Solax x3 hybrid importing from grid

My Solax x3 hybrid g4 15kw + 2 x 5.8kWh battery system is installed in an outbuilding, with supply current monitoring by CT in the house, next to the supply meter. The connection to the inverter is via 20m of (earthed screen) cat6a. We just had the supply meter changed to a UK smart meter so we can get paid for exports.

As far as the inverter is concerned, everything is hunky dory - it thinks it is supplying all the house needs from either PV or battery, and not importing anything, but.... The supply meter disagrees.

When there is insufficient PV input to export, and the batteries may be supplying / supplementing the output, the supply meter says I'm importing 60 to 100W, or around 2kWh per day, which is wiping out any export payments. When the battery is fully charged and the inverter is exporting, the supply meter stops racking up imports. This did not happen with the old meter - that reported about 0.1kWh or less imported per day so it wasn't worth worrying about. This is probably due to the new meter being more sensitive, rather than faulty.

When I compare the phase currents on the inverter and smart meter, there is a difference roughly corresponding to the 100W reported discrepancy.

I've tried changing the Pgrid bias parameter to "grid" and this does reduce the inverter consumption from about 0.5kWh to 0.2kWh / day, but the supply meter still reports 2kWh imports.

The CTs are 100A/33mA so with currents of around 1A in the phases, the feedback voltages in the inverter must be tiny. I'm wondering if I really ought to be monitoring the grid supply currents with a meter and sending RS485 to the inverter instead of using analogue CTs. Any thoughts?

EcoBob said:

I've tried changing the Pgrid bias parameter to "grid" and this does reduce the inverter consumption from about 0.5kWh to 0.2kWh / day, but the supply meter still reports 2kWh imports.

The CTs are 100A/33mA so with currents of around 1A in the phases, the feedback voltages in the inverter must be tiny. I'm wondering if I really ought to be monitoring the grid supply currents with a meter and sending RS485 to the inverter instead of using analogue CTs. Any thoughts?

Click to expand...

How accurately and to what resolution can you monitor grid import / export?

The reason for asking is I'm wondering whether the issue is not so much of a fixed constant bias, but rather whether the inverter is oscillating between import and export as it attempts to keep the grid usage to zero. And, whereas the old meter would not register any import and export within a Wh, the new meter is detecting all import? Just a thought.

Re the CT clamps - if they are true current clamps the voltage won't matter; see

Also, FYI, see this thread that I posted on previously...

I'm recording import & export with the supply meter (edmi es-30b) , which can display import to the nearest Wh and export to the nearest kWh. I timed the interval between 5 successive increments of the Wh import count and got 50, 50, 55, 45, 45 seconds, so a fairly uniform low level import of 65-80W at that time.

I'm using solax cloud as a guide to how much the inverter consumes, but that only updates about every 5 minutes. The statistics state that the inverter hasn't imported anything (except when I charged the battery for 30 minutes) but has exported a few bursts of about 40W, because I have set the Pgrid bias parameter to err on the side of exporting rather than importing when in doubt.

The old meter could display import only to the nearest 0.1kWh, but it took anything up to a week (typically 2 days) for that reading to change. The new meter is recording 30+ times as much import.

I assume the CTs are true current transformers since they are marked 100A/33.33mA. My concern is more about the effect of ambient electrical noise being picked up on the cat6a, rather than voltage drop on the cable (about 5ohms round trip). I've no idea what the burden resistor value in the inverter is.

Oops..... I just checked the CT signal assignments to the cat6a cable and therein might lie a problem. Solax have assigned a T568B cable core colour to pin assignment, with the CT connections on the RJ45 as follows. CT1 - pins 1 & 8, CT2 - pins 2 & 7, CT3 - pins 3 & 6. My installer has followed this on the 20m of cat6a, (there is a cable join about 1m from each end) This means that the O/OW pair would have one CT1 and one CT2 wire, the Br/BrW pair wold have the other CT1 & 2 wires and the G/GW pair would have both CT3 wires. Not sure how critical this would be. Everything is still screened, but there is obviously scope for interaction between phase 1 & 2 feedback. I'll have a good look when I've got time to take things apart

Solved:
The solution to this problem was to enable "phase unbalanced" in the installer settings. Heaven only knows what this has to do with smart meters, but by setting it, the erroneous imports have dropped from about 2kwh/day to about 0.15kWh, which is nearly down to what we got with the old meter. I can live with that.

Our system is currently very unbalanced, with 90% of the load on one phase, pending replacement of the house consumer unit from single phase to 3 phase. The typical background load of TV, fridges, freezers & lighting is usually in the 250 - 500w range with virtually nothing on the other phases