I have noticed that on uphill intervals in particular, the cyclingpeaks software shows power at 15-20 watts less than the powertap software.
It can't be the effect of zeros since there are none on the uphill section.
It is not that big a deal except that I am new to outdoor power training and am relying on the current power readings off the PT for intervals, so I need the downloaded data to correspond.
In that respect cyclingpeaks is a disappointment given the key feature in power training is accuracy and repeatability.
Do you have smoothing turned on in cycling peaks? The rightmost display in the toolbar should read 0:00 if you want the cyclingpeaks graph to display the same thing as power agent. If it doesn't, click on the little down arrow next to the number until it does. Or go to the soothing menu, and select "None." Or just hit the F4 key.
Do you have smoothing turned on in cycling peaks? The rightmost display in the toolbar should read 0:00 if you want the cyclingpeaks graph to display the same thing as power agent. If it doesn't, click on the little down arrow next to the number until it does. Or go to the soothing menu, and select "None." Or just hit the F4 key.
I have noticed that on uphill intervals in particular, the cyclingpeaks software shows power at 15-20 watts less than the powertap software.
It can't be the effect of zeros since there are none on the uphill section.
It is not that big a deal except that I am new to outdoor power training and am relying on the current power readings off the PT for intervals, so I need the downloaded data to correspond.
In that respect cyclingpeaks is a disappointment given the key feature in power training is accuracy and repeatability.
Thoughts?
I don't know the source of the discrepancy, but I do know this: CyclingPeaks WKO+ is telling you "the truth", i.e., the values are straight-up averages of what's in your PowerTap's memory, without any manipulation whatsoever. Thus, if they don't correspond to that shown by PowerAgent, the latter must be doing some post-processing (i.e., fudging) of the data.
Definitely the latter, and I believe the former as well.
I'm not sure why you're getting a discrepency. Have you tried to compare the CSV output files from each program?
In CyclingPeaks use Export Copy As... then select PowerTap Files (CSV) as type.
In PowerAgent use File > Export Activity Data.
This will give you two comma separated value (CSV) files you can easily compare in a spreadsheet. These two files should essentially be identical (just a couple small differences in level of precision of values and the handling of breaks during a workout).
But all the workout and interval stats (min/max/ave power, HR, etc.) should be the same. I've never seen them differ from each other. Though I have seen interval summary data not match the CPU, but only by 1 watt or so. Probably just a rounding issue.
I wondering what the recording rate of the PT CPU is set at? Not that it should make any difference with the data download but just ticking off things. It can definitely have an impact on differences in max power shown between the CPU and software.
I wondering what the recording rate of the PT CPU is set at? Not that it should make any difference with the data download but just ticking off things. It can definitely have an impact on differences in max power shown between the CPU and software.
Even if it's set at 1-sec intervals (which is actually 1.26-secs) there can be differences between the CPU and the software. It's because the CPU calculates data slightly 'wrongly'. (not sure if it's a memory and precision error).
I'm not sure why you're getting a discrepency. Have you tried to compare the CSV output files from each program?
In CyclingPeaks use Export Copy As... then select PowerTap Files (CSV) as type.
In PowerAgent use File > Export Activity Data.
This will give you two comma separated value (CSV) files you can easily compare in a spreadsheet. These two files should essentially be identical (just a couple small differences in level of precision of values and the handling of breaks during a workout).
That might prove helpful as long as the original file was recorded at 1.26 s intervals. If not, small (and sometimes not-so-small) discrepancies may result from the fact that WKO+ will attempt to interpolate the "missing" data when exporting the file as a .csv.
Even if it's set at 1-sec intervals (which is actually 1.26-secs) there can be differences between the CPU and the software. It's because the CPU calculates data slightly 'wrongly'. (not sure if it's a memory and precision error).
Ric
What does that mean?
If the CPU calculates incorrectly, what good is it for doing intervals where you are targeting a certain power range?
The good news is that the averages did match up in the latest download as between Poweragent and Cyclingpeaks. Also, my impression of where my power was based on reading the CPU during 2X20s seemed as accurate as it needed to be ...
If the CPU calculates incorrectly, what good is it for doing intervals where you are targeting a certain power range?
The good news is that the averages did match up in the latest download as between Poweragent and Cyclingpeaks. Also, my impression of where my power was based on reading the CPU during 2X20s seemed as accurate as it needed to be ...
The CPU calculates a running average by 'downsampling' the datastream 'on the fly', which means that it doesn't use precisely the same values to calculate the average for a ride or an interval as is used by the software. The difference is usually only a watt or two, but with short intervals might be slightly larger.
You mean with the money you saved by buying a PowerTap instead of an SRM, you can't afford a laptop computer to take with you on trips? 😉
Cheap laptop: $400 Second license for CyclingPeaks: $49 Ability to analyze data recorded at 1.26 second intervals while traveling: Priceless!
Ok, so my attempts at humor today aren't going too well. :o
🙂 When I last had this problem, I bought a 2nd CPU. But if you're travel is long, then you do need a laptop. I'd just download poweragent for free and upload to CP when you get home again.
For longer aerobic rides, 2.5sec recording intervals are fine if memory is a constraint.
The CPU calculates a running average by 'downsampling' the datastream 'on the fly', which means that it doesn't use precisely the same values to calculate the average for a ride or an interval as is used by the software. The difference is usually only a watt or two, but with short intervals might be slightly larger.
Good. I am not interested in the average until after the ride. I look more to the instantaneous values ...
