Yeah, as Lama noted, that definitely appears to be something amiss with the KICKR. I don’t think I’ve ever seen something that dorked up though (from any unit/company). Specifically, the type of dorking rather than the offset per se. Meaning, it’s just super random. One second its fine, then it loses the plot.
Agree that Wahoo support is the way to go here, especially with as much data as you have. I wouldn’t be surprised if given the data you have assembled they just ship ya a new unit without any argument.
As for my smoothing preference for graphs/etc, it honestly depends a bit on the exact ride type. For outdoor rides I’ll usually do 3-5s for under an hour long, and do around 10s for hour+ rides. Otherwise it’s just hard to see on charts. For SIM style (normal Zwift), I’ll usually do 0-3s at most. And for ERG workouts I don’t typically apply any smoothing. Again, all kinda depends on the situation. My goal with adding smoothing in charts is usually just to make it easier to see disparities. If there’s a lot of shifting power (like a race), it’s often hard to see those differences without the tracks being smoothed a bit.
Nice work! I’ve speculated on this before. Interesting to see that you’ve measured a difference.
FWIW, I think the Assioma pedals are measuring accurately, and others are off.
As I mention in one of the posts below, a similar issue arises with oval vs round chain rings.
Ironic that we all think power meters bring an objectivity to measuring our work effort, whereas we may sometimes be fooling ourselves and RPE is more accurate!
Ah right, i need a crow foot wrench adapter. Any good quality torque wrenches people would recommend? Paying £100+ for a Park Tool kinda hurts. Especially as it appears I need two? One for sub 10nm stuff like seat posts etc, then a 10-50nm tool for pedals etc?