactions are likely to change in non-linear
ways.

In any case, if the hypothesis I advanced earlier is correct, all the
stress is on the most recently engaged link, which is necessarily still
flexing (and, as Tony has pointed out, is rolling up the ramp of the
worn tooth).

Quoted message said:

As a chainring (assuming a round one) has a constant radius, the
only relative motion between adjacent links occurs when the chain
meets the chainring (or sprocket)

Uh-huh. Just so.

--
[email hidden] (Simon Brooke) http://www.jasmine.org.uk/~simon/

;; Life would be much easier if I had the source code.

in message <[email hidden]>, Richard

(') said:

Simon Brooke said:

The experimental method I propose is as follows:

(i) clamp sprocket in the vice

(ii) drape new chain over sprocket

(iii) Attach shock-cord to the pulling side of the chain to simulate
load

If my hypothesis is correct the link on the second engaged tooth away
from the load will be slack and capable of being moved radially of
the order of half a millimetre. Would you agree this proposed
experiment would be conclusive? If not, what experimental method
would you propose?

I'd suggest a few amendments:

- If you can get the chain arranged with a (very) strong pull in the
pulling direction, and a weak pull (jockey wheel spring) in the slack
direction, I'd be happier.

- You would need to do your test at every tooth "engaged" with the
chain - and then rotate the entire thing on by, say, 1/10th of a tooth
spacing, repeat, and keep going ten times. This would check if the
"wear point(s)" are moving.

- your test presupposes (I think) that slack link <==> no wear

Yes, I should have thought so.

Quoted message said:

and thus tight link <==> wear.

Well, not necessarily so. I assume we're agreed that damage to the chain
is largely a consequence of wear to the surface of the pins, bushings
and rollers and that that wear is primarily caused by fine hard
particles of road dirt (but also, probably, by the inherent microscopic
irregularity of the surfaces) as the chain is flexed while under load.

Thus if you hung a straight chain from a ceiling and hung big weights on
the bottom of it and left it there for a month, I assume we're all
agreed that no appreciable wear would occur - and even if you increased
the weight until the chain failed catastrophically, it would not be
wear which had caused the failure.

So as you say it isn't tightness per se which is causing the wear but
flexing while tight. In other words tightness is a necessary condition
for wear but it isn't a sufficient condition. At least, that's my
belief.

--
[email hidden] (Simon Brooke) http://www.jasmine.org.uk/~simon/
"The result is a language that... not even its mother could
love. Like the camel, Common Lisp is a horse designed by
committee. Camels do have their uses."
;; Scott Fahlman, 7 March 1995

in message <[email hidden]>, Pete Biggs

(') said:

Simon Brooke said:

The experimental method I propose is as follows:

(i) clamp sprocket in the vice

(ii) drape new chain over sprocket

(iii) Attach shock-cord to the pulling side of the chain to simulate
load

If my hypothesis is correct the link on the second engaged tooth away
from the load will be slack and capable of being moved radially of
the order of half a millimetre. Would you agree this proposed
experiment would be conclusive?

I don't know. Would you mind if I x-posted this to rec.bicycles.tech?

[sigh]

Well, we probably _would_ get some more light. There are a number of
extremely well informed people who post there. I'm not sure whether I
feel up to an rbt flame fest, though. Yes, go ahead.

--
[email hidden] (Simon Brooke) http://www.jasmine.org.uk/~simon/

IMHO, there aren't enough committed Christians, but that's care
in the community for you. -- Ben Evans

Simon Brooke said:

Well, not necessarily so. I assume we're agreed that damage to
the chain is largely a consequence of wear to the surface of the
pins, bushings and rollers and that that wear is primarily caused
by fine hard particles of road dirt (but also, probably, by the inherent
microscopic irregularity of the surfaces) as the chain is flexed
while under load.

Quoted message said:

Thus if you hung a straight chain from a ceiling and hung big
weights on the bottom of it and left it there for a month, I assume
we're all agreed that no appreciable wear would occur - and
even if you increased the weight until the chain failed
catastrophically, it would not be wear which had caused
the failure.

Quoted message said:

So as you say it isn't tightness per se which is causing the
wear but flexing while tight.

Simon, you surprise me. You seem to be arguing that motion is necessary
for wear to occur - a straightforward assertion you dismissed very recently
as "over-simplistic".

Why the sudden change of opinion?

James Thomson

Quoted message said:

Quoted message said:

n is suspect. Your figure is high, and you're assuming that
the load is shared evenly between n teeth on a new chainring.
It isn't.

Simon Brooke said:

Isn't it? Why not?

Because of the elasticity of the materials.

Quoted message said:

Quoted message said:

If your assumptions were correct (loads higher by an
order of magnitude; superlinear relation between wear
and load) a new chain run on an old chainring would
have a wear life more than an order of magnitude
shorter than one run on a new chainring.

Quoted message said:

No it wouldn't, because as soon as its own wear had
approximated the wear of the drive train, then the accelerated
wearing would decelerate back to something close to
'normal'.

The locus of contact between the new chain and the worn chainring is
determined by the pockets worn by the previous chain. When the new chain
has worn to approximate the wear of the chainring (which will itself
continue to wear, and at an accelerated rate), it's reached approximately
the degree of elongation of the old (discarded) chain - i.e. it's worn out.

James Thomson

in message <[email hidden]>, James

Thomson (') said:

Quoted message said:

Quoted message said:

n is suspect. Your figure is high, and you're assuming that
the load is shared evenly between n teeth on a new chainring.
It isn't.

Simon Brooke said:

Isn't it? Why not?

Because of the elasticity of the materials.

Quoted message said:

Quoted message said:

If your assumptions were correct (loads higher by an
order of magnitude; superlinear relation between wear
and load) a new chain run on an old chainring would
have a wear life more than an order of magnitude
shorter than one run on a new chainring.

Quoted message said:

No it wouldn't, because as soon as its own wear had
approximated the wear of the drive train, then the accelerated
wearing would decelerate back to something close to
'normal'.

The locus of contact between the new chain and the worn chainring is
determined by the pockets worn by the previous chain. When the new
chain has worn to approximate the wear of the chainring (which will
itself continue to wear, and at an accelerated rate), it's reached
approximately the degree of elongation of the old (discarded) chain -
i.e. it's worn out.

Exactly.

--
[email hidden] (Simon Brooke) http://www.jasmine.org.uk/~simon/
;; First they came for the assylum seekers,
;; and I did not speak out because I was not an assylum seeker.
;; Then they came for the gypsies,
;; and I did not speak out because I was not a gypsy...
;; Pastor Martin Niemöller, of Michael Howard.