Key takeaways
- A steady low hum at twice mains frequency is the motor working normally and is almost never the fault you are looking for.
- Rattle means something is loose or air is trapped, and both are free to fix in under five minutes.
- Grinding is the only sound with a deadline attached, because it means the ceramic shaft or its bushings are wearing and the pump has weeks left.
- A whine that rises and falls with flow is cavitation from a restricted intake, not a bearing, and stripping the pump will not cure it.
- Recording ten seconds on a phone held against the housing, then again with the unit lifted clear of the cabinet, separates equipment noise from resonance for free.
A reef keeper sent me eleven seconds of audio recorded at 3 a.m., phone pressed to the sump cabinet door. Underneath the low steady hum of the return pump there was a tick, roughly once a second, that he had convinced himself was a dying bearing.
It was a chipped impeller blade clipping the volute housing once per revolution. The part cost about $18. Nothing about the hum was relevant at all, and that is the usual shape of these calls.
Noise diagnosis is a process of subtraction. You are not identifying the loudest sound, you are identifying the one that was not there last month.
The three minute isolation test
Do this before forming any theory, because resonance fools people constantly. A perfectly healthy Sicce Syncra sitting on bare chipboard can sound like a jackhammer through a cabinet panel.
- Rest a hand flat on the cabinet panel and see whether the noise drops. If it does, the cabinet is amplifying, not the pump
- Slide a folded towel or a 10 mm foam pad under the unit and listen again
- Unplug every other device on the tank in turn until only the suspect is running
- For an internal pump, lift it clear of the glass while it runs, keeping the plug dry and the cable slack
Somewhere between a third and half of noise complaints end at step one or two. Hard surfaces and glass panels turn a 40 decibel motor into something you can hear from another room.
Unplug at the wall before any hand goes into the tank or behind the cabinet. If you need a unit running to hear it, keep your hands on the cable sheath and dry equipment only, and make sure the circuit is behind an RCD or GFCI first.
Hum: the motor doing its job
Every synchronous aquarium pump hums. Mains alternates at 50 Hz in the UK and Europe or 60 Hz in North America, and the magnetic laminations in the stator vibrate at twice that, so you hear a 100 Hz or 120 Hz note that never changes pitch.
That hum is normal. What matters is whether it has grown louder, and whether the housing has become warm enough that you would not hold it for ten seconds.
A hum that has doubled in volume with no change in flow usually means the impeller magnet has picked up ferrous grit, or the rubber mounting feet have hardened and stopped absorbing. New feet are around $5. Grit comes out with a cotton bud and two minutes.
Buzz: a loose part being shaken
Buzz is harsher than hum and less regular, more like a wasp in a jar. It comes from something that is nearly tight: a cracked impeller cover, a strainer cage clipped on one lug only, a lid resting on a hose.
Go around the unit pressing each removable part with a fingertip. If the buzz stops under your finger, you have found it, and the fix is usually reseating rather than replacing.
Old fluorescent ballasts buzz too, and that one is different. A magnetic ballast that has started buzzing and running hot should be taken out of service, not investigated further.
Rattle: air, or something unclamped
Rattle sounds like gravel shaken in a tin. On a canister filter it is nearly always trapped air collecting in the pump head, and it appears the day after a service.
Tilt the whole canister to about forty five degrees, rock it twice, and set it down. The air moves out through the impeller and the noise stops within a minute. If it comes back within a day, you are drawing air in somewhere, which points at the intake being too near the surface or a fitting that is not sealing.
A rattle that survives the tilt test on a canister is worth taking seriously, because a joint loose enough to admit air is often loose enough to weep. Check for damp with the paper towel method for finding a canister leak before you go any further.
Grind: the sound with a deadline
Grinding is dry, gritty, and rises in volume rather than pitch. It is the sound of the ceramic shaft moving in worn bushings, or of sand that has been drawn through the intake and is now circulating in the impeller well.
This is the only noise on the list with a clock attached. In my notes, pumps that start grinding audibly last somewhere between two and ten weeks before the shaft snaps or the impeller seizes, and they usually fail overnight.
Strip and inspect the same week. Pull the impeller, roll the ceramic shaft on a flat surface and look for a bow, then check the rubber bushings at each end for a lip worn into them. The difference between a grind and a loud hum is worth learning properly, because one is a warning and the other is not.
Whine and squeal: cavitation, usually
A whine is high, thin, and it changes with flow. Pinch the outlet hose slightly and if the note shifts, you are hearing water, not metal.
