Inverter Fault - Restart vs Replace vs RMA Decision Tree

Why this matters

An inverter that has thrown a fault is the most common solar service call. The wrong response is to RMA every inverter that flashes a red light - 30 percent of those clear with a restart and the customer pays for a truck roll plus a needless RMA shipping cycle. The wrong response in the other direction is to keep restarting an inverter that has a degraded capacitor stack, watching it slowly cook itself over months. The tree below splits faults by code class into restart-acceptable, restart-once-then-investigate, and RMA-immediate.

Symptom presentation

Pattern A: monitoring shows a fault code with the system back online when tech arrives. Pattern B: inverter is solidly faulted, no production. Pattern C: intermittent production drops with fault code logged repeatedly. Pattern D: ground fault. Pattern E: arc fault.

Quick checks - 12 minute pass

1. Pull the fault code and timestamp from the inverter display or the monitoring portal.
2. Confirm utility power present and inverter has AC supply (a tripped backfeed breaker reads as "grid fault").
3. Confirm DC supply present - if it is sunny noon and DC voltage at the inverter is zero, the issue is upstream (DC disconnect open, blown fuse, broken connector).
4. Check the inverter for visual signs - moisture inside the display, burned smell, swollen capacitors visible through vents.
5. Visual inspection of cooling fan operation (string inverters with active cooling) - clogged fan is the leading cause of repeated thermal faults.
6. Confirm the inverter firmware version. Some persistent fault codes are firmware bugs cured by manufacturer update.

Isolation tree

Step 1 - Restart-acceptable faults

These commonly clear with a power cycle and do not indicate hardware failure:

• Grid Voltage Out of Range, Grid Frequency Out of Range: utility-side anti-island event. Inverter is doing its job. If utility had a known event, restart and document. If recurrent, the issue is at the utility transformer or POI - call utility.
• Excessive Temperature: clean cooling fans, clear obstructions, restart. If recurrent in winter, the inverter is degrading.
• Inverter Internal Communication Timeout: cosmetic or transient. Restart once.

If a restart-acceptable fault recurs within 30 days, it is no longer restart-acceptable. Move to restart-once-investigate or RMA.

Step 2 - Restart-once-then-investigate faults

These need a restart but the root cause is real and must be found:

• AC overcurrent: short circuit downstream, breaker capacity issue, or inverter fault. Inspect AC wiring, breaker condition.
• DC overvoltage: panel string voltage too high (cold weather morning hits theoretical Voc). Confirm string design margin. Adjust string sizing if marginal.
• DC undervoltage: shading, panel failure, broken DC connector.
• Ground fault: see Step 4.
• Arc fault: see Step 5.
• Internal sensor fault: drive board reading out of range. Document. Investigate.

If the cause is upstream, fix upstream. If the cause is the inverter itself, escalate to RMA path.

Step 3 - Immediate RMA faults

These are unambiguous hardware failures:

• Permanent Hardware Error, Internal Fan Failure (after fan replacement), Capacitor Fault, Driver Board Failed, Display Module Failed: all are hardware. RMA per manufacturer process.
• Burned smell or visible damage: do not restart. Power off, photograph, RMA.
• Repeated thermal faults after fan and air-path service: RMA.
• Power output below specification with no DC, AC, or environmental cause: degraded inverter electronics, RMA.

Step 4 - Ground fault deep dive

Ground fault codes (SMA "GROUND FAULT," SolarEdge "Isolation Fault," Enphase "DC Ground Fault Current") signal leakage between DC conductors and ground. Causes:

• Wet conduit, condensation in junction box, water in MC4 connector. Restart after drying, monitor.
• Backsheet failure on a panel (active fire risk).
• Cable insulation damaged by chafing at conduit edge or by rodent.
• Inverter internal isolation sensor failed (rare).

Isolation method: isolate strings one at a time, restart inverter. The string whose isolation removal clears the fault is the location of the leak. Walk that string with the inverter on and an insulation tester (Megohmmeter) to find the leak point.

A ground fault must not be reset and ignored. The reset clears the alarm; the leakage continues. Find the source.

Step 5 - Arc fault deep dive

NEC 690.11 requires arc fault circuit interrupter on residential PV. The inverter watches for DC arc signatures. Faults:

• Loose MC4 connector at end of cable run. IR thermography during high production identifies the hot connector. Replace with field-installed connector of matched manufacturer.
• Series arc inside a junction box (loose ferrule or terminal block screw).
• Series arc inside the inverter itself (internal connection loose, capacitor solder joint failed).
• False positive from inverter EMI - SMA and SolarEdge have firmware patches for known false positive cases.

After an arc fault, the inverter requires a manual reset (it does not auto-reset). Identify the cause before re-energizing.

Step 6 - Manufacturer-specific patterns

Each major inverter has known fault patterns:

• SMA Sunny Boy: State 105 ground fault, State 7202 fan failure. Service via authorized SMA service partner if under warranty.
• SolarEdge HD-Wave (Gen 1): capacitor batch issue in some 2018-2019 units, manifests as Error Code 18x. Manufacturer offered replacement program.
• Enphase IQ7: per-microinverter fault on a single panel is a single-unit RMA. Enphase ships replacement, customer pays labor.
• Fronius Primo: STATE 105 is utility-side; STATE 553 is internal fan. Service network.
• Solis 4G: Permanent Fault Code 21-29 hardware - RMA.

Confirming the diagnosis

Document the fault code, timestamp, weather conditions at fault, restart history, and any associated production data. RMA documentation typically requires panel and inverter serial numbers, fault log printout or screenshot, and customer authorization.

Remediation

• Restart-acceptable, single occurrence: restart, document, monitor 30 days.
• Restart-once-investigate: restart, run upstream diagnosis, fix the cause.
• Ground fault: do not normalize - locate the leak.
• Arc fault: do not normalize - locate the loose connection.
• Permanent hardware fault: RMA per manufacturer.
• Out-of-warranty inverter at end of life: replacement evaluation per system economics.

References

• NEC 690.11 Arc-Fault Circuit Protection (DC).
• NEC 690.41 System Grounding.
• UL 1741 Standard for Inverters, Converters, and Interconnection System Equipment.
• IEEE 1547 Standard for Interconnection and Interoperability of Distributed Energy Resources.
• SMA, SolarEdge, Enphase, Fronius, Solis service manuals and fault code references.