I'm sure someone more knowledgeable will explain it more thoroughly, but generally - the error doesn't come from the torque reading, it's from the measurement of pedal velocity around the circle. By only knowing cadence, the assumption made is that the velocity profile is uniform over each revolution, and this leads to errors in measurement of powers when velocity varies throughout the pedal stroke.

Several questions wrapped up in there, but it breaks down something like this:

- First it's important to differentiate between zero offset adjustment which many vendors and users describe as 'calibration' and actual slope calibration of power meters. They're both important for accuracy but they're very different issues.

- The zero offset represents the unloaded strain gauge bias point when no torque is applied to the torque measuring element of the power meter (e.g. crank spider in crank based PMs, hub torque tube in PT hubs). The strain gauges are relatively simple devices that are basically mechanically variable resistors that change value as the underlying mechanical elements they're bonded to are ever so slightly stretched or compressed. But in addition to changes due actual applied torque the underlying mechanical elements expand and or contract due to temperature changes so the resistance of the bonded strain gauges also varies with temperature changes. In well designed power meters the impact of thermal stress on things like the crank spider and chainrings is minimized by configuring the individual strain gauge elements in bridged configurations where each measurement gauge is paired with a matched gauge. The output of each bridged pair is less sensitive to temperature changes which translates to less zero offset drift due to temperature changes. But there is always some thermal impact so manual and automated torque zeroing approaches are necessary to ensure accuracy. In your slope intercept model the zero offset is the 'b' term. It's also worth noting that even cranksets with mounted rings are pretty complex mechanical structures when it comes to predicting thermal expansion or contraction at the strain gauge points of the spider. So thermal drift is not always in the same direction and it depends on things like chainring material vs spider material and even the material used for chainring mounting bolts as well as things like uniform bolt tightness.

- Actual 'calibration' of a power meter sets the slope or 'm' term in your slope intercept equation. This requires at least two measurements an more if the device is not actually linear across the range of interest but then it could not really be accurately calibrated as the system relies on a device linearity assumption which generally holds for a properly operating PM. Slope calibration is generally performed by taking the unloaded bias point as one measurement and then applying a known torque by hanging a known weight off of the cranks as the second measurement. The resulting slope is simply change in output reading divided by change in applied torque (which is just applied torque if the zero point is used as the first measurement point). Slope can change in some crank based PM designs by changing brand, style or even size of the chainrings. Apparently Quarq is somewhat sensitive to this but IME SRMs less so. It's always a good idea to perform a loaded torque test to measure slope after removing or replacing chain rings but for most of the newer SRM systems I've seen the slope doesn't change a lot with chainring changes but it definitely can after something like a battery change.

- I haven't worked with the Power2Max meters, I understand they have some drift issues and rely on an auto zeroing algorithm to periodically zero out the bias point. But this algorithm only kicks in when the rider coasts which can be an issue during things like trainer workouts, long sustained hill climbs, 20 to 60 minute TTs or sustained intervals or other times when coasting doesn't happen much. Most PMs support some method of manual torque zeroing but it's not clear that they do. I have heard that recent firmware upgrades to their meters have reduced the underlying drift issue but some folks still seem to have concerns about drift and zeroing that drift but I have no first hand experience with that meter. But as mentioned above that isn't really the same as calibration. To support calibration the meter needs some method of displaying crank torque when the cranks are not moving in addition to reporting power when the cranks are in normal use. If their meter supports this then anyone can measure the slope of the meter and field validate accuracy. Whether that slope can then be field adjusted as it can for Quarq or SRM meters or has to be sent to a repair center for actual changes to the programmed slope like PT hubs is another design question which I don't know the answer to.

- In terms of circuitry, designs differ but basically for each resistive strain gauge element included in the design (both torque measuring and bridged reference gauges) there is analog circuitry to actually take the reading (think ohm meter or constant current source and voltmeter), an A/D to digitize the strain gauge output, digital signal processing to do things like average the individual measurements around the cycle, cadence measurement switch and sensing circuitry, and then circuitry to encode and transmit the signal (even the wired SRMs encode the signal and transmit via the inductive loop to the wiring harness). The analog portions of this circuitry have to be designed to be stable and accurate across a range of real world environmental conditions but in practice this is easier to control than mechanical drift issues described above.

