Practical calibration method of airborne ultrasound measurement sys- tems by using acoustic calibrator Hironobu Takahashi 1 , Koto Hirano 2 and Keisuke Yamada 3 National Institute of Advanced Industrial Science and Technology, National Metrology Institute of Japan (AIST/NMIJ) Tsukuba central 3, 1-1-1 Umezono, Tsukuba, Ibaraki, Japan

ABSTRACT Airborne ultrasound appliances have been actively developed and are becoming increasingly popu- lar. To evaluate acoustical performances or to assess the safety of the exposure to airborne ultra- sound, quantitative airborne ultrasound measurement is required. Currently, we ourselves need to build a measurement system by combining various measuring instruments for measuring ultrasound. However, there has been little information on the setup or calibration of measurement systems. Thus, this paper showed an example of the practical measurement system and its calibration procedure that consider the balance between cost-effectiveness and accuracy. The measurement system is com- posed of the WS3 microphone specified in IEC 61094-4. Concerning the calibration procedure, it is practical to calibrate one specific frequency by an acoustic calibrator and the other frequencies by adopting the specification of instruments, for example, the relative microphone sensitivity given by a manufacturer. Since the calibration procedure shown here can be broken down into a few steps, we investigated the step regarding the calibration by an acoustic calibrator. As a result, it was found that the achievable best accuracy in this step was expected to be 0.2 dB when using a feedback type acoustic calibrator.

1. INTRODUCTION

The application of airborne ultrasound has become more common in daily life. The pest repellent system is a widely known example of such an application. Applications developed recently are para- metric speakers and airborne ultrasound tactile devices [1, 2], and these emit extremely high-intensity airborne ultrasound. In some cases, the sound pressure level in their vicinity exceeds 140 dB.

The above is an example of applications that emit airborne ultrasound intentionally. On the other hand, the application that emits ultrasound unintentionally also exists. Resent electric appliances con- tain a switching or inverter circuit to save electric power consumption. These circuits deal with high- frequency electrical signals ranging from around 10 kHz to 1 MHz. In some cases, electrical parts in the circuit are vibrated by a high-frequency electrical signal and emit airborne ultrasound.

1 h.takahashi@aist.go.jp

2 koto.hirano@aist.go.jp

3 keisuke.yamada@aist.go.jp

a 2022

a 2022

As the airborne ultrasound application becomes more popular, the safety of the exposure to air- borne ultrasound is always the subject of discussion [3]. The discussion is still ongoing, and clarifi- cation of the measurement is essential to promote the discussion. At present various types of instru- ments are available for airborne ultrasound measurement. However, sometimes the measurement is made without calibration. This causes the problem of measurement reliability when comparing results measured by different people. In particular, the information on uncertainty or accuracy is quite im- portant for comparison.

Theoretically, if the calibrated microphone is prepared, the sound pressure at the microphone po- sition can be obtained from the open-circuit voltage of the microphone. However, measuring the accurate open-circuit voltage requires special equipment and complicated procedures. Moreover, due to cost-effectiveness of the microphone calibration, this approach tends to be avoided. At present, there has been little information on the measuring system and its calibration for airborne ultrasound. Thus, this paper shows an example of the practical measurement system and its calibration procedure considering the balance between cost-effectiveness and accuracy. Since the calibration procedure can be broken down into a few steps, we investigated the step regarding the calibration by an acoustic calibrator in particular. 2. MEASUREMENT SYSTEM AND ITS PRACTICAL CALIBRATION

2.1. Measurement system

WS3 microphone

Figure 1 shows a typical system for measuring airborne ultrasound. The system consists of a mi- crophone, microphone preamplifier, amplifier and electrical measuring instrument such as an FFT an- alyzer. As the microphone, the WS3 microphone, which is the measurement microphone specified in IEC 61094-4 [4], should be selected considering the frequency range and the calibration procedure explained later. The band-pass filter may be in- cluded in the system to eliminate unwanted fre- quency signals, but it omits in Fig. 1.

Amplifier

Preamplifier

Electrical measuring instrument

(FFT Analyzer etc.)

