There is a lot you'd need to figure out a way to work around, just thinking the top ones to me would be

How will you have a fixed speed/power/cadence without a machine?
Accuracy between two power meters?

I'm sure there is more. I can't think of a reasonable way (although I'd love to test this kind of stuff) to do it :/

Link to an old BTR thread. Ron and Jens are both still active on ST

---

My Blog - http://leegoocrap.blogspot.com

I think the basic idea is correct with 2 powermeters, but if both of them have +/- 2% accuracy that could already be a 12W difference, even if both are perfectly set up and calibrated. I think if you would measure different chain coatings for example, the results would be well within a normal margin of error.

Yes, with a SRM, a PT, and a Computrainer to set the load. We were trying to see whether the data would be clean enough to use to determine differences across chainrings and cogs. They appeared to be.

It might be possible to do this with only a crank-based PM, but it would be a lot more hassle.

Yes, I did it many years ago using a PowerTap and SRM, to determine for myself whether 'big-big' (i.e., 60 x 12) was more efficient than 'small-small' (i.e., 53* x 11), all else (i.e., chainline) being constant. The data indicated a significant difference in favor of the former.

*A 55 would have been better, but I wasn't willing to buy another one to replace the one that I had sold just for this experiment.

Do you mean you think the 55 would have been better than the 60, or just better than the 53? Also did you account for more chain in the equation? Not sure, but maybe a few extra links for the big plates on both ends?

Have you taken a look at the big wheels on the derailleurs yet??

what was the protocol if you don't mind sharing?

---

My Blog - http://leegoocrap.blogspot.com

I think he's referring to the gear ratios. a 55-11 is the same gear as a 60-12 (5.00:1)

---

Swimming Workout of the Day:

Favourite Swim Sets:

2020 National Masters Champion - M50-54 - 50m Butterfly

If you're into Arduino and good at researching stuff, you can probably workout something in the 100w range of motor control for about $200. If you want that level of controlling RPM or input power.

We were interested in whether we could detect differences in drive train efficiency to test Spicer's hypothesis that high chain tension had fewer losses (was more efficient) than low chain tension. Spicer did his tests on a motor-driven jig set up with a single ring and cog. We didn't have that set up, but Greg Steele had a SRM, a PT, and a Computrainer. For the same power, same gear ratio, and same wheel speed the cadence is the same -- but when the chainring and cog are larger, the chain speed is faster. If the power is the same and chain speed is faster, the chain tension is lower since chain speed * chain force (=chain tension) = power.

Greg built up a frankencassette with spacers and cogs so that larger cogs were outboard and in line with the large chainring, and smaller cogs were in line with the smaller chainring. The cogs were chosen so that the overall gear ratios would be very close. (The chainrings were a 42 and a 54, the cogs were 12 and 13 for the 42 ring, and 15 and 16 for the 54 ring) Then he hopped on the bike and did a warm up to stabilize the tire-roller temperature. Then he did some steady power and a ramp up in power for each of the four pre-chosen ring/cog combos, two for each ring.

The PT and the SRM were both checked with a static calibration before the test and both were within spec, but either one read a tiny bit low within spec or the other read a tiny bit high within spec -- I don't remember which it was. However, we were only interested in whether we could determine *differences* in drive train efficiency. If we were really concerned we could've "tared" the calibration a bit better but we weren't.

Anyway, here's the result. The x-axis is minutes of time. The top panel shows the gear ratios. You can see that the 54x16 is between the 42x13 and the 42x12; the 42x12 is between the 54x16 and the 54x15. The middle panel shows the watts that the test was done at (when measured by the SRM); and the bottom panel shows the second-by-second difference between the SRM and the PT.

We concluded that it *would* be possible to use two consumer-level power meters to determine differences in drive train efficiency, and that these data were *not* consistent with Spicer's hypothesis.

^Thanks for sharing

perhaps the takeaway is "possible to do" but you probably need somebody smarter than me steering the ship :D

---

My Blog - http://leegoocrap.blogspot.com

Gawd! My very first post on BTR... over 12 years ago.

Dr Chung gave the right answer. Friction Facts testing also indicated that chain tension did not trump using bigger cogs.

I pondered doing a rollercoaster ride with a Quarq or SRM: coasting downhill so that eta has no effect. Solve VE for the downhill section, then select eta so that the uphill sections solve to the same CdA and Crr. As long as the same gear is used throughout that should hold true.

---

Developing aero, fit and other fun stuff at Red is Faster

Tom A. is somewhere in that thread too. Maybe Andy too, though Kraig used to censor him for reasons I could never fathom. I did a lot of those drivetrain tests back then and was usually frustrated with the lack of consistency. For differences greater than 2 watts, I think it's OK. But less than that and a couple issues come into play:

1) The powermeters drift; and they drift at different rates (SRMs tend to drift more). You could easily get a watt or 2 drift in 20 minutes of testing.

2) When testing different chainring combos, it's not certain that the SRM is measuring consistently. My recollection from hang-test slope calibration of my SRM (decades ago), that it varied slightly depending on both the chainring and the chain angle. That's in addition to the watt differences for slight cross-chaining. So you truly do need a franken-cassette to do a fair test.

hey what ever happened to Greg Steele, anyhow?

---

My latest book: "Out of the Melting Pot, Into the Fire" is on sale on Amazon and at other online and local booksellers

Yup. This is exactly what I alluded to above when I said it would be possible with only a crank-based PM but it'd be a hassle.

I put the 60 on the inside and the 53 on the outside, such that the chain angle was equivalent when used with the 12 and 11, respectively.

I also statically-calibrated both the crank and hub for each combination, and adjusted the data accordingly.

Thank you all for your helpful responses. My take homes from this thread so far are

1. it is possible to measure, but there are a lot of external variables and noise, some of which are very difficult to control
2. it is best to control the input watts precisely in order to remove a big source of variability
3. careful calibration of each power meter is recommended
4. multiple runs are likely required
5. results will probably not compete with a good lab test and resolution will likely be low