Motor winding tolerance & PX4 hover drift logs

I’ve been bench testing matched sets of outrunner BLDC motors for industrial multirotor builds.
Even motors with identical nominal specs show tiny phase resistance deviations below 4%. Static thrust tests look consistent, but long-duration hover ULog data shows subtle yaw drift and uneven motor output allocation over time.

I’m trying to separate two root causes:

1. Minor winding resistance mismatch between individual motors

2. PX4 mixer & low-throttle compensation parameters

It’s hard to tell from raw logs whether the slow drift comes from hardware inconsistency or flight controller tuning limits.
Has anyone compared flight logs from tightly resistance-sorted motor groups vs unsorted ones? Any parameter tweaks to reduce this long-hover imbalance?