That is cavitation: the impeller is trying to move more water than the intake is supplying, so vapor bubbles form and collapse against the blades. The cause is upstream every time. Blocked strainer, collapsed hose, filter media packed too tightly, or a prefilter sponge that has not been squeezed in three months.
Left alone, cavitation pits the impeller face and the noise becomes permanent. Keeping intakes clear is the cheapest maintenance on the tank, and the routine that stops intakes blocking in the first place takes about five minutes a month.
What each sound points at
| Sound | Likely source | Urgency | Typical cost |
|---|---|---|---|
| Steady low hum | Normal stator vibration | None | $0 |
| Louder hum, warm housing | Grit on the magnet, hardened feet | Weeks | $0 to $8 |
| Buzz | Loose cover, cage or lid | Low | $0 |
| Rattle | Trapped air, or a loose clip | Low, unless it returns | $0 |
| Grind | Worn shaft, bushings or sand ingress | Two to ten weeks | $8 to $35 |
| Whine that tracks flow | Cavitation from a restricted intake | Days | $0 |
| Tick once per revolution | Chipped impeller blade | Weeks | $14 to $30 |
| Gurgle or slurp | Water level, not the equipment | Same day | $0 |
Gurgle is a water level problem
A rhythmic slurp from a return or an overflow is almost never a fault in the equipment. Evaporation has dropped the surface below the outlet, or below the point where a weir runs quietly.
Top up with fresh water, not tank water, and see whether it stops. A 200 liter (53 US gallon) open tank in a heated room can lose 1 to 2 liters a day in summer, which is easily enough to change the acoustics inside a week.
Tracing a hum on a four year old canister
An Eheim Classic 2217 on a 350 liter (92 US gallon) cichlid tank started humming noticeably louder about a fortnight after a media change. Flow measured 780 liters per hour against the 1000 that Eheim rates that model at, so it was down but not dramatically.
The impeller well held a teaspoon of fine sand, drawn in when the intake was left sitting low during the service. The magnet had a grey ring of grit around it, and one rubber bushing had a visible groove worn into the face.
Cleaning the well and fitting a new shaft and bushing set, about $16, brought it back to 950 liters per hour and returned it to a normal hum. Had it been left another month, the sound would have moved from hum to grind and the impeller itself would have gone with it.
Noises that mean switch it off now
Most sounds buy you time. Three do not.
- Crackling or frying from any mains device. Unplug at the wall and do not plug it back in.
- A hum accompanied by a smell of hot plastic or varnish. That is winding insulation breaking down.
- Any change in sound that coincides with an RCD or GFCI trip. Treat the unit as faulty until proven otherwise.
None of those three are user serviceable, and none should be opened. Retire the unit and read the symptom by symptom path for aquarium faults to confirm nothing else on the circuit was involved.
Record ten seconds of your pump on your phone the week you install it, and keep the file. A year later that recording is the only reliable answer to whether the sound has actually changed or you have simply started listening harder.
Questions readers ask about this
Yes. Mains powered pumps hum at twice line frequency, so 100 Hz in Europe and 120 Hz in North America, because the stator laminations vibrate as the field alternates. A constant hum you have always heard is fine. A hum that has grown louder over a few weeks, or one paired with a housing too hot to hold, needs the impeller checked.
Air trapped in the pump head. Cleaning breaks the water column and the head refills with a pocket of air that the impeller then chops at. Tilt the canister to about forty five degrees, rock it twice, and set it back down. The noise normally stops within a minute and does not return unless a fitting is drawing air.
Usually between two and ten weeks of continuous running, though it varies with how much grit is in the well. Grinding means the ceramic shaft and its rubber bushings have worn past their tolerance, and once the shaft starts to flex it snaps without further warning. Fit a new shaft and bushing set, typically $8 to $20, rather than waiting.
Cavitation, in most cases. The impeller is moving more water than the intake can feed it, so vapor bubbles form and collapse against the blades. Check the strainer, the prefilter sponge, and whether a hose has kinked or collapsed. If the pitch changes when you slightly restrict the outlet, water is the cause and not a bearing.
Often, because the cabinet is doing the amplifying. A 10 mm closed cell foam pad under the unit, rubber feet in good condition, and 20 mm of clearance between the housing and any panel will cut perceived noise substantially. If pressing a hand on the cabinet panel drops the volume, the pump itself is probably healthy.
