Generator All Readings Nominal But Customer Says It Failed: Reading In Spec Decision Tree

Why this matters

A unit that tests perfectly on the tech's visit but failed for the owner during an actual outage is the most consequential intermittent in standby generation. The owner experienced a real loss of backup power; the diagnostic record shows green. Calling it "no fault found" without a structured pass risks a second failure during the next outage, which is both a liability and a relationship problem. The right move is a methodical search for what the snapshot reading can miss: a transient outside the sample window, a fault that only appears under load or under specific transfer conditions, a sensor that reports normal while the underlying process is not.

Symptom presentation

Owner reports the genset failed to support the house, business, or critical load during a recent utility loss. On arrival, the tech finds the unit on, no active alarms, exerciser history clean, control panel showing nominal voltage, frequency, oil pressure, coolant temperature, and battery voltage. Run hours have accumulated; the most recent self-test or exercise logged a pass. The fault window may have been seconds (a brief stumble that cleared) or minutes (a shutdown that recovered on the next try). Sometimes the owner reports the lights blinked then stayed on utility, indicating the transfer never completed or reversed quickly.

Quick checks

Before assuming an intermittent, rule out the easy explanations. Confirm the unit was actually called to run during the outage by reading the event log or run history with timestamps that align with the outage window. Verify the transfer switch position history if the ATS logs events. Check whether utility loss reached the ATS at all (a service-side fault upstream of the ATS would leave the genset idle by design). Confirm the exerciser is set to a meaningful load test, not a no-load warmup; many failures only surface under load.

Walk the fuel system perimeter for evidence of a recent low-fuel event: tank gauge, regulator vent path, LP tank frost ring during cold operation. Check the battery state of charge and the charger output. Inspect for rodent damage, water intrusion in the enclosure, and any signs of a controller reboot (lost time clock, default settings).

Isolation tree

Start with the event log. If the log shows the unit cranked and ran for less than the outage duration, you have a shutdown that recovered, not a no-start. Read the last fault code even if it has self-cleared; many controllers retain history for the last several events. Common self-clearing trips include low oil pressure on a cold start (sender lag), overspeed on a load step (governor tuning), overcrank if the first attempt failed and a later cycle succeeded, and undervoltage on a heavy inrush.

If the log shows the unit never cranked, the question moves to the start signal path. The ATS must sense utility loss, time out the start delay, and close the two-wire start contact. Bench the two-wire signal at the genset terminal with the ATS simulated to outage; the genset should crank on command. If it does, the ATS sense circuit or the start-signal wiring failed during the actual event.

If the log shows the unit ran but the load never transferred, the ATS transfer mechanism is the suspect. Confirm the ATS saw the genset reach pickup voltage and frequency; some ATS controllers require a stable window before transfer.

If logs are clean across the board and the owner is certain the outage occurred, set a data logger for the next exercise under load. Capture voltage, frequency, oil pressure, and battery voltage at one-second resolution through a full transfer cycle.

Confirming diagnosis

Stage a controlled transfer test that mirrors a real outage. With owner consent and load notification, open the utility disconnect at the ATS or use the test switch to drop utility. Observe the full sequence: start delay, crank, run, warm-up, pickup, transfer, load acceptance, run on load for at least fifteen minutes, retransfer, cool-down, stop. Note any hesitation, voltage sag below regulated band, frequency excursion outside the controller window, or alarm flag that did not store as a code.

If the controlled test passes cleanly and the owner reports no recurrence on the next utility event, document the test and the data and close as confirmed operational. If the controlled test reveals a marginal reading (oil pressure that trends low under load, voltage regulation that hunts, frequency that overshoots on load step), treat that as the lead and pursue the relevant subsystem.

When the test passes but the owner reports another failure, escalate to long-term data logging on the controller's serial or network output. Some controllers expose a continuous event stream that can be captured to a laptop or remote monitor for the next exercise cycle and the next real outage.

Remediation

Treat the visit as a complete commissioning rather than a no-fault drop-in. Verify fuel supply pressure under load, battery condition with a load tester not just a voltmeter, all coolant and oil levels, air filter restriction, and the integrity of all sensor connections. Snug any control panel terminations that have loosened from vibration. Replace any sensor whose reading drifts during the controlled load test. Re-tighten the governor and AVR settings to spec if voltage or frequency wander outside band. Update the controller firmware if the manufacturer has issued a service bulletin for the symptom pattern.

Schedule a follow-up exercise within thirty days under load and review the log with the owner. If recurrence is reported, return with a remote monitor or controller upgrade that captures the next event in full detail.

References

• NFPA 110, Standard for Emergency and Standby Power Systems, current edition (Chapter 8, Routine Maintenance and Operational Testing)
• NEC Article 700 (Emergency Systems) and Article 701 (Legally Required Standby Systems)
• NEC Article 702 (Optional Standby Systems)
• UL 2200, Standard for Stationary Engine Generator Assemblies