Indeed, phase/polarity is one of the most common causes of that problem.
If one of the channels is 180 degrees out of phase with the other, it could cause havoc with the 5.1 processor.
And it wouldn’t necessarily be the whole channel. It could just be a “stem” that is out of phase. Like the music stem or especially the dialog stem.
Did you check the playback when you sum the 2.0 stereo down to monaural? When the 2.0 stereo is played back through a 5.1 processor? lots of things can sound OK when played back on 2.0 (or 2.1) monitor speakers on your editing system. But you never know how your track is going to be played during exhibition. So you must protect yourself by making mixes that will withstand improper playback, exhibition or broadcast. (Like “defensive driving”.)
———————————————————————————
Recording audio without metering and monitoring is exactly like framing and focusing without looking at the viewfinder.