NP vs. AP Trainer Questions

atasic

 A good question came up on JOS group discussion, thought I ask here and seek scientific explanation.

I have gone back and looked at some power files that were recorded this fall.

Equipment:

trainer   Fuild 2 and KK Road Machine

Edge 500

Quarq Cinqo

Software: Race Day Apollo

Intervals observed, mostly 5-10min, some 20min, all at FTP

NP is coming out lower in every one of them, sometimes 6-7W lower than AP. The longer the interval the less difference between the two, NP being lower than AP. 20min intervals show as little as 2W NP lower than AP.

Total time calculations for the ride always give NP higher, so that is no issue.

I can provide some samples from Golden Cheetah on request if needed for an educated answer. RD not friendly enough to be able to post.

Thanks

 

 

William Jenks

Golden Cheetah does this because:

1) It's absolutely essential to do some averaging of the power readings because they do too much fluctuation for the number to be worth anything when you raise them all to the fourth power otherwise.

2) GC's algorithm for averaging is - in effect - "slower" than the one in WKO. It's an exponential tail of I forget what time constant (say, 7 seconds), rather than a strict rolling average, like WKO uses. (In fact GC and RaceDay "brag" about this feature)

Other than this averaging, NP and xPower (which is what GC spits out) are pretty much the same as I recall, but this averaging issue does lead to the issue you're talking about. WKO solves this by not even giving you a NP for an interval less than a certain length in time (5 minutes?)

As a result of the weighted averaging, the beginning of your interval is considered slower (lower power) than it really is in the xPower calculatin.. and it's obviously worse the easier you were going right before the interval. There's no "make up" for it at the back end, because all that happens is that your high power gets averaged a bit into the rest interval, artificially RAISING the xPower there. Because this only affects the first half minute or so, the difference is larger for the shorter intervals.

I'm a little surprised that you get as much as 2 W difference for the 20 minute interval, but I guess it's possible.

tri_antelope

I'm not sure the issue is exclusive to Golden Cheetah.  I also use a Quarq Cinqo with a Garmin Edge 500, and I experience the same thing (NP lower than AP) when I do FTP intervals on the trainer, and I use WKO+ 3.0 to download and review my power data

I've actually even had the same thing happen when I did a 2x20 test outdoors this summer.

I'm trying not to sweat it too much because I tend to ride very low V.I., but it is annoying, especially when doing a 2x20 test that shows NP coming out lower than AP...

 

atasic

 @William, thank you for a thorough answer. Makes sense. I also inquired on ST and got a very nice response along the same lines as what you discussed. 

I will use xPower as NP for the purposes of FTP testing and training and monitor/adjust based on feel. EN protocol combined with Race Day Apollo xPower for 42min may underestimate FTP by a little, but that is ok. I would rather go that way than upwards when getting the estimate.

@Michael, I have not had that situation yet outdoors. I do have a flat loop where I do all of my FTP work and can make it so nice and even that will almost make you cry when you see the power file. I will monitor what happens. I also suspect, but cannot confirm that it may also have to do with one of the firmware updates for Edge 500. I will start comparing outdoor files from this summer as the season progresses and see what happens.

Thanks to both of you.

William Jenks

In reality, the simplest solution is just to use the average power for shorter intervals, at least indoors, where there's little reason for the power to vary much. If you read the fine print, all three of those software support files will tell you that. And, of course, as long as you're consistent, you can track the power changes over time, regardless of the exact calculation.