Cycling UK Forum

UpWrong wrote: ↑16 Nov 2022, 7:50am With some RDs the chain goes slack when there is "too much" chain, with others you get chain wrap with the chain fouling itself. Or does it go from chain wrap to chain slack with increasing excess chain length?

I use around 48 teeth as the max capacity of a long cage rear derailleur as that's about the upper limit before you get to much chain wrap/slack chain.
Once you exceed the capacity by to much, as long as you can get big/big then you just stop using the smallest sprockets when on the smallest chainring.
My max was a 51t difference, 48/36/22 triple and a 9-34 9 block on the back of my bent trike.
22/13 and bigger were fine, the 9/10/11 sprockets weren't.

Luck ........

Dingdong wrote: ↑16 Nov 2022, 7:46am I normally ride a 50/36 on most of my bikes. The 36 seems to give a really nice range across a 10/11 speed set up. Recently tried one of my bikes with a 42 inner and was not at all happy!

The main reason for going sub-compact is to step the whole gear range down a bit to match your riding style.
I see no point in being able to bomb down a hill at plus 30 mph, if I run out of gears down trying to climb it.
My max cadence is around 85 rpm before my efficacy drops and I want to spinout at just under 30 mph.
So given the in a 100" gear spun at 100 rpm you will get you to 30 mph, then a 110" gear spun at 85 rpm will get me to 28 mph.
And a 110" gear is roughly 4x27" or a 4:1 chainring/sprocket ratio.
Given that most cassette start with an 11t sprocket that will gives me a 44t chainring.
Keeping a 3:2 big/small chainring size for the best overlap, that gives me a 44/28 sub-compact double.
Compered to a 50/34 compact double, that drops one gear off the top end where I don't need it and adds it back to the bottom where I do
It was only after I got my first gear low enough such that climbing +20% hills was easy, that I thought about going over a 110" top.

Luck .........

CJ wrote: ↑16 Nov 2022, 6:18pm

AndyK wrote: ↑16 Nov 2022, 1:55pm

pwa wrote: ↑16 Nov 2022, 3:49am I think a difference of up to 6 watts in the worst gear is something you could feel, and weaker riders will feel it more.

Not sure that's right. In that test, the wattage difference was based on a specific (pretty high) level of power input: 250 watts. Reduce the amount of power being put in and I'd expect the losses (in absolute terms) also to decrease as we're mostly talking about the coefficient of friction here.

In every cycle transmission efficiency test that I've seen, where the test was conducted at different power levels, the efficiencies were markedly WORSE at lower power levels, for ALL kinds of transmission, derailleur, hub-gear, single-speed, shaft drive... with the less efficient gears suffering most from comparison at low power. That's because some of the sources of frictional loss are not affected by the level of power being transmitted, or not much affected. So I would expect something similar here too, that the difference in efficiency between 1× and 2× will be somewhat greater at 50W than was measured at 250W.

My guess is the people conducting this test know this and chose to test at a high power so the losses didn't look too bad - and also because they're interested in elite althletes. I am not an elite athlete and I know it. I don't have enough power to waste any more of it on friction than I can possibly avoid, so I ride the easiest rolling tyres I can afford (and never mind if they puncture a bit easier - the occasional rest is quite welcome), avoid the most out-of-line gears and don't usually bother pedalling hard on sprockets with fewer than 15 teeth - except when I judge it to be advantageous to keep in the slipstream of a companion who doesn't understand the futility of pedalling downhill!

Interesting. Do you have some links for those tests? It seems counter-intuitive. As I understand it, frictional losses are usually proportional to the amount of force being applied - at least for dynamic friction. This study seems to indicate that efficiency remains pretty much the same at different levels of torque until you get to very, very low levels.

AndyK wrote: ↑16 Nov 2022, 11:53pm

CJ wrote: ↑16 Nov 2022, 6:18pm

AndyK wrote: ↑16 Nov 2022, 1:55pm
Not sure that's right. In that test, the wattage difference was based on a specific (pretty high) level of power input: 250 watts. Reduce the amount of power being put in and I'd expect the losses (in absolute terms) also to decrease as we're mostly talking about the coefficient of friction here.

In every cycle transmission efficiency test that I've seen, where the test was conducted at different power levels, the efficiencies were markedly WORSE at lower power levels

Interesting. Do you have some links for those tests? It seems counter-intuitive. As I understand it, frictional losses are usually proportional to the amount of force being applied - at least for dynamic friction. This study seems to indicate that efficiency remains pretty much the same at different levels of torque until you get to very, very low levels.

