Microphone Calibration

RUBAT’s calibration factor is a single number — Pa per ADC unit — that maps the raw normalised audio samples (range −1 to +1) to physical sound pressure in Pascals. Once entered, the live waveform display switches to dB SPL re 20 µPa and the Auto-mode threshold is interpreted in the same absolute scale.

How RUBAT uses the factor internally

Every audio sample x arriving from the ADC (in the range −1 … +1) is converted as follows:
p_Pa = |x| × C   # instantaneous pressure in Pa
L_SPL = 20 × log₁₀(p_Pa / 20×10⁻⁶)   # dB SPL re 20 µPa
$$L_{\text{SPL}} = 20 \log_{10}\!\left(\frac{|x| \cdot C}{p_{\text{ref}}}\right), \quad p_{\text{ref}} = 20\,\mu\text{Pa}$$ For the Auto threshold, the inverse is used: given a threshold $L_{\text{thr}}$ in dB SPL, RUBAT computes the equivalent ADC amplitude $$x_{\text{thr}} = \frac{20\,\mu\text{Pa} \cdot 10^{L_{\text{thr}}/20}}{C}$$ and compares each incoming sample against $x_{\text{thr}}$ directly — no per-sample log is needed in the hot path.

What is a “Pa per ADC unit”?

A modern audio interface presents samples normalised so that full-scale digital = 1.0 (and −1.0). The physical voltage that produces that full-scale reading is the interface’s full-scale input voltage, $V_{\text{FS}}$.

The microphone element produces a voltage proportional to acoustic pressure. Given microphone sensitivity $S_m$ (V/Pa) and total preamp gain $G$ (linear, unitless):

\[C = \frac{V_{\text{FS}}}{S_m \cdot G} \quad \text{[Pa/unit]}\]

Three practical routes to obtain $C$ are described below — choose the one that fits your equipment.

Method 1 — From manufacturer datasheets No extra hardware

You can calculate $C$ entirely from published specifications with no acoustic reference source.

What you need:

Step-by-step

1
Find microphone sensitivity $S_m$.
The datasheet will give a value such as −34 dBV/Pa or 20 mV/Pa. Convert to V/Pa: $$S_m \,[\text{V/Pa}] = 10^{S_{dBV}/20}$$ Example: −34 dBV/Pa → $10^{-34/20} = 0.0200$ V/Pa = 20 mV/Pa
2
Convert interface full-scale to Volts peak.
Interface specs often give a maximum input level in dBu (reference 0.775 VRMS) or dBV (re 1 VRMS): $$V_{\text{FS, peak}} = \sqrt{2} \cdot 0.775 \cdot 10^{V_{dBu}/20} \quad \text{(if given in dBu)}$$ $$V_{\text{FS, peak}} = \sqrt{2} \cdot 10^{V_{dBV}/20} \quad \text{(if given in dBV)}$$ Example: +18 dBu full scale → $\sqrt{2} \times 0.775 \times 10^{18/20} = 6.16$ V peak.

Note: RUBAT's ADC unit corresponds to peak amplitude. If your interface datasheet gives an RMS full-scale figure, multiply by $\sqrt{2}$ to get peak.
3
Obtain the preamp gain $G$ (linear).
If the gain knob is set to, e.g., +30 dB: $$G = 10^{30/20} \approx 31.6$$ If gain is set to 0 dB (unity), then $G = 1$.
4
Calculate the calibration factor. $$\boxed{C = \frac{V_{\text{FS, peak}}}{S_m \cdot G}}$$ Example: $V_{\text{FS}} = 6.16$ V, $S_m = 0.020$ V/Pa, $G = 31.6$ $$C = \frac{6.16}{0.020 \times 31.6} = \frac{6.16}{0.632} \approx 9.75 \;\text{Pa/unit}$$
⚠ Limitations of the datasheet method
Microphone sensitivities are measured at 1 kHz under controlled conditions. The actual sensitivity of a given unit may differ by a few dB from the nominal value, and gain knob tracking on analogue preamps is rarely exact. Treat this method as a good first estimate — validate it acoustically whenever possible.

Method 2 — Real-time calibration with a reference sound source Most accurate

A sound level calibrator (e.g., a pistonphone or IEC 60942 Class 1 calibrator) emits a pure tone at a precisely known SPL and frequency — typically 94 dB SPL at 1 kHz or 114 dB SPL at 1 kHz. This is the standard field method.

