Low-frequency noise and hum — finding the source of a sound the meter says is compliant
A low hum that is only there at night, felt as much as heard, that comes through closed windows and disappears the moment you bring someone to listen. This is the hardest category of complaint in acoustics, because the A-weighted sound level meter — the figure the law uses — usually says everything is within limits, while the resident genuinely cannot sleep.
The usual sources are transformers, chillers, water pumps, large extract fans, compressors and pipework vibrating at the machine's frequency. These sounds travel far, pass through structure, and build up in rooms whose dimensions happen to match the wavelength. Geonoise investigates them with measurements designed specifically for low frequency, and reports in a form the complainant, the equipment owner and the authority can all accept.
Where we are engaged
- A resident complains of a hum or a throb, the routine measurement shows compliance, and both sides believe they are right
- A condominium juristic person or hotel keeps receiving the same complaint from the same rooms after changing the air-conditioning and sealing the windows did nothing
- A factory or substation is accused of being the cause and needs evidence either way before agreeing to spend on a fix
- A building owner has to decide whether to replace the chiller, move the pumps or build an isolated plinth, and would rather not pay to guess
- A dispute that needs an expert report explaining why a compliant dBA figure can still be a nuisance
Why the dBA figure misses it
A-weighting removes most low-frequency energy. A 50 Hz transformer hum loud enough to disturb sleep may add so little to the dBA at the receiver that it never trips the 10 dB nuisance criterion, while a C-weighted minus A-weighted difference above 20 dB is a clear indicator of a low-frequency problem. The investigation therefore starts from the spectrum, not from a single number.
- One-third-octave and narrow-band spectra in the complainant's room, at the time the problem occurs, compared with a period when it does not
- Tonality assessment to ISO 1996-2 — the reason a low level can still annoy, and the basis for the penalties several standards apply
- Identifying the dominant frequency and matching it to the running speed of the suspected machine: 50 and 100 Hz point to a transformer; a blade-count times shaft-speed frequency points to a fan or pump
- Vibration measurement on the suspected equipment and on the building structure, to establish whether the path is airborne or structure-borne — which leads to completely different remedies
- Switch-off tests, one machine at a time at night, while recording at the receiver — the evidence that is hardest to argue with
- Acoustic camera to eliminate candidates where there are several machines and not all can be accessed
A room problem, not only a machine problem
Low frequencies have wavelengths of several metres, so an ordinary bedroom has resonances in exactly that range. The same sound can be almost inaudible in the middle of the room and clearly audible in a corner or at the head of the bed. This is why people in the same house disagree about whether the noise exists, and why we measure at several positions rather than one. Understanding it often opens a cheaper remedy than treating the machine — moving the bed, or treating the room — and we will say so plainly when that is the sensible option.
The limits of the method — stated plainly
- Some investigations find no sound that matches the complaint. We say so, with the measured data, rather than shaping a conclusion to suit either party.
- Thai statutory limits are A-weighted. Where the dBA complies but a real low-frequency nuisance exists, we cite the international low-frequency criteria to support a negotiation, not to replace the law.
- Low-frequency remedies cost more than high-frequency ones. Thin absorptive materials do almost nothing; what works is treatment at the source, at the machine base, or along the structural path. That is why we confirm the source before anyone pays.
What you receive
- Frequency spectra at the complaint position during and outside the problem period, with the C–A difference and the tonality assessment
- Source identification with the evidence: frequency matching, vibration results and switch-off test results
- The transmission path — airborne or structure-borne — and the point at which a remedy has the most effect
- Remedy options ranked by effect per baht, at the source, the base, the path and the receiving room, with no stake in any product
- A report written so that complainant, equipment owner and authority read the same thing, signed by a named engineer and usable as an expert report
Standards & method
ISO 1996-2 (including tonality assessment) · Thai National Environment Board Notification No. 29 (B.E. 2550) · Nuisance-noise notification (B.E. 2565) · DIN 45680 (low-frequency criteria, for reference) · ISO 10816 / ISO 20816 (machinery vibration) · IEC 61672-1 Class 1 · instruments calibrated by an ISO/IEC 17025 accredited laboratory
FAQ
Why does the measurement pass but I still cannot sleep?
Because the limit uses the A-weighted level, which heavily discounts low frequencies. A 50 Hz hum that genuinely disturbs sleep may raise the dBA only slightly. A proper investigation starts from the frequency spectrum and the C–A difference, not from one number.
What usually causes this kind of hum?
Transformers (50 and 100 Hz), chillers and compressors, water pumps and the pipework that vibrates with them, large extract fans, and machinery mounted on the floor without vibration isolation. The dominant frequency in the spectrum usually says which.
Why can I hear it in the bedroom but not in the living room?
Low frequencies have wavelengths of several metres. Each room amplifies different frequencies according to its size, and within one room the level in a corner can be far higher than in the middle. So we measure at several positions, and sometimes moving the bed is part of the answer.
Can you prove which machine is responsible?
In most cases, yes: by matching the frequency to the machine's running speed, by vibration measurement on the machine, and by switching machines off one at a time while recording at the receiver. The switch-off test is the clearest evidence and needs the equipment owner's cooperation.
The equipment owner refuses access. What then?
The investigation can still proceed from the receiver side: spectrum, direction, timing and correlation with the operating schedule that can be observed. The result carries enough weight for a complaint to the authority, which has the power to order an inspection.
Can it be fixed with absorptive material in the room?
Almost never. Ordinary absorbers work at mid and high frequencies; at low frequencies they are far too thin relative to the wavelength. What works is at the source, the machine base or the structural path — which is why the source must be confirmed first.
Does the measurement have to be done at night?
Usually, because background noise is lowest and the problem clearest. We set up continuous recording across the complaint period and analyse afterwards, without anyone having to be on site all night.
Can the report be used in a complaint or a court case?
Yes. It cites checkable standards, uses calibrated instruments and is signed by a named engineer. Where the matter goes beyond negotiation, see Expert Witness & Forensic Acoustics.