Power accuracy is REALLY BAD!

I am really dissatisfied with the power measurements on the Flux S.
I wanted to know how my FTP stack upp between mu differnt Power meters since i use different ones on my 3 bikes. The power meter on my day to day bike is a Rotor inPower (left only),  tri bike has a Quarq DZero and the bad weather bike has a 4iiii (left only).
The current setup on the trainer is I have the tri bike on it (with the Quarq cranks paired to Zwift). In order to validate the power accuracy I checked so both Quarq and Flux had the latest SW. Did a new calibration even though Quarq has autoZero/magicZero och did a spin down calibration on the Flux according to instructions.
Pairing: Zwift + Flux
Garmin 1030 + Quarq
I did a few rides and Flux show consistently 40-50W lower than the Quarq, average!  Sprints can diff more than 100W! There also seems to be some kind of smoothing going on on the Flux which might explain the it missing out on shorter sprints.
So in order to dig deeper into this I borrowed a pair of PowerTap P1 pedals from a friend. and threw them into the mix. Worth mentioning the his pedelas stack up perfectly with both his Quarq  and his trainer wattage. I paired the pedals with my Garmin Tactix Charlie and did another test.
Pairing: Zwift + Flux
Garmin 1030 + Quarq
Garmin Tactix + Powertap pedals
The results in the graph speek for themselves after a 10 min ride with both sprints, steady state and a ramp at the end. Data is aligned with HRM which is paired to all 3.
Flux avg: 279W
Powertap avg: 309W
Quarq avg: 317W
This is NOT ok. It is not <3% acuracy, it is more like 10-15% accuracy. How to fix this?