On the Application of Directional Frequency-Tuned Microphones for Studying Engine Noise

V. K. Nechaev

Experimental studies of the noise of high-speed diesel engines, including those carried out in the Internal-Combustion Engine Laboratory of the Tomsk Polytechnic Institute, show that the overall (total) noise produced by a diesel engine (and measured at some point near the engine) usually has a continuous spectrum, differing according to the operating conditions of the engine and the location at which the noise is measured.

However, upon careful examination of noise spectrograms one may notice that the noise spectrum always contains certain peaks, whose positions on the spectrograms remain unchanged under all speed and load conditions of the engine. In many cases these peaks stand out sharply above the general level of the continuous spectrum.

The stable position of these peaks along the frequency axis gives grounds to believe that they are the result of strong forced (resonant) vibrations of certain assemblies or parts of the engine.

Naturally, the question arises of the ways and methods for locating, in the engine, the sources of these strong frequency-fixed vibrations of the air medium surrounding the engine.

As the practice of experimental investigation of engine noise shows, the low directivity of ordinary microphones and the strong nonuniformity of the sound field around the engine do not make it possible reliably to localize individual sources of engine noise.

This greatly complicates the solution of all questions connected with the problem of reducing engine noise.

However, the directivity of a microphone (of the ordinary electrodynamic type) can be substantially increased by fitting the microphone with a cylindrical tube attachment, the inner surface of which is covered with some strongly sound-absorbing material (Fig. 1).

Then, if this tube attachment is sufficiently long, all “lateral” sound vibrations will die out before reaching the microphone itself, owing to their multiple reflection from the sound-absorbing inner surface of the tube attachment. In practice, the microphone will receive only signals directed along the axis of the tube.

But such a tube–microphone acoustic receiver system will have a very uneven frequency response, since the tube—

the attachment will produce a number of fairly sharp resonance peaks in this characteristic, corresponding to the frequencies of free oscillations of the column of air contained in the tube attachment.

In ordinary measurements and investigations of the overall noisiness of an engine, this strong nonuniformity of the frequency response of the receiving-measuring device—the tube–microphone—turns out to be its substantial drawback, precluding its practical use. The presence of strong oscillations of the air column in the tube will distort the frequency spectrum of the noise being measured, the overall noise level, and the levels of its individual components.

Fig. 1.

However, under our conditions, when solving problems of finding sources in an engine of frequency-fixed acoustic peaks, the indicated drawback becomes a substantial advantage of such a receiving device.

It is now sufficient to tune the tube attachment (of course, together with the microphone) to the frequency \(f\) of interest to us, in order to obtain a comparatively sharply directional receiver which, by virtue of its resonance characteristics, has increased sensitivity precisely to this frequency (if possible distortions associated with reflections of sound from the walls of the machine hall, etc., are neglected). The frequency response

Fig. 2.

of the microphone (Fig. 2) will now be transformed into a very nonuniform frequency response of the tube–microphone system (Fig. 3).

However, resonant oscillations in the tube attachment may be produced not only by sound oscillations of frequency \(f\) traveling along the axis of the tube, but also by sound oscillations of the same frequency approaching the mouth of the tube attachment at an acute angle to this axis.

A consequence of this will inevitably be a weakening of the directivity of our receiving device. To mitigate this drawback, a thick grille with numerous perforations and a profiled front surface should be placed at the entrance to the tube nozzle.

To make it possible to tune the tube nozzle precisely to the frequency \(f\) of interest to us, the tube design must allow its length to be varied within the required limits.

Fig. 3.

The length of the tube corresponding to a specified frequency \(f\) can be approximately calculated from the known analytical relations for the laws of free oscillations of an air column in a cylindrical tube closed at one end. However, taking into account the influence of the additional air volumes in front of the microphone diaphragm, this calculation must be refined with the aid of a signal generator, a loudspeaker, and some sufficiently sensitive indicator of the output level from the amplifying device after the microphone.

