In short. Hunting is a regular rise and fall in engine speed, which shows as swinging frequency, and usually swinging voltage and sound too. The commonest causes are fuel starvation — a choked filter or air in the fuel — a governor that is set up badly or worn, and loads that rise and fall in a cycle. If only the voltage swings while frequency stays steady, the problem is the AVR, not the engine.
Who this page is for: For operators whose set surges up and down: separate engine speed hunting from voltage instability and from loads that are themselves fluctuating, and fix the right one.
Key facts
- Watch the frequency meter: if it swings, it is speed hunting; if it holds, it is voltage instability.
- Hunting that starts after a filter change or after the tank ran low points to air or fuel restriction.
- Hunting only at light load is often a governor stability setting.
- Cyclic loads such as compressors and pumps can make a healthy set look unstable.
How speed is controlled
The governor measures engine speed and moves the fuel rack or signals the injection system to hold it steady. Mechanical governors use flyweights and springs; electronic governors use a speed sensor, a controller and an actuator. Too much correction makes the governor overshoot and chase the speed up and down. Too little, and it responds slowly. Anything that delays or limits fuel delivery — a blocked filter, air bubbles, a sticky linkage — has the same effect as a badly set governor.
Diagnostic decision table
| Pattern | Most likely cause | Check |
|---|---|---|
| Frequency and voltage swing together, worse on load | Fuel restriction or air in fuel | Filters, water separator, fuel lines, tank level |
| Swings at light load, steady on load | Governor gain or stability setting | Governor adjustment by an engineer |
| Swings after a mechanical knock or service | Sticky or loose linkage, worn actuator | Linkage free movement; actuator |
| Frequency steady, voltage swings | AVR stability setting, loose sensing wire, failing AVR | AVR set-up and wiring |
| Swings in time with a machine cycling | Cyclic load — compressor, pump, welder, lift | Watch the load; compare with load disconnected |
| Swings with leading power factor reading | Capacitor bank on DG supply | Switch off APFC on DG |
| Erratic speed, sudden drops | Faulty speed sensor or wiring | Sensor gap and signal |
A simple test to narrow it down
- Run the set with the load disconnected. If it is steady, the problem is load-related or appears only under load.
- Add load in steps and note at which point hunting starts.
- Watch frequency and voltage separately.
- Note whether the swings match the cycle of any machine on site.
Worked example (illustrative)
An illustrative case, not a project record. A 160 kVA set at a factory swings between about 48 and 52 Hz every few seconds, but only when the plant’s air compressor is loading and unloading. With the compressor off, frequency holds at 50 Hz. The set is responding correctly to a load that jumps up and down. The remedy is on the load side — soft starting, a different compressor control mode, or allowing for that cycle when sizing — not a governor repair.
When the usual assumptions fail
- “Governor problem.” Most hunting we find is fuel delivery, not the governor. Fix fuel first, then adjust the governor.
- “It’s the engine.” Voltage swinging on a steady frequency is an alternator and AVR matter.
- Electronic engines. Sensor or ECU faults need diagnostic tools; turning adjusters at random makes them worse.
- Parallel sets. Sets running in parallel can hunt against each other if load sharing is set up incorrectly — a job for an engineer with synchronising experience.
Prevention
- Fuel filters changed on time and the system bled correctly
- Tank never run low; water drained regularly
- Linkages kept clean and free
- Governor and AVR set up at commissioning and after major repairs
Inside the speed-control loop
Every governor, mechanical or electronic, works in a loop: measure speed, compare with the set point, change the fuel, wait for the engine to respond, measure again. Hunting is what happens when that loop over-corrects. Three properties of the loop decide whether it settles or oscillates.
- Gain — how much fuel the governor adds for a given speed error. High gain gives a fast response but tips into oscillation easily.
- Stability or reset — how the governor removes the remaining error over time. Too fast a reset makes it chase its own correction.
- Dead time — the delay between a fuel change being requested and the engine producing the torque. This is the property most affected by the condition of the engine and fuel system.
Dead time is why fuel faults so often look like governor faults. A four-stroke six-cylinder engine at 1,500 rpm fires 1,500 ÷ 60 × 6 ÷ 2 = 75 times a second, one injection about every 13 ms, so a healthy engine responds almost immediately. Air in the fuel acts like a spring in the line: the pump compresses the bubble before it delivers fuel, so the response is late and soft. A choked filter caps the maximum fuel available. A sticky linkage or worn actuator adds a dead band where small corrections do nothing and then a large one overshoots. Turbocharger lag adds a further delay on large load steps. In every case the governor, which is tuned for a responsive engine, now over-corrects.
Droop and isochronous control
A governor with droop lets speed fall slightly as load rises; with 4 % droop on a 50 Hz set, frequency falls by 4 % of 50 Hz, or 2 Hz, from no load to full load. Droop makes the loop inherently calmer and is used where sets share load. An isochronous governor holds the same speed at all loads; it gives tighter frequency but needs more careful tuning. A set changed from droop to isochronous, or a replacement actuator fitted with default settings, often hunts until it is re-tuned on load.
