MPA Logo, San Francisco Maritime National Park Association, USS Pampanito, Historic Ships at Hyde Street Pier, Education Programs Maritime Park Association Home Page Maritime Park Association Home Page Events Maritime Park Association Home Page Maritime Park Association Home Page Maritime Park Association Home Page Volunteer Membership Donate Maritime Park Association Home Page USS Pampanito Submarine Historic Ships at Hyde Street Pier Education Programs About Maritime Park Association Home Page Directions to Maritime Jobs at Maritime Facility Rental at Maritime Trustees of the Association Calendar Press Room Store Maritime Map
CHAPTER 3

SOUND RECEPTION AND DETECTION BY LISTENING

Introduction
 

Sound waves can be received only if a device that will absorb a fraction of the incident energy and convert it into a detectable form is placed in their path. Such a device is called a receiver. The proper type of receiver for a particular application depends upon (1) the frequency, amplitude, and form of the sound wave; (2) the type of transmitting medium; and (3) the ultimate object for which the sound energy is required.

A resonant receiver is designed to operate with maximum efficiency at some particular frequency. A nonresonant receiver is designed for use when a reasonably uniform response is desired over a given range of frequencies. If the primary

  concern is faithfulness in the reproduction of waveform, a nonresonant receiver is required. However, if it is necessary to receive sound waves of a particular frequency to the exclusion of other frequencies that may be present in the medium, a resonant receiver is required.

Most sound receivers function to transform the mechanical energy that they absorb directly or indirectly into electric energy. The electric energy representing the sound signals may be portrayed visually, or the sound signals themselves may be reproduced as sound energy by a loudspeaker.

 
Human Ear
 
Sonar equipment that presents sound signals by means of a loudspeaker is useless unless there is an operator to hear and interpret the sound waves radiated to the surrounding air. The capabilities and limitations of the operator, whose task it is to interpret the sounds issuing from the listening gear, are important in determining the success or failure of its mission. For this reason, the following discussion on the physics and psychology of hearing is included, even though it is not strictly a part of the theory of underwater sound. Note that this discussion deals primarily with airborne sound.

Confusion sometimes arises between the objective physical phenomenon of sound and its subjective perception by a listener. The reader is doubtless familiar with a philosophical problem that agitated the ancients, which was formulated somewhat as follows: A tree crashes in a forest, and no living being is present to perceive the fact. Is there any sound?

  Most of the lengthy arguments that were expounded on this question could have been avoided had there been adequate theories of sound and hearing. Today sound means waves, which travel in the air, water, or other medium. Thus the answer to the crashing-tree question is yes. Sound is to be distinguished from the sensation of hearing, or auditory sensation, which is a phenomenon occurring in a human being or animal. There was no auditory sensation in the crashing-tree example. To clarify the distinction between a sound and the sensation produced by a sound, the sound is often called the stimulus. Ultrasonic waves are sound, but they do not stimulate the sensation of hearing in human beings; they are thus not a stimulus of auditory sensation.

THEORY OF HEARING

In this study it is not essential that a physiological study of the ear be made. Of particular interest here is that part of the inner ear called the

 
52

cochlea which has a major part in the hearing process. It is a spiral tube, divided into galleries by a longitudinal membrane-the basilar membrane, which is a sort of carpet of nerve endings. The nerve endings of the basilar membrane are transverse fibers that vary systematically in length. The short fibers respond to sound waves of high frequencies; the long fibers respond to sound waves of low frequencies. That is, the position of the point of maximum stimulation depends on the frequency of the tone.

In response to a complex sound, the basilar membrane vibrates with a certain pattern, perhaps having several maxima, depending on the frequency components in the stimulus. The auditory nerve endings are distributed along the basilar membrane in such a way that they can transmit this pattern to the brain, which interprets it in terms of the pitch, loudness, and quality of the sound. The location of the vibration pattern on the basilar membrane determines the pitch sensation. Loudness is associated with the magnitude of the vibration.

The relation between the perceived loudness of a sound and the magnitude of the stimulus on the basilar membrane is explained as follows: The auditory nerve contains about 3,000 nerve fibers which, analogous to a telephone cable, connect the cochlea to the brain. Each nerve fiber responds according to the "all-or-none" law; that is, when it is stimulated sufficiently to respond at all, it responds at full strength. The response of a nerve fiber is analogous to the discharge of a condenser. The strength of the discharge is independent of the intensity of the sound, but the number of discharges per second does depend on the magnitude of the stimulus in the following manner.

The discharge of a given nerve fiber is followed by a "refractory period" during which the nerve cannot react. This period is about 0.001 second; thus no single nerve fiber can respond at a rate greater than about 1,000 times per second. The refractory period is followed by a "relative refractory period" of about 0.003 second during which the nerve gradually recovers its sensitivity. Thus a very weak tone of, say, 1,000 cycles per second may cause a given nerve fiber to discharge no more rapidly than about 300 times per second, whereas with an intense tone of that frequency the nerve may respond up to 900 times per second.

