Power Meter Accuracy

Just a general question about the accuracy of Power Meters. When the manufacturers quote +/-2% do they mean each individual meter can vary by that amount each individual ride or that each meter they produce will be consistent but they may vary by that amount from unit to unit?

1 Like

As I understand it, it’s the unit might be off by +/- 2% from your true power so if you ride at 100 watts according to your power meter then your actual power could be between 98 and 102.

How much it is off could vary by ride, so on some rides, it might be only 1 watt off the next 1.5. However, it will still be between 98 and 102 for this example.

This is at least what I gather from all the DC rainmaker I read, and I could be completely off and as inaccurate as the zPower readings on Zwift

4 Likes

@tribuddha nailed it, that is correct.

1 Like

You also need to consider drivetrain losses and, if you are using a single sided PM, leg imbalance.

Also, regardless of what powermeter you use, and irrespective of it’s quoted accuracy, as soon as you mention you’re using it on Zwift everyone else seems to believe it’ll over-read by at least 50W.

4 Likes

Thanks everyone. I’m trying to match up the readings on my 2 single sided power meters (Stages & Assioma) with my new Stac Halcyon trainer using the correction factors on the Assioma’s and the Stac and it’s a bit like trying to hit the bulls eye in a dart board with me moving up and down and the board going from side to side. On the question of accuracy then with and FTP of 250W and a PM of +- 2% you would need a bump greater than 10W to know you had an increase.

Reviews of the Assioma dual say they’re very accurate.

I’m not sure that’s how it works. Strain gauges themselves are consistant. I would tend to think that a particular meter reads on average 1-2% (or more) high or low but not both. Curious what he experts think about that.

Any type of measuring equipment has four fundamental properties:

  • Accuracy
  • Precision
  • Resolution
  • Sensitivity

Starting at the bottom of the list:

  • Sensitivity is the smallest change that the device can detect. With a strain gauge this will be very small.

  • Resolution is the smallest increment that the device can output. Typically we see readings that are to the nearest Watt.

  • Precision is how repeatable the results are

  • Accuracy is how close each result is to the benchmark

The problem we have is that powermeter manufacturers only quote accuracy.

As an example let’s take your rider at 250 Watts with a powermeter that has an accuracy of ±2%:

2% of 250 Watts = 5 Watts, giving a total range of accuracy of 10 Watts whist measuring a known input - readings of 245 to 255 Watts would be within spec.

However, it is very unlikely that the precision is as poor as that. Let’s imagine that it’s ±0.4%; a range of precision of 2 watts.

Given the two parameters above, all readings need to be within the 245 to 255 Watt range but should never be more than 2 watts different: 245 to 247 would be ok, as would 253 to 255.

The power meter is unlikely to have a precision equal to the accuracy so from one reading to the next the variation will be very small.

Mike

1 Like

On the topic of accuracy, how do Garmin, Favero, Stages etc validate that the power they report is the actual power, regardless of their ± specifications. I’ve never seen any of them show how they actually do it other than the math. Do they ever validate it in any kind of physical test against a known calibrated high precision tester?

That’s a very good question.

Mike

Most of the stuff I’ve seen, even from well-respected reviewers, are comparison between power meter devices, be it on the bike or smart trainer, so everything is relative. I never seen any kind of dynamometer reports that shows actual power output vs the device under test for an absolute accuracy.

1 Like

This may well be worth a read:

https://www.thieme-connect.de/products/ejournals/html/10.1055/s-0043-102945

Mike

Everyone claims +/- 2%, but not all %s are equal…

https://www.researchgate.net/publication/291351824_Agreement_between_Powertap_Quarq_and_Stages_power_meters_for_cross-country_mountain_biking

https://www.researchgate.net/publication/8474185_Accuracy_of_SRM_and_Power_Tap_Power_Monitoring_Systems_for_Bicycling

Because I lack all sense, I have three Powertap sets – a fixed hub that I built a wheel for on my trainer beast, and a set on each road bike. They all read a little different: the one on my EVO reads a decent 10w higher (and some days, 15w – those tend to screw up the MMP curve) than the one on my CAAD12. The trainer one is in the middle.

I don’t mess with trying to tweak software with the data. I train by feel and look at watts descriptively, not prescriptively, and figure that if I ride all three a decent bunch the charts will work themselves out – and anyway, the important stuff is happening in the muscle cells, not in the pixels.

That’s a good resource. Just because I haven’t seen it before doesn’t mean it doesn’t exist (I should have known better). It looks like it’ll require full attention to see if it covers what I am after so I’ll read it later.

I’ve seen as much as 30w difference between PMs of the same brand, for the same rider, following the same calibrations.

“Accuracy” gets complicated once you get beyond the basic advertising claim.

This!!

I haven’t read it myself yet.

Mike

Interesting read but really does not address the measured accuracy of power meters. The test methodology, in my opinion, has to many variables to be precise and was validating against a mathematical model of power. I’m think more along the lines of what bicycle rolling resistance does for tires.

Conceptually a test jig that consists of a crank on a bottom bracket with a rotating crankarm to drive it through a pedal interface and either an input or output power measuring dynamometer to compare the power meter’s reported power. That way the input force/rotation should be dynamic, consistent, and precise and dynamometer results should be accurate and precise.

Simple in my head anyways…

2 Likes