To add further confusion I did 2 more tests tonight back to back. Calibration not touched since last time gave similar results to last time, then a fresh calibration of Kickr & 4iiii narrowed the gap to only about 10W difference at threshold…i could drive myself mad so I think I’m just going to take a while to get used to the new PM and try to be aware of RPE when using.
I also tried the one leg test noted further up the thread. I’m not sure it’s a fair test and the data was pretty choppy. It actually suggested the PM was reading less than the Kickr (with Kickr power doubled) - that’s the opposite of what I was expecting…I think.
I’m really hoping to find a power meter that can give me an increase of my ftp, I’m hoping for something like a 30W increase which would make me feel a lot better about myself.
I don’t care what marketing calls something. When I hear people claiming that this isn’t really measuring power, I think their point is these units don’t measure forces (or torques). But power is just change in energy over time. And you can easily base that on an energy balance that includes the kinetic energy of the flywheel. Indeed, it seems to me that the trainers have the same accuracy than power meters, with the best of both reaching +/- 1 %. Both, power meters and smart trainers need to be carefully calibrated in factory. As any former employee at one of those numerous power meter startups (and current Shimano/Pioneer employees) will tell you, doing this at scale is hard.
Even virtual power is a method to produce power numbers from measurements, just not a very good one. From a physics perspective, the only difference is that the systematic errors are much larger. Most (all?) sources of virtual power are simply speed sensors that do much the same as inside a trainer. What’s the difference then? Calibration and more sophisticated compensation. Calling it virtual power was a marketing decision, and I am largely ok with it. Once the dumb trainer has warmed up, I reckon the power numbers will be relatively consistent. Especially for efforts like 30-30s where temperature reaches an equilibrium, the power drift isn’t that large. (At least that has been my experience with an Elite Volano, a fluid-based direct drive trainer coupled to a 4iiii single-sided power meter. There was definitely a temperature-component here, across a workout my cadence had to increase by 2-5 rpm to hit the same power numbers.)
All power meters I know of measure power indirectly. A crank-based power meter with strain gauges measures resistances and cadence. The strain gauge is converted into a force, then a torque, and power = torque times 2 pi times cadence. As you can see with 4iiii’s new power meter (I’m referring to @GPLama’s review), if your cadence measurements are not correct, your power numbers will also be off. Inferring power numbers from these measurements requires you to compensate for a number of other factors, including temperature, vibrations and the like.
Excepting trainers like Elite’s more expensive trainers that also have a torque sensor. Most, though, use energy balance, which is probably the most fundamental principle of physics. The spin-down test infers the friction-related losses by measuring how quickly the flywheel spins down. That isn’t really more fundamental than power = torque times angular velocity or power = force times velocity component parallel to force vector.