Hmm.
This can be a complex issue, before you factor DESS.
Typically, in ESS, you would expect a varying setpoint, influenced by loads.
A static negative one is not typical, though the chart shows it pivoting around -100W.
Does it respond to manual changes in the setpoint setting?
In parallel systems I have seen strange behaviour with mismatched units in a pair.
The grid itself, installation can all play a part.
Maybe pick one install and detail it’s configuration, loads and settings, likely easier to try dissect instead of a general problem.
ok, then let’s detail the installation from where I took the above chart.
3xMP-II 5kVA
1xMPPT 250/60
1xMPPT 150/60
1xSolarEdge SE5K-RWB (tried a SE3680H as well)(on AC-OUT1)(integrated into cerbo via Modbus TCP)
2xAPSystems DS3-M 0,8kVA (on AC-OUT1)(measured via shelly and integrated into cerbo)
4xTAB Battery (4*5kWh LiFePO4)
1xVM-3P75CT
1xVictron EV Charger (most of the time in summer in automatic mode)
1xtypical household with general purpose loads
max permissable feed in 800W (by utility)
feed in max setting in Cerbo 500W
feed in max setting in SE5k 200W
individual phase regulation
DESS configured but currently disabled
as all PV power data is gatherd by the cerbo the total actual PV power is correctly known by the cerbo
all inverter and loads on AC-OUT1
All Victron devices with newest firmware - venus OS on newest Beta (3.80~39)
When I change the grid set point the achived grid values are changing, so yes I see the effect of parameter change.
What I recognized as well in this installation, that one of the MPPTs is still feeding in at aorund the amount of power which is calculated for AC loads. In feed in situations this leads very often to exceeding the 800W utility limit. That’s another thing where I’m in the dark. Why does the system not throttle the MPPT when it sees an exceedance of the max feed in setting?
The two DS3-M cannot be throttled, the feed in limit in the cerbo is 500W, therfore the MPPT should IMO be throttled to about 100W to achieve the 500W feed in limit.
You had to say solaredge, that’s another problem child, PV inverters often can contribute when they aren’t responding to controls well enough, the SE has had endless issues with comms to the GX.
You have a complex mix of moving parts with supported and unsupported sources and using shelly etc. Having sources that will not respond to throttling isn’t a good idea either.
To work properly the cerbo needs to be able to control generation, not just some of the generation and rely on maths for it all to work.
I would try isolate all the variables and see if one of them is at fault.
So, no peak-shaving, no AC loads, no AC OUT generation, just let the inverter supply what’s behind it from battery.
Peel it back to basics then add the more complicated stuff.
If problems are going to happen, it will happen in a setup like this due to its complexity.
The issues with SolarEdge communications made me add an SE meter as well. So the feed in limit for the SE inverter is done directly within SE universe without involving the cerbo. This is not a nice solution but at least it is stable now.
no generation, only AC loads on battery is working good as well at night, as you can see from my chart.
Because of the complexity, I’ve already tried to strip the installation on many occasions earlier but unfortunately yet no major problem solving idea came up in the process.
Next step will be to remove the two DS3-M which are not allowed by the utility in an installation > 0,8 kVA. I’ll report then with hopefully new findings.
The PV won’t be helping. The GX will know it’s there and not be able to manage it. Whether it should be able to compensate or not is another question. Not being the way to do things it isn’t something we can exactly raise. It would help to narrow it down.
