Noise Levels in dB(A): How Human Ears Hear Fan Noise vs. Compressor Vibrations

When a manufacturer publishes a noise figure such as 38 dB(A) for a portable split indoor unit, that number encodes specific psychoacoustic assumptions about what sounds humans can and cannot hear clearly — assumptions formalised in the A-weighting filter (IEC 61672) that converts raw acoustic measurements into a single number intended to predict subjective loudness. Understanding how dB(A) weighting handles the low-frequency hum of a compressor differently from the mid-frequency noise of a fan blade helps explain why dB(A) ratings predict real-world sleep comfort well for portable splits, less accurately for monoblocks, and imperfectly for any unit producing tonal noise at a single pitch.

What does dB(A) mean and how is it measured for air conditioners?

dB(A) is the sound pressure level measured in decibels and then filtered through the A-weighting curve, a standardised frequency response that reduces the contribution of low frequencies (below 500 Hz) and raises the contribution of mid-high frequencies (1–4 kHz) to match the relative sensitivity of the human ear. For air conditioners, the measurement standard is EN ISO 5151 or ISO 16358 — a 1-metre free-field microphone measurement with the unit running in cooling mode at rated capacity, typically in a semi-anechoic test chamber with less than 25 dB(A) background noise to ensure signal isolation.

The A-weighting correction values across frequency apply substantial discounts at low frequencies: −26.2 dB at 63 Hz, −16.1 dB at 125 Hz, −8.6 dB at 250 Hz, −3.2 dB at 500 Hz, 0 dB at 1 kHz (the reference point), +1.2 dB at 2 kHz, and +1.0 dB at 4 kHz. This means a compressor vibration producing 60 dB of sound pressure at 63 Hz contributes only 33.8 dB(A) to the final rating — it sounds far quieter to humans than its raw pressure level suggests, because human ears are genuinely much less sensitive to sub-100 Hz sound at quiet ambient levels.

The practical result: A-weighting captures fan noise — which is broadband and centred in the 300–3,000 Hz range where the A-weighting correction is between −8.6 dB and +1.2 dB — with near-to-raw accuracy. It heavily discounts compressor noise centred at 50–100 Hz, where the human ear has its lowest sensitivity. This is why the portable split indoor unit's fan noise dominates its dB(A) figure even when its compressor is running, while the outdoor condensing unit's compressor produces substantial acoustic energy that is heavily discounted by A-weighting and contributes little to its published specification.

How noisy is a portable split compared to other air conditioner types at the same dB(A)?

A portable split indoor unit operates at 36–44 dB(A) at 1 metre in cooling mode, dropping to 26–34 dB(A) in sleep or quiet mode — performance that is roughly equivalent to a true wall-mounted mini-split system and dramatically quieter than a portable monoblock, which typically measures 53–60 dB(A) at 1 metre due to its single-body design combining the compressor, condenser fan, and evaporator fan in the same indoor unit.

The 15–20 dB(A) gap between a portable split indoor unit and a portable monoblock at 1 metre represents a 2× to 4× difference in perceived loudness — a difference that is perceptible as a dramatic improvement rather than a subtle refinement. Stevens' power law approximation — that a 10 dB increase in sound level is perceived as twice as loud — means the 53–55 dB(A) monoblock is perceived as approximately 3–4 times louder than the 36–38 dB(A) portable split indoor unit, despite both being classified as quiet in general consumer descriptions.

Why do two units with the same dB(A) rating sound completely different?

Two units producing identical dB(A) figures at 1 metre can feel very different to occupants because A-weighting measures total perceived loudness but cannot capture tonality: a tonal sound (a pure tone at a single frequency, like a compressor hum at 100 Hz) is psychoacoustically more annoying than broadband noise at the same A-weighted level, and the ISO 1996-2 standard defines a tonality penalty of 3–6 dB to account for this phenomenon. A unit producing a 38 dB(A) tonal compressor hum would need to measure 32–35 dB(A) to feel as acceptable as a 38 dB(A) broadband fan — even though both pass a 40 dB(A) specification.

