Any ideas?
At the moment, my first thought is that since the BMV is being powered via the sense cable and shunt, the additional 5v feed from pin 4 of the ve.direct port on the MPPT controller is unnecessary, and that since something odd is clearly happening there, I will get some JST-PH connectors and make up a little extension that goes between the mppt controller and cable, with pin 4 disconnected and only the gnd, tx and rx wires connected. I'm thinking this is most likely going to fix it, but even so I'd like to know what's going on here.