Post-processing the measurement

A typical result is shown in the figure below where we have used a typical low quality loudspeaker and microphone, thus showing how low quality equipment can influence the measurement (this can be corrected).

 

In the plot in the figure below there are a few changes compared to the settings in the setup:

      The plots have switched order, showing time data on top.

      The time window used in the figure is rectangular and 50 ms. The time window in the setup is half-cosine and 200 ms. The limits can be changed by clicking and dragging, or making adjustments in the Time Window Toolbar shown in the middle of the figure above. Decreasing the window size will reduce the noise in the frequency response, but it will also give a decreased frequency resolution and thus false results for very low frequencies.

      The Frequency Resolution Line shown to the left in the figure below depends on the window size and is supposed to warn the user where the results may not be trusted. For frequencies at the thick part of the line, the results should not at all be trusted. For frequencies at the thin part of the line (which is twice the length of the thick line), the result is not certain. To display the frequency resolution line, go to Plot->Plot Type Settings->General Frequency Domain... and make sure the check box  is checked.

      We have used a MLS noise signal instead of the sine sweep that is default. This MLS signal causes the spurious peaks that are shown at the end of the impulse response. When using the sine sweep signal, these spurious peaks should not be shown.

 

Figure 4 Multimedia microphone and loudspeaker measured on axis at a distance of ~20 cm. Time window 50 ms.

Note that if the frequency resolution is not sufficient, two resonance peaks may not be resolved, but shown as one peak. I am not sure if this statement refers to the above figure or is a general statement describing one problem resulting from the poor resolution These resonance peaks can be resolved if the time window size is large enough. But this is a tradeoff since the frequency response would then include reflections from boundaries near the speaker/microphone.