Firmware rollback to deal with power inaccuracy

I have used 4-5 earlier updates, all of them have same power difference against power2max.

power2max has almost 5% higher readings, whole range.

Tested two  neo2t units, same result.

it is strange that older neo 2 unit the difference is almost zero. not tested myself but readed other internet sites.

are You using oval chainrings or round rings? 

Jone88 which model of P2M do You have? newer NG series or older TypeS? (and with oval or with round chainrings?)

I’m using round chainrings on a P2M NGeco

If your other power meters are measuring power at the crank or at the pedals, then the 4% is likely drive loss. If drivetrain loss is 2-3% or more (and it depends on which gear combinations, drivetrain age, drivetrain cleanliness, type of lube, amount of lube, etc.) and if both power meters are 1% accurate, then 4-5% difference is what you'll see. You're measuring the power at different locations. If, the VeloNews article below is correct, 9.45 watts lost to drivetrain friction in a 2x setup equals to 4.75% at 200 watts. 

In automotive settings, they measure horsepower at the crank and at the wheels.The two are very different. The marketing department uses horsepower at the crank. People who race, etc. measure power at the wheels. Same applies to bicycles. Drivetrain loss can add up. Why do you think people spend hundreds of dollars on ceramic bearings in bottom brackets, oversized derailleur pulley wheels with ceramic bearings, and obsess over which lube or wax they use?

Ceramic Speed found "losses ranged from a minimum of 6.26 watts, seen in a new chain & new ring/cogs, to a maximum of 10.27 watts, seen in the highest-elongation chain & worn ring/cogs" (https://www.ceramicspeed.com/en/cycling/inside/test-data-reports/drivetrain-efficiency-test-old-vs-new/).

VeloNews and Ceramic Speed found "The average friction of the 2X drivetrain was 9.45 watts, computed as the sum of the power losses" (https://www.velonews.com/gear/gear-issue-friction-differences-between-1x-and-2x-drivetrains/). The same article notes "The frictional losses of the system were highest for each chainring when the chain ran on the smallest cog. This is also where the difference in frictional losses were greatest between the two drivetrains: at the highest gear ratio (4.8) the 48 X 10-tooth combination consumed six watts more than the 53 X 11-tooth combination." 9.45 watts of drivetrain loss at 200 watts is 4.75%.

that VeloNews article is Ceramicspeed-sponsored... Obviously excessive, usually it is 1.5-2.5% apprx, not 4.75. The internet is full with comparisons of different trainers and powermeters, all highend Elite, Tacx , Saris, Wahoo have this 2% (+-0.5%) difference with a reliable dual powermeter. If the difference is higher, could be: #1: the bike powermeter measures higher than it should, #2 the bearings in the trainer (or in the drivertrain, so in BB) are worn and it could add a few extra watts (as the oval chainrings have a "modifier" too) . #3 the trainer measures lower than it should

can You give a direct link to ZP comparison/s/? the summary page is not enough for deep analysis. 

Yes, it's sponsored, but do you have any links to the internet comparisons? What were their testing protocols?

the ZwiftPower is full with them. (dual / triple recordings) But just watch a few GPLama videos on YT, he tried a "few" trainers with multiple reliable and weight calibrated powermeters , with good testing protocols (he is not newbie )

You're right, VeloNews is sponsored by CeramicSpeed and both have vested interests, but who is sponsoring people like GPLama? If VN and Ceramic Speed have a bias, it's likely others do too.

And if everyone is measuring the same difference, that says it's likely that the 9.45 average watts of drivetrain loss isn't just marketing by Ceramic Speed.

And if you measure power at two different locations, you'll get different readings. The automotive industry notes this. In automotive settings the difference between at the crank and at the wheels can be roughly 10% in stock settings. Now, bicycle drivetrains are different, but a 4-5% difference given the variables isn't a problem.

Even if you halve the 9.45 watts loss from VeloNews, you're at 2.75% at 200 watts. Add in that both power meters have a 1% accuracy, and you're at 3.75% (or more if both at the opposite ends of their 1% accuracy). And that's likely with a very clean drivetrain with no wear.

Now maybe the Neo2T is using cheaper bearings (given some reports of how the bearings in some units failed early that may be the case) and that means there's more dirvetrain loss. Maybe given the way Tacx reconfigured the magnets and wiring is resulting in more drivetrain loss.

Now, given that everyone has biases DC Rainmaker has a good rundown on power meter accuracy. It's a few years old, but worth a read: https://www.dcrainmaker.com/2018/12/how-to-troubleshooting-power-meter-and-trainer-accuracy-issues.html.