Figure 1: Typical airborne ultrasound measurement system

10

2.2. Influence of microphone protection grid

5

Relative Sensitivity [dB]

Ordinally, the protection grid of a microphone is never removed to protect the microphone mem- brane. Figure 2 shows an example of the relative sensitivities of a WS3 microphone with and with- out the protection grid. The sensitivities with the protection grid do not have flat frequency charac- teristics. It reaches about 6 dB at around 40 kHz. Above 40 kHz, the sensitivity drops shapely and less than -10 dB over 80 kHz. In contrast, the sen- sitivities without the protection grid have flat fre- quency characteristics of 1 dB up to 100 kHz.

0

-5

-10

-15

With grid Without grid

-20

10 2 10 3 10 4 10 5 Frequency [Hz]

Figure 2: Example of the relative free-field sensitivities of WS3 microphone (BK 4939, SN: 2451478), ( - ) With protection grid, ( - ) Without protection grid (From CD-ROM data attached to the microphone)

Thus, when using the microphone with the pro- tection grid, the frequency characteristics of the microphone have to be corrected. This correction is a rather complicated process. If the tolerance of

approximately 1 dB is allowed for the sensitivity flatness, the correction is not necessary by removing the protection grid from the microphone.

2.3. Practical Calibration Procedure

If the calibrated microphone is not used for the measurement, the measurement system has to be calibrated by ourselves. The calibration procedure we propose is as follows: firstly, the system is calibrated by an acoustic calibrator with a specified frequency to determine the absolute sound pres- sure level. Secondly, the frequency characteristics of the microphone are determined by the relative sensitivities of the microphone, such as shown in Fig. 2. In many cases, the information on the relative sensitivities is attached to the microphone or provided by a manufacturer. The frequency character- istics of the rest instruments are adopted from their specifications, which are provided as, for example, frequency flatness or tolerance level.

2.4. Factors Influencing Accuracy When Calibrating the System by Acoustic Calibrator

Of interest here is the accuracy achieved by this calibration procedure. The calibration procedure in Sect. 2.3 can be broken down into a few steps. In this paper, we investigated the step “the calibra- tion by an acoustic calibrator”.

When calibrating the system by an acoustic calibrator, the improper fixing of the WS3 microphone to the acoustic calibrator influences the calibration accuracy. Regarding the microphone protection grid, its existence becomes the factor when measuring ultrasound without the protection grid. Usu- ally, fitting the acoustic calibrator to the microphone without the protection grid is avoided to prevent damage to microphone membranes. This means that the conditions when calibrating and measuring are different. Therefore, we investigated the factors, the improper fixing of the microphone and the influence of the protection grid.

3. IMPROPER FIXING OF MICROPHONE TO SOUND CALIBRATOR

3.1. Examining Acoustic Calibrator

Four different models of acoustic calibrators were examined. Two acoustic calibrators were Brüel & Kjær (hereinafter BK) 4226 (124 dB, 250 Hz) and Rion NC-72A (114 dB, 250 Hz), which were pistonphones classified as class LS/C in IEC 60942 [5]. The other two acoustic calibrators were BK 4231 (94 dB, 1 kHz) and Rion NC-75 (94 dB, 1 kHz). They were called the feedback type acoustic calibrator and classified as class 1 in the IEC standard. Unlike pistonphones, the feedback type acous- tic calibrator needs no static pressure correction because it has a feedback control circuit to keep the generating pressure constant irrespective of atmospheric pressure.

3.2. Measurement Method

Figure 3 depicts the measurement system for examining the improper fixing of the microphone to the acoustic calibrators. The microphone fitted to the sound calibrators was BK 4939, classified as a WS3 microphone in IEC 61094-4 [4].

In this system, the microphone was connected to a preamplifier BK 2670. The signal from the preamplifier was amplified by a conditioning amplifier BK NEXUS. The sound pressure generated by the sound calibrators was 94 dB to 124 dB, differing from the models. The conditioning amplifier adjusted the signal level in such a way that the AC voltage observed at a DMM, Agilent 3458A, was around -14 dB (re. 1 V/dB), independent of the models. The signal except ranging from 210 Hz to 1200 Hz was suppressed with a band-pass filter NF 3625. The SN ratios, the difference in voltage observed by the DMM when the acoustic calibrator was switched on and off, were more than 50 dB in all cases.

a 2022

a 2022

Acoustic calibrator

Band-pass filter (210 Hz to 1200 Hz)

