Generator Frequency Drifts Under AC Load After Board Replace Decision Tree

Why this matters

After a control or speed board replacement, a genset that holds a clean 60 Hz at no load but sags to 57 to 58 Hz when the air conditioner compressor pulls in - and then wanders - is telling you the speed-control loop was not fully commissioned. Frequency stability is not cosmetic. Equipment with switching power supplies, variable-speed HVAC, and well pumps tolerate a brief dip on a hard motor start, but a frequency that drifts and never recovers under steady AC load means the governor gain or droop is mistuned, the new board did not inherit the old calibration, or there is an actual fueling or speed-sensing fault that the board swap masked. Diagnosing in order keeps you from re-flashing a board when the real issue is a vacuum leak or an MPU gap.

Symptom presentation

No-load frequency is correct (about 60 to 62 Hz). When a large motor load (typically a 3 to 5 ton AC compressor or a deep-well pump) starts, frequency drops sharply, then instead of snapping back it either recovers slowly, settles low, or hunts up and down by 1 to 2 Hz. The customer reports lights dimming and pulsing or sensitive electronics dropping out under AC load that the set handled before the board was replaced.

Quick checks

• Confirm no-load frequency first. If it is wrong at no load, fix base speed before chasing load behavior.
• Verify the new board's firmware/configuration matches the model and that any model-select or governor parameters were set, not left default.
• Check fuel supply under load. On natural-gas and LP sets, a regulator or line that cannot deliver rated flow starves the engine exactly when load steps up, which reads as frequency droop. Measure gas pressure at the genset inlet while running at load (typically 5 to 7 in WC for NG, 10 to 11 in WC for LP vapor at the appliance inlet, per OEM spec).
• Confirm the speed signal source the new board uses (MPU vs stator-derived) and that its gap/leads are correct.

Isolation tree

Branch 1 - No-load frequency wrong after the swap. The board was not calibrated to OEM base speed. Set no-load governed speed to the OEM target (about 61.5 to 62.5 Hz on 60 Hz air-cooled droop sets, or the isochronous setpoint on electronic-governor sets). Many replacement boards ship with generic defaults that must be tuned per model. Correct base speed, then re-evaluate load behavior.

Branch 2 - No-load is right, frequency droops and recovers slowly under AC load (governor gain too low). On an electronic governor, low proportional gain makes the actuator respond sluggishly to a load step, so frequency sags and crawls back. Raise gain in small steps until recovery is crisp without inducing hunting. On a mechanical-governor set, slow recovery points to weak governor spring response or sticky linkage rather than the board - verify the linkage moves freely and the spring is correct.

Branch 3 - No-load is right, frequency hunts (oscillates) under load (gain too high or stability mistuned). If frequency swings rhythmically up and down by 1 to 2 Hz under steady load, the loop is overdamped or stability/derivative is set wrong on the new board. Back gain down and adjust the stability pot until the oscillation settles to a flat line. Hunting is the classic over-correction signature of a freshly swapped electronic-governor board left at aggressive defaults.

Branch 4 - Frequency droops and stays low only under heavy load, regardless of gain (fuel starvation). If the loop responds correctly but the engine simply cannot make power - frequency holds at no load and light load but collapses under the AC compressor and will not come back no matter how you tune gain - the engine is fuel-starved. Measure running gas pressure at the inlet under load: a sag well below the OEM spec under load confirms an undersized line, a clogged regulator, a partially closed valve, or vapor-draw limits on a small LP tank in cold weather. The board swap did not cause this; the symptom was likely there before and got blamed on the new board.

Branch 5 - Frequency unstable and speed signal is dirty. If the board reads a noisy or intermittent speed signal it cannot govern smoothly. Check MPU air gap (about 0.020 to 0.030 in / 0.5 to 0.75 mm), inspect the tone wheel/ring gear for damaged teeth, and verify clean grounds and shielded speed-signal routing away from the alternator and ignition leads. A marginal speed signal makes any gain setting hunt.

Confirming diagnosis

You are done when no-load frequency sits in the OEM band, an AC-compressor start produces a brief dip that recovers to 59.5 to 60.5 Hz within a couple of seconds with no sustained hunting, and the set holds steady through a full HVAC cycle and a deep-well pump start. Confirm running gas pressure stays within OEM spec under the worst-case load step. Record the final gain/stability values and the loaded Hz so the calibration survives the next service.

Remediation

For an uncalibrated board, set base speed and governor parameters to OEM, then tune gain and stability against a real load step. For fuel starvation, correct the gas pressure problem (line size, regulator, valve, tank vaporization) before declaring the governor done. For a dirty speed signal, fix the MPU gap, tone wheel, grounds, and routing. Never compensate for a fueling defect by cranking governor gain - it only trades droop for hunting.

References

• Cummins Onan electronic governor (EFC/ISC) Service Manual, gain and stability tuning and isochronous setpoint procedures.
• Generac Air-Cooled Standby Generator Service Manual, governor calibration, no-load speed setpoints, and fuel-pressure specifications.
• Kohler Generator Service Manual, governor adjustment and gas-pressure requirements at the genset inlet.
• NFPA 54 / ANSI Z223.1 National Fuel Gas Code, gas-supply sizing and delivered-pressure requirements.
• NFPA 110, Standard for Emergency and Standby Power Systems, frequency stability and load-acceptance criteria.