Microinverters Report but One Row Shows Night-Only Faults: Decision Tree

Why this matters

A microinverter array where every unit produces energy during the day but one specific row logs faults after sundown looks like a hardware problem and gets a truck roll with replacement units on board. It almost never is. Production microinverters stop transmitting once they lose DC bus voltage at night, so a fault that appears only after dark is a communication or gateway interpretation issue, not a power-stage failure. The units producing all day are proving their power electronics are fine. Chasing this as a hardware RMA wastes parts and a return trip when the real cause is power-line-communication degradation, a gateway report-interval artifact, or a phase-coupling weakness on one branch circuit. Diagnosing it correctly turns a multi-unit swap into a single coupler or wiring fix.

Symptom presentation

The monitoring portal shows full daytime production across the array, including the row in question. After sunset, that one row flags "no data," "device not responding," "DC voltage low," or "microinverter offline" while the rest of the array reports a normal nighttime sleep state cleanly. The faults clear at sunrise when the units wake and re-establish strong power-line comms on a freshly energized circuit. The affected row often shares one branch circuit, one subpanel, or one phase leg. Severity tends to track ambient load on the building: more household loads on that circuit at night, more reported faults.

Quick checks

Confirm the row is geographically or electrically grouped, not random. In Enphase, map the serial numbers to their branch circuit; in any system, trace the affected units to a common AC home-run. Check whether the gateway (Envoy/IQ Gateway or equivalent) sits far from that branch in the panel, since power-line communication weakens with distance and across phase legs. Pull the comms-level or signal-strength report if the platform exposes it. Verify the fault is purely nighttime by comparing the event timestamps against local sunset. If every fault lands within an hour of sundown and clears at dawn, you have ruled out a power-stage defect already.

Isolation tree

Branch one, is it real loss of communication or a reporting artifact? Check the gateway report interval and last-successful-report timestamps. Some gateways flag "offline" simply because a unit's last expected nighttime heartbeat did not arrive within the polling window while the array is asleep. If daytime data is complete and energy totals reconcile, the nighttime flag is cosmetic; document and monitor. Branch two, power-line-communication strength. If signal-level data shows the affected row at the bottom of the range, the night-time household loads (switching power supplies, LED drivers, variable-speed motors) inject noise that swamps the weak PLC carrier on that branch after dark. Branch three, phase coupling. On a split-phase or three-phase service, the gateway must couple its signal across legs. If the affected row sits on the opposite leg from the gateway and there is no phase coupler, comms degrade, worst when cross-phase noise peaks at night. Branch four, a single bad AC connection or long home-run on that branch raising impedance to the carrier.

Confirming diagnosis

Confirm by correlating three facts: the affected units produce full energy by day (power stage good), lose comms only after sunset (DC bus de-energized, PLC marginal), and share an electrical path distinct from the healthy rows. Those three together exclude a power-stage failure, because a failed power stage would drop daytime production first, not nighttime telemetry. Add a temporary load on that branch midday and watch comms hold, then repeat the same load at night and watch comms drop, which isolates noise-versus-distance: if the same load kills comms only at night, the night-time accumulation of household switching loads is swamping the carrier, not the distance. If a phase coupler is missing on a multi-leg service, installing or relocating it and rechecking signal level proves phase coupling was the gap, since the affected row will jump from the bottom of the signal range up to par. Verify energy totals reconcile to the revenue meter so you can confidently call the nighttime flag non-impacting if it is purely a reporting artifact; if daytime kWh matches the meter, no production was lost and the nighttime "offline" is cosmetic.

Remediation

If signal level is marginal, relocate the gateway closer to the affected branch in the load center, or add the manufacturer phase coupler so the carrier bridges legs on a split-phase or three-phase service. Install a noise filter on the offending noisy load if one circuit (a switching power supply, LED driver bank, or variable-speed motor) is the culprit. Re-terminate any high-impedance AC home-run lug that raises impedance to the carrier on that branch. If the gateway report interval is producing false nighttime "offline" flags and daytime energy reconciles to the meter, update gateway firmware and adjust the report window per the manufacturer guidance rather than swapping units, and set the customer's expectation that a nighttime sleep flag with full daytime production is not a fault. Reserve a microinverter RMA for a unit that actually under-produces during the day with a logged power-stage fault, which is the exact opposite of this symptom and the only case where replacing hardware is warranted.

References

• Enphase IQ Gateway (Envoy) installation and operation manual, power line communication and phase coupling guidance
• Enphase Microinverter troubleshooting guide, DC voltage and nighttime sleep behavior
• UL 1741 SB, grid-support inverter functions and operating states
• IEEE 1547-2018, interconnection states and ride-through behavior
• NEC 2023 Article 690.4 and 690.6, microinverter and AC module circuit requirements