Conditioning amplifier

(BK NEXUS)

(NF 3625)

Preamplifier (BK 2673)

Digital multimeter

WS3 microphone (BK 4939)

(HP 3458A)

Fig. 3: Schematic diagram of the measurement system for examining fixing difference of microphone to sound calibrator

The acoustic calibrator was placed on a ta- ble. As for the fixing condition of the micro- phone to the acoustic calibrator, two different conditions were examined. One was that the microphone was inserted vertically into the acoustic calibrator but did not fix each other tightly, simulating an improper fixing (called No fixing). The other was similar, but the mi- crophone and sound calibrator were fixed tightly with a rubber band (called Fixing).

Table 1: Mean and standard deviation for each fix- ing condition and acoustic calibrator

Observed voltage by

Fixing condition

DMM Mean [dB] SD [dB] (a) BK 4228 No-fixing -14.457 0.010 Fixing -14.412 0.001 (b) NC-72A No-fixing -14.626 0.006 Fixing -14.401 0.008 (c) BK 4231 No-fixing -14.480 0.002 Fixing -14.483 0.004 (d) NC-75 No-fixing -14.640 0.002 Fixing -14.651 0.003

3.3. Result

Each of the two fixing conditions was measured five times for the four acoustic cal- ibrators. Generally, the microphone sensitivi- ties vary with temperature, static pressure and humidity. During the measurement, the tem- perature and static pressure varied 0.2 ℃ and 0.2 kPa. Such a level change did not affect the sensitivity. Since the humidity coefficient of the microphone was quite small, the humidity change was also ignored.

Table 1 summarizes the mean and standard deviation obtained from these five-time measurements for each acoustic calibrator and fixing condition. The standard deviations are less than 0.010 dB in all cases, indicating stable measurements. Focus on the mean value, the influence of fixing condition can be found with the pistonphones, BK 4228 and Rion NC-72A. The mean value differences are 0.05 dB for BK 4228 and 0.23 dB for Rion NC-72A. In contrast, no obvious differences are found with the feedback type acoustic calibrators, the differences being less than 0.01 dB.

This means that the pistonphones need careful treatment. The improper fixing of the microphone to the pistonphone affected the acoustic compliance of the pistonphone cavity, resulting in an incor- rect calibration. From a different point of view, this can be interpreted that the improper fixing only made a difference of at most 0.3 dB. In the case of the feedback type acoustic calibrators, stable calibration was realized even if the microphone was fitted to the acoustic calibrator a little rough. This is because the sound pressure at the sensor, which was installed in the cavity of the acoustic calibrator, was feedback-controlled to a constant and not affected by the fixing condition.

Considering the on-site calibration, improper fixing is likely to occur. Thus, the feedback type acoustic calibrators are appropriate for this purpose and easy to use. Even if the microphone is roughly fitted to the feedback type acoustic calibrator, an accuracy of about 0.01 dB can be achieved. Addi- tionally, it has a very small value, and this factor may be negligible under certain situations.

4. INFLUENCE OF MICROPHONE PROTECTION GRID

4.1. Measurement Procedure

The sensitivity difference with and without the protection grid was obtained based on the relative sensitivity to the reference microphone. The determination of the relative sensitivities was followed by IEC 61095-5 [6].

Suppose that the pressure sensitivities of the WS3 microphone with and without the protection grid are denoted as 𝑀𝑀 𝑤𝑤𝑤𝑤 and 𝑀𝑀 𝑤𝑤𝑤𝑤𝑤𝑤 , respectively. The pressure sensitivity of the reference micro- phone is denoted as 𝑀𝑀 𝑟𝑟𝑟𝑟𝑟𝑟 . When the WS3 microphone with the protection grid and the reference microphone are set face to face with a narrow space in a free field, the difference in the output voltage between both microphones 𝐻𝐻 𝑤𝑤𝑤𝑤 is written as follows:

𝐻𝐻 𝑤𝑤𝑤𝑤 = ൫𝑀𝑀 𝑤𝑤𝑤𝑤 + 𝐺𝐺 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 ൯−൫𝑀𝑀 𝑟𝑟𝑟𝑟𝑟𝑟 + 𝐺𝐺 𝑟𝑟𝑟𝑟𝑟𝑟 ൯ . (1)

