Car Audio SPL Measurement Explained: Sensor Placement, Burn Windows and Vehicle Classes
Sound pressure level is the most quoted and least understood number in car audio. A single figure printed on a shop window or a social post tells you almost nothing on its own, because the number is a function of the measurement almost as much as it is a function of the system. This article explains what a decibel is, where the sensor has to sit, what a burn window does, why vehicles must be grouped into classes, and which variables a written test protocol has to nail down before two results can honestly be compared.
Two cars can produce identical acoustic output and post numbers several decibels apart. Nothing dishonest need happen: different microphone positions, weighting curves, averaging windows and states of charge are enough. A measurement means something only when the method behind it is written down and repeatable — which is why a testing body publishes its protocol before it publishes a result.
What a Decibel Actually Is
Sound pressure level is a logarithmic ratio, not a linear quantity: twenty times the base-ten logarithm of measured pressure over a reference pressure of 20 micropascals, roughly the quietest 1 kHz tone a healthy young ear detects. Because the scale is logarithmic, the intuitive arithmetic is wrong — levels do not add, and a bigger number is not proportionally louder.
- A doubling of acoustic power is +3 dB. Moving from a 500 W amplifier to a 1,000 W amplifier, everything else unchanged, buys about 3 dB — not twice the number on the meter.
- A doubling of sound pressure is +6 dB, which requires four times the power.
- Roughly +10 dB is needed before most listeners call a broadband sound twice as loud.
- Two uncorrelated sources of equal level sum to +3 dB; two correlated sources in phase can sum to +6 dB, and out of phase they partially cancel.
Sensor Placement: Where Centimetres Are Worth Decibels
At low frequencies a car cabin does not behave like a room. The speed of sound is about 343 m/s at 20 °C, so a 30 Hz wave is roughly 11 m long and a 50 Hz wave about 6.9 m — far larger than the longest cabin dimension, typically 2.5 to 3 m in a sedan. Below the first axial mode, approximately the speed of sound divided by twice the cabin length and so near 55 to 60 Hz in that example, the cabin pressurises more or less uniformly. That is the cabin gain which makes small enclosures viable in cars at all.
Above that region the picture changes. Standing waves put pressure maxima at boundaries and minima at nodes, and the difference between the two in one cabin at one instant can exceed 10 dB. Move a microphone 10 cm and the reading moves. That is physics, not meter error, and it is why placement must be defined rather than chosen.
| Placement variable | Why it changes the reading | What a protocol must specify |
|---|---|---|
| Position in three axes | Modal pressure varies steeply near nodes; boundaries add pressure doubling | A fixed vehicle datum with X/Y/Z offsets in millimetres |
| Capsule orientation | Directivity and near-field pressure vary with angle at higher frequencies | Axis and tilt relative to the datum |
| Mounting and coupling | Contact with glass or trim couples structural vibration into the capsule | Decoupled mount, no panel or glass contact |
| Cabin aperture state | Open windows or hatch destroy the pressure-vessel condition | Sealed, partially open or open, declared per class |
| Occupancy and cargo | Bodies and soft goods absorb and change effective volume | Occupant count, seat positions and cargo fixed |
Burn Windows: What the Meter Is Actually Integrating
A level meter reports whatever its detector integrates over whatever window it is given. Three numbers come out of the same run and all three can legitimately be called SPL: instantaneous peak, maximum hold, and an average over a defined window. They are not interchangeable, and the gap between a 250 ms peak and a 30 s average on one system is routinely one to three decibels.
Why duration is a test in itself
A short burst measures the system while everything is cold and the battery sits near resting voltage. Sustain the signal and two things degrade together: voice coil resistance climbs as the copper heats, cutting the power delivered into it, and terminal voltage sags under continuous current draw. Since output power for a fixed load scales with the square of supply voltage, a drop from 14.4 V to 13.0 V is roughly a 19 percent power loss on its own — a little under 1 dB before any thermal effect.
- 1Precondition the vehicle: engine state, alternator state and battery state of charge recorded, not assumed.