Reading the swing: period and amplitude
A phone video of the frequency meter, with a clock or stopwatch in frame, turns “it is surging” into evidence. Two numbers matter: how far the frequency moves (amplitude) and how long one full swing takes (period).
| Swing pattern | What it usually indicates | Who acts |
|---|---|---|
| Fast and small — under about 1 s, a few tenths of a hertz | Governor gain too high, speed-sensor signal noise, loose sensor or wiring | Engineer |
| Slow and regular — about 2–10 s per cycle | Stability setting, air in fuel, fuel restriction, sticky linkage | Fuel-side checks first, then engineer |
| Irregular, with occasional deep dips | Water or debris reaching the injection system, intermittent sensor, fuel pick-up uncovering | Engineer; check tank level and water separator |
| Matches a machine’s cycle on site | Cyclic load | Load-side solution |
| Starts only above a certain load | Fuel delivery limit — filter, lift pump, return restriction | Engineer |
| Engine derates or logs fuel-pressure codes while swinging | Electronic fuel-system fault on a CPCB IV+ engine | Engineer with diagnostic tool |
Indian fuel and climate: why hunting peaks in certain seasons
- Monsoon water. Condensation in partly filled day tanks, leaking tank vents and water in delivered fuel all rise in the wet months. Water reaching the injection system causes erratic delivery and, on high-pressure systems, rapid injector and pump wear.
- Microbial growth. Where water sits under stored diesel, microbial growth forms a slime that blocks filters gradually. Hunting that worsens over weeks and returns soon after a filter change points to the tank, not the filter. Our fuel cleaning and polishing page covers tank treatment.
- Summer fuel temperature. Fuel returned from the engine heats a small day tank, and in a 45 °C canopy fuel can run very warm. Warm diesel is thinner and less dense, so less fuel energy is delivered per injection; a set near its limit can lose stability on the hottest afternoons.
- Sediment after deliveries. Filling a tank stirs settled debris. Hunting that begins within hours of a delivery is a fuel-quality signal.
Electronic engines
On common-rail and other electronically controlled CPCB IV+ engines, the ECU controls rail pressure and injection timing as well as speed. It logs fault codes for low rail pressure, rail-pressure deviation, low fuel supply pressure, water-in-fuel and speed-sensor faults, often with time stamps. Read and record these codes before clearing them: a rail-pressure deviation logged at the same moments as the frequency swings is strong evidence of a fuel-supply problem rather than a tuning one.
Fuel-side checks before anyone re-tunes the governor
- Tank level well above the pick-up, and no recent delivery that may have stirred sediment
- Water separator bowl drained and the volume of water noted (with the set stopped)
- Primary and secondary filters within their change interval, and of the correct type
- Fuel system bled correctly after the last filter change
- Supply and return lines free of kinks, crushed sections and loose joints that can draw air
- Tank vent clear, so the tank is not pulling a vacuum as fuel is used
- Engine fault codes read and recorded before any reset
Hunting observation log
| Load step | Frequency min–max (Hz) | Period of one swing (s) | Voltage min–max (V) | Notes, codes, machines running |
|---|---|---|---|---|
| No load | ||||
| About 25 % | ||||
| About 50 % | ||||
| About 75 % | ||||
| Normal site load |
Worked example 2: monsoon hunting on an electronic engine
Illustrative case. A 250 kVA CPCB IV+ set at a hospital runs steadily through summer. In August it begins to hunt on load, with frequency moving between 49.2 and 50.8 Hz roughly every 3 seconds. Voltage swings with it, so this is speed, not AVR.
- In engine speed terms the swing is 1,500 × 49.2 ÷ 50 = 1,476 rpm to 1,500 × 50.8 ÷ 50 = 1,524 rpm, a band of 48 rpm, or ±1.6 % about rated speed — enough to trouble the hospital’s UPS and imaging equipment.
- The 3-second period places it in the slow, regular group in the table above: fuel side first.
- The ECU log shows repeated rail-pressure deviation and one water-in-fuel warning, all at times matching the hunting.
- With the set stopped, the water separator gives up a large volume of water, and the secondary filter is discoloured. The day tank’s vent cap is found cracked, letting rain in.
The vent is replaced, the day tank drained and cleaned, filters changed and the system bled. On re-test the frequency holds within about ±0.1 Hz at every load step, and no further codes are logged. The governor settings were never touched — re-tuning them would have masked the water problem while it continued to damage the injection system.
Common mistakes
- Retuning the governor while the real cause is air in the fuel or a sticking linkage.
- Judging stability on no load only; hunting often appears with a particular load.
- Ignoring voltage hunting that is actually an AVR or sensing problem, not speed.
- Leaving a set to hunt on UPS or drive loads, which can trip sensitive equipment.
What to send us
Send the set’s kVA, governor type (mechanical or electronic), how much the frequency swings and how often, the load when it happens, and a phone video of the panel meters during the hunting. Use the service and AMC enquiry, or WhatsApp the details and photographs.
Questions engineers ask
What is generator hunting?
A regular rise and fall in engine speed that shows as swinging frequency, voltage and engine sound. It is usually caused by fuel restriction, governor settings or a cycling load.
How do I know if it is the engine or the AVR?
Watch the frequency. If frequency swings, the engine speed is hunting. If frequency is steady but voltage swings, the AVR or its sensing is at fault.
Can a dirty fuel filter cause hunting?
Yes. A choked filter or air in the fuel limits delivery, so the governor cannot hold speed steady, especially under load.
Why does my generator hunt only when a compressor runs?
The compressor’s load rises and falls as it loads and unloads, and the set follows it. The fix is usually on the load side rather than in the generator.
Related pages
- Generator Running but Low Voltage or No Output
- Why DG Sets Trip When the APFC Panel Is On
- Generator Not Starting? A Step-by-Step Diagnosis for DG Sets
- Generator Troubleshooting
- Breakdown Diagnostics
Technical content reviewed by D. Srinivas Reddy, Managing Partner and service engineer. Last updated 3 November 2026.