Inverter Fault Service Call

Purpose

An inverter fault call is won or lost before the truck rolls. The code, its exact timestamp, whether it repeats at the same hour and what the weather was doing decide which parts and skills go on the truck, and a tech who arrives without that ends up doing the one thing this procedure exists to prevent: pressing reset, watching the green light, and leaving.

That reset is not a repair. On a grid over-voltage fault it hides a condition that will trip again at noon tomorrow. On an arc fault it re-energizes a circuit that may contain a live arcing termination, the one code in the set where a reset can start a fire rather than clear a nuisance. This separates the codes where a reset is a legitimate diagnostic action from the codes where it is not.

Scope

Covers service calls on a reported inverter fault, alarm or zero-production condition on grid-tied residential and light commercial PV: string inverters, microinverters and DC optimizer systems.

Does not resolve the diagnosis itself; the Solar troubleshooting shelf owns the isolation-versus-arc-fault and grid-voltage-rise reasoning, and this routes to it. Does not cover battery and backup-island faults. Does not cover the energized DC work practice every step assumes, which is the energized DC array safety standard.

Roles and responsibilities

Procedure

1. Pull the evidence before the truck is booked. From the monitoring portal capture the fault code and its exact timestamps, whether it recurs at a consistent hour, production for the seven days either side, firmware version, equipment serials, and the weather at each fault time. Ask the customer what they saw, whether the system was already reset and how many times, and whether anything electrical changed at the property. Acceptance: a code, at least one exact timestamp, and a yes or no on recurrence, on the ticket. Wrong is a ticket reading "inverter error," which sends a tech to read a screen the office could have read. Stop rule: a report of a burning smell, a burn mark, smoke or a tripped main is not a normal service call - dispatch it as an emergency and tell the customer to open the AC disconnect if it is outdoors and reachable, keep everyone away from the equipment, and touch nothing else until the tech arrives.

2. Look and smell before you meter. Inspect the inverter, disconnects and any combiner before touching anything: scorch marks, melted or discolored insulation, a bulged enclosure, staining below a connector, a burnt or acrid smell, buzzing or crackling. Acceptance: no visual or olfactory evidence of thermal damage anywhere. Wrong is reading a code off a screen while standing next to a combiner with a brown streak under it. Stop rule: any of those signs stops the call - open the AC disconnect, do not open the enclosure, move yourself and the customer back, and call the service manager, because an enclosure showing thermal damage can hold an active arcing fault that opening it feeds with air. Hazard: the array is at full voltage regardless of what the inverter's screen says, and this is the step where that is easiest to forget because nothing has gone wrong yet from the tech's point of view.

3. Capture the state before you change it. Photograph the fault screen, the labels with serials and firmware, and every termination visible without opening an enclosure, and export the fault log if the platform allows. Acceptance: photographs and a log export timestamped before any reset, power cycle or setting change. Wrong is a reset first and a photograph after, which destroys the evidence the manufacturer will want in a warranty case. Stop rule: if the unit will not export a log, photograph the on-screen fault history before touching anything, because a power cycle clears volatile history on many platforms.

4. Classify the fault family and measure the one thing that family turns on. Sort the code into grid, isolation or ground fault, arc fault, thermal derate, or communication, then take that family's measurement. For a grid fault, measure AC voltage at the inverter terminals, at the main panel and at the meter, at the time of day the fault occurs, with the system exporting. Acceptance for the grid family: service voltage inside ANSI C84.1 Range A, which puts a 240 V nominal split-phase service between 228 and 252 V, measured at the meter, the rise from meter to inverter terminals inside the shop's own target for its AC run, and the terminal voltage itself inside the inverter's listed operating window, because the terminals are where the inverter makes its own trip decision. Wrong is measuring at 8 a.m. when the fault happens at 1 p.m., because export current is what raises the voltage at the inverter and there is none in the morning. Stop rule: a service voltage above Range A at the meter is a utility condition, and the inverter's grid-support trip settings are not the fix - those settings are part of the equipment's listing and the interconnection agreement, so changing them to stop a trip is not a repair the shop makes on its own authority. Hazard: measuring AC at inverter terminals is energized work, so it follows 29 CFR 1910.333(b)(2) for general industry or 29 CFR 1926.417 on construction, with PPE per NFPA 70E-2021 in the edition your electrical safety program adopts.

5. Apply the reset rule, which is different for arc faults. For a non-arc code with no thermal evidence and no prior occurrence, one reset is a legitimate diagnostic action: reset once, observe through at least one full production ramp, and record the result. Acceptance: the code does not recur within the observation window. Wrong is a second reset on the same code, which converts a diagnosis into a habit and puts the shop on record as having cleared a fault it never explained. Stop rule: an arc fault code is never reset before a physical inspection of the DC terminations, connectors and conductor runs, because the reset re-energizes a circuit whose detected condition may still be present and arcing; and a second occurrence of any code goes to physical inspection and to the service manager rather than to another reset.