- 2Verify sensor position against the datum and calibrate against a reference acoustic calibrator.
- 3Record resting voltage at the amplifier terminals, not at the battery.
- 4Run the defined signal for the defined window, logging voltage and level continuously.
- 5Report peak, maximum hold and windowed average separately, each with its weighting.
- 6Record the cool-down interval, so a repeat run is not measuring a hot system.
Vehicle Classes: Comparing Like With Like
Cabin volume is a term in the acoustic equation, so comparing a compact hatchback with a large SUV as if they were the same test object is meaningless. The pressure a given air displacement produces is inversely related to the volume it has to pressurise, which is why a small sealed cabin flatters low-frequency numbers. Classing holds that variable roughly constant across a comparison group.
- Cabin volume band, derived from body type rather than from a claim on a form.
- Aperture condition — sealed, defined partial opening, or open — since it changes the low-frequency loading regime entirely.
- Electrical configuration: number of batteries, alternator capacity, engine running or not.
- Structural intervention: whether panels, glass and door skins are reinforced, since panel losses are acoustic losses.
- Displacement and power bands, so a class does not pit a single driver against eight.
Why Dashboard Meters and Phone Apps Mislead
Most numbers circulating online come from a phone app or a handheld meter placed wherever there was room. They are not fraudulent so much as physically incapable of being what they claim.
- Acoustic overload. Typical MEMS microphones in phones overload somewhere around 120 to 130 dB. Above that the capsule itself clips, and the app reports a compressed number as if it were valid.
- Automatic gain control. Phone audio paths apply AGC and noise suppression the app cannot fully disable, so the input is level-processed before measurement.
- Wrong weighting. A-weighting attenuates 31.5 Hz by roughly 39 dB and 63 Hz by roughly 26 dB, modelling hearing at conversational levels. Applied to a bass-dominated system it discards most of the energy being measured; low-frequency work needs C- or Z-weighting.
- No traceable calibration. A class 1 or class 2 instrument is calibrated against a reference and drifts predictably; a consumer app has no reference, and microphone sensitivity varies unit to unit.
- Undefined placement. Without a datum the reading describes a position nobody can reproduce, which makes it unusable for comparison even with a perfect instrument.
How a Written Protocol Controls the Variables
MCAA's approach is not to invent new acoustics — the physics is settled — but to remove discretion from the steps where it silently changes the answer. The protocol fixes the sensor datum and orientation, fixes the burn window and signal, requires calibration before and after a session, requires electrical conditions logged at the amplifier rather than assumed at the battery, and requires every figure to carry its weighting and window. Conditions are published with the result, so a number can be challenged on method.
A result that cannot be reproduced by a second team, in a second vehicle, on a second day, is not a result. It is an anecdote with a decimal point.
The practical takeaway is simple: treat the meter as an instrument, not a scoreboard. Record the conditions with every number, keep the microphone in one defined place, use a weighting appropriate to the bandwidth, and log supply voltage during the run. Done consistently, your readings become engineering data rather than screenshots.
Frequently asked questions
Why does doubling amplifier power only add 3 dB?
The decibel scale is logarithmic, and a doubling of acoustic power corresponds to a 3 dB increase. To gain 6 dB you need four times the power, which is why driver efficiency usually delivers more level per riyal than a larger amplifier.
Should car audio SPL be measured with A-weighting or C-weighting?
A-weighting is inappropriate for bass-dominated measurement because it attenuates 31.5 Hz by roughly 39 dB. Low-frequency SPL work should use C-weighting or unweighted Z-weighting, and the weighting must always be stated with the number.
How much does microphone position change an SPL reading?
Above the cabin's first acoustic mode, standing waves create pressure maxima at boundaries and minima at nodes, so moving a microphone by 10 cm can change the reading by several decibels. This is why protocols define placement against a fixed vehicle datum.
Can a phone app measure car audio SPL accurately?
No. Phone microphones typically overload somewhere around 120 to 130 dB, the audio path applies automatic gain control, and there is no traceable calibration. Phone readings are useful for rough relative comparisons at moderate levels and nothing more.