- In terms of elliptical chainrings and crank based PMs, the issue is that current crank based PM designs use a single (or in the case of track SRM, two) reed switch to measure cadence. Then there is an assumption made that crank angular velocity is constant for the complete revolution of the cranks. For round rings this is a decent assumption but for elliptical rings this assumption breaks down. So in typical modern elliptical rings crank angular velocity is above the cycle average as the cranks pass through the top and bottom of the pedal stroke during a time where little to no crank torque is being applied. But the crank angular velocity is below the cycle average during the power phase of the pedal stroke where maximal crank torque is being applied. Since power = torque*angular velocity this means power is over reported during the actual portion of the pedal stroke where power is being generated. Sure the power is being 'under reported' during the top and bottom of the pedal stroke when the angular velocity is below the cycle mean but since there's little to no torque being applied through these dead spots in the cycle the result is the per complete pedal cycle power reported is inflated. It all comes down to the constant angular velocity assumption and once per pedal cycle average torque reporting. This is a crank based PM issue and elliptical rings don't inflate power measured on hub based PMs.

Hope that helps,
-Dave

Torque is normally sampled between 60 and 200Hz. As Jeff mentioned it's the angular velocity that is sampled only once per crank/wheel revolution.

Power2Max is similar to Powertap in that there is no ability to adjust for a change in the slope. You can check it on a powertap but if it is out of spec you have to send it back.

I would love to see a test comparing power readings using non-circular (preferably osymetric) and circular chainrings fitted on a crank-based pm vs. power tap. I've been looking but haven't been able to find anything online. If anyone has seen this test, please post link. I think that test would answer a lot of questions about non-circular chainrings.

Get in touch with Tom Anhalt. He's done this test and published the results.

Tom A. on this forum has done that. IIRC he found 3-4% inflation with O.symetic rings compared to round ones on his Quarq. You can probably find it if you search or else I'm sure he will share his results if you ping him.

Tom's tests:
http://bikeblather.blogspot.com/...ose-funky-rings.html

Yep. That's what I was looking for. Thanks, everyone.

So, I went ahead and gave the Osymetrics a go. And I'll admit that it was mostly because Chris Froome rocked them like a superstar at the TDF. Now, like a lot of people on this forum, I've been riding for a while and have been a cat 1 for a couple decades (yeah, I know, I never went pro because I'm a weenie). I rode them for a few weeks and here is what I experienced:

1. Everyone says that you'll get used to them within five minutes, or, at the most, a couple rides. That was not the case for me. I expected them to feel awkward at the beginning and they were. While that feeling went away for the most part, they never felt completely normal to me. They never felt like I was pedaling in a smooth circle, no matter what the intensity was.

2. You have to get used to the noise in the little ring. Twice a revolution, the chain will slap onto the flatter, smaller portion of the 'oval'. It didn't bother me so much, but, I can't speak for those around me.

3. These rings were the newest version of the osymetrics and came with shift ramps and pins. I will say this, I was using a 52x38 with an older version SRAM Red steel cage front der and the shifting was awesome. They shifted better than with my circular rings.

4. I did notice slightly inflated power numbers as described in previous posts. I am using an s-works quarq crank based pm.

5. A lot of people said that they like the non-circular rings because it is easier on their joints as well as helps them get over injury. I did not find this to be the case. I felt like the peak load on my knees was higher, and I started to notice a bit of knee pain. The cranks spin quickly over the dead spot and then ramp you right into the power section of the ring. Hitting that harder section of the ring with each leg rev was noticed by my knees. Also, I felt like my body employed some new muscles with the new setup. I had some leg soreness for a bit.

6. I sent an email to Thomas Craven asking why Osymetric didn't allow a tweaking of the major axis in reference to top dead center like some non-circular chainrings. He responded that shifting the relationship will make you weaker. I sent him a message back asking if he had any data to show or if I was just supposed to take his word for it. He never replied.

After all that, I went back to my circular rings (with hesitation. Many say that once you go oval you don't go back to round). I am very glad that I did go back to round. I feel like my smooth spin is back and I feel strong. It was a fun and slightly expensive experiment, but, whatever. I was riding my bike the whole time and there's little in the world that compares with that.

Stumbled on to this post, doe this help?

http://www.me.utexas.edu/~neptune/Papers/job41(7).pdf

---

Thomas Craven
Osymetrican
Owner/Evangelist
OsymetricUSA.com
info@osymetricusa.com

I ride with that guy, his sprint is insane. Set a 30 second power record just following in his draft when he attacked one day.

