I recently upgraded my RK100 with upgrade pedal to have RK200, and I also bought XC200 rebuild kit to be able to check both my power meters few times a year to match them. Recently I noticed that readings from XC200 are fine at the beginning of the ride, but then there is always a drop in power after long descends or coasting (basically after auto recalibration is performed). When analysing the fit files after rides I see that left and right leg readings are fine after manual calibration and after a while, right leg readings are 10-30W lower and I can see that immediately also when comparing with quarq.
You can see example file here:
https://analyze.dcrainmaker.com/#/public/3e2324b2-cbcb-4b4b-6450-32341f3b23c8
The first hour of data is fine, then I assume auto-calibration happened, and from 1:30h of ride it's visible that right leg readings are much lower. It's then visible also in comparison to quarq data. If I would have only one power meter, it would be only noticeable by sudden change in L/R balance ratio.
I tried to switch off auto-calibration on rally, and it seems that data is now correct. Although it would mean to perform several manual calibrations during the ride if conditions are changing - which they are on longer rides with lots of ascents. Here is example file from ride with auto-calibration off:
https://analyze.dcrainmaker.com/#/public/3c37a122-142d-42f6-428b-23bd3919888b
I have firmware version 3.0 on Garmin rally. I use Quarq paired with my Epix 2, and Rally XC200 (RK200) with edge 830. Everything cleaned and reset recently, everything installed with torque wrench to prescribed torques. My weight is 68kg, so difference up to 30W is huge! I have dataset supporting my assumptions from several rides. I initially though that ovals might be the problem, but I switched to round chainring, reinstalled and re-cleaned everything properly and did couple of rides again, but that didn't help at all. It's quite annoying to deal with such issues on device which cost that much....