Power Meters vs Power Estimators aka The Case Against Single-Sided "PMs"

Triathlon Forum
1

(Separating this topic from the Shimano 4-bolt PM inaccuracy thread)

I think it’s appropriate to draw the line between a power meter and power estimator based on the data received by the device.

If the device receives all the relevant data (i.e. data from both the right and left leg), then it’s a power meter.

If the device receives only half the relevant data (i.e. data from just the left leg), then it’s a power estimator. It can “measure” the power produced by your left leg, but then the total power it reports is estimated (as it doubles the left leg data).

Now sure some power meters are more accurate and/or precise than others, because they are better designed and engineered to measure and report accurate power, whether from better data capture or processing. But they are still measurement, rather than estimating devices.

Consider a tape measure. I think we would all agree it is a measurement device, not an estimating device. But some tape measures will be better than others at providing accurate measurement. Sure, these days a tape measure is so easy to produce that the differences between them may only be measurable in micrometers, but the fact that there is some deviation in accuracy doesn’t make them estimators rather than measurement devices. In the case of a tape measure, where you are measuring length, you just need two data (start point and end point) and you can use your measurement device (your tape measure) to calculate distance.

Now let’s think about measuring our height with a tape measure. If we measure from the bottom of the foot (start point) to the top of the head (end point), we’re providing our measurement device with a full data set and it can then provide us with a measurement (not an estimate) of height.

But if we take away one of our data, our tape measure become an estimating device rather than a measurement device. Let’s suppose we are allowed only to measure from the bottom of the foot (start point) to some intermediate point on the body (e.g. hip bone) and then apply some multiple to that measurement to calculate height. In this case, we are no longer measuring height; instead we are estimating it. In this example, our tape measure is perfectly capable of acting as a measurement device, but we are providing it with incomplete data and as a result the best that it can provide us is an estimate of height.

Same with “power meters” - if your device is working from a complete data set, then it can be called a power meter (again, with some being more accurate/precise in measurement than others). But if your device is operating from an incomplete data set and then “calculating” your power from that incomplete data, then it’s a power estimator.

2

If you find one side is significantly putting out more power what would you do?

3

They are power meters with a lower precision. A total power PM may hit its published ±1% precision. A single sided PM is probably more like ±6% to ±10%. It’s not estimating; it’s merely less precise.

4

Consult a doctor, physical therapist and/or bike fitter to attempt to diagnose the issue.

5

Consult a doctor, physical therapist and/or bike fitter to attempt to diagnose the issue.

Interesting

I posed the same question to Dr Coggins
.

6

They are power meters with a lower precision. A total power PM may hit its published ±1% precision. A single sided PM is probably more like ±6% to ±10%. It’s not estimating; it’s merely less precise.

Meaning that the average value provided over the course of a workout approaches the actual power produced?

7

They are power meters with a lower precision. A total power PM may hit its published ±1% precision. A single sided PM is probably more like ±6% to ±10%. It’s not estimating; it’s merely less precise.

I think you’re conflating results with process/methodology. An estimating device can indeed provide a more accurate result than a measurement device, but it doesn’t mean it’s not estimating.

In the end though, it’s like saying LeBron with one arm tied behind is back is better at basketball than most human beings. Which while true, doesn’t mean that a two-handed LeBron isn’t better than a one-handed LeBron.

8

“If the device receives all the relevant data (i.e. data from both the right and left leg), then it’s a power meter.”

No, most definitely not. See iBike (or whatever they call it now).

9

If you find one side is significantly putting out more power what would you do?

Irrelevant. The OP isn’t saying you need to know your left/right balance, they are saying you need to know your total power.

10

If you find one side is significantly putting out more power what would you do?

The issue isn’t with whether the data is actionable, it’s whether it’s accurate or more importantly consistent. If your L:R balance was consistently 45:55, then a one-sided PM would be inaccurate but consistent. 300W might actually be 270W or 330W on a different PM, but at least 300W one day is the same as 300W the next day, so the data is usable - you can set target power and perform tests with it.