He had round rings that day too =)

---

---

Kat Hunter reports on the San Dimas Stage Race from inside the GC winning team
Aeroweenie.com -Compendium of Aero Data and Knowledge
Freelance sports & outdoors writer Kathryn Hunter

Good article.

"The optimization successfully identified an elliptical
chainring with an eccentricity (i.e., the ratio of major to
minor axes lengths) of 1.29, which increased the average
crank power by 3.0% at 90 rpm"

Is that elliptical chainring with an eccentricity of 1.29 close to a Rotor Q-Ring?

Just remember, it is a theoretical optimization.

---

---

Kat Hunter reports on the San Dimas Stage Race from inside the GC winning team
Aeroweenie.com -Compendium of Aero Data and Knowledge
Freelance sports & outdoors writer Kathryn Hunter

I'll take an extra 10W!

Side note, has anyone experimented with just replacing the big ring with a non-round ring? I don't use the little ring often enough to justify spending $100+ to put an ovalized ring on it, but it'd still be nice to know whether the two will play nice or if I may as well just make my bike a 1x10 at that point...

---

__________________________

I tweet!

Doval rings are much cheaper:

Love you post real_time,

Had the exact same experience you had with the Osymetric, definitely on my TT bike (with triathlete non UCI position) the chanrings were getting too big too early and the was putting pressure on my knee. I still decided to race 70.3 UK in June with the setup just to see how it would go in race conditions and during 90km (with cold and rain so maybe I didn't feel my body much anyway) my knees took it ok but then when I started the run I could really feel some pain from "over working" the knee cap.

Then I decided to give the Rotor a go, first Qrings, then XL. Didn't like the XL, rode them once only, it was not a matter of getting used to them, the shape clearly wouldn't work for me and I didn't enjoy riding them. The standard Qrings worked quite well, though being only slightly ovalized they feel comfortable BUT not sure there is anything to gainwith such a small difference in shape compare to round.

So just like you I'm back to round and happy, I sold everything except for the standard Qrings which are the only one I may want to try again at some point, I think I could even switch back and force between Qrings and round rings as the difference in feel and muscles work seem minimal. I wouldn't do it though (that would be a waste of time) but that's to say that I'm confident I can ride both safely with the same power. As for the other two with more ovalization, I'm glad I try but they were not for me, maybe what Rotor says about the XL being for stronger animal with more fast twitch fibers is true ???

They look cheaper...and they look to be sort of oval. The osymetric ring has a dual cam design hence the even number of teeth. Where does the extra tooth go on the Rotor and the Duvals?

---

Thomas Craven
Osymetrican
Owner/Evangelist
OsymetricUSA.com
info@osymetricusa.com

tcbiker,

Have you ever considered making two versions of Osymetric rings :
1) actual version, standard road/MTB position with riders "seating behind the bottom bracket".
2) triathlete version with riders seating "on top of in front of bottom bracket", that would probably take rotating the mounting holes a few degrees from actual position in order to get the chainring at max radius around horizontal position of crank arm instead of earlier --> would be a huge difference for triathlete. I was almost liking the Osymetric rings on my TT triathlon bike but my knees DID NOT... maybe rotating by a few degrees for triathlete would be all it would take for BOTH my and my knees liking them ;-) .

Thanks for your thoughts on this.

The problem with non-round chainrings is not measuring torque, it's measuring rotational speed. Power = torque * rotational speed. Most (all?) crank-based powermeters assume constant rotational speed over one revolution (or half a revolution). This is not true with non-round chainrings (also not completely true with round chainrings, but truer).

If rotational speed (cadence) was measured accurately every time torque is measured, then power would be accurate regardless of chainring shape.

Looks like Van Lierde uses the Osymetric USA rings and won Kona on them.

---

Thomas Craven
Osymetrican
Owner/Evangelist
OsymetricUSA.com
info@osymetricusa.com

Never said there are no good athletes on Osymetric. It would be stupid to say that when you consider Froome, Sancher, Van Lierde, etc.

What I said is that orientation is more road oriented than triathlon, and if you look at Van Lierde position you can see his seat is quite on the rear for a triathlete... is it because it's his favorite position and the Osymetric rings happen to well oriented for his "almost inside UCI rule" position... or is it that he adapted his position to feel better with the Osymetric rings... THAT... I don't know ;-) !

Congrats on one more nice win on your rings by the way ;-) .