Sunteți pe pagina 1din 14

Review: Assess the microphone and data acquisition system provided by Sonitus systems.

Introduction The aim of this report is assess the microphone and data acquisition system provided by Sonitus systems. This will be achieved through a number of steps. The microphone characteristics are compared with a high quality GRAS microphone (BF40). The sensitivity and frequency response of the microphone is examined. The tests conducted, the procedure and aim of the specific tests are outlined below. In addition the data acquisition system was also compared. Finally the output from the system, i.e. the A-weighted values such as LAEq was calculated and compared. Objectives Microphone sensitivity Threshold noise level of the microphone Linearity of the microphone Frequency response of the microphone Maximum measurement range of the microphone A-weighting comparison Quality of Sonitus DAQ

Procedure & Results Microphone sensitivity In the first instance the sensitivity of the microphone was obtained using a Bruel & Kjaer Sound Level Calibrator Type 4231 at a frequency of 1000Hz and pressure of 1Pa (94dB). Calculating the RMS level of the voltage signal the sensitivity was found to 0.214V/Pa. Threshold noise level of the microphone The threshold noise level of the Sonitus microphone was measured and compared with the high quality GRAS microphone. This was achieved by placing the microphones in a box insulated from extraneous sound sources and acquiring the signals using the National Instruments DAQ (Labview).

Figure 1: Background noise test GRAS (red) and Sonitus (blue) Microphones 8th and 12th Aug

Figure 2: Background noise test GRAS (red) and Sonitus (blue) Microphones 8th and 12th Aug Figure 1 plots the threshold noise level noise as a function of sound pressure level against frequency. The noise floor of the Sonitus microphone is less than that of GRAS microphone. However, there are two spurious narrow band peaks as well as a discrete single tone peak. It is not clear where these arise from. However, they exist irrespective of whether the data was acquired using the Sonitus system or the National Instruments system. This would indicate that they origin of these frequencies is due to the microphone amplifier circuitry. Furthermore, examining Figure 2, the peaks changed when the tests were repeated.

Linearity of the microphone The next series of test are used to examine the linearity of the microphone. This series of tests will also be used to compare the frequency response of Sonitus microphone against the high quality GRAS microphone. Finally, using these tests the quality of the Sonitus data acquisition system will also be examined. The test setup requires that both microphones experience the same fluctuating pressures at all frequencies. This was achieved using the concept that in a cylindrical duct plane waves propagate below a certain cut-off frequency (Eqn. 1). If two microphones are located in the same plane along the duct they are exposed to the same pressure wave distribution for frequencies below the cut-off frequency. f cutoff = 1.84c 2a (1)

where c is the speed of sound and a is the radius of the impedance tube. A schematic of the test set up is shown in Figure 3. The setup included an impedance tube with a diameter 0.05115m. At the end of the impedance tube the GRAS and Sonitus microphones were flush mounted. At the other end of the impedance tube there was a speaker connected to a white noise1 generator.

Figure 3: Schematic of test setup The tests were conducted at a sample frequency of 49152Hz for 8 seconds which was equivalent to the parameters of the Sonitus system. Five quantities were measured simultaneously in this test series and are detailed below along with the respect data acquisition system. National Instruments (Labview) Channel 1 Trigger from Sonitus DAQ Channel 2 Output from white noise generator Channel 3 Output from Sonitus Microphone (before A-D converter) Channel 4 Output from GRAS Microphone Sonitus Wireless connection from Sonitus DAQ with output from Sonitus Microphone. Column 1: Output from Sonitus microphone Column 2: A-weighting value from Sonitus microphone
1

White noise is random signal used to generate energy at all frequencies.

Tests were repeated for a number for a range of speaker loudness i.e. a range of outputs from the white noise generator.

Figure 4: Sonitus microphone - linearity plot

Figure 5: GRAS microphone - linearity plot In the first instance the linearity of the microphone was examined. To examine the linearity of the microphone, the time domain voltage data was converted into the frequency domain pressure data. The RMS of the white noise generator signal and the microphone signal where plot against each other (Figure 4). It is clear from Figure 4 that the microphone responds linearly which is further verified by an R2 2 value close to 1. The linearity of the Sonitus microphone is comparable to GRAS microphone
2

R2 is the proportion of variability in a data set.

