task4_barrier

Barrier Insertion Loss

For these measurements, a loudspeaker was placed at one end of the table and the receiver (microphone) at the opposite end. First the direct sound was measured without the barrier, then a long barrier was placed between the source and the receiver.

The barrier must be long enough to be sure that "all the incoming sound" will be altered by the barrier effect. But since the barrier is long, the window, used to compile the results in Matlab, must be longer than the one used before as well.

Figure 1: Barrier insertion loss measurement set-up

First of all, when we measure the impulse response of the loudspeaker without the barrier and we take a 8 ms snippet, we get: without.txt

Figure 2: Impulse response without barrier

Using the same snippet length, but this time for the measurement with the barrier, we get: with.txt

Figure 3: Impulse response with barrier

As a general result we can see that the impulse response of the case with the barrier has a much lower level than for the case without the barrier (about 150 compared to 1000 Pa). In Figure 3 we recognize the necessity of using a snippet which is long enough in time as the response flattens out completely only after around 10 ms. The reason why the signal flattens out more slowly could be the scattering which takes place at the edges of the barrier when the sound passes it.

Data processing

In the Matlab code barrier.m we compensated the measured sound pressure of the direct sound because of the longer sound path over the edge, we applied an FFT and calculated the corresponding sound pressure levels with and without the barrier.

Figure 4: Frequency response with and without the barrier

The frequency response level of the case without barrier is clearly higher in most of the frequencies. That means that the barrier successfully deviates the sound propagation upwards so that a less important amount of sound energy hits the microphone. We can see this result directly on the figure which shows the difference in level of both cases:

Figure 5: Without-with spectrum

There is no need to investigate the effect of the table itself (not the edges), since in the difference Without-With barrier they would be canceled.

From around 300 Hz upwards the insertion loss is positive. That means that from this frequency on the barrier reduces the sound arriving at the mirophone. We can recognize several peaks which can go up to 30 dB. That suggests that at certain frequencies the barrier reduces some frequency components of the incoming sound wave of 30 dB which is quite efficient. As a mean value the barrier has an insertion loss between 5 and 10 dB which is not huge, but interesting considering the small height of the barrier.

For further studies we plot the frequency response also in third octave bands

Figure 6: Frequency response with and without barrier in 1/3 octave bands in free field

What's important for us is to get the insertion loss of the barrier in third octave bands so that we can compare it to the insertion loss curve which can be found in the ATA book (figure 7.17):

Figure 7: Insertion loss of a barrier in free field as a function of the parameter X (ATA book fig. 7.17)

For the direct comparison to the figure in the ATA book, we use the frequency response in third octave bands. Let's compare all the results in one figure:

Figure 8: Insertion loss of a barrier in free field

The calculated and the measured insertion loss have the same mean level over the whole frequency range of interest. That's a pretty good result. It would be difficult to get a better match even if we changed its material properties or the position of the barrier (which would increase the insertion loss).

Return to Task 4 main page