Power Data Comparison Sheet

Based on the continued questions that we see from people comparing one power device to another, I decided to try and make something useful. Here is a very rough start at a Google Sheet to compare power values.

See the [Power Tolerances] tab in my google sheet.

  • It features 2 areas, with 2 fields for data entry (both in bright yellow)
    1. Device Power Value
    2. Device Power Tolerance

You enter one data value and the tolerance related to the measuring device. The manufacturers of smart trainers and power meters list a tolerance, and I plan to populate a list for quick reference. This is the other value. Then do a second one and you can see the variation between them, with the tolerance in effect.

The current calculations take the Power Value and Tolerance, and calculate the Maximum, Middle, and Minimum values. It reports 3 sections as if the data value is either the Max, Mid, or Min with respect to the total range.

I envision taking this and comparing the calculated values to identify if the gaps between data from two devices meet within some acceptable range, or if they are separated. Long term, adding in considerations for things like drivetrain loss could be good.

I decided to start with this and see if people think it has some value for continued refinement?

1 Like

In short, I am confused. What are you trying to show?

Initially, I thought you were trying to show that given the recorded wattage from two different power meters for the same value of time, whether or not the values were within the realm of acceptability based on the +/-% accuracy.

For example: Kickr 200W 3% accuracy vs Quarq 185W 1.5% accuracy. This yields a +/-4.5% accuracy swing. Note that the percentage diff between 200W and 185W is 7.8%

Maybe it’s the presentation format that is confusing me or have I misinterpreted what you are trying to show🤔

p.s… Looking at your Google spreadsheets, someone has been a busy bee :smile::+1:

1 Like

Yeah, I am trying to have a simple way to input data from 2 sources and maybe see if the combination of their tolerance ranges overlap, or if they don’t.

I don’t have that level of intelligence setup yet.

All I have now is a known power value and 3 possible ranges of power for that data point given a claimed tolerance range.

Why 3 ranges per power?

  • When we look at a power value from a device, we have no idea if that value is is at the max, mid or min of it’s tolerance range.
  • Without specific validation to “prove” any value, any 3 of those could be true (or really, any value between the Max and Min.
  • I had just max and min originally, but added mid because I thought it might be informative.

I am open to reformatting and this is already on the 3rd iteration. I don’t know the best way to show the data in a non-confusing way yet, even for a single value.

The whole thing gets messier when you consider how to combine and evaluate both values. It may be that a graphical method is useful for this.

  • This is the kind of example I want to evaluate.
  • The challenge is that it’s not a simple addition of the % tolerances to know if the split is good or bad (within acceptable range combinations).
  • Each tolerance must be applied to the data from the related device, then compare the results.

I will continue to refine the info to see if I can make something helpful. But it isn’t there yet.

1 Like

I too was thinking of how to best display this information, but have come up with nothing.

But it is this simple. You are comparing the two values from two different meters, neither of which is a known true value. To do so we use the percentage difference between the two numbers and see if that %diff falls within the acceptable range.

I believe you are saying to do the following: Using 200W 3% and 185W 1.5%. Let’s say the real value is between the two, one meter reading high and the other low. So, 200W is +3% (your max) and 185W is -1.5% (your min). This yields 194W and 188W. The % diff here is 3.14% which would be within spec.

The problem with this is you are fudging the numbers before you start. You do not know if a meter is reading higher/lower and cannot arbitrarily make the assumption. Both values are experimental values. There is no known true value in this scenario.

1 Like

Just going to leave this here:


So, Chad, the two power meters read 200+/-3% and 185 +/-1.5%. Thus the “true” value for PM1 lies in the range (194, 206). The “true” value for PM2 lies in the range (182.23, 187.78).

Estimate Accuracy high low
PM1 200 3 206.00 194.00
PM2 185 1.5 187.78 182.23

Simply plug the estimates and accuracy into the spreadsheet and bingo. In this case the power meters are reading significantly [in a non-statistical sense] differently.