  The number of responses of a given nerve fiber depends on the strength of the stimulus; moreover the number of nerve fibers excited increases with the intensity of the stimulus because (1) a greater area of the basilar membrane is activated and thus the stimulus pattern on the membrane takes in nerve endings over a wider area, and (2) the high intensity excites nerve fibers having higher normal thresholds of stimulation. It seems reasonable, therefore, to correlate the sensation of loudness with the total number of nerve impulses arriving at the brain.

NUMERICAL DATA CONCERNING THE EAR

The preceding theory of hearing suggests how the structure of the ear enables it to respond to frequency and intensity characteristics of a sound. Although it is a theory that has not been verified in all details and is subject to revision, it should help in understanding some of the pages which follow. However, the following facts are independent of the correctness of this theory.

Frequencies of from 20 to 20,000 cycles per second can be heard by a normal, young ear. A change in frequency of less than one-half of 1 percent results in a perceptible change in the pitch of pure tone. This phenomenon takes place at 10,000 cycles per second only if the listening level is comfortably loud. As the duration of the tone signal becomes shorter, the ability to hear pitch changes decreases. This relation is shown in figure 3-1, where the least-perceptible frequency change plotted against the signal duration. It is interesting to note that at 1,024 cycles per second

Threshold of frequency discrimination for several frequencies as a function of signal duration.
Figure 3-1. -Threshold of frequency discrimination for several frequencies as a function of signal duration.

 
53

the length of the signal affects pitch discrimination only if the signal length is less than 0.1 second. This fact is important in doppler discrimination in echo ranging.

The ear is most sensitive at frequencies between 1,000 and 5,000 cycles per second, where a sound intensity of approximately 10-16 watt/cm2 can be heard. A sound intensity of approximately 10-4 watt/cm2 produces a sensation of pain rather than of hearing. Thus the ear has a dynamic range of about 120 db at frequencies around 1,000 cycles per second.

A rapid change of 1 db, or slightly less, in the level of a pure tone can ordinarily be perceived at all frequencies between 50 and 10,000 cycles per second if the listening level is comfortably loud.

The ability to detect changes in level is less for randomly fluctuating sounds, such as noise, than for pure tones. However, a simple rhythmic variation is very easily perceived, particularly if it is cyclic at the rate of about 3 per second.

The ear requires approximately 0.2 second for the sensation of loudness to catch up with a sudden increase or decrease of sound level. These dynamic properties seem to be determined by neural rather than mechanical processes. They influence the response of the ear to tones of short duration such as those used in echo ranging.

Sounds having the same pitch and loudness may produce different sensations if their spectra are different. The general term "quality" is used to describe the difference in the complex sensations they stimulate. These differences may be sufficient to influence the masking of one sound by another. Because masking is a primary factor in preventing the detection of signals, its general principles will be discussed in greater detail than has been accorded to the other aspects of hearing.

THRESHOLD OF HEARING

The threshold of hearing may be illustrated by the following experiment. A microphone is placed near a sound source which produces a pure tone of controllable intensity. Apart from this sound the experimental location is to be very quiet. The microphone converts the mechanical energy of the sound into electric energy which can be used to operate some device, such as an oscilloscope.

Beginning with a sound intensity of moderate value, the intensity of the tone is gradually reduced.

  The oscilloscope fails to operate properly before the sound intensity has reached zero. This minimum intensity to which the oscilloscope responds depends on two factors. One is the amount of energy dissipated in the various parts of the microphone; the other is the self-noise of the oscilloscope, the microphone, and the circuit. The oscilloscope will not operate properly unless the signal is at least as intense as the self-noise. The minimum sound level that will cause the device to operate properly is its threshold.

Suppose that the receiver is now replaced by a human ear, and the same procedure is followed. A precisely analogous situation results, and for much the same reasons. The ear receives the sound energy incident on it, is stimulated mechanically, and the mechanical energy then is converted into some form of nerve energy which activates the brain. Some of the incident energy is dissipated in this process. Corresponding to the self-noise of the receiver, there are sounds generated by breathing and by the circulation of the blood. Thus there is a minimum level which must be exceeded by a sound before it can be heard. This threshold of hearing corresponds to the threshold of the microphone-oscilloscope system.

The value of the threshold of hearing differs among people. We say that their acuity is different. The average value of the threshold of hearing also depends on the frequency. At 64 cycles per second the pressure of the threshold of hearing is 0.12 dyne/cm2; it decreases more or less uniformly with increasing frequency up to about 3,000 cycles per second, at which frequency the pressure is 0.000041 dyne/cm2. This value corresponds to the lowest limit of sensitivity mentioned earlier. Above 5,000 cycles per second it increases with frequency until at 18,000 cycles per second it is 4.1 dynes/cm2.

MASKING

Under all ordinary circumstances, we hear many sounds at once but are usually able to concentrate on the wanted sounds and ignore the unwanted background. This background is always present. Even in a very quiet place the self-noise produced by the normal internal processes of the human body becomes audible. Thus there is complete analogy between the ear and an electronic receiver of sound. This analogy is

 
54

close enough to permit the frequent use of the word "receiver" with reference to the ear as well as to electronic devices.

