Transformer Sound Level Measurement: IEC 60076-10 Microphone Positions, Background Noise Correction, A-Weighted Sound Pressure & Sound Power, Load vs. No-Load Noise

Abstract

Transformer noise is a regulated environmental parameter — most grid codes and local environmental permits set maximum noise limits at the substation boundary (typically 40-55 dBA at night). IEC 60076-10 defines the standardized methods for measuring transformer sound levels during factory acceptance testing, establishing a repeatable and comparable basis between manufacturers and purchasers. This article explains the physics of transformer noise generation (core magnetostriction and winding electromagnetic forces), the measurement procedure per IEC 60076-10 including prescribed microphone positions and the measurement surface definition, how to apply background noise correction when the test environment is not perfectly quiet, the distinction between A-weighted sound pressure level (L_pA, dBA) and sound power level (L_WA, dBA), and the difference in spectral characteristics between no-load (core) noise and load (winding) noise.

1. Physics of Transformer Noise

1.1 No-Load (Core) Noise

The dominant source of transformer noise under no-load conditions is magnetostriction — the physical dimensional change of the silicon steel core laminations under alternating magnetic flux. Grain-oriented silicon steel expands and contracts by approximately 1-3 μm per meter of core length at 1.7 T flux density, at twice the power frequency (100 Hz for a 50 Hz system, 120 Hz for 60 Hz).

Key characteristics of core noise:

Fundamental frequency: 2× power frequency (100/120 Hz)
Harmonics: Integer multiples of fundamental (200/240 Hz, 300/360 Hz, 400/480 Hz, etc.), all even harmonics of the power frequency
Tonal character: Pure tones — the noise is perceived as a "hum" or "buzz" at clearly identifiable frequencies
Amplitude: Proportional to core flux density B^2, and to core mass
Cooling method influence: Core noise dominates in naturally cooled (ONAN) transformers because there is no competing noise from fans or pumps

Typical A-weighted sound pressure levels for core noise only: 55-75 dBA at the prescribed measurement surface, depending on transformer rating.

1.2 Load (Winding) Noise

Under load conditions, the leakage flux between HV and LV windings produces electromagnetic forces on the winding conductors. These forces alternate at twice the power frequency and their harmonics, causing the windings to vibrate against their clamping structure.

Key characteristics of load noise:

Frequency: 2× power frequency fundamental (100/120 Hz), plus harmonics up to 1-2 kHz
Amplitude: Proportional to the square of load current (I²)
Spectrum: Broader spectrum than core noise because the vibration transfer path through the winding clamping structure to the tank wall involves multiple mechanical resonances
Dominance: Usually 5-15 dB lower than core noise at rated current; load noise can equal or exceed core noise in large generator step-up (GSU) transformers with very high current (several thousand amperes per phase)

1.3 Cooling Equipment Noise

For transformers with forced cooling (ONAF, OFAF, ODAF), the cooling fans and oil pumps generate broadband aerodynamic noise (fan blade passing frequency, flow turbulence). Cooling equipment noise typically dominates the total noise at frequencies above 500 Hz and can increase the overall A-weighted level by 3-10 dB compared to core noise alone, with a distinctly different spectral character (broadband rather than tonal).

2. IEC 60076-10 Measurement Procedure

2.1 Measurement Surface and Microphone Positions

IEC 60076-10 prescribes two methods:

Sound Pressure Method: Microphones are placed at prescribed positions on a measurement surface surrounding the transformer. Measured sound pressure levels are averaged and converted to sound power level using the measurement surface area.

Sound Intensity Method: A sound intensity probe (two closely-spaced microphones) scans the measurement surface. The sound power is derived directly from the integrated intensity, without requiring a specialized acoustical environment (this method is inherently immune to background noise and room reflections).

2.2 Sound Pressure Method — Measurement Surface

The measurement surface is a hypothetical box (rectangular parallelepiped) at a distance d from the transformer's main radiating surfaces. The prescribed distance is:

d = 0.3 m for transformers with length/width ≤ 4 m (distribution transformers)
d = 1.0 m for transformers with length/width 4-10 m (most power transformers)
d = 2.0 m for transformers with length/width > 10 m (large power transformers, GSU units)

For a transformer with a measurement surface at d = 1 m:

Measurement surface area S = 1.25 × h × L_m

Where h = transformer height (m), L_m = perimeter of the measurement contour (m). The 1.25 factor accounts for sound reflected from the floor (assuming a hard reflecting floor). S is the "equivalent absorption area" used to convert from sound pressure to sound power.

