jhay

It looks like a rebadged  InfiniSolar 3kW plus.

The only stuff I found was on this Czech site:
 
http://www.ostrovni-elektrarny.cz/index.php?page=podpora
 
Interesting there is also a battery voltage calibration file.
 
I have not tried any of this myself.

The way I implemented mine was that the processor on the battery would turn off the FET after a while. So if the master wanted to keep the FET on, it would need to keep on sending the command. That also handle the case where the master die. Actually instead of just a plain on or off, I have the master send a PWM value to the battery processor, so it does not need to be 100% on. It can be anything between 0-100%.
 
My implementation does use a ring, but with only 1 opto between every battery. About the reliability, my thinking was that there is no use trying to balance the rest if there is a problem with one, so the master should rather just send a cry for help if there is a permanent comms failure.

Here are a few lines during different load sheddings:
 
QPIGS (000.0 100033 00.0 0002.6 230.5 00559 50.0 002.4 020 393.8 393.8 050.0 ---.- 057 00120 00526 ------ 377.4 ---.- ---.- 036.0 D---110001
QPIGS (000.0 100046 00.0 0002.8 232.4 00603 49.8 002.5 022 415.8 415.8 050.5 ---.- 058 00645 00557 ------ 406.0 ---.- ---.- 037.0 A---101001
QPIGS (000.0 100131 00.0 0010.4 231.3 02401 49.9 010.3 080 452.9 452.9 054.7 ---.- 100 02572 02270 ------ 297.4 ---.- ---.- 044.0 F---101001
QPIGS (000.0 100034 00.0 0001.6 230.3 00287 50.1 001.2 012 460.1 460.1 054.4 ---.- 069 01845 00253 ------ 416.2 ---.- ---.- 038.0 A---101001
All documentation that I could find on the web says it should be grid current, but the actual values deny that. In the above cases it was impossible to draw current from the grid. Even the example in my previous post had 6.2A, but I'm pretty sure it was not coming from the grid. The two currents are not the same but normally not too far from each other.
 
Interesting that even during load shedding, it thinks it can draw a few watt from the grid.
 
About where I got the info, capturing the serial streams and searching the web for those keywords. I have sent a few emails to VoltronicPower, but have not received any answers back. I have emailed MPPSolar (where the Ausies got their info from), they replied, but said I should contact my supplier.

OK, as per jhay's suggestion, the earth-neutral link is moved closer to the inverter and in bypass (utility) mode it would be out of the circuit. I've also moved the optional "Inv. Out" breaker to also be out of the circuit when the changeover switch is in bypass (utility) position and I've added a separate isolator feeding the inverter input.

I see in your diagram that you pick the utility power for the change-over switch from after the isolator. The drawback is that if you need to work on the inverter and switch off the isolator, your circuits will be dead. If you pick the utility power from before the isolator, they can have power even if you work on the inverter.
 
Another thing I did differently was to make the earth-neutral connection closer to the inverter output, so that if you are in bypass mode, that connection is not in circuit anymore and then you depend on the main earth-neutral connection. It seems a little closer to the spirit of the SANS 10142-1.