Although unwanted sounds can be ignored to a considerable extent, their presence does interfere with the ear's ability to detect another sound. This interference is called masking. Masking is the increase of threshold level caused by the unwanted sound.

The level at which a particular sound becomes audible differs from the threshold of hearing by an amount depending on the extent to which the background noise masks the signal. This level is the masked threshold; it is the level of the signal when it is audible above a particular background noise 50 percent of the time. The masked threshold therefore applies to the signal-noise pair, not to the signal alone, although it is measured by the level of the signal alone. The value of the masked threshold is, however, determined by the level of the noise. Raising the level of the noise raises the masked threshold of the signal.

The variable acuity of a listener introduces the need for the phrase "50 percent of the time." Not only does the threshold of a signal under identical conditions vary from individual to individual, but the same individual sometimes hears a signal and sometimes not, even though the levels of signal and masking noise are the same on the various occasions.

This problem may be clarified by describing a typical experiment designed to measure the masked threshold. Arrangements are made so that a number of listeners will hear the background noise at a constant and known level. Other arrangements are made for producing a series of signals at various levels. Care is taken so that the listeners cannot determine when or at what level a signal is produced except by hearing it; they receive no cues from the person administering the test nor from each other. The administrator records the level of each signal and, after a suitable interval, instructs each listener to vote yes or no as to whether he heard the signal.

Each level of the signal is presented 5 times to 10 listeners, so that the total number of votes for each level is 50. The recognition probability is the percentage of yes votes for a given level. This probability is plotted as a function of level in figure 3-2.

  Probability of recognition of a pure tone in a
background of a noise at a constant level of 12 db.
Figure 3-2 -Probability of recognition of a pure tone in a background of a noise at a constant level of 12 db.

Note that, there is no abrupt transition from inaudibility to audibility. Instead, the probability of hearing the signal increases gradually from zero to 100 percent over a 5-db range of levels. This complication was not considered in discussing threshold levels in the preceding pages. Fundamentally, there is no one level at which the signal is "just audible." To avoid confusion, threshold levels are usually defined as the level at which the recognition probability is 50 percent; but, when necessary, other percentages may be used, provided they are specifically indicated. Figure 3-2 shows that the 50-percent masked threshold is 14.5 db, the 90-percent threshold is 16.4 db, and the 10-percent threshold is 12.6 db.

This difference between the threshold level of the signal and the level of the background is called the recognition differential. In the example the recognition differential for 50-percent recognition is thus 2.5 db (14.5-12.0).

PSYCHOLOGICAL CHARACTERISTICS OF SOUND

How does the ear distinguish between a specific sound and all the other sounds that form a background for it? Everyday experience suggests the answer. A boatswain shouting orders must rely chiefly on his ability to produce sounds of an intensity great enough to override the clamor of winches and other noises. A shrill whistle produces a sound that is audible, even though the intensity of the background is incomparably greater than that of the whistle. In this case the perception is due partly to the pitch difference between the signal and the background noise, and partly to a decided difference in the quality of the two sounds. A rhythmic drumbeat is audible over many noises. Before the days of telephone and radio the common method of transmitting

 
55

orders to masses of troops was to use drumbeats of various rhythmic patterns. Bugle calls with very decided rhythm utilized the advantages of all the factors mentioned.

To sum up, the sensations produced by sound have at least four distinctive characteristics: (1) Loudness, (2) pitch, (3) quality, and (4) time pattern. In the recognition of a particular sound, all four of these characteristics probably contribute to differentiate it from others heard simultaneously. In experiments, however, the effect of each characteristic can be isolated.

Loudness, pitch, and quality are psychological, rather than purely physical, terms. That is, they directly characterize the sensation and only indirectly the sound. It is customary to say loosely that loudness is determined by the level of a sound, pitch by the dominant frequency, and quality by the spectrum. This explanation is over-simplified. A more careful examination discloses that in determining any one of the three, all the physical characteristics of the sound play a part. Loudness, it is true, is determined primarily by the level of the sound, but it is influenced also by the frequency and spectrum. It has been demonstrated experimentally that a moderately high frequency is perceived as being louder than a low frequency of the same intensity. This fact is almost implicit in the discussion of the threshold of hearing given above. If the frequency exceeds about 14 kc the reverse is true, and ultrasonic sound of any level is inaudible. Pitch, in its turn, is determined largely by the dominant frequency of the sound waves but is influenced also by the level and the other characteristics of the spectrum. Quality is principally a matter of spectral distribution; and the time pattern may consist of systematic changes in any of the other three psychological characteristics.

One point is worthy of particular emphasis. Ignoring the fact that intensity is not the only

  factor that determines loudness, we may inquire as to the mathematical relation between intensity and loudness. It appears that this relation is not a simple proportionality-that is, when one sound is said by most people to be "twice as loud" as another, the intensity of the one is not twice, but approximately 100 times, the intensity of the other. In general, loudness is more nearly proportional to the level of the sound in decibels. A barely perceptible increase of loudness usually accompanies a sudden increase of 1 db in sound level, whether the original level was 5 or 50 db.

