I have a poorly maintained road bike – cassette, chainrings and chain are all ready to be replaced after a long, messy season. And the chain is dirty right now.
Tonight I was messing around with my kickr a bit and comparing the watts in erg mode vs my quarq. The good news is that they are basically the same. There is a little bit more noise in the quarq data, so the NP is smidge higher, but the average power was exactly the same after 1:20 of riding.
The interesting thing was that that I was able to tell the difference between when I was cross chained and when I was in a good chainline. I noticed it subjectively at first, then did a mini experiment.
Set at 250 watts in erg mode (wahoo fitness app). All were 5 minute tests. 53/39 chainrings. 11-26 cassette. Spindown kickr and rezero quarq between each test - although all offsets were identical (140 for quarq and 254 for kickr)
53 x 12 - quarq read 249 watts
53 x 26 - quarg read 263 watts
39 x 23 - quarq read 251 watts
39 x 11 - quarq read 261 watts
I had no idea that a good chainline vs cross chaining was worth 5%. That’s huge.
Andy
Edit to add: The cassette on the kickr is basically new. It wasn’t clear the way I wrote it. Chain is dirty and worn. Chainrings are just worn.
Wouldn’t rezeroing the quarq add MORE variability? I think you need to repeat the test 3 times, in random order and add one or 2 more ratios as a baseline.
Plus as mentioned, a worn chain and sprockets probably have proportionally even more more drag when crossed up than a good chain and cassette.
There most likely is a small difference between what your Quarq reads in the small and big chain rings in the first place. What does the output of the Qalvin app say?
Oh…and what do we know about how well the trainer produces torque at varying “rear wheel” speeds?
I guess I’m saying that perhaps what you saw is not all due to the drivetrain condition…but yeah, a clean, straight drivetrain is a happy drivetrain
Tried it again to try to answer some of these questions.
Re-zeroed the quarq and did spindown on the Kickr once, not between each set – even though there was really no difference in the offset between trials, the point about doing frequent calibrations upsetting results is well taken.
This time I had a clean, well lubricated chain (finish line dry lube). Like I said, chain is a little worn. Chainrings are quite worn. Cassette is pretty good.
Bike is a trek madone with bb90 BB. Shimano ultegra chain. Sram X-glide chrinrings on the quarq. Sram pg1170 cassette on the kickr. Otherwise sram force drivetrain. You could argue that the chain is a bit too long since the 53:26 is a usable gear. But otherwise nothing too weird.
I measured the chainlines to make sure they were as straight as could be for the straight chainline trials. Turns out that i didn’t have the straightest chainlines for the straight chainline trials last time. Straightest are the 53:13 and 39:21.
This time at 250 watts for 2 minutes in erg mode on the kickr, the quarq read:
There doesn’t seem to be any problem with the quarq reading differently between the small and large chainrings
The kickr doesn’t seem to be terribly sensitive to “wheel” speed
The cross chain effect seems to be even bigger than I saw the first time.
My takeaway: we don’t spend enough time thinking about maintaining a good chainline when we ride. These differences are as big or bigger than the differences we discuss in aero framesets/wheels/helmets/etc
53 x 12 - quarq read 249 watts
53 x 26 - quarg read 263 watts
39 x 23 - quarq read 251 watts
39 x 11 - quarq read 261 watts
I had no idea that a good chainline vs cross chaining was worth 5%. That’s huge.
.
One of the top amateur cyclists in the country told me he races a 55 tooth front chainring for just this reason. He said that in the 55x14/15 (or somewhere right around there) it’s dead straight. No wasted watts and maximum efficiency when every little bit counts.
I just had a go at this myself, using the combination of an SRM and the Lemond Wattbox. I have previously established that the two do not drift relative to each other over a long session. I went through the entire cassette, riding for 5 minutes at 230W in each sprocket, and then as a final check I did another 5 minutes in sprocket 5. I subtracted the delta for sprocket 1 (smallest) from all the values, as sprocket 1 had the highest drag, so the chart shows how much less drag each sprocket had relative to the smallest sprocket:
As you can see, I didn’t find anything close to 10W difference between any sprockets, the biggest difference is 1.4W between sprocket 1 and sprocket 5.
The repeat test of sprocket 5 gave a result 0.2W different to the first test (giving it 1.6W reduction). I don’t know why sprocket 7 looks a bit out of place, but on the whole, my testing suggests any differences are <1W provided you avoid the smallest sprocket and the two largest sprockets.
Good research, but you may have overlooked something.
I doubt that you have discovered friction differences between chainline angles. Instead you are finding cog/chain friction differences. Research that I have read indicates that for the same size gear (i.e., the same mechanical advantage), bigger cog to bigger chainring beats smaller cog to smaller chainring by a significant margin. I bet that is what you’re seeing.
I care! Thanks for taking the time to do this Steve. Those results are a lot closer to what I thought they should be - actually less loses than I thought. In a road race, I often use my 50-28 and 50-24, and they haven’t “felt” inefficient to me.