Similarly, the difference when the protection grid is removed from the microphone 𝐻𝐻 𝑤𝑤𝑤𝑤𝑤𝑤 is given as

𝐻𝐻 𝑤𝑤𝑤𝑤𝑤𝑤 = ൫𝑀𝑀 𝑤𝑤𝑤𝑤𝑤𝑤 + 𝐺𝐺 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 ൯−൫𝑀𝑀 𝑟𝑟𝑟𝑟𝑟𝑟 + 𝐺𝐺 𝑟𝑟𝑟𝑟𝑟𝑟 ൯ , (2)

where the parameter 𝐺𝐺 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 represents electrical factors, such as a preamplifier and signal amplifier, other than the sensitivity of the WS3 microphone. The parameter 𝐺𝐺 𝑟𝑟𝑟𝑟𝑟𝑟 is also the factor other than the sensitivity of the reference microphone. By combining Eqs. (1) and (2), the sensitivity difference with and without the protection grid Δ is written as

Δ ≡𝑀𝑀 𝑤𝑤𝑤𝑤 −𝑀𝑀 𝑤𝑤𝑤𝑤𝑤𝑤 = 𝐻𝐻 𝑤𝑤𝑤𝑤 −𝐻𝐻 𝑤𝑤𝑤𝑤𝑤𝑤 . (3)

4.2. Measurement Method

The sensitivity difference with and without the protection grid was investigated for three BK 4939 microphones and three GRAS 40 BF microphones, classified as WS3 microphones in IEC 61094-4 [4]. The reference microphone was BK 4180, classified as an LS2 microphone in IEC 61094-1 [7].

Figure 4 shows the schematic diagram of the measurement system for examining the sensitivity difference with and without the protection grid. The measurements were made at an anechoic cham- ber in our institute, the inner dimension of which was 5.4 m in width, 3.8 m in depth and 4.6 m in height. The wire mesh floor was located 0.9 m above the bottom of the chamber. The cut-off fre- quency of the chamber was 80 Hz.

The WS3 and reference microphones with a preamplifier were each attached to a 1 m long metal pipe. They were then hung from the ceiling of the anechoic chamber in such a way that the micro- phone membrane of both microphones faced each other apart from 1 cm. The microphone height from the wire mesh floor was set to 80 cm. The loudspeaker radiated band-limited swept sine signal ranging from 100 Hz to 3 kHz, which was generated by a data acquisition device, National Instru- ments USB-4431. The sound pressure level at the microphone position was set to 80 dB by adjusting

a 2022

a 2022

Anechoic chamber

Ref. microphone (BK 4180)

WS3 microphone (BK 4939)

Conditioning amplifier

1 [m]

(BK NEXUS)

Sig. in

Preamplifier

(BK 2669)

Sig. out

Power amplifier

Data acquisition device

(Luxman B-1)

(NI USB-4431)

Fig. 4: Schematic diagram of the measurement system for examining the influence of microphone grid

the volume of a power amplifier Luxman B-1. The signals from both microphones were amplified by a conditioning amplifier, BK NEXUS, and ac- quired and processed 500 times synchronous addi- tion by the data acquisition device. The relative sensitivity was determined based on this synchro- nous adding data.

Table 2: Mean and standard deviation of the sensitivity difference with and without the protection grid for WS3 microphone.

The sensitivity difference

with and without protection grid ∆ Mean[dB] SD [dB] BK 4939 (250 Hz)

Microphones

In the measurement procedure, firstly, the rela- tive sensitivity when the protection grid was at- tached was measured. Next, keeping all the posi- tions unchanged, only the protection grid of the microphone was removed. Then, the relative sen- sitivity without the protection grid was measured. The time which took the relative sensitivity with and without the protection grid for one micro- phone was approximately 15 minutes. During this period, since temperature and static pressure fluc- tuations were little observed in all cases, the vari- ations of all microphone sensitivity were ignored.