Problem is that in my experience L:R balance isn’t consistent. Mine varies from 45:55 to 51:49. There is some correlation with intensity and fatigue, but I only know that because I can measure it. If I had a single sided PM then the same effort might show as 300W or might show as 265W depending on whether I was having a 45:55 day or a 51:49 day.

I don’t care that much that on some days I have a 45:55 L:R balance (that’s not quite true, I’m doing some single leg exercises to see if I can get it closer to 50:50 more of the time, but it’s within normal range and nothing to really worry about) but a >10% variance in recorded power would basically render a PM worse than useless. I’d have no issue buying something like a hub-based PM which measures total power without distinguishing between L and R. But no way would I consider a single-sided one.

11

Meaning that the average value provided over the course of a workout approaches the actual power produced?Not necessarily. If your power balance averages to 0%, then it might be close to true total power. However, if your average power balance is not 0, then the average would be wrong by 2x that imbalance.
I think you’re conflating results with process/methodology.Not conflating, and the LeBron analogy is a non sequitur. A left side power meter is still a measuring device, it is just less precise than a total power measuring device. An estimating device would be something like Strava, which takes no power measurement, but estimates based on other data points.

12

(Separating this topic from the Shimano 4-bolt PM inaccuracy thread)

Now let’s think about measuring our height with a tape measure. If we measure from the bottom of the foot (start point) to the top of the head (end point), we’re providing our measurement device with a full data set and it can then provide us with a measurement (not an estimate) of height.

But if we take away one of our data, our tape measure become an estimating device rather than a measurement device. Let’s suppose we are allowed only to measure from the bottom of the foot (start point) to some intermediate point on the body (e.g. hip bone) and then apply some multiple to that measurement to calculate height. In this case, we are no longer measuring height; instead we are estimating it. In this example, our tape measure is perfectly capable of acting as a measurement device, but we are providing it with incomplete data and as a result the best that it can provide us is an estimate of height.

Same with “power meters” - if your device is working from a complete data set, then it can be called a power meter (again, with some being more accurate/precise in measurement than others). But if your device is operating from an incomplete data set and then “calculating” your power from that incomplete data, then it’s a power estimator.
So you know that your spine compresses during the day, is your height measured in the morning the correct one or the height measured last thing in the evening, or after a period of zero G a real problem if you are a astronaut? https://knowledgestew.com/...ing-and-shorter.html
A single sided powermeter is close enough for me and I have a powertap as well. If my left leg power goes up I am sure that my right leg is going in the same direction.

13

If you have to start a topic on ST to justify spending more money…that’s on you.

There’s the same mental gymnastics and justifications in the automotive “tuner” world centered around whether a “Mustang” or a “Dynajet” or a “Dynapack” is the BEST and why all the others are worthless trash.

End of the day, it’s the same conclusion…it’s all about consistent use of the same measurement device in the same conditions.

Your tape measure analogy…guess what…there are indirect measurement methods that are more accurate/better than breaking out a ruler. You really think that a modern surveyor is going to break out a 4-mile long tape measure? No. But it works, and is consistent.

Ya’ll need to get off this high horse. Graeme Obree had his indoor dumb trainer stationary bike setup consistent enough that it probably was for training purposes as good as any modern power meter.

But, by all means…hop on your Velotron and startup a 50 page topic on ST about crank length.

14

(Separating this topic from the Shimano 4-bolt PM inaccuracy thread)

Now let’s think about measuring our height with a tape measure. If we measure from the bottom of the foot (start point) to the top of the head (end point), we’re providing our measurement device with a full data set and it can then provide us with a measurement (not an estimate) of height.

But if we take away one of our data, our tape measure become an estimating device rather than a measurement device. Let’s suppose we are allowed only to measure from the bottom of the foot (start point) to some intermediate point on the body (e.g. hip bone) and then apply some multiple to that measurement to calculate height. In this case, we are no longer measuring height; instead we are estimating it. In this example, our tape measure is perfectly capable of acting as a measurement device, but we are providing it with incomplete data and as a result the best that it can provide us is an estimate of height.