I assume the efficiency of the larger cog combos, makes up for most of the inefficiency of the cross chaining?
apparently no one cares about this data :)-
Perhaps it’s more exciting if I do the graph the other way round, showing the increase in drag rather than decrease?
But fundamentally, being told there’s only ~1W to be saved by moving one sprocket across from the smallest isn’t all that exciting. My enthusiasm for getting a 66T chainring made has definitely been diminished a bit.
I’ve been beating this drum for years, pretty much since compact cranks became commonplace. You keep hearing people say that a 50x11 is all the gear you need, and dismiss the traditional large TT rings as simply a macho delusion. I’ll grant there is a psych visual aspect to a big-ass 54T ring, but really if you’re spending a lot of time in a 50x11 or 12 then you’d be better off with a bigger chainring and move up a cog or two, unless you really need the 34T on the other end for a lower low on big climbs (or if the smaller big ring will enable powering over small rollers without having to drop into the small ring). The compact nazis will then point out that you’ll hardly ever need a 54x11 so again that’s a wasted combo, but to that I say on a flat course I have zero use for a 21 or 23T either; all but the middle 4-5 cogs are essentially just spacers.
It’s cool to actually try to quantify it w/ a PM, but anecdotally it’s always been plainly evident to anyone who does their own maintenance when it’s right in front of your face on the workstand, simply from the fact that it’s louder running in the smallest cogs than it is towards the middle of the cassette. Seriously, that noise is nothing but metal-on-metal contact; how could more of it possibly be anything but a detriment? How much of it is lateral due to a sharper angle of incidence vs how much is a function of more acute bending due to a smaller cog radius (as DSW points out) I can’t say, but it’s undeniable it’s a negative either way.
I’ve been beating this drum for years, pretty much since compact cranks became commonplace. You keep hearing people say that a 50x11 is all the gear you need, and dismiss the traditional large TT rings as simply a macho delusion. I’ll grant there is a psych visual aspect to a big-ass 54T ring, but really if you’re spending a lot of time in a 50x11 or 12 then you’d be better off with a bigger chainring and move up a cog or two, unless you really need the 34T on the other end for a lower low on big climbs (or if the smaller big ring will enable powering over small rollers without having to drop into the small ring). The compact nazis will then point out that you’ll hardly ever need a 54x11 so again that’s a wasted combo, but to that I say on a flat course I have zero use for a 21 or 23T either; all but the middle 4-5 cogs are essentially just spacers.
It’s cool to actually try to quantify it w/ a PM, but anecdotally it’s always been plainly evident to anyone who does their own maintenance when it’s right in front of your face on the workstand, simply from the fact that it’s louder running in the smallest cogs than it is towards the middle of the cassette. Seriously, that noise is nothing but metal-on-metal contact; how could more of it possibly be anything but a detriment? How much of it is lateral due to a sharper angle of incidence vs how much is a function of more acute bending due to a smaller cog radius (as DSW points out) I can’t say, but it’s undeniable it’s a negative either way.
I too have been beating this drum ever since I started working with Atomic whom made me think about this. At one time, I know that Tony Martin and Mirinda Carfrae were the only two athletes who had a 10 tooth rear cog for this exact reason. However chainring teeth really still comes down to the rider. Obviously a rider doing a sprint versus and Ironman and whether their ftp is 150 or 400 is going to matter, their cadence also matters as well. A 50x11 may be all someone uses, but it is the reason I have a 28 on the back regardless. I don’t care if I ever use the 28, I look it as backup, but I primarily have it on there to push the 24 and 21 over a bit.
I decided to repeat this test. The previous test used a 10 speed cassette with 11 speed Di2. This worked without any obvious problems and wasn’t noticeably noisy in any gear, but when we’re looking at ~1W differences I thought it was worth eliminating the small accumulated pitch difference across the cassette. So as I recently got a friend with a lathe to convert my race wheel freehub to 11 speed, allowing me to fit 11 speed cassettes to that and the Lemond, I had everything in place to be able to repeat it with an 11 speed cassette.
The result was only a little different to last time. I repeated the tests with the 12T and 16T after going all the way along the cassette from 12T to 25T, and the figures from the repeat tests differed from the first pass by 0.3W and 0.5W, in opposite directions, so it looks like the level of noise in the testing method of 5 minutes per sprocket is up to ~0.5W.
The repeat test of sprocket 5 gave a result 0.2W different to the first test (giving it 1.6W reduction). I don’t know why sprocket 7 looks a bit out of place, but on the whole, my testing suggests any differences are <1W provided you avoid the smallest sprocket and the two largest sprockets.
The two lab tests I’ve seen didn’t show a significant loss from a crossed chainline either.
Research that I have read indicates that for the same size gear (i.e., the same mechanical advantage), bigger cog to bigger chainring beats smaller cog to smaller chainring by a significant margin. I bet that is what you’re seeing.
What research? Spicer’s testing showed the opposite. Higher chain tension was better down to the 12t anyway.
