Firmware rollback to deal with power inaccuracy

I’m well aware of the fact there will be drivetrain losses, but the Neo is reading significantly out versus my previous turbo (from another big brand!) and the 4% example I posted the image of is the best I have seen it (it has been as far adrift as 7%). But hey, thanks for bombarding the thread with random stats and not answering my question

Sure Luko79, it’s here www.zwiftpower.com/analysis.php

• And here’s another www.zwiftpower.com/analysis.php

ahh i remember for this comparison! :) (i checked the Vegan's result yesterday :D) . my 1st thought was yesterday that it is a Neo1, but as Your wrote, it is a 2T with 0.0.36 fw, right? or any older firmware in it? What was your gearing during race? Big ring front i assume and middle at rear or even lower? (trainer difficulty 50%? or lower for racing? )

4% drivetrain loss plus 1% difference (high or low on each end) equals 6%. And it's likely your drivetrain loss is 4.5% or more. It's basic math. 4+2=6.

Firmware is 98.0.38

trainer difficulty was set at 50%, and spent the race in the big ring and middle of the cassette. However, I have also done workouts in ERG mode using various gearing combinations and seen the same results

just a quick sheet with different trainers (i removed the single sided powermeters because those are not relevant), where is the 4-7%...? 

Or ALL TRAINER MANUFACTURERS (WITH SILENT AGREEMENT) ARE USING FAKE UPSCALED POWER NUMBERS WITH A GIVEN % TO ELIMINATE THE LOSS, DECREASE THE DIFFERENCE AND MAKE THE CUSTOMERS HAPPY. OR EVERY USERS ARE USING BETTER WAX/CHAINS/LUBES/BEARINGS THAN CERAMICSPEED BECAUSE THEY COULD REACH LOWER DIFFERENCE THAN THEM...  COME ON :)  

But i can send You more evidence (and i checked the history of these comparisons, lots of users have more trainers and more powermeters at home and can cross-checked them , and the differences usually the same , if not , that is suspicious :)