(Figure 5). However, these results do not provide information about the frequency response or whether unwanted artefacts exist in the measurements. Frequency response of the microphone To examine the frequency response of the Sonitus microphone, it was compared with the GRAS microphone which has a flat frequency response up to 20kHz (see Figure 6). The response of the Sonitus microphone was compared to the GRAS microphone by examining the transfer function between the two microphones. The test setup is described in the previous section i.e. impedance tube with white noise source input. As mentioned previously, in a cylindrical duct plane waves propagate below a certain cut-off frequency. The dimensions of the impedance tube result in a cut-off frequency of approximately 12kHz. Hence, the transfer function between the two microphones will be examined up to the 12kHz limit. Examining the transfer function between the two microphones it is clear that the response is not flat and varies as a function of frequency. At low frequencies the transfer function is poor. This is as a result of a limitation of the speaker range to excite the lower frequency components. Using a speaker capable of producing the lower frequency components the transfer function curve is cleaner. At the higher frequencies the transfer function changes considerably. This is due to the fact that higher frequency range is more susceptible to the position of the microphones. Figure 7 shows the transfer function with the Sonitus microphone position shifted. The Sonitus microphone was moved to several positions to find the flattest response. However, the position resulting in the flattest response was not flush mounted. Having a frequency response that is not flat is problematic such that the sensitivity measured at 1kHz will be different at other frequencies. The difference between the two microphones is further demonstrated when one examines the power spectral density (Figure 8) as demonstrated by the difference observed in the higher frequency ranges.

Figure 6: Transfer function between GRAS and Sonitus microphones Test 10 8th Aug

Figure 7: Transfer function between GRAS and Sonitus microphones Test 17 and 19 12th Aug

Figure 8: White noise PSD for GRAS (red) and Sonitus (blue) Microphones Test 8 8th Aug Maximum measurement range of the microphone A second series of tests examined the magnitude at which clipping 3 occurs. These tests used a signal generator connected to the speaker producing a single tone at frequencies of 0.5, 1, 2, 4, 7, 10, 13, 15, 17, 20kHz for five different speaker levels.

Clipping is a form of waveform distortion that occurs when an amplifier is overdriven

As stated previously the aim of this series of tests was to determine the magnitude of acoustic pressure clipping occurs at. It was observed that before clipping occurred there was significant distortion when the amplitudes were high. It must be noted in the overall scheme, the magnitude although high for Sonitus microphone it would be considered small in the broader understanding of loudness amplitude levels. Figure 9 shows the Sound pressure level against frequency for an input frequency of 2kHz. Examining the power spectra density plots it is seen that both the Sonitus and the GRAS microphones have energy at the input frequency and its harmonics. However, when the magnitude reaches a certain level the Sonitus microphone displays additional spurious tones not observed in the high quality GRAS microphone. It must be noted that these spurious frequencies occur at magnitudes significantly lower than the magnitudes required for clipping.

Figure 9: 2kHz tone at various levels. SPL against Frequency. Shows spurious frequencies from Sonitus microphone (blue) GRAS microphone (red) Test 33-36 11th Aug The maximum sound pressure level that can be achieved is frequency dependant. Table 1 shows a number of frequencies with the respective sound pressure levels, in some cases before spurious frequencies (sp) were observed, in other cases were spurious frequencies have just appeared. Also detailed is the corresponding RMS voltage level. In is noteworthy that the voltage level is below the 2.5V peak to peak the microphone and the respective circuitry can measure.