1 Like

Right, and I am not making an assumption. That’s the reason I have the Max/Min version with the known data value at the top or bottom of the tolerance range.

I am explaining my goal terribly, but I was looking for a way to see if it is “possible” that two different values from two power data sources “could” overlap. Effectively trying to see if the data and related tolerance ranges from each “match up” enough to be considered “good”.

We see posts frequently from people along the lines of “my trainer say X watts, but my power meter says Y watts… which is right?”

I hoped to make a quick analysis tool to take those two data points and the know devices, and see if there is “common ground” between the tolerances. We know the two values don’t match, but are the differences “in tolerance and possibly shared” was my goal.

I know that we don’t know the disposition of the value and where it really falls within the tolerance range. As such, I made the Max/Min as a way to see the absolute range in either condition. Then I added the Mid out of curiosity, but I think it is more confusing than helpful.

Maybe this is all unnecessary and not worth the time?

1 Like

Yeah, that was how I started. I had the known value in the middle of the range and just a single evaluation.

Then I got off to a different train of thought (max min), that is likely wrong.

I need to review and digest all of the comments here and revisit my sheet. I think I have some simplification that can be done.

That’s not how this works: The ranges need to overlap for both to be within spec.



For the record. The data I have in the sheet right now is junk. It am just using it to evaluate my equations and validate I am getting the right calculations.

I may grab some real examples from posts in the forum, once I get something that makes any sense at all :stuck_out_tongue:

The simple answer is you can’t answer the “which is right” question.

What you can do is:

  1. Use bike’s power meter both inside and outside, for one source of “truth” about power. Just ignore trainer and move on.
  2. Measure the difference between two power sources, assume the difference stays constant (almost certainly a false assumption), and then use a power offset as a correction factor.

For someone with one bike, or a power meter that moves between bikes (e.g. crank or pedal or hub), item 1 is the only way to proceed IMHO. Therefore I can’t see the point of having an analysis tool that shows what you are proposing.


I think the “right” way to look at this is how I started. Take the reported power value and simply apply the device tolerance.

The whole Max/Min thing seems like a mistake now that I look at it. So, my “Mid” value is probably the only value and calculation to keep, for the overall evaluation.

  • I totally agree. “Right” is not my term, but one I paraphrased from the many posts with these types of comments.
  • As a mechanical designer, and one who lives with tolerance issues on a daily basis, I know that everything is relative. I am just trying to make a tool that makes some level of evaluation possible, but not claiming it is concrete or absolute in any way.

not sure that will help non-engineers. If you have two sources of truth, and they don’t agree, about the only thing you can say is “pick one and always use it.” And that gets to my point:

  1. For one bike / one power meter, or multiple bikes and one power meter, use bike’s power meter both inside and outside. Just ignore trainer’s power.
  2. For multiple bikes and multiple power meters, measure the difference between two power sources, assume the difference stays constant (almost certainly a false assumption), and then use a power offset as a correction factor.
1 Like

Indeed. If nothing else, I plan to use this myself when power questions come up on occasion, because I think there is a distinct lack of knowledge about power meter & smart trainer tolerance ranges for power data.

It might help show the simple range of power that “could be real” based on the reported value and the device tolerance. So it might be useful from a purely information and example standpoint.

FWIW as you know this gets into an accuracy / precision discussion, which is often lost on people.


I was thinking along the same lines.


Yup, Pandora’s box.

I just got tired of running the basic tolerance calculation manually every time I hopped into one of these discussions. So I decided to make a quick resource for the calculation.

Beyond that, I started wondering about other ways to make a sheet like this useful. It may well remain something for myself, but I wanted to share and get some feedback. And the discussion around this has helped me refine it and my thoughts about how it might work, so I am happy :smiley:

1 Like

Not at all. I already do this in a notebook with my calculator.

I know none of the figures are accurate as both sources will fluctuate constantly within their +/- values if working correctly. If they never overlap…
It’s total BS to say a power meter, or trainer, is OK because it is consistently out of whack. Screw that. I’d like to get what I paid a premium for.

1 Like