Another characteristic that can be used to differentiate sounds is their direction of arrival. In simple cases, this direction coincides with the direction of the source from the listener. The binaural effect is the ability of a human with two ears to determine the direction of a sound source. This sense of sound direction depends primarily on the difference in phase (or time) of the waves reaching the two ears, although it depends partly on the difference in intensity of the sound received in the two ears. The binaural effect is similar in principle to the split transducer used with bearing deviation indicators (BDI).

In the early days of sonar, attempts were made to use the binaural effect to determine the direction of underwater sound. These listening devices used two receivers placed along a baseline varying from several feet to several hundred feet. This procedure virtually increased the baseline between the two ears.

An early device of this type, designed for underwater listening, consisted of two hollow rubber spheres mounted on the ends of a pipe about 4 feet long. Projected through the hull of the ship, the receivers were separately connected over lines of equal length to the two ears. The tube might then be turned until the sound appeared centered in the head; at which time it should be on a line perpendicular to the baseline of the receivers.

 
Doppler Effect
 
DEFINITION

The Doppler principle applicable to all wave motion was developed by the Austrian physicist, Christian Doppler (1803-1853). This principle shows that when there is a relative motion between the source of a wave motion and a receiver the

  apparent frequency at the receiver differs from the frequency at the source. The Doppler principle has important operational applications in sonar.

If an observer is moving toward a source of sound, he hears a tone the pitch of which is higher than when he is at rest. If the observer is moving

 
56

away from the source of sound, he hears a tone the pitch of which is lower than when he is at rest.

Thus the frequency of the sound appears to increase when a observer moves toward a source and appears to decrease when he moves away from it. Similarly, if the source is moving toward the observer, the frequency is higher; if the source is moving away from the observer, it is lower.

The apparent frequency of the sound is found as follows: When the observer is at rest, the number of waves he receives each second is Fo, the true frequency of the sound. When the observer moves toward the source, he receives more sound waves in each second than when he is at rest. If his mean range rate is dR (in feet per second), the additional number of waves received per second are those that occupy the distance by which the range is changed in 1 second. Because the distance between successive waves is the wavelength λ, this number is dR / λ.

If the relation for the velocity v of the sound,

v= Foλ,  (3-1)

is used, the number of additional waves received is FodR / v. The apparent frequency, F, is the total number of waves received each second and is therefore given by

F=Fo(1 + dR/v).  (3-2)

When the observer is in motion away from the source, the plus is replaced by a minus-

F=Fo(1 - dR/v). (3-3)

If the source is receiving echoes from a target, the Doppler effect occurs twice, so that the frequency of the echo FE, received at the source is

FE=Fo(1 ± 2dR/v).  (3-4)

Equation (3-4) gives the apparent frequency of the echo when the range rate is dR; the positive sign is used if the receiver and the source are moving toward each other, the negative if they are moving away from each other.

The equations apply to the ultrasonic frequency of the sound in the water. To make this sound audible, the received waves are heterodyned in the receiver. This heterodyne receiver reduces the frequency by a constant amount. Note that

  this reduction is subtractive and not proportional-that is, the receiver subtracts a constant amount, FH, from the received frequency, FE, so that the audio frequency of the output is

fE=FE-FH  (3-5)

If this equation is applied to equation (3-4) the audio frequency of the echo is

fE=Fo-FH±(2FodR)/v,  (3-6)

or

fE=fo±(2FodR)/v,   (3-7)

Here fo=Fo-FH is the audio frequency of the echo for a zero range rate. The difference fE-fo; that is, the quantity ±2FodR/v; is called the absolute doppler shift. It is proportional to Fo, and independent of FH and fo. This fact is very important because the transmitted frequency, Fo, is much greater than the heterodyned audio frequency fo. Because the Doppler effect is to shift the frequency by 0.7 cycle per kilocycle per knot of range rate, if dR is expressed in knots and Fo, in kilocycles, the doppler shift is

fE-fo=0.7FodR cps (approx).  (3-8)

If Fo, is 24 kilocycles,

fE-fo=17dR cps (approx).  (3-9)

This shift can be very appreciable. If the sonar ship and the target are on opposite courses, and one is moving at 25 knots and the other at 20, the shift is 45 X 17=765 cycles per second, and a band pass of twice this quantity, or 1,530 cycles per second is required. Because fo, is commonly 800 cycles per second, this frequency shift is important in determining the width of the band pass of the sonar receiver. Circuits may be used to eliminate this shift when it exceeds the band pass of the receiver. One such circuit is called own doppler nullifier and the other, target doppler nullifier. These circuits will be discussed later.

APPLICATION TO ECHO RANGING

In echo ranging the operator does not hear the outgoing ping, because the equipment is on send and the receiver is blocked. Therefore, he cannot compare the frequency of the returning echo with that of the outgoing ping. However, he can compare the frequency of the echo with that of the

 
57

reverberation heard immediately after the ping is emitted. This comparison has an important effect. The difference between the reverberation and echo frequency depends only on the target's absolute motion through the water and its direction relative to the sound beam. It is independent of own ship's motion.

