Boom. There it is on the XCadey, cadence sensing issues in peak power acceleration/sprints. I'd consider that in-line with my testing scope. And unfortunately for XCadey, in-line with my results.
I have talked to xcadey for months about the calibration and try to improve their discrepancies with power. What they dont understand is that to be accurate there are a minimum of 3 pts in calibrations is needed not just 1
My explanation is
Power = Torque x "rpm"
Torque from sensor is accurate but its assumed a linear relations ship [ Y=mX+C ] where C is the offset at zero - That is the calibration number.... And m (slope) is fixed at the factory and its fairly accurate but when manufacturing deviates a tad that "m" is off and cannot be changed - Thats why i propose a 2 point calibration (like the old Quarq) by placing a known weight to get the 2nd point.
now for cadence where the biggest issue with new power meters that dont use a magnet for rpm... the rpm can be all over the place and delayed and if using the peak of the sensor and average the time between a small amount of peaks that is problematic.....
I gave up trying to explain my case to them