Now the tube–microphone measuring system is ready for carrying out measurements of the noisiness of the engine installation (Fig. 4).

By changing the angle \(\alpha\) of the direction of the microphone axis and simultaneously measuring the readings of the output indicator (calibrated in decibels), one can construct a graph of the type shown in Fig. 5, giving indications of the position in the engine of the principal source of sound oscillations with frequency \(f\).

Measurements of this kind (with the construction of the corresponding graphs) can be made by changing the angle \(\alpha\) in both the horizontal and vertical planes.

However, it should be borne in mind here that the tube–microphone system:

1) will, as a rule, have several resonance peaks (in accordance with the multiplicity of forms of free oscillations of the air column in the tube nozzle), that is, the tube–microphone system will receive, amplified, not only oscillations directed along the tube axis at the frequency \(f\) of interest to us, but also oscillations at multiple frequencies \(2f, 3f\), etc.;

2) will also receive all other sound oscillations directed along the tube axis that are produced by the engine; these oscillations may prove to be strong and, consequently, will be a source of masking of the oscillations of interest to us at frequency \(f\).

All this can substantially distort the picture we are investigating of the dependence of the level of sound vibrations, directed along the axis of the tube, with frequency \(f\), on the direction of the axis of the tube-microphone, if one uses for this purpose only an ordinary summing (averaging) sound-level meter, to which, as a rule, one always tries to impart as uniform a (horizontal) frequency response of sensitivity as possible.

Obviously, in our case it is more expedient to have a sound-level meter with a highly nonuniform, selective sensitivity characteristic,

Fig. 4.

Fig. 5.

with a maximum of sensitivity at the frequency \(f\) and with a sharp falloff on both sides of this frequency.

This can be achieved, for example, by connecting before the sound-level meter a frequency filter tuned to the same frequency \(f\). However, the creation of a sufficiently selective filter at audio frequencies with a variable tuning frequency is a difficult task.

Therefore, the following ways of ensuring high frequency selectivity of our tube–microphone system are more expedient:

Fig. 6.

1) the use, instead of a sound-level meter as the meter of the level of sound vibrations with frequency \(f\), of a ready-made heterodyne analyzer of the sound spectrum, tuning this analyzer each time precisely to the frequency \(f\) that interests us;

2) use, instead of the noise-meter analyzer, an analyzer with a “rocking frequency” (for example, a domestically produced instrument of the ASChKh type), determining the intensity (level) of sound oscillations of frequency \(f\) from the height of the resonance peak on the screen of the instrument (Fig. 5).

However, in the second case the picture on the screen of the ASChKh instrument will also be cluttered by harmonic components of the total engine noise that has passed through the tube-nozzle into the microphone along the axis of the tube (see the dashed line in Fig. 6).

to the horizontal plates

Fig. 7.

To points 1 and 2 of the bridge in Fig. 7

Fig. 8.

True, these “interferences” will prove to be comparatively strongly smoothed, since, unlike oscillations of frequency \(f\), \(2f\), etc., they will not be ampli-

occur in the tube-nozzle owing to resonance phenomena in the column of air contained in it.

But these interferences can be eliminated. For this it is necessary, by appropriate intervention in the circuit of the AFC instrument, to eliminate (switch off) the sweep of its tuning frequency and to set this frequency to a constant value corresponding to the tuning of the microphone tube to the frequency \(f\) of interest to us. Now the magnitude of the vertical deflection of the beam on the screen of the instrument will give a value proportional to the intensity of the sound oscillations of frequency \(f\). The sweep along the abscissa axis will be absent here.

Continuing the intervention in the circuit of the AFC instrument, one can use this horizontal sweep for visual (or photographic) recording of the level of the sound oscillations of frequency \(f\) as a function of the angle \(\alpha\) of the direction of the axis of the microphone tube (Fig. 7).

In the same way, on the screen of the AFC one can obtain a graph of the dependence of the sound level of frequency \(f\), for example, on the coordinate \(x\) along the axis of the engine (Fig. 8).