For example, suppose a ship is moving with its sound beam directed dead ahead and with a velocity, V1, which is also the range rate, dR, if the echo is from stationary objects (scatterers). Just as with an echo from a moving target, the relative motion between the source and the scatterers causes the reverberation frequency to increase. From equation (3-7), the reverberation frequency after heterodyning is

fR=Fo+(2FoV1)/v (3-10)

If a submarine is approaching the echo-ranging ship with a speed V2, the relative speed or range rate is

V=dR=V1+V2  (3-11)

and from equation (3-7), the audio frequency of the echo is

fE=fo + (2FoV1)/v + (2FoV2)/v cps.  (3-12)

A comparison of equations (3-10) and (3-12) shows that the audio frequency of the echo exceeds that of the reverberation by an expression that does not contain V1, the speed of the sonar vessel.

If Fo, is 24 kilocycles,

Δf=17V2 cps;  (3-14)

  thus, for an approaching 20-knot submarine, the frequency of the echo is 340 cps above the reverberation frequency. The quantity Δf is known as the target doppler. Because operationally it is much more important than the absolute doppler shift, it is frequently called simply Doppler. It is "up-doppler" if the submarine is moving toward the echo-ranging ship and "down-doppler" if it is moving away from the echo-ranging ship. Another useful characteristic of target doppler is that it is proportional to the speed of the target. Hence it can give information concerning the motion of the target. A trained operator can estimate also the probable aspect of the target with considerable accuracy from the change in target doppler.

In the foregoing example, it s assumed that the course of the target is directly toward (or away from) the echo-ranging gear. t may be shown that, in general, V2 is not the actual speed of the target, but is its range rate relative to a stationary point, P. This point, P, momentarily coincides with the sonar projector but must be considered stationary even though the sonar is moving.

The importance of target doppler in echo ranging is immediately evident. It is a common experience that a difference in pitch between two tones is a great aid in hearing them; and even a very weak tone can often be distinguished from others if its pitch differs markedly. Thus target doppler is a great aid in detecting echoes against a reverberation background but not against noise. The ability of the operator to estimate the difference in frequency between reverberation and echo depends on the ping length.

Many "false" echoes are received from floating debris, kelp, and unknown causes. These echoes do not show the effect of target doppler. Thus a final important application of target doppler is in the identification problem.

 
Ear in Underwater Detection
 
LISTENING

Detection of underwater objects by listening for the sounds they emit is known as listening. Sounds made in the sea are easily detected by the use of listening equipment. Listening, the oldest method of detection, was used in World War I in a very crude, but nevertheless effective, form. The success of detection by listening is primarily

  dependent on the ability of the operator to hear and properly evaluate these sounds delivered by the listening equipment.

Any listening system must consist of (1) a hydrophone, (2) an electronic receiver, (3) a bearing indicator, and (4) a speaker or headphones. The sound-listening problem for the operator consists primarily of learning to distinguish between (1)

 
58

sounds emitted by another ship's machinery through the hull and from the propeller and (2) the multitude of other sounds that exist in the ocean.

There is always the problem of background noise, which may make the sounds to be detected unrecognizable. As pointed out earlier, the characteristics of the ear enter into this problem. During World War II many persons were found to have hearing that was defective for sonar work.

Echo ranging and the listening problem differ materially in several ways.

In echo ranging, the searching vessel projects a sound signal into the water intentionally with the expectation that the sound will strike a target and that enough of the energy will be returned by the target to the transducer to activate the receiver so that the operator can recognize the echo. The primary source of the sound is in the searching vessel; the target is only a secondary source. The transmission of the sound is a two-way process. In listening, on the other hand, the sound signal is emitted by the target itself, which therefore is the primary source. Listening is hence a one-way process.

This fact suggests that losses by transmission should be smaller in the case of listening, and that detection should be possible at longer ranges by listening than by echo ranging, provided that the sound output of the target is comparable to that of the standard echo-ranging transducer. however, the noise output of most targets is less than the output of a standard transducer. Even the noisiest type of ship, a large battleship moving at high speed, has an over-all output of sound of about the same level as a standard transducer. Furthermore the sound from a transducer is a pure tone, because the echo has frequencies that are restricted to a band of about 200 cycles. On the other hand, the sound from a battleship has components of a wide range of frequencies, and hence is more easily masked by the background noise.

Nevertheless, conditions are frequently such that ships are detected by listening at ranges of 10,000 yards and more, whereas echo ranging is rarely effective beyond 3,000 yards. Echo ranging enables the range and bearing of the target to be determined accurately; listening gives the bearing quite accurately, but provides little or no information on the range except in specialized equipment.

  Listening is used chiefly by submarines. A surface vessel produces considerable noise, and this noise interferes with the detection of the sounds of other ships-especially the low sounds of submarines. On the other hand, this difference in the noise output enables a submarine to detect the presence of a surface vessel rather easily. An anti-submarine vessel, moreover, will generally not use evasive tactics. Therefore it will not hesitate to emit a powerful signal into the water, and thus gain the advantages of echo ranging; whereas a submarine will hesitate to reveal its presence by echo ranging except in the last stages of an attack.

