The inverter is the component most often replaced and most often replaced unnecessarily. That is not because technicians are careless; it is because the inverter is where every other part of the system reports its problems. Understanding what is happening inside it — and what its display is actually measuring — turns a fault code from a verdict into a starting point.
What the box is doing
A modern string or hybrid inverter is several converters in one enclosure, each with its own protection logic.
- Input stage / MPPT. Presents an adjustable load to each string and hunts for the operating point where voltage times current is greatest. That point moves continuously with irradiance and cell temperature.
- DC bus. An internal high-voltage rail that everything else feeds from and draws on. Storage, where fitted, usually couples here through a bidirectional DC-DC stage.
- Inversion stage. Switches the DC bus into an AC waveform at grid frequency, synchronised in phase and voltage with the utility.
- Grid interface and protection. Anti-islanding, voltage and frequency windows, ramp rates, and whatever the local grid code requires. This layer is legally obliged to disconnect in conditions that are not the inverter's fault.
- Monitoring. Sensors, logging, and the display or portal you are reading — which reports derived values, not raw truth.
MPPT explains most "underproduction" calls
Two facts do a lot of diagnostic work. First, nameplate is defined at 1000 W/m² and 25 °C cell temperature; on a hot roof cells sit well above ambient and every degree costs roughly 0.3–0.4% of power, so 75–85% of nameplate on a clear hot afternoon is normal, not a fault. Second, strings on a shared MPPT are pulled to a common operating voltage, so one shaded or mismatched string drags the others off their own optimum. A single soiled or partially shaded string on a shared tracker can cost far more than its share of the array.
When a customer reports underproduction, compare measured output against a condition-corrected expectation, not against nameplate. If you cannot state irradiance and temperature, you cannot yet say whether there is a fault.
What a fault code is and is not
A fault code names the protection that operated. It does not name the cause. Some common patterns worth internalising:
- Grid overvoltage / voltage rise. Very often the network or the AC cable run, not the inverter. Check AC voltage at the terminals against the point of connection under export.
- Insulation resistance / isolation fault. Usually moisture in a connector, a damaged cable at a clip, or a compromised junction — frequently weather-correlated, appearing after rain and clearing by midday.
- Arc-fault detection. Sometimes a genuine marginal termination, sometimes nuisance tripping from electrical noise. Timestamps and repeat patterns distinguish them.
- Thermal derate. Airflow, ambient, mounting position, dust in the heatsink. Check the install before the electronics.
- Battery / BMS communication. The inverter reports a fault it did not cause; the storage side refused an operation.
In every case the diagnostic move is the same: pull the fault history with timestamps, line it up against weather, time of day, and household load, and look for correlation. A fault that only appears at midday in summer and one that only appears after overnight rain are different faults even when the code matches.
Hybrid inverters add arbitration
Once storage is present, the inverter is arbitrating between PV input, battery state of charge and health limits, house load, grid import and export limits, and any tariff or backup schedule the owner configured. A large share of "battery not charging" calls are configuration outcomes rather than faults — a charge window, an export limit, a reserve setting, or a BMS temperature limit doing exactly what it was told.
Read the configuration before you read the hardware. It costs five minutes and resolves a meaningful proportion of storage calls.
Safety notes specific to power electronics
DC capacitors hold charge after isolation — respect the manufacturer's discharge time and verify rather than assume. Strings are live whenever there is light on the array; the AC breaker does nothing about that. Battery terminals can deliver very high fault currents even at modest voltage. Treat the DC side, the storage side and the AC side as three separate isolations that each need proving.
Practical takeaways
- Compare against a condition-corrected expectation, never against nameplate.
- Treat fault codes as the protection that operated, then hunt for the correlating condition.
- Check the install before the electronics on thermal and isolation faults.
- Read hybrid configuration first when storage behaviour is the complaint.
- Prove three isolations — DC, storage, AC — not one.
Closing
Power electronics reward technicians who reason about conditions rather than components. The equipment-depth phase of the Solar Service Technician curriculum works through inverter and storage behaviour in scenario form, and EcoPowerHub Academy grades the decisions rather than the recall. Recording those findings cleanly — conditions, photos, measurements against a template — is what EcoService OS is for once you are back in the van.