The tests I'm most familiar with are those conducted for Fichtel & Sachs by Keller, J. (1983) and the place I go for a summary of Keller's results is my 3rd Edition copy of Bicycling Science, p343. Keller compared Single-speed, Three-speed hub and 6-speed derailleur gears and in another study compared shaft with chain drives: worn and lubricated or dry and rusty! He tested at power levels of 50, 100, 200 and (first study only) 400W. All drives were most efficient at the highest power tested. The six-speed derailleur in top for example (the gear which combines the loss-making factors of misalignment and low tooth-count, but not so very low in 1983 at 13T), was 97.8% efficient at 400W, 96.9% at 200W, 94.9% at 100W, but only 92.1% at 50W. Note that the percentage loss increased by only 0.9% with the first halving in power, but by 2.8% with the last. It might not be too much of an overstatement to say that efficiency 'goes off a cliff' below 100W! Bicycling Science cites several other studies of cycle transmission efficiency and all of those that test the same transmissions at differing power levels find them markedly more efficient at high power.

I'm sorry I can't link you to a facsimile of this book, but in searching the web for Keller's work I found a pdf of this PhD paper by Matthew Donald Kidd. Go to pages 37-40 in the pdf (numbered 24-27) to read how effciency was found to vary with power in all the same papers referred to in Bicycling Science. MD Kidd concludes from those observations that: "efficiency is high and it rises as power throughput increases".

AndyK wrote: ↑16 Nov 2022, 1:55pm

pwa wrote: ↑16 Nov 2022, 3:49am I think a difference of up to 6 watts in the worst gear is something you could feel, and weaker riders will feel it more.

I suggest that weaker riders will feel the difference less because at lower power output levels it's so small as not to be noticeable.

At this point of the discussion, I tend to wander off, simply because I very much doubt I can feel the loss of 2.5% or even 5% in a twenty five mile journey when other factors include wind, weather, how well I slept the night before, how much shopping i'm carrying, and whether I'm cycling before an ad hoc cake shop or after.....

AndyK wrote: ↑16 Nov 2022, 11:53pm As I understand it, frictional losses are usually proportional to the amount of force being applied - at least for dynamic friction.

MD Kidd, on the other hand, on p28 of his dissertation, concludes from the data of Spicer et al and Keller, that: "the losses from friction remain fairly constant, independent of power". I think there must be both constant and variable components to the frictional losses: constant on the slack side of the chain and increasing according to the power applied on the top side. So I made a little table of Keller's data and did some calculations on it to tease out the likely sizes of these two components. The 'Loss Factors' in italics are calculated by comparing losses at each higher power with those at 50W. Using those as a basis I then iterated the factors in bold so as to achieve a best fit between calculated losses (calculated using the factors in bold) and the losses measured by Keller.

As one would expect, fixed losses dominate at low power, variable at high power. More interesting to find is that fixed losses seem to be increased mostly by the greater link pivot articulation that comes with fewer teeth, whereas variable losses are increased by the greater side friction that comes from out-of-line running. I don't know about you, but I'd have expected it to be the other way round.

My take-home thought is that it seems relatively more important for strong riders to avoid out-of-line running, and for weaker riders to avoid tiny sprockets. As a weaker rider, I spend very little time and even less energy in high gears anyway, so I'm rather glad about that

Hi Chris - interesting stuff.
Another way to understand what is going on is to use a least-squares fit to Keller's data. For each sprocket I did a linear regression calculation on the Loss v Power data (columns D and B in this spreadsheet): This gives, in columns F and G, the Fixed loss (W) and the Variable rate (%) needed for Calculated Loss in column E. The results are quite similar and the conclusions exactly the same, of course.
Best wishes,
Chris

cycle tramp wrote: ↑20 Nov 2022, 9:08am

AndyK wrote: ↑16 Nov 2022, 1:55pm

pwa wrote: ↑16 Nov 2022, 3:49am I think a difference of up to 6 watts in the worst gear is something you could feel, and weaker riders will feel it more.

I suggest that weaker riders will feel the difference less because at lower power output levels it's so small as not to be noticeable.

At this point of the discussion, I tend to wander off, simply because I very much doubt I can feel the loss of 2.5% or even 5% in a twenty five mile journey when other factors include wind, weather, how well I slept the night before, how much shopping i'm carrying, and whether I'm cycling before an ad hoc cake shop or after.....

On the other hand, if you are on a 60 mile journey and you are getting a bit weary due to a headwind, an additional 5% efficiency loss (making you work 5% harder) is something you could do without.

Hmm - but Kidd concludes from his own data and calculations that "The slack side tension and pressure of the derailleur mechanism was found to reduce efficiency by up to 5% at low power and 1.5% at high power throughputs in comparison with the same gear combination with no rear derailleur." and that "For all gear ratios with a derailleur gear, a varying amount of non-zero slack slide [sic] tension is present and this has its greatest effect on efficiency, where it can account for up to 4% losses."

