In all branches of industrial activity, in every form of transport and in all fields of engineering steps are being taken to eliminate unwanted noise, which is injurious to health and efficiency. Remarkable devices have been developed to measure the strength and intensity of noises
TRAFFIC SOUNDS IN BERLIN being recorded by police officials. The noise of passing vehicles is picked up by the microphone attachment on the kerb and its strength is recorded in phons (units of noise) on the meter on the pavement. This is the objective method, contrasting with the subjective method illustrated later in this chapter. The typical noise level of a loud motor horn is 80 phons, that near concrete drills 110 phons.
SCIENTIFIC investigation has substantiated what common sense suggests, that noise is harmful to health and impairs efficiency. In certain types of vehicles, such as aircraft, railway trains and motor cars, a lower noise factor is a good business asset, as the public will appreciate in terms of pounds, shillings and pence any practical contributions to quieter modes of transport. Further, Government anti-noise legislation has focused attention on the problem.
Directly it is proposed to limit noise it is required to know by how much the intensity of a sound exceeds a predetermined tolerable maximum. This apparently simple task is really complicated. The simplest acceptable definition of noise is “any unwanted sound”, although this does not take into consideration the varying degrees of annoyance or nervous strain caused by such sounds. These last factors are determined by the pitch and intensity of the sound, and they are also dependent to some extent upon the purity of the tones comprising the noise.
Any sound is conveyed to the ear by a series of air waves. The rapidity with which these waves rise and fall determines the pitch of the note. The note struck, say, on a piano is called the fundamental, and its frequency is expressed in cycles per second. For instance, middle C on the piano represents a fundamental frequency of about 256 cycles a second, which means that 256 complete vibrations travel through the air to our ears in one second. The same note played on a violin, or on any other instrument, would sound entirely different. The reason is that few tones are pure. What are known as harmonics are superimposed on the fundamental. A harmonic is a multiple of the fundamental. When the engineer speaks of a second, third or fourth harmonic, he means a wave having a frequency of twice, three or four times the frequency of the fundamental respectively. The musician uses the term in a rather different sense.
It might be supposed, if a harmonic is simply a much higher pitched edition of the original note, that two or three separate notes would be heard, corresponding to the number of harmonics present. This is not so. The harmonics add on to the fundamental and produce an entirely new and distorted wave form. A violin tone is full of harmonics, which provide its characteristic sound, but a flute yields a pure tone, completely free from harmonics. A knowledge of the harmonic content of noise is necessary to the engineer when he is investigating trouble of this nature in certain types of rotative machinery.
Among the peculiarities of the ear which make the assessment of noise difficult is the fact that it is not equally sensitive tor all notes. Roughly, the ordinary ear can hear tones anywhere within the range of 50 to 30,000 cycles a second. Some people can detect notes as low as 30 cycles, and a number have a higher limit of 40,000 cycles.
The ear is most sensitive between 1,000 and 3,000 cycles. At 1,000 cycles, the softest tone which may be heard is one which impinges on the drum with a pressure of 200 microdynes per square centimetre. The dyne is an extremely small force, being equivalent to 1/445048 of the weight of a pound. These figures give a striking picture of the delicacy and sensitivity of our hearing mechanism. A pressure of one pound on each square inch is small in everyday usage, yet the minimum pressure perceptible to the average ear is one three-hundred-and-seventy-millionth part of it, that is, 200 microdynes per square centimetre.
This figure is scientifically exact, having been derived from much research and practical experiment. It is also taken as the datum line in all noise measurements. Because a note of any pitch is sounded with an intensity corresponding to this datum pressure it will not necessarily be audible. It is correct only for a note of 1,000 cycles. At other frequencies the pressure must be much greater, which is equivalent to playing the note much louder, before the ear can detect it at all. A low organ note, say of 100 cycles, would have to impinge on the drum with an intensity of 3,500 microdynes before it could be heard. Thus, tones differing by a few octaves in pitch require to be sounded with an intensity varying in strength by twelve or fifteen times before they convey to the listener the same sensation of loudness. Similarly, in the range above 1,000 cycles, the pressure must again be increased for audibility. By plotting a number of points correlating pressure and minimum auditory sensation a curve is obtained known as the threshold of hearing.
