SolarEdge Inverter Error 36 — DC Injection Fault Explained

What this code means

Everything downstream of your inverter — your home's wiring, the street's transformer, your neighbors' equipment — is built for alternating current. Direct current leaking onto the AC side is not just untidy; a DC component flowing through a distribution transformer pushes its magnetic core toward saturation, which overheats the transformer and distorts power quality for everyone on it. That is why every grid code on the planet sets a hard limit on DC injection, and why inverters are required to watch for it continuously.

Error 36 is the tripwire: the inverter constantly measures its AC output for DC content and disconnects the moment it crosses the threshold. Single-phase inverters show 36 on older firmware or 2x24 on v3.19xx and later; three-phase inverters report per line — 72, 74, 75 (older) or 8x27/8x29/8x2A (newer). The meaning is identical everywhere: DC where only AC should be.

What makes this fault unusual is the thinness of its troubleshooting row. Most rows name a cause and checks; the DC Injection row carries only 'Contact SolarEdge support.' That absence is the message: no external condition to wait out, no setting to correct, no useful homeowner measurement. It is either a transient support clears from diagnostics or an internal power-electronics problem ending in a service case.

In practice the outcomes split cleanly: a one-off event that never returns after a single careful power cycle was likely transient; a 36 that returns or never clears means the inverter keeps measuring real DC, pointing at the power stage and often a warranty replacement.

Common causes

• An internal power-electronics fault in the inverter's output stage. This is the cause most consistent with a persistent or recurring 36: something in the conversion circuitry is genuinely passing a DC component, and only SolarEdge's diagnostics — usually run remotely — can confirm it.
• A transient measurement excursion. Inverters sample their output thousands of times per second, and a sufficiently violent grid event can produce a momentary reading that crosses the DC threshold without any underlying defect. The single power cycle described below is the test for this case.
• A firmware-level measurement issue. Support occasionally resolves DC injection flags with a remote firmware action when their diagnostics show the "injection" was a sensing artifact rather than real DC on the output. This is diagnosed from the inverter's logs, never from anything visible at home.