There is a good chance a 3kVA would work. No guarantees, but I have a gut feeling it’s a good try.
The only other idea I had… was maybe you could arrange some sort of backup for the GX device, to get better data. Run it off a 12V alarm battery for example 
I’d love to see if the battery raises an alarm, I would expect at least a High Discharge Current alarm to show up in this case.
I can arrange a temporary battery connected to get all the juicy logs when it goes out, 6pm today.
On a side note yesterday running @ 2250W caused reboot/disconnect, after that I set back to 2000W and the 4pm went smoothly.
There is a good chance a 3kVA would work
Will be some work to swap these out, however doable.
Edit:
Stage is set, will test tonight keep batteries charged and GX on 12vdc battery (will try get the BMU on if it works on 12v as well)
Test done, and failed…
As suggested the BMU and GX (and internet) were powered by a 12V battery and indeed we got more logs, from my trained eye nothing shouts out other than the low battery voltage alarm.
Things that I would have expected to be more pronounced that weren’t,
DC voltage from 54.94V took almost 2 minutes to only drop to ~53.5V, not a massive drop as expected (Both vebus and battery agree)
Moderate DC amps (nothing before shutdown) only after reboot as expected ~13A ~800W for a tiny bit.
Any insight?
Nothing useful in the data. Only thing I can say with some certainty is that the BMS disconnects the cells from the DC-bus. That causes the low voltage and the ripple warning. The event is extremely short-lived.
One thing to know about BYD. The BMS only sends a voltage and current reading every 2 seconds. Anything intermittent may well be missed unless you get lucky.
Either the BMS is faulty and disconnects way too soon, for no good reason, or it has a reason that it is not telling us about. If it was my system and I had all the time in the world, I’d probably put a scope with a current clamp on the battery cable… or a DC-clamp meter with peak-hold activated. But then you have to own that kind of kit… so I’m a little stumped. Best I can offer is the existing theory: System is underbatteried and when running ESS this causes problems. Get rid of the ESS assistant, make the inverter smaller, or make the battery bigger. At this point, sans solar panels, getting rid of the ESS assistant is the simpler option.
In the data, look at the battery voltage the multi reports. Looking at the bms might mislead you. If the BMS is disconnecting it will still report it’s internal voltage while the multi might report something totally different. Just guessing but your answer might lie there.
Do you know how to download the csv files to give you all the data you might need? Sorry did not read the full post…
Thanks @JacoDeJongh , the loadshedding was at 2022/11/22 18:11:57 (the missing data earlier was me testing the GX on the 12V battery)
The voltages seem to be similar mV apart
Parents_20221122-1258_to_20221122-1857.zip (13.9 KB)
@plonkster Petty sure my clamp meter can do hold (or actual only hold the value… ), well at least think, will try.
So I have for now done as suggested, many tests back and forth…
1st tested a BSL 100AH battery on the 5kVA, it worked 100%, then tested the BYD 50AH battery on my 3kVA MultiGrid, it worked, long story short I have now paired the battery with a MultiGrid 3kVA and its working flawlessly now. Thanks