How Loud Is Two Machines in One Room
Researched
Seeded with published specifications and cited sources rather than results measured here. Every seeded figure links to where it came from.
Silencing everything except the loudest machine would win you 1.8 dB. Moving twice as far away would win you 6.
This calculator needs JavaScript. The formula is one line and it is written out below.
Runs in your browser. Nothing you type here is uploaded, logged or sent anywhere. The page is a static file and the arithmetic happens on your machine. Same principle as the rest of the site.
How this is calculated
Convert each level back to sound energy, add the energies, convert back. Correct for distance first if the ratings were measured somewhere other than where you sit.
level at your position = level − 20 × log10(your distance ÷ measured distance)
combined = 10 × log10( sum of 10^(level ÷ 10) )
The 20 in the distance line and the 10 in the combining line are not a typo. Sound pressure falls with distance, and pressure is squared to get energy, which is where the factor of two comes from. The result is the familiar 6 dB per doubling of distance, and the equally familiar 3 dB for doubling the number of identical sources.
| Situation | Effect | Why |
|---|---|---|
| Two identical sources | +3.01 dB | 10 log10(2). Independent of how loud they are |
| Adding a source 10 dB quieter | +0.41 dB | A tenth of the energy of the one already there |
| Adding a source 20 dB quieter | +0.04 dB | Effectively nothing. Do not bother |
| Doubling your distance | −6.02 dB | 20 log10(2), point source in free field |
Worked example, the one above: 38 dB(A) is 10^3.8 units of energy and 41 dB(A) is 10^4.1. Added and converted back, that is 42.76 dB(A). The naive sum, 79 dB, would be a noise level closer to a chainsaw.
What this can't know
- Free field is not a room. The distance correction assumes sound radiating outward with nothing to bounce off. A real room reflects, so levels fall off more slowly than 6 dB per doubling and the combined figure here is the optimistic end. Hard floors and bare walls make it worse; carpet, curtains and soft furnishings make it better.
- A single number hides the character of the noise. Two machines rated identically can be very differently annoying: a tonal whine or a low rumble is far more noticeable than broadband hiss at the same A-weighted level. dB(A) also deliberately discounts low frequencies that a rack in a bedroom will still transmit through the floor.
- Specification figures are measured, not promised. A published dB(A) rating is usually taken under load in a controlled chamber at a stated distance. Your machine in your room, with dust in the heatsink and a fan curve that ramps, will differ.
- Directivity is ignored. Fans are louder in the direction they blow, and a chassis wall between you and the fan is worth several dB that this page does not model.
- It cannot tell you what you will find annoying. That is the actual question and it is not a number.
Sources
- Logarithmic addition of sound pressure levels and the inverse-square distance law are standard acoustics, and are the basis of the measurement methods in ISO 3744. iso.org
- For a sense of scale against everyday sounds, the US Centers for Disease Control publishes a noise level chart with common sources in dB(A). cdc.gov
- The levels you type in come from your own hardware's specification or your own meter. No typical machine noise figures are published here, because nothing on this bench has been measured yet.
- When runs are published they will carry a measured noise field, and this tool gains a machine picker. The method is on the methodology page.
FAQ
How loud are two machines in one room together?
Much quieter than adding the numbers suggests. Sound pressure adds on energy, not on decibels, so 38 dB(A) and 41 dB(A) together make about 42.8 dB(A), not 79. The quieter machine contributes only 1.8 dB over the louder one on its own. Two identical sources always add exactly 3 dB, however loud they are.
Will a second machine make my room noticeably louder?
Usually less than people expect. Adding a source equal to the loudest one you already have costs 3 dB, and adding one 10 dB quieter costs about 0.4 dB, which is below what most people notice. A rough rule is that 3 dB is just perceptible, and about 10 dB is what people describe as twice as loud. The practical consequence is that quietening the loudest machine is nearly always worth more than removing a quiet one.
Why does distance matter so much?
Because sound pressure falls with distance, and fan noise is usually quoted at a fixed distance that is not where you sit. For a point source in the open, every doubling of distance costs about 6 dB. Moving a machine from one metre away to two metres is the same as making it 6 dB quieter, which is more than most fan swaps achieve. A real room reflects sound back, so the real fall-off is gentler than 6 dB, and this tool gives you the optimistic end.
Where do I get the dB(A) figure for my machine?
From the manufacturer specification, which usually states a measurement distance, or from a sound level meter. Phone apps are convenient but their microphones are not calibrated and they are poor below about 40 dB(A), which is exactly the range a quiet home lab lives in. Whatever the source, record the distance it was measured at, because a level without its distance is not a level.
Is dB(A) the right weighting for fan noise?
It is the usual one and it is what specifications quote, so it is what this tool uses. The A weighting approximates how human hearing rolls off at low frequencies, which is reasonable for broadband fan noise. It is less useful for the low-frequency rumble of a rack or a tonal whine from a specific fan, both of which people find more annoying than the A-weighted number suggests. If a machine bothers you more than its rating implies, that is likely why.
Where to go next
- The other cost of leaving machines running: the power cost calculator.
- A disk array is usually the loudest thing in a home lab and the one that never switches off: NAS capacity and power.
- Everything else in the section: the tools index.