Fan noise is broadband: a well-balanced cross-flow fan distributes its acoustic energy across a continuous spectrum from 200 Hz to 4,000 Hz with no dominant single frequency. The ear interprets this as a smooth hiss or whoosh — less annoying than a tone of equivalent A-weighted level. Compressor noise is tonal: the reciprocating or rotary compressor produces energy at its fundamental rotation frequency (typically 50–100 Hz in fixed-speed units running at 50 Hz mains frequency) and harmonics (100, 150, 200 Hz), creating a discernible hum that the A-weighting scale heavily discounts from the measurement but that occupants find more intrusive than the dB(A) number implies.

Variable-speed inverter compressors add a further complication: as the inverter modulates compressor speed, the fundamental tonal frequency shifts — from 50 Hz at minimum speed to 120–140 Hz at maximum load in typical European 200–240 V units. The shifting tonal pitch is more noticeable than a steady tone at the same dB(A) level, because the auditory system is highly attuned to tonal changes that historically signalled motion or approaching threat. Well-engineered inverter units mount the compressor on vibration-isolating rubber feet and wrap the outdoor condensing unit in acoustic baffling to attenuate tonal energy before it couples to the building structure.

Edge case: transient compressor speed-change events cause brief dB(A) spikes

Variable-speed inverter portable splits occasionally produce transient acoustic events when the compressor modulates speed in response to thermal load changes — a 3–8 second event where the compressor passes through speed ranges where tonal harmonics temporarily align with room resonance frequencies or the natural frequency of the outdoor unit enclosure. These transients typically measure 42–48 dB(A) at the indoor unit against a steady-state background of 28–34 dB(A) in sleep mode — a 10–18 dB(A) spike lasting under 10 seconds. Because the A-weighting standard measures steady-state noise at a single setpoint, manufacturer dB(A) specifications do not capture these transient events. Owners sensitive to sudden sound changes in quiet sleeping environments report them as more disruptive than the higher steady-state noise of a fixed-speed unit, which at least provides consistent masking sound rather than sudden interruptions.

I measured my portable split with a phone-based SPL app. Got 34 dB(A) in sleep mode at one metre — genuinely quieter than the background of my house at night. Then it ramped up for about 5 seconds to 44 dB(A), just long enough to wake me. The average rating looks great but those ramp-up moments are the problem.

How can you verify a portable split AC noise level before buying?

Verifying a portable split dB(A) rating before purchase requires understanding what the specification was measured under: confirm the figure references EN ISO 5151 or ISO 16358 (the standard used by reputable European manufacturers) rather than an informal measurement; confirm measurement distance as 1 metre; and confirm which operating mode was measured — many units publish a best-case sleep-mode figure rather than the normal cooling mode rating. A verified 38 dB(A) at 1 metre in cooling mode from a manufacturer following EN ISO 5151 is meaningfully comparable between models; unverified figures from product marketing copy frequently understate normal operation noise.

• dB(A) at 1 metre in cooling mode under EN ISO 5151 is the standard comparison figure — confirm mode and measurement distance before comparing products.
• Sleep mode dB(A) figures (26–34 dB(A) for quality portable splits) represent fan-only or very-low-capacity operation; they are achievable, but the unit will not actively cool the room efficiently in this mode.
• A 10 dB(A) difference represents approximately 2× perceived loudness under Stevens law — the difference between 38 and 48 dB(A) is very noticeable in a quiet bedroom.
• Tonal noise at a single frequency sounds more annoying than broadband noise at the same dB(A) level — a unit producing a compressor hum at a discrete pitch may feel louder than its specification implies, even if the dB(A) figure is accurate.
• Transient compressor speed-change events are not captured by steady-state dB(A) measurements — read owner reviews specifically mentioning sleep use to assess real-world night noise before purchasing.

The spec said 42 dB(A). My phone measured 44 dB(A) in cooling mode at one metre — pretty close. But it also measured 54 dB(A) for about two seconds when it changed speed at night. That brief spike was enough to wake me up twice. Now I know to check for inverter ramp noise specifically.

Low-noise portable split units with verified sub-40 dB(A) indoor ratings and effective compressor vibration isolation represent the performance tier that European buyers seeking bedroom comfort require — and they are consistently among the fastest-selling units during summer heat events.