This measurement column addresses frequently asked questions received by our customer support center and provides answers to those questions.
When analyzing audio signals recorded with microphones or similar devices using analysis equipment such as FFT analyzers or data stations, acoustic calibration is performed using an acoustic calibrator.
The REF signal and acoustic calibrator of sound level meters and sensor amplifiers are signals at specific sound pressure levels (e.g., 114 dB, 94 dB). By checking the magnitude of the voltage observed by the analysis device when this signal is input, the sensitivity (mV/Pa) of the microphone and sound level meter can be determined. This also serves to verify the operation of the microphone, sound level meter, and analysis device, and corrects for sensitivity deviations due to environmental factors (temperature, etc.) and aging, as well as deviations in the input characteristics of the analysis device. Therefore, this type of acoustic calibration is usually performed when measuring sound.
If acoustic calibration is performed without noticing setting errors, equipment malfunctions, or broken cables, the displayed values may appear correct, but in reality, the measurements are not accurate. This time, following on from last time, we will introduce several examples of incorrect acoustic calibration.
Normal calibration signal
Figure 1 shows an example of connecting a measurement microphone to an analysis device.
Figure 2 shows the results of measuring the signal of a 124dB, 250 Hz acoustic calibrator (pistonphone) by connecting our Measurement Microphone and Preamplifier to a DS-3000 data station. The overall sensitivity of the microphone preamplifier used is approximately 27.6 mV/Pa. Since the magnitude of the calibration signal is 124dB (31.6Pa), the calculation shows that the voltage of the calibration signal is 0.872 V in RMS and 1.233 V in amplitude. Looking at the upper part of Figure 2, the actual maximum value of the time-domain waveform is 1.255 V. The sound pressure signal in the middle part is calculated to be 31.6 Pa in RMS and 44.7 Pa in amplitude. Looking at the middle part of Figure 2, the actual maximum value of the time-domain waveform is 45.49 Pa. Acoustic calibration was performed with this signal applied, so the 250 Hz component and overall value in the lower power spectrum are 124 dB.
-
Figure 2. Normal calibration signal (Top: Voltage waveform, Middle: Sound pressure waveform, Bottom: Power spectrum)
There are two types of acoustic calibrators: pistonphone type and speaker type. The pistonphone type is a calibrator that generates a constant sound pressure by moving a piston inside a cylinder back and forth. Because it generates sound pressure through mechanical piston motion, the power spectrum of the pistonphone type is more prominent compared to the speaker type. In the lower part of Figure 2, the power spectrum shows 72.2 dB at 750 Hz (3rd order component) and 67.9 dB at 1000 Hz (4th order component). Due to the effects of these harmonic distortions, the maximum value of the time-domain waveform deviates slightly from the calculated value, but this is not a problem as acoustic calibration is usually performed using the overall value of the power spectrum.
When CCLD (constant current drive) is turned OFF
CCLD (Constant Current Drive) is a mechanism for supplying power from analysis equipment to sensors and preamplifiers. Some manufacturers call it IPC or IEPE, but it's the same mechanism. When using a CCLD type microphone preamplifier, the analysis equipment must supply CCLD power.
Figure 3 shows the calibration signal measured with the CCLD power supply from the DS-3000 turned OFF. The maximum value of the calibration signal voltage waveform (upper panel) is 0.88 mV. This is a very small value compared to the original value (1.233 V). However, if the vertical axis of the graph is set to autoscale, a waveform close to a sine wave will be displayed, so there is a possibility of performing acoustic calibration without noticing a setting error.
The sound pressure waveform and power spectrum obtained when acoustic calibration is performed in this state are shown in the middle and lower sections of Figure 3. Compared to Figure 2, the sound pressure waveform has noise, and the noise floor of the power spectrum is also high at 80 dB. The EU value obtained by acoustic calibration should be around 0.0276 V/EU, but the actual EU value is 0.0000178 V/EU. This indicates that the microphone sensitivity is only 0.0178 mV/Pa.
By checking the voltage waveform before performing acoustic calibration, you can notice these setting errors because the amplitude is abnormally small. Furthermore, by intentionally performing such incorrect acoustic calibrations and observing the resulting EU values, sound pressure waveforms, and power spectra, you can detect setting errors through abnormal EU values and waveforms.
-
Figure 3 Calibration signal with CCLD OFF (Top: Voltage waveform, Middle: Sound pressure waveform, Bottom: Power spectrum)
When the characteristic filter is turned ON
Figure 4 shows the calibration signal measured with A-weighting applied using an analog filter in the input settings of the DS-3000.
The sensitivity of the human ear varies depending on the frequency. A-weighting (frequency weighting: A) is a characteristic used to measure the perceived loudness of sound to a person. A-weighting is based on 1 kHz, and the correction value at 250 Hz is -8.6 dB.
When performing acoustic calibration, if A-weighting is applied using analysis equipment, sensor amplifiers, sound level meters, etc., the EU value will be shifted by the amount of A-weighting. The standard for frequency weighting is 1 kHz, so there is no problem if the frequency of the sound generated by the acoustic calibrator is 1 kHz. However, when using acoustic calibrators that generate 250 Hz, such as pistonphones, or acoustic calibrators that can switch between generating sounds of multiple frequencies, applying A-weighting or similar effects during acoustic calibration will not yield accurate results.
Looking at the voltage waveform in the upper part of Figure 4, the maximum value, which was 1.255 V in Figure 2, has decreased to 0.462 V, which is about 0.37 times smaller. The EU value obtained by acoustic calibration should be 0.0276 V/EU, but it is a small value of 0.0095 V/EU. However, because the EU value decreased due to the application of A-weighting, the sound pressure waveform in the middle part of Figure 4 and the power spectrum in the lower part appear to be the correct waveforms, so it is not possible to notice the calibration error by looking only at the calibration result waveforms.
If you perform an actual measurement in this state, the phenomenon will occur if the measurement result deviates by 8.6 dB from the previous measurement. If you experience a deviation of 8.6 dB in the measurement result, please check whether the acoustic calibration was performed correctly using the Z-weighting (or C-weighting) method.
-
Figure 4 shows the calibration signal for A-type characteristics (top: voltage waveform, middle: sound pressure waveform, bottom: power spectrum).
summary
When analyzing sound signals recorded with microphones or sound level meters using analysis equipment such as FFT analyzers or data stations, acoustic calibration is performed using the REF signal (CAL signal) of the acoustic calibrator or sound level meter. If there are setting errors, a calibration signal different from a normal calibration signal will be observed.
However, simply being told that a signal is "not a normal calibration signal" can be difficult to discern, so this time, following on from the previous article (Frequently Asked Questions about Measurement - Part 19: "Acoustic Calibration Using Calibration Signals (In the Case of Sound Level Meters)"), we have introduced acoustic calibration using intentionally "incorrect calibration signals."
(Excerpt from the email newsletter issued on January 21, 2018)