2.3 Microphone Positions

Microphones are placed on the measurement surface at predetermined positions. The basic grid consists of:

Positions at every corner of the measurement contour
Additional positions along each side such that adjacent microphone positions are spaced ≤1 m apart
Height: Microphone at half the transformer height; for transformers taller than 2.5 m, two heights are used — 1/3 and 2/3 of the tank height

For a typical 50 MVA transformer (6 m long, 3 m wide, 4 m high), at d = 1 m, the measurement contour has a perimeter of approximately 28 m. With microphone spacing ≤1 m, approximately 28 microphone positions are measured along the perimeter, at two heights (14 + 14 = 28 measurements total).

2.4 Measurement Conditions

Parameter	Requirement
Background noise	At least 3 dB below the transformer noise at ANY measurement position; if less than 3 dB but greater than 1 dB, apply correction per Section 2.5
Wind speed	<5 m/s; windscreen on microphone required for outdoor measurements
Rain/snow	Measurements not permitted during precipitation
Measurement instrument	Type 1 (precision) sound level meter per IEC 61672, calibrated before and after each measurement session
Frequency weighting	A-weighting (standard); C-weighting optional for low-frequency content assessment

2.5 Background Noise Correction

If the background noise level (L_bg, measured with the transformer de-energized) is 3-10 dB below the transformer-plus-background noise level (L_total), the corrected transformer sound pressure level is:

L_corr = 10 × log₁₀ (10^(L_total/10) - 10^(L_bg/10))

L_total - L_bg (dB)	Correction to Subtract from L_total (dB)
3	3.0
4	2.2
5	1.7
6	1.3
7	1.0
8	0.8
9	0.6
10	0.5
>10	0 (background noise negligible)

If L_total - L_bg < 3 dB, the measurement is invalid — the background noise is too high for a reliable correction. The test must be moved to a quieter time (night, early morning) or a quieter location.

3. Sound Pressure Level vs. Sound Power Level

3.1 Definitions

Sound Pressure Level (L_p, dBA): The RMS sound pressure at a specific measurement point, relative to the reference pressure p₀ = 20 μPa
Sound Power Level (L_W, dBA): The total acoustic power radiated by the transformer, independent of measurement distance. Reference: W₀ = 10⁻¹² W (1 pW)

3.2 Conversion

L_W = L_pA_avg + 10 × log₁₀ (S / S₀)

Where:

L_pA_avg = 10 × log₁₀ [(1/N) × Σ10^(L_pAi/10)]

L_pA_avg = energy-averaged A-weighted sound pressure level over all microphone positions:
S = measurement surface area (m²)
S₀ = reference area = 1 m²

Why sound power is preferred for specification: Sound power is independent of the measurement distance — it is a property of the transformer. Sound pressure depends on distance, room reflections, and nearby surfaces. A transformer specified as "65 dBA at 1 m" is ambiguous (measurement surface not defined). A transformer specified as "L_WA = 85 dBA" is unambiguous and verifiable.

4. Load Noise Measurement

IEC 60076-10 includes a method for separating load noise from total noise:

Measure total noise with the transformer energized at rated voltage and rated current (or the maximum current achievable with the available load bank)
Measure no-load noise immediately after de-energizing and before the core cools (the core noise remains constant while the load noise drops to zero)
Load noise (L_WA_load) = L_WA_total minus L_WA_no-load, applied as a logarithmic subtraction

Alternatively, a short-circuit test is used: energize one winding at reduced voltage sufficient to circulate rated current in the short-circuited other winding. The resulting noise is predominantly load noise (core flux is reduced because the applied voltage is only u_k% × rated voltage). The load noise at rated current is then scaled by (I / I_test)².

FAQ

Q: Why is transformer noise measured in dBA instead of dB?

The A-weighting curve (IEC 61672) approximates the frequency response of the human ear, which is most sensitive between 1-4 kHz and progressively less sensitive at low frequencies (<500 Hz) and very high frequencies (>8 kHz). Transformer core noise is dominated by 100/120 Hz and its harmonics (200/240, 300/360 Hz) — frequencies that are attenuated by 15-20 dB by the A-weighting curve. A transformer that measures 85 dBA would measure approximately 100 dB linear (unweighted). The dBA measurement reflects the "perceived loudness" and is the metric used in environmental noise regulations (residential noise limits at night are typically 35-45 dBA). When evaluating noise impact on substation personnel, C-weighted (dBC) or unweighted measurements are more relevant because low-frequency energy contributes to auditory fatigue even if it is not perceived as "loud."