Frequency (kHz) 0.5 1 2 4 7 10 13 15 17 20

SPL (dB) 104 103.5 93 94sp 106 97sp 99 102sp 100sp 101sp

Voltage RMS (V) 0.93 0.85 0.23 0.26 0.92 0.34 0.32 0.42 0.33 0.32

Table 1 The question now remains as to whether these spurious frequencies a significant effect on the overall sound pressure level. The calibration transfer function was applied to the second series of test data. The calibration transfer function was good up to 11kHz due to limitations in the setup. The RMS of the pressure signal up to 11kHz was examined for both the GRAS and Sonitus microphones. Frequencies of 500, 1k, 2k, 4k, 7k and 10kHz were examined at various loudness levels. Results are presented in Table 2. Typically the difference in the overall pressure was 4% at most. It is noticeable that at the lower loudness levels the Sonitus microphone has an RMS level below that of the GRAS microphone. This is due to the lower background noise level of the Sonitus microphone. However, with increasing loudness the RMS level of the Sonitus microphone is greater than the GRAS microphone. This results from the spurious frequencies observed in the Sonitus microphone. At 10kHz the pressure difference is of the order of 20%. This difference was not solely caused not by the spurious frequencies as the RMS level of the pressure signal from the GRAS microphone is always below that of the Sonitus microphone. This outcome can be explained by the fact that there is a difference in the pressure at 10kHz, irrespective of whether spurious frequencies are present. This could be as a result of an error in the calibration transfer function. As shown in Figures 6 and 7 the transfer function is sensitive to microphone positioning in the impedance tube. Secondly, for this test series the two microphones where located close together and at higher frequencies the microphones may not have been subject to the exact same pressure field (i.e. not plane wave conditions). In the next section A-weighting values for measurements conducted in the impedance tube show a smaller difference between the Sonitus and GRAS microphones suggesting that the latter (test conditions not plane wave) explanation is the most probable.

Frequency 500Hz

Sig Gen (RMS) 0.1643 0.5266 0.8691 1.2571 0.2437 0.4854 0.6817 0.9255 0.2459 0.5876 0.8821 1.2758 0.4008 0.9582 1.5994 2.399 0.1684 0.3369 0.5907 0.8451 0.2542 0.5076 0.8484 1.2738

Sonitus (Pa-RMS) 0.8127 2.6175 4.3764 6.4229 1.9859 4.0084 5.6511 7.7902 1.0679 2.5469 3.8286 5.5843 1.1989 2.7817 4.6446 6.9849 1.4923 2.9756 5.2524 7.6056 2.1983 4.3508 7.3505 10.8641

Gras (Pa-RMS) 0.8214 2.6217 4.3232 6.2343 2.0293 4.0356 5.617 7.6044 1.1534 2.741 4.0898 5.8684 1.3068 3.0192 4.9682 7.2974 1.4954 2.9736 5.1808 7.3772 1.906 3.7648 6.1999 9.1674

Difference (Pa) -0.0087 -0.0042 0.0532 0.1886 -0.0434 -0.0272 0.0341 0.1858 -0.0855 -0.1941 -0.2612 -0.2841 -0.1079 -0.2375 -0.3236 -0.3125 -0.0031 0.002 0.0716 0.2284 0.2923 0.586 1.1506 1.6967

1kHz

2kHz

4kHz

7kHz

10kHz

Table 2 A-weighting comparison This section examines the difference between the A-weighting sound pressure level for the GRAS and Sonitus microphone. It also examines the difference between the A-weighting sound pressure level for the calibrated Sonitus Microphone (Calibration transfer function applied) and the GRAS microphone. Finally the A-weighting value output from the Sonitus system is compared with the value from the raw microphone data calculated offline. The data used for the analysis in this section is from the first series of tests. This was calculated in the frequency domain and was performed in accordance with BS EN 61672-1:2003. The A-weighting formula is shown in Equation 2 and the Aweighting curve as a function of frequency is shown in Figure 14. f 42 f 2 A( f ) = 20 log10 2 2 2 2 1/ 2 f 2 + f 32 f + f1 f + f 2 A1000

)(

) (

) (f
1/ 2

+ f 42

(1)

where f1 = 20.6 Hz f 2 = 107.7 Hz f1 = 737.9 Hz f1 = 12194 Hz

A1000 is -2dB, the gain needed to provide frequency weightings of 0dB at 1kHz.