In order for listening to be a tactical aid, the sound operator by use of his ear must be able:

1. To distinguish the sound emitted by the target from the usual background noise.
2. To distinguish between the various kinds of ship sounds with a view to possible identification of the type of vessel emitting them and to obtaining information on the ship's operating conditions.
3. After detecting and perhaps partially identifying a target, to obtain information concerning its approximate location and motion while it is still at comparatively long range.

These considerations suggest the value and purpose of the investigation of ship and submarine sounds. Such information will aid in the problem of the control or possible elimination of revealing noises. The basic principle in this problem is the same as that underlying visual camouflage-to render the target inconspicuous by making it resemble its background. Thus the sounds that are unintentionally and unavoidably emitted should, in the ideal case, have spectra that are very similar to that of the background noise.

Another application is in the design and operation of acoustic mines and in the prediction of their actuating ranges. This application, as well as the defense against mines of this type, requires a knowledge of the sound emitted by the vessels against which they are to be used.

BACKGROUND NOISE

There are two principal sources of background noise-airborne noise and amplified noise. When using listening equipment, the operator depends almost entirely on his ears, unaided by any form

 
59

of recorder or other apparatus. Occasionally a decibel meter or "magic eye" is available for supplementary quantitative information. His task is reduced to detecting and recognizing a wanted signal against the background of all the other sounds that impinge on his ear. These sounds are many and complex.

In the discrimination process, the operator distinguishes between wanted sounds of the signal from the target, and the unwanted sounds that are picked up or generated by the receiver as well as airborne sounds from his surroundings.

Airborne sounds often may be a limiting factor. Listening in an airplane for the signals from a sonobuoy sometimes is limited by this type of noise, which often is referred to as "local noise" or "room noise." The signal can be made more perceptible by increasing the amplification of the receiver; for in this case the airborne noise is not amplified and the signal-to-noise ratio is increased.

The desired signal is but one of the many sounds that are amplified and heard by the operator.

  These sounds originate in the sea and in the listening vessel itself, and they constitute a masking background for the signal. Increasing the gain of the receiver in this case does not help, for the background noise also is amplified with the signal. Noises that are created in the receiver itself also are amplified and, mask the desired signal, the same as those sounds that are picked up by the hydrophone.

The sources of the circuit noise are (1) thermal agitation of electrons in the tuned input circuit, (2) tube noise, (3) hum due to man-made disturbances, and (4) vibration of tube elements resulting in "microphonics."

Figure 3-3 shows the complete classification of background noise. This figure shows that self-noises are (1) circuit noise, (2) hydrophone motion, and (3) noise from own ship such as vibration and turbulence caused by the ship's motion.

The other important sources of background noise are classified as ambient noise. Ambient noises are (1) sea noise, due principally to the

Figure 3-3. -Classification of background noise.
Figure 3-3. -Classification of background noise.
 
60

TABLE 5. -Over-all Levels of Amplified Noise (0.1 to 10 kc)
Types of noises Decibels
Self-noise:  
  Circuit noise -30 to 0
  Submarine self-noise 0 to 20
  Surface vessel self-noise
(DD or DE) 10 to 25 knots
5 to 40
Ambient noise:  
  Sea noise:  
    Deep sea -5 to 6
    Near surface -17 to 9
  Biological noise:  
    Snapping shrimp 5 to 7.5
    Croakers 36 (max.)
    Porpoises 40 (max.)
    Evening noise 8.5 (max.)
  Traffic noise
(includes sea noise)
0 to 22

wave motion at the surface of the water; (2) biological noise, caused by many species of marine life; and (3) traffic noise, which exists when many ships operate at the same time, such as in a harbor. The noise of fish and marine life is not always undesirable but in the detection of ships or submarines is usually a source of trouble. Because this type of noise is rather peculiar, it will be discussed in some detail.

  Table 5 is a summary of the average values of background noise of all kinds. This table, which gives some interesting information regarding the intensity of noise made by fish, will be referred to from time to time.

SOUNDS PRODUCED BY OBJECTS IN THE SEA

Biological Noise

Surprisingly large numbers of species of marine life produce sounds of various sorts. They are mostly crustaceans and vertebrates. Biological noise is an important factor in limiting listening ranges in shallow water only in tropical and subtropical regions. To discuss the complicated subjects conveniently, it is customary to group the various sounds from marine life into three categories, which in the order of their importance from an operator's viewpoint are (1) shrimp noise, (2) periodic fish choruses or croaker noise, and (3) miscellaneous biological noise.

Early in World War II it was observed that as a listener approached shallow water, the ordinary ambient noise was sometimes replaced by sounds resembling the sizzle of frying fat. As he came closer to the shore, he noticed that the sound approximated the crackle of burning twigs or the

World globe showing distribution o snapping shrimp.
Figure 3-4. -Distribution o snapping shrimp.
239276°-5-5
 
61

crashes of static noise heard in a radio receiver. This noise was encountered only in tropical and subtropical regions, and was more common over boulder-strewn or cobble-strewn bottoms. It was sometimes confused with noise due to surf. Investigation discovered the source of this noise to be colonies of certain species of snapping shrimp (not to be confused with the ordinary edible species) that close their pincers with a loud audible click, similar to that caused by snapping a fingernail. The rate at which a single shrimp produces clicks and the reason for this activity are not known. The combined activity of hundreds of thousands of shrimp is required to produce the observed sizzle.