Q: What is a typical sound level specification for a power transformer?

Typical contractual limits: Sound power level L_WA ≤ 85-95 dBA for a 50 MVA transformer (ONAN cooling), increasing to L_WA ≤ 95-105 dBA for a 250 MVA unit. Sound pressure level L_pA at 1 m: typically 65-75 dBA (ONAN), 70-80 dBA (ONAF with fans). These values decrease by ~6 dB per doubling of distance in a free field. For a transformer located near a residential boundary, the specification should be set based on the predicted noise level at the boundary (using acoustic modeling that includes barrier effects, ground attenuation, and atmospheric absorption), not at the transformer itself.

Q: How can transformer noise be reduced?

Noise reduction strategies, in order of effectiveness: (1) Specify low-noise core steel — domain-refined (laser-scribed) Hi-B grain-oriented steel has 2-4 dB lower magnetostriction than conventional CGO steel (at an additional cost of 10-20% on core material), (2) reduce core flux density — designing at 1.6 T instead of 1.7 T reduces core noise by 3-5 dB but increases core mass and cost by 10-15%, (3) improve core joint design — step-lap joints reduce magnetostriction at the joint regions, (4) use sound-insulating enclosure or acoustic panels (10-20 dB insertion loss, expensive and complicates cooling), (5) install the transformer in a sound-insulated building (most effective for indoor substations), and (6) active noise cancellation — experimental and rarely cost-effective for transformers.

Q: If the background noise is too high for a valid measurement, what alternatives exist?

Three options: (1) Use the sound intensity method (IEC 60076-10, Annex B) — sound intensity is a vector quantity (magnitude + direction), so the probe can discriminate between sound radiating from the transformer (outward) and background noise (incident from all directions including inward). Sound intensity measurement is valid even when background noise exceeds the transformer noise by 10+ dB — making it the preferred method for busy factory testing environments where electrical background noise (motors, cooling fans from adjacent test stands) cannot be eliminated. (2) Perform the test at night when background noise is lower (factory ventilation systems off, no adjacent test activity). (3) Move the transformer to an outdoor test pad remote from noise sources.

Q: Does load noise matter, or is core noise always dominant?

For ONAN (natural cooling, no fans) transformers up to ~100 MVA, core noise typically dominates by 5-15 dB at rated load. Load noise becomes significant for: (1) large generator step-up transformers (200-1,000 MVA) where the phase current can exceed 10 kA, producing winding forces proportional to I², (2) transformers with low flux density design (1.55-1.60 T) where core noise is already reduced — then load noise may equal or exceed core noise at full load, and (3) transformers with series reactors or phase-shifting windings where the internal leakage flux pattern is more complex. Always specify that the guaranteed sound level applies under the specified loading condition — "no-load" and "rated load" guarantees may differ.

Q: How accurate are factory noise measurements compared to on-site measurements?

Factory measurements are typically 2-5 dB lower than on-site measurements. Reasons: (1) the factory test bay is a hard-floored semi-anechoic environment with minimal reflections; the substation has nearby walls, adjacent transformers (which contribute background noise), and hard ground that increases the effective radiating surface, (2) transformer core noise increases slightly over the first weeks of energization as the core laminations settle under operating temperature and flux (the "run-in" effect can add 1-2 dB), and (3) on-site measurements are often taken at the substation boundary rather than at the IEC 60076-10 measurement surface, with additional attenuation from distance and barriers but less controlled conditions overall. When specifying noise limits, account for a site adjustment factor of +2-4 dB above the factory guarantee, depending on the specific site acoustic environment.

References / Standards

Reference	Title
IEC 60076-10:2016	Power transformers — Part 10: Determination of sound levels
IEC 61672-1:2013	Electroacoustics — Sound level meters — Part 1: Specifications
ISO 3744:2010	Acoustics — Determination of sound power levels and sound energy levels of noise sources using sound pressure — Engineering methods
ISO 9614-2:1996	Acoustics — Determination of sound power levels of noise sources using sound intensity
IEEE C57.12.90-2021	IEEE Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating Transformers
CIGRE TB 659	Acoustical Noise Emission from Transformers, Reactors, and Associated Equipment

*Authored by Du Fu, Production Engineer at ZY POWER. Transformer noise measurement is both a contractual acceptance test and an environmental compliance requirement — understand the measurement method, the correction factors, and the difference between sound pressure and sound power before signing a noise guarantee.*