24V Battery issues with Hoymiles

A Hoymiles inverter provides confusing power and efficiency data on a 24V battery with high input current

When operated on a 24V battery and with currents of more than 10 A per input line, which corresponds to about 240 W per input, the internal current metering gets saturated, such that no higher currents are reported. So even if the actual input and output power increases, the reported power stays at this level, and the efficiency calculation may go wrong.

To obtain realistic data for high load situations, one cannot rely on the DTU but needs to use external measurement equipment, such as a Shelly Plus 1PM on the AC connector.

Below you will find some links discussing about the findings of forum members.

A very lengthy (german) discussion about this is here: Nach 6 Monaten Trial & Learnings - Meine 51,2V / 100Ah / HM-1500 onBattery Anlage

And here is another lengthy discussion with surprising findings: LifePO4 25,6V Leistungsanzeige (Watt) inkorrekt

Joe6899 commented Jan 25, 2024 I have also observed the same behavior for dual 300W inverter inputs on 600W inverter.

I believe the reason why the output power appears to be limited to 500W is due to the way the input DC current sensor is designed. Basically at an input current >~10A it reaches a saturation point. No matter ho much more current is drawn, the measured input current value stays the same.

The DC input current is measured using a LS shunt resistor. The shunt resistor has been set to provide maximum accuracy over the intended operating range. Intended operating range is 35V PV panel. So this means that when peak output power is delivered, the maximum DC input current will always be less than 10A. This is also true at low light conditions - low input PV voltage is seen at low power.

When using a 24V battery the peak output power results in DC input current more than 10A. The input DC current reading is no longer correct. But importantly the control loop on the output power stage still limits it to the max rated output power - 600W. You really do achieve the max rated power even at 24V. The problem now is that if you used the "measured" input current together with the actual output power, you would get the magical situation of a an efficiency level or more than 100%. To counter this, I believe the software guys at hoymiles have been a bit clever. They cannot know the actual input current for currents >10A. But to make the efficiency seem plausible, they have limited the declared output power with 10A current based on an efficiency of ~95.5%.

On the discussion on whether it is better to use 24V or 48V. 24V clearly will cause measurement issues at max power, but I think it does have some reliability advantages.

The input electrolytic capacitors on the hoymiles inverters are rated to 65V. Electrolytic capacitor lifetime is strongly affected by voltage. Rule of thumb - half the voltage doubles the lifetime. The next issue is the max transient input voltage. e.g assume inverter is running at max power and BMS/switch/ or relay is opened. In this instance inductive kick will result in far higher voltage. It is plausible to vastly exceed the 65V rating. So if 48V is used I would be careful not to disconnect the input voltage when under output load.

There is one more helpful hint, courtecy of member Tehzap:

I had a similar problem initially when operating both mppts on a 24V battery. I observed different power input on both channels, although obviously the same is expected. Also the AC power output was fluctuating between the real and an odd number. The real output was measured by a Shelly in series, which showed that actually only the measurement was fluctuating but not the real power output. What solved the issue for me was connecting the negative channels using a Y cable instead of leading two separate negative cables to my busbar. Since then I got identical power output and no more power fluctuations in openDTU.