For a offgrid solar powered house with 48v battery, I've tested the FCHAO 3500w-48 inverter for a bit, and it seems to be the best option I can find in terms of idle consumption.

However, the ground on this thing is a quite powerful 12-30Vac voltage compared to battery negative.
This is causing issues with other components that has their chassis connected to the negative DC terminal. All of a sudden I have two different ground potentials that has 0Vdc between them, but upto 30Vac between them. And to make matters even worse, the inverter is only 3500w, so I need two of them. If they happen to power on so that their frequencies becomes opposite of each other, I get 3 ground potentials with up to 60Vac between them.

If they are mounted in the same metal cabinet, disaster is guaranteed.

How should I deal with this?

The unit itself is aluminum, connected to the PCB ground through a bunch of screws. The metal is covered in an isolated paint that makes it safe to touch, except for the screws. The problem is however, what to do with the mounting screws and outlet ground connections? I don't want three different ground potentials in the house - suddenly there is a risk of fire if ground touches ground!

The ground voltage (compared to battery negative) is steady 12.9Vac at idle, increasing by about 1Vac for every 200W of load connected.

The solution that comes to my mind is to connect the outlets ground internally to battery negative rather than inverter chassis. Then create some isolation mounts so that inverter chassis never gets in contact with the cabinet it's mounted into. Does this introduce some issues I have not realized yet?

(For all inverters I've used before this, the inverter ground has been floating relative to the battery negative, so connecting them together has been a solution to make a common ground.)

 

Your description is a challenge to follow, so please forgive me if my interpretation is way off.

The inverter likely uses an output transformer, and thus is isolated unless the output is explicitly connected back to the input.

The 12VAC could be a “phantom” voltage between two isolated systems. Have you tried putting a load between the those points? Say 100 Ohms 5W? The voltage might drop to near zero if I am right.

 

I put a load on, while watching the voitmeter - it did not change.
What you describe is what has been the case with the inverters I've been using prior to this. This one is different.

 

The "ground" connection to the case could also be related to some heat sink connected to the case. That would be sloppy design, but it does happen. I suggest checking for other connections to the case before isolating the ground. from the case.
There might also be an electrical noise issue that was dealt with that way. (JUST A GUESS)

 

There is nothing but the mounting screws connecting the ground to the case.
Undoing those screws (there are lots of them. I didn't count, but guessing they are between 12 and 15)
I did unscrew them all - that left the case floating.
The ground at the outlet is connected to the pcb.

My idea of connecting the outlet ground to battery negative rather than inverter pcb ground does not isolate ground from the case, but it shifts ground to a different potential. The two are related: 0Vdc, 12.9Vac (unloaded), 30Vac (max load). Are there any disadvantages going this path?

 

I recommend a single point solid PE ground for Neutral output at the breaker panel.

The battery must not also be PE grounded however the DC cables must be shielded and the shield only connected to the single point PE ground.

Install toroidal ferrite chokes on both the DC input cables and the AC output cables, as close to the inverter terminals as possible. Clip-on types are available for troubleshooting.

The manual should define the proper PE bond requirements to avoid ground loops or RCD faults.

The batteries must not share current between multiple inverters at different voltages.

 

If one should claim the "neutral" to be ground as well, I get 5 different ground potentials when using 2 inverters! Lets call the AC wires that actually powers devices for L1 and L2, and leave ground to be either the battery (negative) ground or inverter (1 or 2) grounds. L1 and L2 are about 70Vac and 160Vac (+/- 10V) from the inverter ground.

The batteries must not share current between multiple inverters at different voltages.

Are you saying that two inverters cannot be connected to the same battery? What kind of problem would that introduce? (assumed we don't define neither L1 nor L2 from any inverter as ground)

 

First, for those who are not familiar with some jargon, "PE Ground is slang for "Protective Earth Ground", the same as I reference as Green wire Ground. THAT ground is intended as shock prevention.
Unfortunately it seems that folks say "ground" when they really mean "common neutral" or Common Return Side of a circuit

Really, L1 and L2 makes the most sense of all, The only possible problem would be sloppy penmanship turning the "L's"into ones.

I suggest caution and careful thinking before tying any significant DC circuit to an actual hardware "ground, (Additional comment
deleted to avoid offending some folks.)

Remember: The Jolt of the volts is proportional to the Large of the Charge.!!

 

First, for those who are not familiar with some jargon, "PE Ground is slang for "Protective Earth Ground", the same as I reference as Green wire Ground. THAT ground is intended as shock prevention.
Unfortunately it seems that folks say "ground" when they really mean "common neutral" or Common Return Side of a circuit

Goimg back, before your time here, I had quite a few posts on this very subject.
When I came to NA, I had to get used to not only the term 'Ground' used indiscriminantly, but also the GND symbol's indiscriminant use, e.g. all the example circuits used in a well known book, "The Art Of Electronics' has the power Common indicated by an earth gnd symbol.
My previous experience in the UK we made the distinction, by using the symbol and terming it as Earth.
Most posting schematics here use the earth symbol indiscriminantly for all power common terminations.
p.s. One reason I like the Schematic pgm Kicad, it has all the power symbol shown and labeled as they should be as in the European fashion.

 

I'm in Europe and see that translation of terms is required, as you explain.
Also, our power grid does not have a neutral as seems to be the case everywhere else in the world. Our terminology would in English translate to L1, L2 (, L3) and Earth for our grid. Each phase has ~120Vac to Earth (each phase shifted 120degrees), and further translated into Hot, Neutral and Ground for internet forum usage.

When it comes to inverters, the ourputs are labeled L and N, but I have the feeling the wires have more similarities to our power grid (L1, L2), than to any other power grid where the term Neutral is used.

While we're on the subject of Earth, I got a surprise today. I have battery negative connected to copper rod hammered into the ground (the physical ground you walk on). Today I were about to lift a steel plate that had gotten frozen to the ground. As I had a voltmeter in my pocket, and a DC outlet quite close, I measured the voltage between battery positive and the steel plate - it was 46.7V. Battery voltage was 48.1V. Never before have I seen anything near such a low voltage drop through the physical earth. The distance to the copper road was about 20m.

Back to the topic of this thread - as no one has pointed out any issue with the idea of connecting house earth/ground to battery negative rather than inverter ground, that will be the path I go. This will require isolating the inverter chassis from the cabinet it's mounted into.

 

I'm in Europe and see that translation of terms is required, as you explain.
Also, our power grid does not have a neutral as seems to be the case everywhere else in the world.

Most power grids around the world do not include a neutral over HV line transmission, i.e a balanced three-phase system,
neutral is created at the termination points.