Reading a noise specification comparatively

Why do two suppliers quote different dB figures for the same product?

Because a decibel figure without its measurement conditions is not a specification. Ask for the microphone distance, the weighting (A-weighted or not), the background noise floor, the meter class and whether the unit was run at maximum speed; a 30 cm class-1 measurement and a one-metre phone reading are not comparable. This catalogue publishes a ceiling of 60 dB at maximum with the method stated alongside it. The noise test method records the conditions used.

Which three variables make two decibel figures comparable?

Published noise figures in this category are almost impossible to compare, and not because suppliers are dishonest — because they measure different things. Three variables explain nearly all of the difference, and a noise specification without all three stated is unusable rather than merely incomplete. The rest of this guide is those three variables and how to read a noise specification that actually names them.

Here is the judgement the table implies. A catalogue figure that names none of the three is not a specification, it is a floor for a conversation — and the error is not small. Sound pressure falls by roughly 6 dB per doubling of distance, so a unit measured at 30 cm and the same unit measured at 1 m differ by about 10 dB. That is not a rounding difference: 10 dB is a factor of ten in sound power. Two suppliers can therefore publish the same "50 dB" and be describing products that are an order of magnitude apart, with neither of them lying. Until the distance, the speed setting and the weighting are on the line, the number cannot be compared, cannot be specified, and cannot be enforced after delivery.

Our own range demonstrates the same point. The single-motor bodies publish 45–55 dB(A) at 30 cm against up to 60 for the dual-motor ones, and that gap is mechanical rather than a tuning choice — one motor instead of two. The Slim, A0187 publishes ≤55 dB(A) at 30 cm in a 151 mm, 97 g body, the smallest in the range: the model with room for one motor is also the one that fits a smaller enclosure.

What does the decibel scale do to a straight comparison?

The decibel is logarithmic, so intuition misleads: a 10 dB increase is perceived roughly as a doubling of loudness, and 3 dB is close to the smallest difference most people notice. That is why chasing small differences between two suppliers' specifications is wasted effort, while a 10 dB gap is immediately audible.

A-weighting matters for the same reason. It de-emphasises frequencies human hearing is less sensitive to, so it reports what is perceived rather than what is present. A measurement quoted without its weighting is not comparable with one that used it — [Beranek & Mellow, 2012] set out the underlying acoustics if you want the derivation.

Under what conditions are our noise figures measured?

Our noise specification is taken at 30 cm from the motor axis, A-weighted, maximum intensity, unit held free of contact surfaces, instrument specified to IEC 61672. The full conditions — including why the distance is 30 cm rather than the more flattering one metre — are laid out in the noise test method, which is the noise specification every unit on the range is measured against.

Published figures: dual-motor models measure at or below 60 dB(A), single-motor models 45–55 dB. Every unit is checked against the limit rather than a sample being drawn, because motor noise is the defect a customer notices first and cannot be argued with afterwards. The same figures sit alongside the other measured values on the platform page.

One case sits outside that two-row taxonomy. The Forge clamp drives two textured pads and the flexible band between them from three independent motors, so its figure is the sum of three sources at once rather than one — quoted at or below 60 dB(A) on the same 30 cm, A-weighted basis as everything else on the range, and comparable only with another unit measured the same way at the same motor count. Ask how many motors are running before you compare two figures: a number quoted at one motor says nothing about a body that runs three.

Where does the noise come from, and what actually reduces it?

In a motorised device, noise is partly mechanical — bearing quality, rotor balance, gear meshing — and partly structural, transmitted through the housing and amplified by resonance. That second part is why two products with the same motor can measure differently, and why improving noise often means changing the chassis rather than the motor.

It is also why we state a noise specification rather than promising quiet: the failure mode is a unit-level variation that only shows up when you check every unit. Ask any supplier what percentage of finished units are checked for noise, then decide whether their figure is a specification or an average.

What makes a motor quiet: the motor, the mount or the housing?

Noise in this product is not one sound. It is four, and they scale differently: the motor itself, the gear train between motor and shell, the housing acting as a sounding board, and the imbalance of whatever the motor drives. Treating them as a single problem is why so much of the category sounds identical at 55 dB and yet feels different in the hand.

The design consequence is that quietness is bought at the bottom of the speed range, not at the top. A unit that is quiet at maximum but hums at minimum has a resonance problem; a unit that is even across its range has an isolation problem that somebody solved. When a supplier quotes one figure, the useful follow-up is which speed it was measured at.

What should a noise clause in a supply contract say?

A noise claim that cannot be adjudicated is not a term, it is an advertisement. Three pieces of information make it enforceable, and all three cost nothing to agree before the order.

1. The limit and the condition: "≤60 dB(A) at 30 cm, A-weighted, at maximum intensity", not "quiet".
2. The basis of measurement: which instrument class, what room, whether the unit is hand-held or fixtured. A hand-held unit reads differently from a clamped one.
3. The remedy: a unit above the limit is a defect at final inspection, so it is reworked or rejected before shipment rather than argued about after arrival.

The third item is the one that needs a process behind it. A limit with no gate is a sentence nobody enforces, and the quality system behind the meter is what turns a decibel figure into a stop on the line rather than a line in a document.

How do you run the comparison yourself in one afternoon?

Two samples and a phone are enough to rank two suppliers, provided the conditions are held fixed rather than conveniently chosen. The protocol below is deliberately blunt: it measures nothing to an accredited standard, and it answers the only question a buyer has before an order, which is whether one unit is quieter than another by a margin a customer would notice.

1. Same room, same hour. A hard-surfaced room adds reflections that swing readings by several decibels. A soft-furnished room is closer to a bedroom, which is where the product will be judged.
2. Same distance, measured not estimated. Thirty centimetres from the housing, not from the end of the box. Mark it once with a ruler rather than eyeballing it per unit.
3. Same surface. Resting a unit on a table couples its vibration into the furniture, which is louder than the motor. Hold it free, or hang it the same way for every sample.
4. Same intensity step. Compare like with like: minimum against minimum, maximum against maximum. A maximum-intensity figure compared with a low-intensity one is the most common way a comparison is accidentally rigged.
5. Same instrument setting. If the app offers A-weighting, use it, and use it for both. Fast or slow response, pick one and stay on it.
6. Three readings each, then the range. A single reading is one sample of a noisy quantity; the spread across three tells you whether the difference you heard is real.

What this protocol does not do is produce a number you can publish. A-weighted dB(A) readings from a phone are indicative at best, and the acoustic environment of a warehouse is not the one in the measurement method. It is a shortlisting tool: it tells you which two of five samples deserve a laboratory measurement, and it costs an afternoon rather than a purchase order. The measurement conditions we publish are the ones a figure should be quoted against; anything else is a comparison, not a specification.

References

1. Beranek, L. L. & Mellow, T. J. (2012). Acoustics: Sound Fields and Transducers. Amsterdam: Elsevier.
2. Rossing, T. D. (ed.) (2014). Springer Handbook of Acoustics, 2nd ed. New York: Springer.
3. International Electrotechnical Commission (2013). Electroacoustics — Sound level meters — Part 1: Specifications, IEC 61672-1. Geneva: IEC.