The inverters should be tested and operated with output Line, Neutral not connected to anything. Then each individually only connected to a load, not to each other.
It is possible the way the two are connected together shorts grid output, or has inverters fighting.
One guy wired Sunny Island with input and output both connected to grid, because they were in an out building and he wanted to power the house where service entrance panel was located. Output circuit breakers tripped without damaging it; maybe wire resistance limited current sufficiently. I explained that either wires should be run back from the battery inverter to a transfer switch, or the inverter located at the transfer switch. He wanted to keep it as it was, but dynamically switch relays so the wire run could be used bidirectionally. I never heard back if that was accomplished.
Several parallel paths with objectionable current flowing everywhere.
Perhaps, but what harm does "objectionable current" cause?
I have seen the expression used to refer to return current flowing in ground, when neutral is bonded to ground in more than one place. So a portion of the return current which was supposed to flow in neutral splits off to flow through ground, then back to neutral. Everything works, and you would likely be unaware of it.
I haven't thought of any way that connecting input neutral to output neutral would cause harm. The inverter itself does that for pass-through mode. Then it opens a relay connecting them and instead connects output neutral to ground. Whether or not neutral is hardwired to grid, what harm could that cause?
One inverter popped when I first set them up. The problems started then. I think there was a problem with that was not on the main board. There is a possibility of a bad data cable. But that would not explain why they both popped when switching from SUB to BSU.
Bad data cable or missing data cable could let the two inverters run independently. When on grid, each would just pass its phase of the grid through to output, and L1/L2 on output would look just like grid, no problem.
When off grid, each inverter would produce its L output unsynchronized with the other, so might be 240Vrms between them, then drift to zero, then back to 240Vrms. Output swings from zero to +170V to zero, switches H bridge, swings to -170V, and back. Possibly, after H bridge switches and an inverter is trying to pull output up to say +25V, if 240V load between that inverter and second inverter pulling output up to +170V, that pulls mosfet transistor source higher than gate, killing it.
But inverters need to be able to handle inductive and capacitive loads, which do just that.
If line and neutral were wired backwards, that would do funny things.