He also notes that slack side tension increases as the chain is moved to larger sprocket combinations, due to the design of the rear derailleur. Using the examples in your table, the slack side tension will be higher for the 24t sprocket than for the 13t sprocket (assuming the front chainring remains the same). In real-world combinations this could offset the increased chain efficiency found when using larger sprockets.

So it doesn't sound like that's a fixed component of friction losses, though it does support the idea that losses are greater at lower power outputs.

Also it's interesting that the losses due to slack side tension are potentially more than those due to chain mis-alignment. In the example Kidd gives in fig 8.4 of "typical" usage (250W, 52:12, 80rpm, which admittedly isn't my idea of typical!) slack side losses are more than triple the losses due to chain offset.

The messages I take from this are that
(1) It's complicated. Far more complicated than many people make out.
(2) Chain alignment, though a factor in friction losses, is by no means a major one. Other factors can have equal or more effect.
(3) Assuming reasonably-well maintained components and lubrication, we're talking about losses broadly in the 2%-5% range here.
(4) When you're in a road race and trying to keep up with the pack, a 2% loss may well be noticeable. However when you're on tour or going to the shops, a 5% difference is probably not noticeable. (Though next time I'm doing a 100-mile ride and it's getting dark and cold and I still have 5 miles to go, I reserve the right to reconsider this view.)

I therefore conclude that cycle tramp is correct.

Slack side losses:
Slack side chain tension is provided by the tension spring, which is probably linear, i.e. the chain tension varies linearly with jockey cage angle. Now also the total angle through the which the chain articulates in going through the jockey cage also varies linearly with jockey cage angle, or very nearly so. (The lower chain run is nearly straight in big-big, and almost does a complete doubling back in small-small.) Whether the increase in friction loss due to increased slack side chain tension in big-big is balanced by the decrease in loss due to smaller total articulation angle I would not care to guess, but the two effects are certainly operating in opposite directions.

Chris Jeggo wrote: ↑23 Nov 2022, 2:01am Slack side losses:
Slack side chain tension is provided by the tension spring, which is probably linear, i.e. the chain tension varies linearly with jockey cage angle.
...

Second thoughts: chain tension varies linearly with jockey cage angle only while the jockey cage is roughly vertical (within 45 degrees, say).

Also, jockey sprockets are usually only 11 teeth, giving large articulation losses.

This new video just popped up on youtube ,where the 'Path Less Pedalled' site's Russ Roca lists some of the sub-compact double alternatives available (in the U.S.A anyway) , Spa Cycles gets a mention towards the end, 9 minutes long.

https://www.youtube.com/watch?v=z3GvzJF_sEY

tooley92 wrote: ↑5 Nov 2022, 2:40pm 2E7A5211-0430-4F93-8037-DB47E5619279.jpeg

Front mech swapped out for XT786 follows the shape of the rings much better, shifts very well and didn’t require any filing like CJ’s

I didn't realise this was possible. What is the BCD of the inner ring?

Tripletail52 wrote: ↑15 Feb 2023, 5:07pm

tooley92 wrote: ↑5 Nov 2022, 2:40pm 2E7A5211-0430-4F93-8037-DB47E5619279.jpeg

Front mech swapped out for XT786 follows the shape of the rings much better, shifts very well and didn’t require any filing like CJ’s

I didn't realise this was possible. What is the BCD of the inner ring?

74 bcd

l have recently fitted a sub compact 30/46 chainset, coupled with a 11 speed 11/34 cassette.

I appreciate that cross chaining isn't really advisable to do, but sometimes it's done.

With the current set up l can utilize every possible gear option. However, when on the 30 at the front and the 11 on the cassette the chain is far to slack.

I have now removed one chain link and the only gear ratio not possible is using the 46 on the front and 34 at the rear, the chain is simply not long enough now to accommodate it.

In truth its not really a problem as it's a less than ideal chain line. My only issue is that if l do go to select that ratio it will either snap the chain or worse case rip the mech off the hanger.

I've looked at maybe fitting a rear mech extension, but l dont think that's the answer as l don't have an issue with the clearance between the jockey wheels and the largest sprocket on the cassette.

I also need to lower the front mech but as it's a braze on bracket I need to probably get a front mech extender. What does confuse me, I've seen front mech extenders with 0 degree and 4 degree angles, what's the difference, other than the angle of course

That said, I went out today and did 50 miles with the front mech still in the original position and it worked exactly as it should. The current gap between the teeth on the outer ring and the lowest edge of the front mech is around 10mm.