The Ear as the Final Judge
If the phenomena already considered comprised the sum total of the ear’s idiosyncrasies, then the task of the acoustic engineer and designer of measuring equipment would be straightforward. There are, however, two further properties to be taken into account, both connected with the sensation of loudness. If two instruments of exactly similar type are played simultaneously, the ear does not receive the impression that the two together are twice as loud as one by itself. They will appear only slightly louder because the response of the ear is roughly logarithmic. A hundredfold increase in sound energy is necessary to produce double the loudness sensation.
The other remaining difficulty is due to the response of the ear changing as loudness is increased. A low note must be sounded with twelve times the intensity of a medium pitched note for them both just to be audible. When, however, they are sounded extremely loudly, they are both equally audible at the same intensity.
There are two methods of noise measurement, termed respectively subjective and objective. The subjective system is simple and overcomes many of the peculiarities of the ear by allowing the ear to be the final judge in the assessment of noise level. The apparatus used is readily portable and comprises a self-contained oscillator generating a constant tone of 1,000 or 800 cycles. This tone is applied to one, or a pair of headphones.
Consideration of the usual procedure adopted in practical application of the subjective noise meter will make the principle clear. Assuming that it is desired to make tests in the vicinity of a generator, the operator takes up a position about a yard from the machine, standing sideways to it. The noise meter, contained in a small box, is slung from his neck by a leather strap. Use may, in this particular instance, be made of only one earphone, which the operator applies to the ear farthest from the machine.
SUBJECTIVE NOISE METER in use near an electric motor. A valve oscillator emits a ton? with a frequency of 800 cycles a second. The operator stands facing the source of noise and quickly places the telephones over his ears, noticing whether the test tone is louder or quieter than the sound of the motor. By adjusting the test tone until it is only just less loud than the noise being measured, the operator can find the phon value by reading a meter.
He now adjusts the oscillator output through the medium of a calibrated dial until the loudness of the tone in the earphone appears to be just equal to that of the machine under test. His decision as to when the two are exactly equal is reached by first regulating the oscillator tone until the noise from the machine is just masked. Then the intensity of the tone is reduced until it is just inaudible above the machine noise. By ever finer corrections of this nature the final figure is reached. The whole process is then repeated, the other ear being used, and the mean of the two readings is taken as the ultimate noise level figure.
The dial of the noise meter is calibrated in decibels or phons. These are logarithmic units which roughly correspond with the minimum change in noise intensity which the ear is able to detect. The decibel was formerly the unit used in the United States, but the phon was adopted as international standard by the Paris Conference of 1937. The following figures give some typical noise levels in phons:
Near concrete drills — 110 phons
Loud motor horn — 80 phons
In express train — 70 phons
Busy office (with typists) — 60 phons
Quiet home — 30 phons
Still night in the country — 10 phons
If the measuring outfit is intended for use with two earphones, the operator stands facing the source of noise with the phones arranged so as to be slightly spaced from his ears. In this manner, the test tone and machine noise reach his ears simultaneously and the process of adjusting the tone until the two appear equal is carried out as for the single headphone method.
At the National Physical Laboratory, Teddington, successful estimates of noise level have been obtained with the aid of a tuning fork only. The tone emitted by a tuning fork dies away in accordance with a logarithmic law. This rate of decay, or decrement, can be calibrated for any given fork and the procedure in measurement is to note the time taken for the note to die down to a level when it appears equal in loudness with the noise under test.
Objective measuring apparatus attempts to record noise level direct by arranging for a microphone to pick up the sound which, after amplification and passage through special circuits, is applied to a meter reading in phons. The microphone is non-directional, so that it is equally sensitive to sounds coming from all directions and must have a good frequency response. Except for the necessity for providing a much higher standard of quality, the amplifier is rather similar to that in a radio set.
It is in the design of the weighting circuits, or filter networks, that the main difficulties are encountered. Somewhere in the chain between microphone and meter the engineer must introduce an artificial ear. The microphone will pick up the sounds faithfully, and normally the output of the amplifier will represent those sounds equally faithfully only in stepped-up form. They cannot, however, be applied to a meter as being exactly representative of the original sounds. Unfortunately, although such a reading would be a true record of the original noise, it would not be a true record of the loudness sensation which the ear would experience. Because the ear must be the final judge in all noise measurement, it is essential that the meter should record what the human ear would experience in similar circumstances.
