IAero and PT SL+ side by side comparison

[turbomentor]

I do regular testing to determine "FTP". There are lots of protocols for determining this value; the method I use is a warmup followed by 2x20 min with 2 min rest in between. I download the data and take the Normalized Power for that 42' segment and use that as my FTP. I then base all of my training rides off of that value (usually intervals targeting 85-100% of that value).

In any case, on Tuesday I rode an FTP test on one of my typical routes. This was the first ride where I got to use my newly updated ANT+ SPORT PT hub (now and SL 2.4+ hub).

Bottom line is that the iAero picked up the PT signal right away and I had no issues whatsoever. Furthermore, while there are definitely some outliers, the PT data matched up very well with the iAero data. I think that the Pavg for the ride was within 4W of each other or so.

I plan to conduct my next FTP test on my TT bike, I will provide another side by side ride file for that ride.

[djconnel]

There was a good data comparison done in a BikeTechReview thread. The iBike data tend to lag the PT data by around 1 second. So the best comparison optimizes an offset between the two to maximize the correlation coefficient.

Then the iBike assumes a fixed 2% drivetrain loss. This is a bad model of actual drivetrain losses, but since the Powertap doesn't measure drivetrain losses, a proper comparison of the two would subtract 2% from the iBike data before the comparison. A complication is the iBike tends to drop small powers if it thinks they're coasting. So before comparing drop, for example, data where the PT measures under 50 watts or the iBike measures under 50 watts. Actually, the drivetrain loss issue can be a big deal when comparing SRM/Quarq/Ergomo to Powertap numbers: easily a difference of 0.1 W/kg. If the iBike had a user-programmable model for drivetrain losses, one could set it to zero to generate "PowerTap power". I think this needs to go on the short list for future firmware upgrades.

Another issue for data comparison is the iBike data tend to be smoother than PT data. If upward spikes in power tend to be sharper than dips in power, you can easily imagine this would result in the iBike tending report lower than PT at higher powers, while tending to report higher at lower powers. This is what I see in your data. A better comparison would be to smooth the PT data to maximize the correlation coefficient before comparing. This would need to be done simultaneously with the optimization of the delay, but doing it after the delay optimization would probably be okay, as well. Just don't do the smoothing optimization first.

Instead of doing all of this, which would take more work than I wish to invest, I just took your raw data and did a simple y=ax regression for data where both the PT and iBike report at least 50 watts. Equivalently, I averaged the ratio of the iBike to PT data. My answer: 1.019. How about that? As I noted the drivetrain loss multiplication factor in the iBike is 2.0%. That means after dividing by 1.02, the iBike power agrees with the PT power to around 0.10%, on average. That's well within the precision of these units.

It would be interesting to see a more rigorous comparison, with both the smoothing of the PT and the time shift.

[Velocomp]

Very interesting analysis!

Drivetrain losses can be set to 1.00 (no drivetrain loss) in the Advanced Settings menu of the Edit Profiles menu.

[R Mc]

John--

They can?? When I call up the advanced settings in edit profiles, the Cm figure is one of the few that I can not edit.

[Velocomp]

OOPS! It looks like Travis changed this! I will find out what is going on...

[LouisB]

So ?

There is a way to manually set the drivetrain losses to 1.00 in a profile ?