Honestly, I have not run the system off grid for any real length of time. I have it connected to my main panel off of a 20 amp 240 volt split phase circuit. The output is feeding a sub panel for my "essential loads" that will get backup power. The Enphase solar is also connected in that sub panel to keep the solar operating in a power failure. I currently only have about 150 watts running on the sub panel. During the sunshine, I have it charging the battery bank up to 57 volts, 58.8 would be 100% on 14S NMC, so this is about 85% full. Then from 4pm to 9pm I have it run in "Grid Sell" mode to push 6 amps back to my main breaker panel. My power meter runs backwards, slowly, just like it does off the Enphase inverters at 1pm. For about the first hour, it is only pulling 300 watts from the battery, and it ramps up as the solar output falls to about 1,400 watts as it is limiting the "Grid Sell" power, so if the solar is pushing back more than 1,400 watts, it actually goes into shut down. That part works great. I can't notice any change or flicker in my LED lights during any phase of operation. From grid power only, Grid + Enphase, Grid + XW-Pro.
I have done 3 short "Off Grid" power fail tests. The longest was maybe 15 minutes. In all 3 tests, a couple of the Enphase iQ7's would drop offline, but the rest were running fine with no error. At one point I did have the Enphase solar producing over 2,500 watts and it was charging the battery bank at over 50 amps. I then turned on 1,000 watts of load, and saw the charge current fall to about 30 amps as expected. I did not have any LED lights connected, so I can't say how stable the power was, but my fan running and meter were not noticeably fluctuating. The frequency output was dead solid on 60.0 Hz and the voltage was holding steady at about 122 volts per leg. At that point, 3 of the iQ7 inverters were offline, which is why I was only making 2,500 watts. I top out over 3,500 watts on a good day near solar noon. The microinverters would not come back online until the light was blocked from hitting the solar panels. I was able to get to two of them and cover with a piece of hardboard. My other tests were done close to sunset, so total production was much lower, and as many as 7 panels went offline from the switchover. I was only getting 500 watts from solar. I turned on my roughly 1,000 watts of load, and saw the battery go from charging to discharging to make up the other 400 watts. Even in that mode, it was all working fine. The next morning, all of the panels came up again no problem. Enphase is looking into the issue. It seems to be a problem with the California Rule 21 frequency/watt profile. Even thought the power was completely stable, the inverters were showing "grid frequency out of range". Even after well over the 5 minute lockout, they would not come back on. Enphase said they have also been seeing this on Tesla Power Wall systems. The very odd part, is that even shutting off the AC disconnect so that the microinverters have to go to full grid fault shut off, and then connecting them back to a solid grid, the locked out inverters would still not come back. The DC input from the solar panel had to go away to reset it. That is a problem.
I took a pretty bad power glitch 2 days ago. The UPS on my PC kicked over, and the lights went full off and back on. Still less than a full second, but I would say the power was dead close to 1/2 second. 14 of my 16 Enphase inverters showed "Grid Instability" but they all continued to make power. They did not even do the 5 minute reset. The Schneider XW-Pro did not even store an error in the log. I totally expected it to do a 5 minute anti islanding, but it didn't. We had a huge rainstorm yesterday, and had no glitches at all, even with some huge lightning not too far away.
The Schneider XW series is essentially a basic dumb H bridge inverter. Over the years, the + and now the Pro versions have just added capability to the control board driving the H bridge. It measures the voltage and current at the two inputs and the inverter output. I have a socket for my generator connected to the second input, so it shows zero now. It has a 60 amp contactor to connect the input to the output, but the inverter bridge, is always connected to the output terminals. The input to the H bridge is directly connected to the battery bank terminals. The output side of the H bridge goes to the primary winding of a HUGE toroid transformer. The transformer is fully isolated so the core has to handle the full 6,800 watts of the inverter. I would not be shocked if the transformer alone is a solid 80 pounds, or more than half the weight of the entire inverter. The secondary windings are 2 separate coils. Default mode has them in series to make the split phase, but they can be changed to a parallel config for 120 volt only output. It is very well built with very solid components. My only complaint now is the software not allowing it to switch back to charge from AC on it's own. The high pitched wind out of it seems to have gone down quite a bit, but maybe it's because I am running a little less current and getting used to it. But I remember the whine being real bad, even at fairly low current. Oh well. I no longer cringe going into my garage.