Why Does a Generator Lose Voltage Under Load? Common Causes and Fixes

Why voltage drops only when the load comes on

A generator that looks fine at no-load but loses voltage as soon as equipment is connected is usually telling you something useful: the machine can still build voltage, but it cannot hold excitation and terminal voltage when current demand rises. In the field, that distinction matters. It helps you avoid wasting time on the wrong side of the system.

For maintenance teams, the first practical question is not “Is the generator bad?” but “What changes under load?” Current increases, magnetic flux changes, speed may dip, weak connections heat up, and any weakness in the excitation circuit becomes much more visible. A light bulb test may pass, while a motor start or process load makes the problem obvious within seconds.

In power equipment service, especially on units tied to gas or steam combined-cycle applications, voltage instability under load is rarely caused by one dramatic failure alone. More often, it is a chain of smaller issues: slight engine droop, marginal AVR sensing, aging windings, or a connection that was “good enough” until reactive load exposed it.

Start with the simplest split: overload, speed loss, or excitation loss

If terminal voltage falls together with frequency, suspect prime mover speed first. On an engine-driven set, low speed directly reduces frequency and usually drags voltage down as well. This can come from governor response, fuel supply issues, air intake restriction, or a load step that exceeds what the prime mover can pick up cleanly. A common mistake is adjusting voltage before confirming rated speed under load.

If frequency stays stable but voltage drops sharply, the investigation shifts toward the excitation system, sensing circuit, or stator/rotor condition. That is the more typical pattern when the AVR, exciter, rectifier components, or winding insulation are no longer performing as they should.

And yes, sometimes it really is overload. Not only total kW overload, but poor power factor load, large motor starting current, or an uneven phase distribution that creates a local problem before the nameplate limit is reached.

AVR problems are common, but not every voltage drop is an AVR failure

Automatic voltage regulators are often blamed first, and sometimes correctly. If the AVR cannot increase excitation fast enough, or if its sensing input is unstable, the machine will sag under load. You may see hunting, delayed recovery, or a steady low-voltage condition after load acceptance.

Before replacing the regulator, check the basics around it:

• Loose or corroded sensing leads
• Incorrect AVR settings after previous service work
• Blown fuses or weak supply to the excitation circuit
• Poor grounding or noise affecting voltage feedback

Field experience says many “bad AVR” calls are actually wiring or calibration issues. If the AVR is adjustable, compare settings with the manufacturer’s commissioning data before turning pots blindly. Random adjustment can mask the root cause and create unstable operation later.

Winding condition and rotating parts deserve more attention than they usually get

A generator can still produce acceptable open-circuit voltage even with winding deterioration, then collapse under load because current stresses the weak area. That is why insulation resistance alone does not tell the whole story. You may need winding resistance comparison, polarization index, surge comparison where applicable, and a close look at hot spots, discoloration, contamination, or varnish breakdown.

On brushless machines, failed rotating diodes are a classic cause of reduced excitation under load. The machine may build voltage, but not enough field current is available when demand rises. On slip-ring machines, worn brushes, poor brush pressure, dirty rings, or uneven contact can create the same symptom in a different way.

This is especially relevant on larger industrial units, where the cooling method and internal structure affect inspection access. In practice, machines ranging from 1.5 MW to 400MW can show very different fault behavior depending on whether they use air cooling, air internal cooling, double water internal cooling, or water-hydrogen-hydrogen cooling. Inspection steps should match the machine design, not just the symptom.

Bad connections create real voltage loss, not just nuisance alarms

Loose terminals, oxidized lugs, cracked bus joints, and undersized cables can all produce measurable voltage drop once current flows. This is one of the most overlooked causes because the generator itself gets blamed while the real loss is in the outgoing circuit or neutral path.

A good habit is to compare voltage at the generator terminals and at the load side during the same event. If the machine terminal voltage holds better than the downstream reading, you are dealing with distribution loss, not purely an internal generating problem. Thermal imaging, torque verification, and millivolt drop checks across suspect joints are often faster than deep electrical teardown.

When load type is the real problem

Not all loads are equally friendly. Large motors, crushers, compressors, and certain variable frequency drive arrangements can pull heavy inrush or harmonic-rich current that drags voltage down even if average running load seems acceptable. In these situations, the complaint “voltage drops under load” is true, but the deeper issue is load compatibility and transient response.

This is where knowing the machine rating and application matters. In combined-cycle and other efficient clean power generation modes, units are often selected for specific reactive behavior, cooling arrangements, pole configuration, and grid or process stability requirements. Two-pole and four-pole machines can behave differently in installation constraints and dynamic response, so service decisions should always be matched to the actual design file and operating profile.

For teams working across mixed fleets, it helps to keep a record of which units are more sensitive to motor starting, leading power factor, or sudden block loading. That saves a lot of repeat troubleshooting.

A practical troubleshooting order that usually works

In the field, a sensible sequence is worth more than a long theory list:

• Confirm speed and frequency under load, not just at idle.
• Measure terminal voltage phase-to-phase and phase-to-neutral, then compare with the load end.
• Check load level, power factor if available, and whether the drop happens on all loads or only certain equipment.
• Inspect AVR supply, sensing leads, settings, and excitation response.
• Examine connections, breaker contacts, bus joints, and neutral points for heating or looseness.
• Test rotor and stator condition if basic checks do not explain the symptom.

That order avoids the common trap of pulling the machine apart before confirming that the prime mover is actually holding rated speed or that the cable run is not the source of the drop.

Repair decisions should reflect machine class, not just fault symptoms

On small standby sets, replacing an AVR or fixing a terminal issue may be straightforward. On larger industrial and utility-support units, the decision is broader: excitation tuning, rotor inspection planning, outage window, spare parts availability, and compliance with machine standards all come into play. Equipment built to IEC60034-3 or Chinese GB/T7064 should be evaluated against the relevant design and test expectations rather than by generic rules of thumb.

Manufacturers with long turbomachinery experience tend to approach this differently. SINO-QNP, for example, works across generators, gas turbines, steam turbines, and compressors, so troubleshooting is usually connected to the whole train rather than the electrical side alone. That matters when a voltage complaint is actually linked to process conditions, coupled equipment behavior, or package-level control interaction. For reference, their Generator range covers different structural types and capacity levels, which is a reminder that service strategy should follow the machine configuration, not a one-size-fits-all checklist.

If a unit loses voltage only under load, do not treat it as a vague electrical symptom. It is usually a narrow diagnostic clue. Check whether the machine is slowing down, whether excitation is failing to rise, whether current is being lost across bad connections, and whether the load itself is unreasonable for the setup. In many cases, the answer shows up quickly once measurements are taken during the actual load event instead of after it has already tripped out.