I have had this problem before and again it happened today.The xert user should never have to wonder if the data shown in xert matches what is in the fit file. I have yet another file where xert imagines power numbers that do not exist in the fit file.
Data fidelity is important. The 5-10 seconds of weird data is not significant. What is significant is the feeling that xert is messing with data that was written to the fit file.
Attached is a screenshot from wko5 showing the values, and the view from xert.
Leave the data as is, if the data is really suspect, flag the file and ask the user to look at it. I have a zillion files in Xert as do others, do you want people wondering what else has been changed?
Thought so. Stick power is a problem some power meters have. When pedaling stops, power and cadence will get repeated by the power meter for 2-3, perhaps even more seconds. These along with power spikes or stuck power have to be corrected in order to avoid signature miscalculations. It’s an error on the part of some power meters and trainers. Our method works well in the many thousands of rides we process every day. Users never have to worry about their power meter errors.
We did have to put in exception logic for some trainers though that use power smoothing. Super dumb feature on the part of trainers. If this is a smart trainer in smoothing mode, suggest you remove the smoothing so that we can properly detect sticky power.
We’ll be adding an option to view the corrections made by the system. That’s coming with a larger update we’re planning for the overall system. Fortunately, problems like this are pretty rare and our engineers have done a remarkable job in codifying all the various issues and corrections to errors that power meters make. If it is from power smoothing, do disable it. If not, reach out to support and we can see about having a way to reverse the corrections it makes if these are impacting your training.
It’s from a powertap on a bike on rollers recorded directly to the garmin, not from zwift. 4 seconds of the same power number, 5 seconds at same cadence does not seem to be implausible at all. For 20 seconds here the cadence is within 1 or 2 rpm I’m pretty sure my outside rides are not like this but indoors why would this be flagged as bad data? I’m just trying to understand the logic here, not complaining.
It wasn’t flagged per se. The 3 repeating power/cadence values tripped the sticky-power logic and that last one was set to 0/0. The chart shows a rolling average, hence the dip. We’d have to thihk further about how the data would indicate it wasn’t sticky power.
Is that data generated by an original Powertap hub, CR, or P1/P2 pedals?
I imagine most riders learn to pedal very smoothly on rollers.
On a trainer you can be very sloppy and never fall off.
It would be cool if you guys published a note pointing out which trainers have what sort of issues and what you do about it instead of waiting for someone to find something odd and point it out in the forums.
We don’t have every trainer nor the time to hop on every trainer to test it nor the time to generate a comprehensive document nor the time to keep it up to date with each new trainer that comes to market.
The forum is a great place for people to share what they have and what experiences they are having. A community of users will share far more insightful information than a few people with only 24 hours available in each day and need time to sleep and eat.
I don’t mean to be rude or insulting just pointing out that many expect things we simply don’t have the capacity or means to do.
I just thought that if you’ve run into these issues and did things to address them, posting a note somewhere really wasn’t any extra work. After all, someone had to figure out the problem in the first place and then there was agreement to do something about it and then do it. I get it though - there’s just never enough hours in the day!