SN: 2451477 0.007 0.012 SN: 2451478 -0.011 0.009 SN: 2546599 -0.009 0.014 GRAS 40BF (250 Hz)

SN: 320944 0.004 0.008 SN: 321078 0.011 0.006 SN: 334003 0.001 0.007 BK 4939 (1 kHz)

SN: 2451477 -0.002 0.009 SN: 2451478 -0.017 0.016 SN: 2546599 -0.020 0.012 GRAS 40BF (1 kHz)

4.3. Result

The sensitivity difference with and without the protection grid was measured five times for each microphone. Table 2 summarizes the mean and standard deviation obtained from this five-time measurement. Since the emitting frequency from acoustic calibrators is ordinally 250 Hz or 1 kHz, the analysis is done for these two frequencies.

SN: 320944 0.006 0.017 SN: 321078 -0.002 0.014 SN: 334003 0.001 0.006

Focusing on the mean values, all are uniformly within ± 0.020 dB, not depending on the micro- phone model or frequencies. The standard deviation of each microphone varies from 0.006 dB to 0.017 dB, and distinctive characteristics cannot be found. From the mean of ± 0.02 dB and maximum standard deviation of 0.02 dB, the sensitivity difference with and without the protection grid is esti- mated to be ± 0.06 dB, providing an approximate 95 % confidence level.

5. DISCUSSION

So far, the calibration by an acoustic calibrator, which is a part of the procedure for calibrating airborne ultrasound measurement systems, has been investigated. The influence of the improper fix- ing was found to be less than 0.01 dB if applying the feedback type acoustic calibrator. In terms of the protection grid of the WS3 microphone, the sensitivity difference with and without the protection grid was estimated to be 0.06 dB. This means that a 0.06 dB discrepancy will occur if the protection grid is attached when calibrating and removed when measuring ultrasound.

Currently, the calibration of sound calibrators is possible with the expanded uncertainty of 0.1 dB to 0.3 dB with the coverage factor of 2, corresponding to the combined standard uncertainty of 0.05 dB to 0.15 dB. From the information obtained here, let us estimate the accuracy of the step “calibra- tion by acoustic calibrator”. Suppose that the calibration is performed by the feedback type sound calibrator and the microphone protection grid is removed when measuring ultrasound. Under this condition, the dominant factors are the generating sound pressure from the sound calibrator of ap- proximately 0.15 dB and the protection grid of 0.06 dB. The influence of improper fixing is so small and negligible. In total, the accuracy is estimated to be around 0.2 dB.

Following the calibration procedure proposed in this paper, the remaining factors that affect the calibration accuracy involve relative free-field sensitivity and frequency characteristics of measuring instruments. Though their accuracies have not been investigated yet, they appear to be on the order of 1.0 dB. The accuracy of the step related to the calibration by an acoustic calibrator seems to be higher than the accuracy of the rest factors. It is premature to conclude at this stage, however, the influence of the calibration by an acoustic calibrator may be negligible. Further investigation is nec- essary to clear this prospect.

6. CONCLUSION

This paper showed an example of the practical measurement system for measuring airborne ultra- sound and its calibration procedure that considers the balance between cost-effectiveness and accu- racy. Since the calibration procedure is composed of a few steps, we investigated the step regarding the calibration by an acoustic calibrator in particular and found that the achievable accuracy in this step was estimated to be 0.2 dB by using the feedback type acoustic calibrator. Here, only the accu- racy in a part of the calibration procedure has been investigated, and other parts still remain. Investigating the remaining factors is future work. 7. REFERENCES

1. Hoshi, T., Takahashi, M., Iwamoto & Shinoda, H. Non-contact tactile display based on radiation pressure of airborne ultrasound. IEEE transactions on haptics, 3(3) , 155-165 (2010). 2. Gan, Woon-Seng., Yang, J., & Kamakura, T., A review of parametric acoustic array in air. Ap- plied Acoustics , 73(12) , 1211–1219 (2012). 3. For example, Leighton, T. G., Are some people suffering as a result of increasing mass exposure of the public to ultrasound in air? Proceedings of the Royal Society A , 472(2185) , 20150624 (2016). 4. IEC, Measurement microphones - Part 4: Specifications for working standard microphones , IEC 61094-4: 1995. 5. IEC, Electroacoustics - Sound calibrators , IEC 60942: 2017. 6. IEC, Electroacoustics - Measurement microphones - Part 5: Methods for pressure calibration of working standard microphones by comparison , IEC 61094-5: 2016. 7. IEC, Measurement microphones - Part 1: Specifications for laboratory standard microphones , IEC 61094-1: 2000.

a 2022