The chief habitats of these shrimp are in coral formations and on rocky sea bottoms where the water is less than 30 fathoms deep. Few are found on mud or sand bottoms. The map in figure 3-4 shows that they are widespread throughout tropical and subtropical regions of the world. In this figure, shaded areas show regions where shrimp occur when water depth and bottom are favorable.

Shrimp noise is a serious masking noise in listening, both because of its intensity and because of its spectral distribution. Although it has a

Spectra of shrimp noise for daytime and nighttime.
Figure 3-5 -Spectra of shrimp noise for daytime and nighttime.

  Diurnal variation of shrimp noise, over-all level
at various locations.
Figure 3-6 -Diurnal variation of shrimp noise, over-all level at various locations.

measured frequency range of from 1.5 to 45 kilocycles the main components lie between 1.5 and 20 kilocycles. The spectrum level at 10 kilocycles may be of the order of -39 to -29 db, as can be seen from figure 3-5. In this figure the dots indicate average values; the dotted curves show the spread of the spectrum levels. It is evident that shrimp noise is a serious complication in both sonic and supersonic listening.

Shrimp noise is remarkably constant throughout the year. There is a small diurnal variation-the noise is from 2 to 6 db higher at night than in daytime, small maxima occurring about 1 hour before sunrise and about 1 hour after sunset. (See figure 3-6.)

The chief noise makers among fish are certain species of croakers and drumfish, which are common, especially on the Atlantic coast. An individual croaker emits sounds resembling 4 to 7 rapid blows on a hollow log.

At certain periods of the year large schools of croakers infest certain localities. In the Chesapeake Bay the croaker season extends from May to July. During this season there is an evening chorus of croaker noise lasting several hours, with a peak just after sundown. Over-all levels of croaker noise showing seasonal and diurnal variation are shown in figure 3-7.

The spectrum levels of a sample of croaker noise are shown in figure 3-8. The solid curves show the difference in average level between early evening and the period after midnight during July.

 
62

Seasonal and diurnal variation of over-all levels
of croaker noise.
Figure 3-7 -Seasonal and diurnal variation of over-all levels of croaker noise.

The dotted curve is the average spectrum for early June. When it comes, croaker noise may completely mask desired signals, for the frequency range of croaker noise lies almost entirely below 1 kc, the region where the most prominent components of ship sound occur.

In and near busy harbors the ordinary sea noise and biological noises are overlaid with the sounds associated with the movements of ships, especially small high-speed craft, and by the noise of industrial operations on the beach. Listening in harbors thus becomes extremely difficult; hence installations

Spectra of croaker noise.
Figure 3-8 -Spectra of croaker noise.

  off the harbor entrance have been devised to ensure protection of harbors against sneak attacks by enemy submarines.

Traffic noise is essentially variable, but a certain periodicity can be expected. Measurements made in New York Harbor and its approaches are shown in figure 3-9. Curve A shows the spectrum level of the noise in the harbor in the daytime, and curve B, the average levels measured in upper Long Island Sound near the ship lanes. Curve B is about 9 db below the harbor level at all frequencies. For comparison, the curve of sea noise for sea state 2 is included as curve C. In the region of sonic frequencies the harbor noise is from 10 to 18 db above this level. Over-all sound levels (0.1 to 10 kc) for the noise in the harbor itself is about 16 db, compared with 6 db in the harbor approaches and 0 db for water noise with sea state 2.

Spectra of traffic noise in New York Harbor and
its approaches during the daytime.
Figure 3-9 -Spectra of traffic noise in New York Harbor and its approaches during the daytime.

Nighttime levels of ambient noise in the approaches to New York Harbor are shown in figure 3-10, with a curve showing average daytime levels added for comparison.

Submarines

From the standpoint of antisubmarine operations, a knowledge of the sound output of submarines is needed for the prediction of maximum listening ranges. The design of listening gear, in particular the choice between sonic and ultrasonic devices, depends on the spectrum of the sound to be detected.

From the standpoint of submarine operations, it is important to know the relative sound output of various submarine maneuvers, so that evasive action is not nullified by excessive detectable

 
63

Same as figure 3-9 but for nighttime.
Figure 3-10. -Same as figure 3-9 but for nighttime.

sound. The problem of noise control, and the design of propellers, engines, and auxiliaries, all demand measurements of sound output.

The machinery of the submarine is extremely diversified and complicated. The submarine has more than 50 auxiliaries, all of which are potential sound sources. Figure 3-11 lists a few of these sources, shows the source levels that have been proposed as best naval practice, and gives the maximum permissible limits.

