Search

How 'good' is your power meter?

I was determined not to write a CV-19 blog post. However, with everyone spending more and more time on their indoor trainer I am increasingly getting asked (or told!) by athletes how the power numbers from the power meter on their bike correspond to those they are seeing on their smart trainer.


Now, I appreciate that you can get your smart trainer (in most cases) to read power from the power meter attached to your bike - however, that is only useful when you are riding them together. What if you are lucky enough to have 2 bikes and wanted to compare the power from one power meter with your smart trainer?


I wanted to share was a great way to compare power output values between a smart trainer and a power meter - or multiple power meters if you are lucky enough to have multiple bikes with different power meters.


It's all just an estimation:


The first thing I want to stress is that your power meter - whatever brand, model, shape or size it is is only ever estimating power! In fact, whenever we 'measure' something we are actually just asking a piece of equipment to provide an estimate.


Take for example the altimeter on your bike computer! It 'measures' altitude in one of two ways - either using barometric pressure (pressure reduces as you go higher) OR using location and then cross referencing that with the altitude on a (digital) map. Now air pressure can be affected by weather, so if the weather changes during the day so does the air pressure at a given altitude and that can effect the altitude 'measurement'. Likewise, a GPS location is only accurate to ~15m. Stand on the edge of a cliff and that 15m can have quite the effect on your altitude! So as you can see - the altitude measurement is actually only an estimate.


In reality, what your bike computer does, is it combines the two measures to enable a more accurate and reliable estimate (measurement). Later on we will see how we can use multiple power meters to do the same. However, first to understand how good an estimate is, we need to talk about what those two different concepts: accuracy, and reliability, actually mean. To illustrate this we will use the example of throwing darts and aiming to hit the bullseye.


Accuracy:



Accuracy is how close to the bullseye the darts land. An accurate player will be able to get every dart fairly close to the bull, however the darts may be scattered around on all sides of the target. Accuracy is therefore how large the difference is between the altitude on your bike computer and the '1909m' sign at the top of Mount Ventoux! One time it might read 1919m and the next time it might read 1899m!


Reliability:


Reliability is where the darts, although not very close to the bullseye, are very close to one another. Essentially you can rely on this particular dart player to get each dart they throw close to the last few darts...however that might be nowhere near the bullseye! Reliability is how much variation you would see between your bike computer and the '1909m' sign at the top of Mount Ventoux if you cycled up there 10 times! A very reliable bike computer might read 1898m every time!


Obviously in a perfect world our imaginary dart player / bike computer would be both accurate and reliable....


Accurate and Reliable:


This is what would happen if your bike computer was to read '1909m' each and every one of the 10 times you visited the top! It would be accurate and reliable!


What about measuring power:


Let's equate these concepts to power meters....


To do this we are going to compare 3 different fictional power meters;


1, A smart trainer

2, A left hand crank based power meter

3, A pedal spindle based power meter.


We are going to compare the power figures for these 3 power meters for a period of 30s while our fictional rider pedals but doesn't necessarily hold a constant power.


We are going to record the power value every second (1Hz) for all 3 power meters during the same effort! Here is the data from the 3 power meters:





The average power for the power meters was as follows:


Smart Trainer - 205w

Left hand crank based - 200w

Pedal spindle based 190w


Since our fictional rider was just riding along and not necessarily holding a fixed power we have no way of actually knowing what power they were putting out either at any given second, or what that average power was for this effort - we only have 3 different estimations from 3 different power meters.


Essentially we have 3 different dart players all aiming for the same point on a dart board but we can't know exactly which point they were aiming for! All we can do is make an educated guess based on where the darts landed.




So how then can we possibly know which power meter (or dart player) is the most accurate and which power meter (or dart player) is the most reliable?


Since this is a cycling blog lets focus on the power meters from now on...


To calculate the power we can use something called a Bland - Altman plot! (sorry this is where the stats comes in!).


A Bland - Altman plot is a graph that compares the average power output between the two power meters for any given second with the difference between the two power meters for the same second.


Lets start by comparing the smart trainer and the left hand crank:


The solid yellow line shows the average difference between the two - 5w. We already calculated this from the difference in average power. However, the important lines are the dotted yellow lines. These show what we call 95% confidence intervals - I won't bore you with how they are calculated here. But 95% confidence intervals allow us to say what, 95 times out of 100 (ie we can be 95% confident), what the average of the 2 power meters will be.


Here you can see the 95% confidence intervals are -7w and +17w. So that tells us at 200w, 95% of the time we will get an average value of between 193 and 217w between the left hand crank power meter and the smart trainer - thats a range of 24w.


What this tells us is the two power meters are, overall quite accurate to one another (only a 5w difference on average), but they are not reliable with one another - a range of 24w!



Comparing the left hand crank and the pedal spindle power metes we see an average difference of 10w. This time the 95% confidence intervals are -1.5w and +21.5w. This is a range of 23w! So only slightly better than above and still not great!


This means that at 200w the average between the left hand crank and the pedal spindle will, 95% of the time, be measured between 198.5w and 221.5w.


What this tells us is that the two power meters are not that accurate to one another, 10w difference on average, and are also not that reliable to one another - a 23w range!

Now we come to the comparison between the smart trainer and the pedal spindle. Here we can see a big difference average between the two of 15w. However the 95% confidence intervals are +12w - +17w! That is a range of only 5w! So for the same 200w, 95% of the time, the average between the smart trainer and the pedal spindle power meter is going to be between 212-217w.


What this tells us is that the two power meters are not accurate to one another - an average difference of 15w. But that they are very reliable to one another with a range of only 5w!



How does this information help us?


When it comes to power meters reliability is king. It doesn't actually matter if the 200w we measure today is in reality 210w or 196w. What matters is that the 200w you measure today is exactly the same as the 200w tomorrow.


The reason for this is because we use training zones. When you go out training and your coach says ride at 200w we need to know that 200w today is the same as 200w tomorrow. It's no good if the 200w today is in reality 197w and the 223w tomorrow!


Because reliability is king we cannot assess power meters purely on the average difference between them - as this doesn't tell us how much variation we are likely to see. In the example above, the two power meters that were the most accurate to one another - the smart trainer and the left hand crank power meter were also the two power meters with the lowest reliability to one another.

Therefore, in the example above it is quite clear that for training purposes the smart trainer and the left pedal spindle power meters would provide the best option as they are the most reliable to one another - despite not being particularly accurate to one another.


Using this in training


If you were to use the smart trainer inside and the pedal spindle power meter outside any your power would be on average 15w higher on the smart trainer. However, so long as you took that difference into account you could quite happily use the same training zones on both power meters and only see a small +/- ~3w variation from one session to the next.


However if you were to use the left hand crank based power meter and the smart trainer together, although the overall difference might be smaller (only 5w) from one effort to another might expected at different of +/- ~12w swing. This means that one day your left hand crank power meter might read 17w higher than the smart trainer and the next it might read 7w lower!! Thats quite a range to be dealing with and could impact the quality of your training.


When it comes to training reliability is king! Therefore, make sure you know not only how ACCURATE your power meters are to one another but also how RELIABLE they are to one another when comparing values!

71 views

Home          Profiling          Performance Plan          Performance

Screenshot 2020-01-03 at 11.22.07.png