And people rejoiced...

As long as the Garmin Forerunner 310xt has been out, it has received praise and criticism across the board.  My main selling point to upgrade from the Forerunner 305 was the ability to receive ANT+ signals and thus act as a computer for my power meter of choice.  When I broke into the power market, I started with a HED Jet 60 rim laced into a Powertap SL 2.4 that had been upgraded to transmit via ANT+ protocol.  The Garmin 310xt picked up signal and away I went.  The ability to act as a head unit is what draws many people to the 310xt.  At the time, Garmin only had the Edge 705 out which would also read ANT+ signals and had the ability to do 3 and 30 second power averaging, a staple in using power.  Since then Garmin has released the Edge 500 and now the Edge 800 which also do rolling power averages.  Saris has also released the Joule 2.0 which does the rolling averages as well, thus putting the 310xt at the bottom of the list for ANT+ receivers and useability.

Why would one want rolling power averages?  Well if you are reading this, I am sure many understand, but for those who do not, let us dive into it a bit deeper.  Each data point records at a specific point in time, and that point has a reading from the power meter.  Most people are not 100% efficient in their pedaling, just as in lifting weights, they may have a dominate side.  Think about doing a bench press.  How many times do people push the bar up slightly higher on one side, and then use that as a pivot point as they push the other side up?  This would indicate that they have a dominate side.   Let’s take that and apply it to biking.  Since you have two legs and two sides of the crank, each leg can be considered individually.  A truly inefficient biker will use the pushing motion(down in direction) to generate power peaks and when they are in the recovery (upwards direction) they are not applying any force upwards. 

 When one leg is going down, the other is going up.  So overlaying the two waves that each pedal sees would result in something like this, taken from Metrigear blog as an example:

Since each leg generates a spike, if one leg is stronger or applying more force, you can see the black spike in the middle of the graph applying a higher force than the other leg ever does.

If your sample comes at one of these moments, you will see a higher number reported, but if the sample is taken somewhere towards the bottom of your weak leg pushing down, you may see a drastically lower number.  This results in your computer screen showing very different numbers at each interval.

Is the problem becoming clear?  If you are doing an interval set, and look down to see only 180 watts, the gut instinct is to get your behind in gear and push harder, often overshooting the mark. 

The 3 or 30 second averaging allows the computer to take samples across the selected interval of time and compute the average. 

Second	Current Power	Average Power (3 Second)
1	200
2	200
3	200	200 (200+200+200)/3
4	208	202.67 (200+200+208)/3
5	209	205.67 (200+208+209)/3
6	201	206 (208+209+201)/3
7	199	203 (209+201+201)/3

You can see the smoothing that is done by taking the average, and it would be seen even more if we did the 30 second average.  The averages help to eliminate spikes or dips in power if they only exist for a second or two since the majority of the points are closer to the desired range. 

For intervals, I still like to see actual power especially if they are shorter intervals.