Figure 14: A-weighting curve against frequency The Sonitus system including the microphone and DAQs A-weighting values are less than the Sonitus microphone and Labview DAQ. This is as a result of the different anti-aliasing filters used by the system. Firstly the sonitus system filters after 20kHz whereas the labview system filters after 22kHz. Secondly, the filters are different resulting in different attenuation characteristics especially at the higher frequency range. The difference was demonstrated in Figure 12 where a comparison of the spectrum of the microphone from both systems was plot. Finally, the time domain filter differs slightly to the standardised frequency domain filter. Incidentally both Aweighting values are lower than the value obtain by the GRAS microphone. When the calibration transfer function is applied to the Sonitus microphone and the A-weighting value is obtain via the frequency domain analysis the value increases beyond the level recorded by the GRAS microphone. It might be expected that it would be below the level of the GRAS microphone as it has a lower threshold noise level. There are a number of probable reason why this was not the case. Firstly, the calibration transfer function was incorrect. Secondly, the spurious frequencies discussed earlier are providing and additional contribution to the overall A-weighted sound pressure level.

Gras

Time Domain Analysis Sonitus Sonitus S

Sonitus S AW 0.0604 0.0797 0.1279 0.1844 0.2439 0.2979 0.6307 0.9329

Gras

Gras AW 0.0795 0.0968 0.1516 0.217 0.2883 0.3466 0.735 1.0967

Frequency Domain Analysis Sonitus Sonitus - AW

Sonitus Cal 11k 0.0975 0.1166 0.1607 0.2227 0.2951 0.3517 0.7318 1.0902

Sonitus Cal 11kHz AW 0.0786 0.0949 0.1477 0.2109 0.2803 0.3366 0.7127 1.0653

0.1034 0.1325 0.1948 0.2426 0.3114 0.3809 0.7822 1.1466

0.0922 0.1164 0.1576 0.2106 0.2744 0.3354 0.6863 1.004

0.0916 0.1152 0.1546 0.2055 0.2672 0.3266 0.6651 0.9715

0.0993 0.1207 0.1699 0.2355 0.3119 0.3731 0.7801 1.1596

0.0962 0.1137 0.1531 0.2107 0.2786 0.3313 0.6855 1.0198

0.0766 0.0903 0.1368 0.1937 0.2568 0.3076 0.6477 0.9677

74.27 76.42 79.77 81.68 83.85 85.60 91.85 95.17

73.27 75.30 77.93 80.45 82.75 84.49 90.71 94.01

73.22 75.21 77.76 80.24 82.52 84.26 90.44 93.73

69.60 72.01 76.12 79.29 81.72 83.46 89.98 93.38

73.92 75.61 78.58 81.42 83.86 85.42 91.82 95.27

71.99 73.70 77.59 80.71 83.18 84.78 91.31 94.78

73.64 75.09 77.68 80.45 82.88 84.38 90.70 94.15

71.66 73.09 76.70 79.72 82.17 83.74 90.21 93.69

73.76 75.31 78.10 80.93 83.38 84.90 91.27 94.73

71.89 73.52 77.37 80.46 82.93 84.52 91.04 94.53

Table 3

Quality of Sonitus DAQ The quality of the Sonitus DAQ was analysed using the data from the first series of tests used to define the linearity of the microphone and the transfer function between the microphones. The analysis was based on the signal acquired was that from the Sonitus microphone. Two signals were used, the first was from the National Instruments DAQ. It was connected to an output after the A-D converter in the Sonitus system. The second signal was the wireless output from the Sonitus system. Figure 10 plots the two signals. From Figure 10(a) it appears that the signals are in reasonable agreement. At the start of the acquisition the signals are similar. However, further on in the acquisition a delay becomes apparent and this is magnified towards the end of the acquisition. Figure 11 shows the transfer function between the two signals. It is observed that there is significant magnitude and phase distortion. However, examining the power spectral density of the signals, both signals contain similar frequency information. At higher frequencies there is a difference in magnitude. Dividing the spectrum yields a basic transfer function (Figure 13) which shows better correlation between the two systems. It is not clear why the complex transfer is so different.

Figure 10: Time series Labview (blue) against Sonitus DAQ (red) Test 10 8th Aug

Figure 11: Transfer Function Magnitude and Phase between data from Sonitus microphone acquired using Labview and Sonitus DAQ (Complex Transfer Function) Test 10 8th Aug

Figure 12: PSD for Sonitus microphone; Acquired using National Instruments and Sonitus DAQ systems Test 4 8th Aug

Figure 13: Transfer Function Magnitude between data from Sonitus microphone acquired using Labview and Sonitus DAQ got by diving PSDs (Basic Transfer Function) Test 10 8th Aug

S-ar putea să vă placă și