Same with “power meters” - if your device is working from a complete data set, then it can be called a power meter (again, with some being more accurate/precise in measurement than others). But if your device is operating from an incomplete data set and then “calculating” your power from that incomplete data, then it’s a power estimator.
So you know that your spine compresses during the day, is your height measured in the morning the correct one or the height measured last thing in the evening, or after a period of zero a real problem if you are a astronaut? https://knowledgestew.com/2015/03/why-youre-taller-in-morning-and-shorter.html
A single sided powermeter is close enough for me and I have a powertap as well. If my left leg power goes up I am sure that my right leg is going in the same direction.

This is basically what Andrew Coggins has said

If I am Mia quoting him perhaps he can come in this thread and state it more accurately

15

so, the definition would change depending on usage, correct?
Since all relevant info changes.

Im pretty sure a good number of folks would say that in aero testing, anything not measuring at the hub, is estimating power to wheel, and so is an estimator.

Any power meter that has temperature calibration 100% has an estimator build it (its changing the response curve of strain guages based on temp)

Crank arm mounted ones that can be mounted to either carbon or steel , probably have some defaults built in.

What about the difference between lower accuracy strain guages… is +/- 4% (but dual leg) a power meter or estimater… what about if you are comparing it to one that has +/- 1%?

I get that I choose not to have a 1 sided pm, and never would take one… but… is this anything other then a flame? Do you just want to have a list of everyone on ST who has a 1 legged meter so you know when they complain about it?

16

All power meters are estimators. I thought all the histrionics over single-sided power meters was settled like 5 years ago.

17

Im pretty sure a good number of folks would say that in aero testing, anything not measuring at the hub, is estimating power to wheel, and so is an estimator.

Crank measures higher. They don’t estimate a lower value to align with a hub meter. They simply display the number that is there.

The aero testing part also would only matter if you cared about the ACTUAL CdA value and not the delta change in CdA between different runs.

If you want to know “my CdA is actually .218”, then yeah. If you want to know “my change in CdA from helmet A to helmet B was .00X”…then having the crank versus hub won’t matter since you’re dealing with a delta and not an absolute.

18

Whats up with one sided powermeters?http://www.isitt.org.uk/resources/Koppenberg+cyclist.JPG
.

19

If you find one side is significantly putting out more power what would you do?

The issue isn’t with whether the data is actionable, it’s whether it’s accurate or more importantly consistent. If your L:R balance was consistently 45:55, then a one-sided PM would be inaccurate but consistent. 300W might actually be 270W or 330W on a different PM, but at least 300W one day is the same as 300W the next day, so the data is usable - you can set target power and perform tests with it.

Problem is that in my experience L:R balance isn’t consistent. Mine varies from 45:55 to 51:49. There is some correlation with intensity and fatigue, but I only know that because I can measure it. If I had a single sided PM then the same effort might show as 300W or might show as 265W depending on whether I was having a 45:55 day or a 51:49 day.

I don’t care that much that on some days I have a 45:55 L:R balance (that’s not quite true, I’m doing some single leg exercises to see if I can get it closer to 50:50 more of the time, but it’s within normal range and nothing to really worry about) but a >10% variance in recorded power would basically render a PM worse than useless. I’d have no issue buying something like a hub-based PM which measures total power without distinguishing between L and R. But no way would I consider a single-sided one.

This is exactly my experience as well. Having been able to measure and monitor L/R balance for several years now, I know that it varies day by day and within a workout based on intensity and fatigue. Knowing what I know now about my L/R balance throughout a workout and day to day, I would never trust that a single-sided PM could produce accurate and repeatable power measurements for me. Others may be different, but would you really know without first riding for years with a power meter that measures L/R balance? No.

20

I have a pelvic instability and my L/R has been as been as 30/70 when its completely out of whack, and normally is 48/52 (rarely are people 50/50). Additionally, when people ride in z1 really one leg tends to ‘go on vacation’ so you can have a 40/60 that’s perfectly normal. L/R isn’t a critical stat to look at daily, but its nice to have just to make sure everything is working properly.

So depending on which leg you are drawing power from there are huge differences. Even if you don’t have any leg issues its still a 4-20% difference depending on the ride and is going to affect everything from race estimates to CTL. Its better than nothing, but you don’t know your actual output.