1
Couple the calibrator to your microphone (most come with adaptors for standard capsule diameters — 1/4", 1/2", 1").
2
Start the RUBAT stream with Calibration = 0 (uncalibrated mode). The waveform displays dBFS.
3
Activate the calibrator. Let the level settle for a few seconds, then note the peak amplitude visible in the waveform. Alternatively, record a short clip (Tap or Continuous) and compute the RMS in MATLAB:
[y, fs] = audioread('calibration_clip.wav');
x_rms = rms(y); % ADC units (RMS of the 1 kHz tone)
4
The reference calibrator produces a known RMS pressure. For a 94 dB SPL tone: $$p_{\text{ref, RMS}} = 20\,\mu\text{Pa} \times 10^{94/20} = 1.000 \;\text{Pa RMS}$$ For 114 dB SPL → $p_{\text{ref, RMS}} = 10.000$ Pa RMS.
5
Calculate the factor directly from the ratio: $$\boxed{C = \frac{p_{\text{ref, RMS}}}{x_{\text{RMS}}}}$$ Example: measured RMS = 0.0412 ADC units during a 94 dB SPL tone: $$C = \frac{1.000}{0.0412} \approx 24.3 \;\text{Pa/unit}$$
6
Enter $C$ in the RUBAT Calibration field. The waveform should immediately display ~94 dB SPL (or 114 dB SPL, depending on your calibrator). Fine-tune if needed.

Method 3 — Two-channel comparison with a reference microphone No calibrator needed

If you have one calibrated microphone with a known factor $C_{\text{ref}}$ but want to calibrate a second (test) microphone, you can derive the test factor by recording the same sound source on both channels simultaneously.

1
Mount both microphones as close together as possible, pointing at the same sound source (a loudspeaker, or any stable broadband source works). Distance between capsules should be much less than the shortest wavelength of interest.
2
Record a clip with both mics on separate RUBAT input channels (Continuous mode, same gain settings).
3
Compute the RMS amplitude on each channel over the same time window:
[y, fs] = audioread('comparison_clip.wav');
x_ref = rms(y(:, ch_ref)); % reference mic channel
x_test = rms(y(:, ch_test)); % test mic channel
4
The pressure at both capsules is (approximately) the same, so: $$p = x_{\text{ref}} \cdot C_{\text{ref}} = x_{\text{test}} \cdot C_{\text{test}}$$ $$\boxed{C_{\text{test}} = C_{\text{ref}} \cdot \frac{x_{\text{ref}}}{x_{\text{test}}}}$$
5
Validate the result. Enter $C_{\text{test}}$ into RUBAT, then record the same source again. The dB SPL reading should match what the reference channel shows with $C_{\text{ref}}$ applied. If they agree within ~1 dB, calibration is satisfactory.
⚠ Assumptions and caveats
This method assumes both microphones occupy acoustically identical positions — valid in a free-field if capsule separation ≪ wavelength, but not valid near a reflecting wall or at high frequencies where even a few centimetres of separation introduces a phase difference. For ultrasonic work (>20 kHz), co-locate capsules to within ~1 cm.

Quick reference — dB SPL ↔ Pa

dB SPL Pa (RMS) Typical source
20 dB 0.000 2 Pa Rustling leaves (threshold of hearing)
60 dB 0.02 Pa Normal conversation
94 dB 1 Pa Standard calibrator (Class 1)
114 dB 10 Pa Standard calibrator (high-level)
120 dB 20 Pa Threshold of discomfort
140 dB 200 Pa Jet engine at 30 m
\[L_{\text{SPL}} = 20 \log_{10}\!\left(\frac{p}{20\,\mu\text{Pa}}\right), \qquad p = 20\,\mu\text{Pa} \times 10^{L/20}\]

Entering the value in RUBAT

The Calibration field in the recording panel accepts a value in Pa per ADC unit. Enter 0 to disable (dBFS display). The field is available at all times — you can update it while the stream is stopped or running without restarting.

  Uncalibrated (C = 0) Calibrated (C > 0)
Waveform Y axis dBFS (0 dB = full scale) dB SPL re 20 µPa
Waveform range −120 dBFS … auto 0 … 140 dB SPL … auto
Auto threshold Relative amplitude Absolute dB SPL