Threshold of Hearing
The object of a weighting circuit is to imitate the response of the ear. It is comparatively simple to design a circuit which will have a stated response curve throughout the audible range. Thus an amplifier can be built with a response exactly corresponding to the threshold of hearing curve. But the ear response changes with every variation in the intensity of the sound, so that the amplifier also must have a variable response for all intensities, which is technically impossible. A compromise is therefore adopted whereby various filters may be switched in circuit so that the response curve is right for four or five intensity levels. In practice this arrangement meets most of the requirements of those concerned with noise measurement.
NOISE ANALYSIS in a United States transcontinental air liner. Obnoxious noises in aircraft are responsible for much discomfort. Experiments with an acoustic analyser in an aeroplane in flight indicate how noises originate and how they may be eliminated.
After the quantity of a noise has been discovered it is generally necessary to know its quality or the constituent tones which are jumbled together to produce the discordant whole. High notes are more upsetting to the ear than low, and the engineer therefore aims first at a reduction of the total noise level, or, failing this, elimination of the higher frequencies. Before he can decide what component part of a machine is probably responsible tor the generation of any particular frequency appearing in the noise, he must have a means of splitting the total noise up into its individual parts. This requirement is met by a process of analysis.
Some types of equipment for analysing noise operate on the heterodyne principle. Probably most people have noticed, when listening to an organ being played, that when two notes are played simultaneously a rhythmic “beat” is sometimes audible. This is the heterodyne effect. If a note of A cycles a second is mixed with one of B cycles a second, two new notes are obtained, equivalent in frequency to the sum (A+B) and the difference (A—B) of the two tones respectively. Most modern radio receivers make use of the effect, which is extremely useful, because a frequency which is too high to be audible can be heterodyned with another high-frequency wave, so that a comfortably audible low or medium-pitched note is produced.
The reason for adopting the principle in a noise analyser, and in a radio circuit, is that a single resonant circuit can be used for all frequencies. By suitable variation of the heterodyne frequency any wave can be made to provide a sum or difference frequency corresponding to that required for resonance. In one type of analyser operating on the heterodyne principle the resonator takes the form of a steel bar whose length is such that it has a natural period ol vibration of either 11,000 or 16,000 cycles. With the latter frequency use is made of the sum of the heterodyne and noise tones, and the 11,000-cycles resonator is used when it is intended to pick out the difference frequencies. When any frequency other than its natural frequency is applied to the bar no appreciable effect can be noted, but when the applied wave approximates to the resonant frequency the bar vibrates in sympathy. In addition to the bar’s ability to detect a given frequency in this manner, an indication is also obtained of the magnitude of the wave — that is, its loudness, because the extent of the vibration varies accordingly.
The Heterodyne Method
To analyse noise by the heterodyne method, the operator arranges for the total noise to be picked up by a microphone as before. It is amplified and then the heterodyne oscillation is introduced. Suppose, for example, a high-pitched tone were prominent in the noise and it was desired to discover the exact frequency of this particular component. The operator may estimate the tone mentally as being roughly 10,000 cycles; so he rotates the heterodyne oscillator control until the dial indicates that it is tuned to 21,000 cycles, which, if his estimate is correct, will provide a difference tone of 11,000 cycles. The strength of this last frequency will be recorded on suitable meters, thus affording the required information.
By rotating the dial on the heterodyne oscillator control the whole frequency range of the noise may rapidly be explored in a similar manner to the tuning of a radio set. In another analyser two tuned circuits take the place of the heterodyne oscillator, these having a frequency discriminating action rather similar to the once common band-pass arrangement in radio receiver design. These two circuits may be tuned simultaneously to any frequency between 20 and 25,000 cycles, and they will always be most responsive to frequencies corresponding to that to which they are tuned.
Once he has discovered the principal components of the noise the engineer can, by calculation and practical experiment, decide which item, or items, in the machine are the major offenders. Cooling fans used in motors and generators are frequently a source of trouble.
Applications of the noise meter are widely extending, tests having been conducted in nearly all branches of industrial activity. The results so obtained have proved of undoubted benefit to the designers of machinery, internal combustion engines and the like. This, in turn, is directly beneficial to the general public and a quieter existence is brought nearer to realization.
SOUND ANALYSIS MEASURING APPARATUS. On the left is the microphone, connected to the amplifier by a flexible lead. On the right is the frequency analyser, covering a range of from 24 to 6,000 cycles a second. Accurate tuning is indicated by the galvanometer on top of the analyser. For measuring extremely feeble sounds, an additional amplifier and a more sensitive galvanometer may be used.
