Cycling Peaks power doesn't match PT

I've been noticing that when I download my ride data the average power that shows up in Training Peaks WKO+ doesn't match what my PowerTap computer shows. I can't understand why they'd be different but they haven't matched up on any of the rides I've checked. Has anyone else noticed this and is there an explanation or it?

 

The typical answer is that you have the PT CPU set to ignore zero power data when it displays its average. That's a custom menu setting that you can change through the setup menus. If you ignore zeros and coast a bit during your ride the displayed average power will be higher than the actual average power that you'll see when you upload the data.

-Dave

 

I usually see 1 or maybe 2 watts different in the average (PT CPU is lower). Are you seeing a greater difference than that?

I'm not sure where the minor difference comes from, but I'd guess it's most likely between the way the averages are calculated and rounding. I'm pretty sure the software averages the discreet 1-sec recorded power values, while the CPU probably keeps a running total of energy expended and calculates the real-time average power from interval duration.

 

Yup, it's in the setup menu #3. You can set PT cpu's to ignore zero's for avg's for watts, speed and cadence inpedendantly. This only affects the avg that gets displayed on your CPU screen and doesn't affect the downloaded file at all. Hence any program will show a different value cause most of them include zero's.

I typically see a difference of about 10 or 15 watts at the most.

 

I see exactly the same thing - PT CPU is 1-2 watts lower than what WKO+ shows after download. I've never messed with figuring it out since it's a small difference AND the WKO+ number is higher!

 

There should be no difference. An average of a given set of numbers should be the same whether you use a abacus, pen/paper, computer, PT head unit or WKO+

Then again there are lies, damned lies and statistics....

 

Actually it does make a difference - to the ability to do the calcs at speed and constantly update them every second new data is added, which is what the PT CPU does vs. a once only calculation on a non-changing data set by software on a PC. Try doing the calcs on your abacus or with pen/paper every second that more new data is added.

Due to limitations of memory capacity and processing power/speed of the on board CPU, the firmware uses some processing shortcuts that introduce small rounding errors in the calculation of average power. It is usually no more than 1 or 2 watts at worst.

It is no more sinister than that. The downloaded data is correct and the average calculated from that data in the software is correct.

 

Alex - thanks for explaining the difference! Your explanation makes perfect sense. And - I always knew the higher number had to be the better number!