s5100e wrote:
both what GP Lama and Ai_1 say seem to make sense. on the latter point if you assume each has an error of +/-3% then it could be that one is on the high side the other the low side and the resultant effect is a deviation of 6%, so if in you example one was reading low 3% say the trainer, and the other was high 3% then what you see is the error of 6% difference. At 300 watts that is 18 watts difference between the two. I realize this may be unlikely but that can add to the complication. I suspect instead of erg mode try slope mode and hold a consistent wattage and then compare, also vary which gearing so that you can see how fly wheel speed and pedalling cadence comes into play. For me it is just easier to use the data from my bike power meter because it should still be more similar to power I see outside vs inside on the trainer. The trainer just changes the effort.
It doesn't even need to be half that total error to explain his data in my opinion.
Let's assume those 100W, 200W and 300W figures are exact according to the trainer's measurment? There could easily be say 10W lost in the power train at the 300W range. That only leaves a 8.7W differential for his biggest gap. 8.7/300=2.9% total error. I think the Flux and Quarg have declared accuracies of 2.5% and 1.5% repectively. That's 4% when comparing data between the two. It only takes 2.9% to explain the OPs results without anything being out of spec, assuming 10W lost in the drive train.
The difference between small and large ring may be a separate issue, likely due to a flawed test method (ERG), and that can easily be checked.