ata2k6

Aaaand the winner is @Coulomb
The upgrade to 76.03 did the trick, the inverters are connected and works in parallel, no more voltage difference on the display.
Thanks man, you saved the whole week, not just the day!

No, only 0.1 V difference and even that only some of the time is as good as it gets. One firmware that I looked at seems to have a threshold of 4.0 V (in a 48 V system), and 6.0 V if the other machine is in a different mode (battery versus line mode for example). So it seems to me that only the most gross battery voltage differences should trigger fault code 83, and in fact I suspect that in practice it means that the firmwares disagree about the interpretation of can message formats. So the two machines are comparing completely different things, resulting in what appears to be more than 4.0 V discrepancy in battery voltage readings.
Of course, iGrid models are on-grid models, and I'm only familiar with off-grid firmware, so I might be way off base. I find that I do have Infini V IV firmware version 76.03, which might be close enough; use at your own risk, I don't recall where I found it:
4564-00H7 FW upgrade_28335_DSP INFINI V IV 5.6K 76.03.7z
I also have 56.00, which might match 76.01 better than 56.05 matches 76.03:
INFINIVIV5.6KReflash5600.rar