We see many questions about parallel systems - what is a normal delta between inverters, how can I improve the imbalance and even how do I measure this? So I thought I would share some of my experiences.
You aren’t always onsite and a clamp meter isn’t always on hand (though this is an important tool for working on systems like this).
I created a flow here, that helps, using nodered, to display variance locally or remotely, in real-time using VRM or the nodered dashboards.
No system will be perfectly balanced, there are too many variables and component tolerances involved, before you consider inverters of different ages.
Anything in the mid to low single digits is reasonable, so, 6-7% is respectable.
Bear in mind that measurements are less accurate at low loading, so check across a broad range of loads.
Read the manuals and wiring guides. It is counter-intuitive, but parallel systems do not want low resistance AC wiring, it is the resistance that helps the system balance, so too short, or too thick a gauge wiring, can amplify an imbalance.
If you have to RMA inverters, this can also cause additional complications by introducing a younger system.
Even with a 6-7% delta, one inverter will “wear” more than another, or rather its internal characteristics will change faster than the other.
On my personal system, after 18 months or so of use, I noticed the delta had crept up to 9-10%, so I wanted to get them closer together.
In the thermal images you can see the one multi running hotter than the other (by almost 10%).
Due to location, my wiring was quite short, so I decided to increase the length of both AC IN/OUT to 2m per inverter, adding an additional 10% to the multi that was working hardest.
The result was that the inverters came down to 2-3%.
An observation was, that, with grid disconnected, the imbalance was a bit higher at 5-6%, but again well within the expected tolerance.
Hopefully this is of use to some of you starting out with these types of systems.