6. Establish why a protective device operated before calling it the fault. An arc-fault or ground-fault detector that has tripped is reporting something until proven otherwise. Inspect connectors for discoloration or heat marking, check terminations for looseness, and where an isolation fault is indicated measure insulation resistance at the test voltage the inverter manual specifies, against that manual's threshold. Acceptance: either a physical cause found and corrected, or measurements clearing the array side against the manufacturer's stated limits. Wrong is ordering a detection board because the array checked out on a dry afternoon when the log shows every occurrence on a wet morning. Stop rule: no protective device is replaced or bypassed as a nuisance until the array-side cause is ruled out under the conditions the fault occurs in, because replacing a correctly operating device reaches the same end state as jumping it, one step slower and with a part number on the invoice.

7. Escalate to the manufacturer with a complete case. Open a technical support case before removing any unit, supplying the serial, firmware version, exported log, photographs and the step 4 measurements. Acceptance: a case number on the ticket and the manufacturer's written disposition. Wrong is pulling a unit first and calling later, which is how a warranty claim becomes a purchase. Stop rule: no inverter leaves a site without a case number or an explicit instruction from the service manager to proceed without one.

8. Close with the customer and set the watch. Say in plain words what was found, whether the system is in service or out, what happens next, and what they should call about. Set a monitoring watch for the same code with a named owner and a review date. Acceptance: the customer can state whether their system is running and what happens next, and a dated watch exists. Wrong is "we reset it, keep an eye on it," which delegates the shop's job to someone with no monitoring login. Stop rule: a system left out of service is written into the ticket as out of service with the reason, so nobody closes the job as complete.

The record this produces

• Pre-roll evidence: code, timestamps, recurrence pattern, seven-day production either side, firmware, serials, weather at fault time, customer statement including prior resets.
• On-site findings: inspection photographs, log export, family classification, every measurement with its location and time of day.
• Actions: reset taken or refused and why, physical corrections, protective device disposition.
• Escalation: manufacturer case number, written disposition, RMA if any.
• Close: in service or out, what the customer was told, watch owner and review date.

The service manager reviews repeats across the fleet, because the same code on three systems from one install month is a workmanship or firmware pattern rather than three unrelated calls. The manufacturer reads the log and photographs when the claim is decided. The next tech reads step 4's measurements to know whether the condition was ever explained or merely cleared.

A worked pass, including the step that failed

Residential system, string inverter, reported as "shuts off every afternoon and comes back."

Step 1: the portal shows a grid over-voltage code with occurrences at 12:40, 13:10 and 13:55 across three clear days, no occurrences on the two overcast days between them, firmware current, no customer resets. That recurrence pattern books the call for midday rather than for the first open slot. Step 2: no scorch, no smell, no discoloration at the inverter, disconnects or the AC junction. Step 3: fault screen photographed, log exported, serials and firmware captured before anything is touched.

Step 4 passes on the meter reading and on the rise, and fails on the terminals. Measured at 13:20 with the system exporting: meter 249 V, main panel 251 V, inverter AC terminals 253 V, on a 240 V nominal split-phase service. The meter sits inside Range A, whose ceiling is 252 V, and the 4 V rise from meter to terminals is 1.6 percent of 249 V, inside the shop's 2 percent target for its own AC run, so nothing the shop installed is at fault. What fails is the terminal criterion: at 253 V the inverter is above its own listed window and is tripping correctly. The problem sits upstream, because a 249 V baseline against a 252 V ceiling leaves only 3 V of room for any rise at all. The stop rule runs and the trip settings are not widened.

Steps 5 and 6 do not apply: the fault is explained so a reset would clear nothing, and no DC-side protective device operated. Step 7: a case is opened with the log and the three voltage readings, and the manufacturer confirms the unit is operating to its listed settings. The shop separately opens a voltage complaint with the utility, attaching the meter reading, the timestamps and the production data showing the correlation with midday export, and requests a transformer tap review.

Step 8: the customer is told the array and inverter are healthy, that the trip is the inverter protecting itself against a supply voltage near the top of its allowed range, and that afternoon production will keep dipping until the utility acts. A watch is set on that code with a named owner and a two-week review date.

Outcome: three afternoons of lost production explained with numbers rather than reset away.

References

• ANSI C84.1 for nominal service voltage ranges, in the edition your utility and equipment documentation reference.
• NEC 690.11 for PV DC arc-fault circuit protection, applicable to DC circuits at 80 V or greater on or penetrating a building, in the edition your AHJ has adopted.
• 29 CFR 1910.333(b)(2) and 29 CFR 1926.417 for energized electrical work practice, with PPE per NFPA 70E-2021 in the edition your electrical safety program adopts.
• Inverter manual for fault code definitions, isolation resistance test voltage and threshold, and warranty case requirements.
• See related: the energized DC array safety standard, and the Solar troubleshooting articles on grid over-voltage, arc fault and ground fault conditions.