recordacefromnew wrote:I think you have explained very well indeed given the lack of evidence (and indeed imho with overwhelming evidence against). I have just looked at Spicer’s data again, it seems they all demonstrate no more than c1%/1W offset loss to me.
well amongst other things it says;
52/21 (no offset, 60rpm 100W) 95.0%.
52/21 (offset, 60rpm, 100W) 93.8%.
Or to put it another way the losses (the thing we are really interested in) increased by about 25%. I have no idea why he ignored this, but he did. I don't think it would be fair to lump in the data with the other tests without allowing for the degree of offset; this would have been a lot smaller for the 52/11T gear, even so the efficiency was lower with offset here, too. Spicer also ignored time-dependant changes in losses. He (and everyone else it seems) has also ignored the possibility that even-numbered chainrings (or sprockets) can give two distinctly different types of result with any slightly imperfect chain, depending on which teeth see inner and outer sideplates.
In Kyle and Berto's work we see the following results @ 200W;
sprocket...Eff 22T....Eff 44T...difference%... differenceW
12.............91.9......93.7......+1.8%............3.6W
16.............93.9......95.5......+1.6%............3.2W
26.............94.5......93.9......-0.6%.............1.2W
34.............95.0......94.2......-0.8%.............1.6W
you can see there is a trend of increased efficiency with improved chainline in both directions. Other factors are certainly at work too but would not be excepted to show the same pattern. I think you will see the same kind of thing (comparing efficiency on each sprocket) in the Rohloff data too, but it is less clear-cut because they did not report all the data from all the gears.
None of this data is at loads that represent real pedalling of course, but even so the data suggests significant variations in efficiency with chainline.
It is true that chains do have wear at side plates, but then gear change, especially the front, is a brutal, metal on metal, crunchy business. One can also not rely on gears running perfectly indexed all the time.
the chain is under a low load at the rear mech. At the front mech the wear would be concentrated in one place on the front mech (vs spread over many links on the chain) but the only time you see worn front mechs is when they have been rubbing in normal running, not through shifting. The chain wear during shifting is likely very small. The chain wears laterally most of all on the top run under tension, obviously.
Perhaps the relevant question is if frictional loss being a major loss for a bicycle chain offset regularly what should we have seen? Side plate failures perhaps, i.e. so worn that a VERY worn chain would not simply be “stretched”. I have seen a number of VERY worn chains with each and every pin having a clear chunk worn away to give rise to the “stretch”, but I have never seen a side plate worn down to anywhere near approaching mechanical failure.
another false syllogism. 0.05mm wear on the sideplates will not compromise the chain strength, yet constitutes more material loss than you would need to cause 2% 'stretch' in the bushings.
So mechanically how can one explain the lack of such wear/inefficiency from chainline offset? I confess I am now only guessing, but perhaps it is due to the shape of the teeth - laterally they are all pyramidal, so the relatively wide base of each tooth typically would always guide and line up the next chain link along to ensure minimal rub of the next tooth on the next side plate at each initial link/tooth engagement. I run an 8x1 full size folder with 43mm chainline (for no better reason than I happened to have a decent 115mm UN72 at hand) on a 135mm hub, I have never noticed any difference in chain noise in different gears.
In spicer's work the chain was almost certainly coated with some kind of anti-friction coating; that much is clear from the alleged 'no lube' results. It would have mitigated the rubbing in the offset gears, too, for a while, anyway. If you set up a singlespeed with derailleur chain and sprockets, offset, and apply a decent tension to it, it is very noisy; you mightn't hear it going down the road though. I'm sure that tooth shapes make a difference, and so do chain sideplate shapes. But you can't entirely escape the lateral loading; this has to cause rubbing, and this is why sprockets, chains and chainrings wear as they do.
If your conjecture is true, the other thing I would expect to have seen is a queue of researchers lining up to broadcast Spicer’s conclusion being wrong. Instead 15 years have now passed, and no one has. I trust you would agree: a) it would have been perfect ammunition for Rohloff, and b) he did try hard to find ammunitions to support his hub gear; it would also have been valuable to Jason Smith at Friction Facts to highlight the precision and utility of his rig(s)/experiment(s) being able to debunk such a long-standing conclusion. I was a researcher once - few things are more desirable to a researcher than being able to slaughter a sacred cow or two...
Does it not occur to you that it would be surprising/unlikely indeed for someone to jump up now and say a decent bicycle chain in a high power, high tension scenario expends a significant % of power due to typical offset within a rear cluster, given their continual application in racing costing £XXX millions p.a. where winners and losers are often separated by mere fraction of a second? Wouldn’t you have thought road racers, especially sprinters, would all want their large ring lining up with their small sprocket?
Far from being a sacred cow, certain of Spicer's results may well have been deemed so ludicrous by others that they are not deemed worthy of slaughtering. His report does contain some useful stuff, just nothing useful about chain offset. That offset chain runs are inefficient falls into the realms of 'obviousness'; the real question is 'how much' and to date no-one has really grasped that nettle in a meaningful way. I'd expect Jason to demonstrate this at some point, provided his rig can be adapted and he is still keen.
Rohloff's tests were simple tests to compare a derailleur system with a hubgear, rather than to understand exactly why derailleur systems are inefficient per se. Arguably they cynically used the worst (least efficient) derailleur ratios in their report to show the IGH in a good light. They had no interest in explaining why derailleurs are inefficient, or improving the derailleur results in any way; to have done so would have exposed their (rather clumsy) sleight of hand.
Kyle and Berto's report was effectively sponsored by Browning, and their pitch was 'like a derailleur but shifts better' more or less. Again no real interest in exposing or explaining derailleur losses there! Looking at figure 12 (titled 'Derailleur type transmissions compared with hub gears') it looks as if the derailleur transmissions are the most efficient by far. However
elsewhere in the same report there are
other hub gear results which are omitted from figure 12, which clearly show that a humble three speed IGH is actually just as efficient on average (and perhaps moreso when you account for the typical duty cycle in each gear) than most derailleur systems. Again I have no doubt that the sponsors of the work had no interest in exposing the existence/origins of any losses in a derailleur system; arguably they did their best to hide these losses, judging by the crazy way the data is presented in that report; (it took me over half an hour to decipher their data to produce the simple table above, BTW).
Racers have long worried about chainline and for good reason; one or two percent makes a difference which is why pro mechanics have long used special setups for time trials etc which are aimed at reducing losses to an absolute minimum. They don't go on about it because they wish to retain their competitive advantage and they are after all, in league with the manufacturers who want you to buy one 'just like Wiggos' etc...
It seems to me that on top of the sets of published data, circumstantial evidence also strongly suggests chainline offset being meaningfully inefficient is a myth.
well I disagree on all counts. Maybe I have not explained myself very well but the signs are there for anyone who can be bothered to look for them. I think the subject definitely deserves some further study.
cheers
) but it makes for interesting reading.
