Microphone Qualification Tests

A brief comparison between normal conditions, 5 mbar and -40°C

2015-03-11T11:44:54Z

Here is a comparison between tests under normal conditions, 5 mbar and -40°C.
It shows that :

a lower pressure decreases the sound power received by the microphone while temperature has no effect on it
the microphone is well qualified at least above 2000 Hz in extreme conditions

Test at -40°C

2015-02-09T16:14:35Z

As we had already tested the microphone at low pressure, the objective of this test was to determine the frequency response of the microphone under a temperature of -40°C (Martian surface temperature), still under with an Earth-like atmosphere for now.
The test consisted in recording 60 samples of 1 second at a sampling frequency of 100 kHz, for each frequency. The shift of frequency and the record of samples were automatically done by the Matlab script. The Matlab script made all the work and tested 38 frequencies (16 between 100 Hz and 1000 Hz, 17 between 1000 Hz and 10 kHz, and 5 between 10 kHz and 15 kHz, all evenly spaced). The goal of this test was also to analyse the effect of the distance between the emitting source and the microphone, so we changed the gauge.

What did we retrieve ?

We obtained 60 noise records, plus 60 records for each of the 38 frequencies that we tested.
Although we used a square wave below 1000 Hz (in order to have stronger power at low frequencies, since the previous tests have shown that measures at low frequencies were hard to extrapolate), the buzzer was then supplied by a sine signal whereas the datasheet's advice was to use a square one. That was done to prevent harmonics inherent to a square wave from appearing.

Analysis

We used the same algorithm to reduce the noise background on our graphs.
In previous tests, when measuring low frequencies inferior to 1000 Hz, the signal emitted by the buzzer was still too weak for the microphone to provide a significant signal at the expected frequency. The same phenomenon tends to appear at low temperature.
We observe directly on the microphone recordings that the amplitudes of the voltage due to the sound are quite the same at -40°C as at 20°C. One can notice a peak around 6500 Hz, another one at 11 kHz.
We obtained the following graph:

: PSD of the peaks agains its frequencies,
d = 4.5 cm

It is a bit less regular than before but it may be due to the several manipulations of the buzzer. The distance still does not affect the microphone sensitivity.

Conclusion

The microphone is well qualified above 1 kHz. The two components have quite well resisted to the long stay at low temperature but we will probably have to change the buzzer for future experiments. Moreover, we will try to find another component or system to test the low frequencies because it has not been done yet.

Test with a 5 mbar pressure

2015-02-09T15:32:24Z

Introduction

We made the first tests at IRAP, in a vacuum chamber. The goal of these tests was to qualify the microphone under a 5 mbar pressure (Martian surface pressure) but still with an Earth-like atmosphere composition.
The test consisted in recording 60 samples of 1 second at a sampling frequency of 100 kHz, for each frequency. The shift of frequency and the record of samples were automatically done by a Matlab script that used the LabView software. The Matlab script was supposed to make all the work and test 30 frequencies. However, a license problem forced us to do the tests manually… That's research work! Thus it modified the outputs: number of samples, frequencies…

What did we retrieve?

We tested 17 different frequencies in a range from 100 Hz to 15 kHz. We also did noise background in order to eliminate (or at least reduce) the noise in the records.
Let's note that we did two tests: the first one with a 4.3 cm gauge, the second one with a 27.5 cm gauge, in order to analyse the influence of distance on the signal amplitude.

Analysis

Deleting the noise background

The first part of the analysis was to delete the noise background of the vacuum chamber from the records for each frequency. To do so, we assumed that the background noise is not correlated with the sound produced by the magnetic buzzer. Therefore, the PSD of the measured signal is the addition of the PSD of the noise and the PSD of the desired signal. By having a lot of recordings for the noise background and for each frequency that we want to test, we were able to subtract the noise from the records.

Results

We obtained graphs such as the following one (the buzzer emits a 5 kHz sinusoid):

As the microphone is less sensitive at low frequencies, the deleting noise algorithm is more efficient at those frequencies. Higher, we still have some noise which is not deleted.
What the tests show is that under 1000 Hz, the power of the emitted sound by the buzzer is not strong enough to be clearly heard by the microphone, or the microphone has too low sensitivity, thus one cannot distinguish it from the noise background. This behaviour is similar to the one at 1 bar. Above 1 kHz however, the results can be taken into account.
We observe directly on the microphone recordings that the amplitudes of the voltage due to the sound is the same at 5mbar as at 1bar. Nonetheless, the buzzer was fed with a sine tension whereas its datasheet gives the emission power against frequency for a square tension. In addition, we do not know the emission power of the buzzer at 5mbar. Therefore, the normalized curves might not be correct.
Let's note that the distance does not drastically affect the signal amplitude at this scale.

: PSD of the peaks against its frequencies,
d = 4.3cm

Conclusion

Globally, the microphone seems to be well qualified above 1 kHz. Moreover, once back to normal pressure, it appears that both the microphone and the buzzer have successfully resisted to the stay at low pressure.