In general, these sounds have a continuous spectrum, with a maximum at low frequencies. Sometimes, however, the machinery produces a

Suggested limits of over-all sound level representing best
naval practice.
Bow Plane Hand 15-30
Bow Plane Power 32-42
Steering 12-22
Drain Pump 15-32
Periscope 12-28
Trim Pump 15-35
Stern Planes Hand 15-28
Stern Planes Power 32-44
Starboard Astern Port Ahead 40 RPM 30-42
Starboard Ahead Port Astern 40 RPM 30-42
Figure 3-11 -Suggested limits of over-all sound level of several auxiliaries on submarines, and the levels representing best naval practice.

  strong line spectrum that is superimposed on the continuous spectrum.

Propeller sounds are of two general kinds- (1) singing, due to vibrations of the propeller blades, and (2) cavitation. Cavitation sounds are the most important of all submarine sounds. Vibrations of the propeller blades may be due to faulty design or manufacture and are generally not difficult to eliminate.

Dependence of over-all source levels of submarine sounds on depths of submergence h (feet) and speed
V (knots). A, 0.1-kc to 10-kc; B, 10-kc to 30-kc.
Figure 3-12 -Dependence of over-all source levels of submarine sounds on depths of submergence h (feet) and speed V (knots). A, 0.1-kc to 10-kc; B, 10-kc to 30-kc.

Cavitation results when the propellers turn so rapidly that the water does not close in behind the blades. Thus, a stream of bubbles resembling those in a boiling kettle is formed. These bubbles may be caused by reduced pressures on the backs of the propeller blades or by vortices at the tips of the propeller blades.

 
64

The steep rise between the value of 0.4 and the value of 0.6 for V/h½ is due to cavitation (figure 3-12). The smooth curve is drawn on the assumption that the speed at which cavitation occurs is inversely proportional to the square root of the hydrostatic pressure. Figure 3-12, A, plots the levels measured in the 0.1-kc to 10-kc bandwidth; figure 3-12, B, the levels in the 10-kc to 30-kc band. Acoustically, tip cavitation appears to be much more important than blade cavitation. This condition may exist because blade cavitation has a more serious effect on propeller thrust and is usually prevented by the designer of the ship.

Besides these two main sources of submarine sounds, there are some minor sources, such as splashing of water at the bow and in the wake when the submarine is at the surface; when submerged, the fittings of the vessel, such as handrails, may be set into vibration by the turbulent flow of water past them. These sounds are considered to be of small significance compared with those due to cavitation.

The activities of the crew are a source of incidental sound. It is interesting that, according to some British measurements, over-all source levels of from 45 to 50 db may be produced by dropping a wrench or by the use of the engine-room telegraph-levels comparable to those produced by the submarine itself under conditions of evasive operations. The transitory character of such sounds makes them comparatively unimportant, except when the submarine is evading detection by an alert enemy.

The sound output of a submarine varies widely with the size and type of submarine. For a given submarine it varies with speed and operating conditions. If the submarine is submerged, its sound output at a given speed decreases as the depth increases.

The over-all source level may range from about 40 db under evasive conditions to more than 75 db at top speeds. An average based on a large number of measurements gives the following values: (1) Running submerged at 6 knots, or on the surface at 12 knots, the over-all source level is about 72 db; (2) at top surface speeds, the over-all source level is about 77 db.

The dependence of the over-all source level on speed is shown for two submarines in figure 3-13.

 

Over-all source levels of submarine sounds.
Figure 3-13 -Over-all source levels of submarine sounds. A, Submerged variation with speed; B, two submarines, surface operation, illustrating the variability between ships.

In figure 3-13, A, the over-all source level is plotted against the ship speed for a submerged submarine, and in figure 3-13, B, for two submarines operating at the surface.

The variability of source level from ship to ship is indicated by the curve of submarine B in figure 3-13, B. The values of source levels of various submarines may vary by as much as 15 db under identical operating conditions.

The curve pertaining to operation at periscope depth is typical of ship sounds in general. At very low speeds the source level is quite low. At a certain critical speed-in this case 4 knots-the sound output increases very rapidly with speed, so that an increase of 2 knots is accompanied by an increase in the source level of 30 db. If the speed is increased beyond 6 knots, the curve levels off.

This abrupt increase in the sound output at the critical speed is due to cavitation, which is related to many factors but chiefly to the shaft rate or speed and to the hydrostatic pressure. If other

 
65

factors remain constant, the speed at which cavitation occurs is inversely proportional to the square root of the static pressure. Hence the sound output at a given speed is less when the submarine submerges to greater depths. This fact is shown by figure 3-12, in which over-all source levels are plotted against V/h½ where V is the speed in knots and h is the total hydrostatic pressure head. The value of h is calculated from h=33+d, where d is the depth in feet and 33 feet is the head of sea water equivalent to 1 atmosphere. The experimental points fit the theoretical curves fairly well.

The speed required for cavitation to set in is, in general, higher for submarines of new design because of a persistent effort to decrease the sound output of American submarines. It has been decreased, on the average by about 20 db; however, a few submarines still produce prominent and undesirable single-frequency tones below 1,000 cycles per second. There is considerable evidence that these sounds originate almost entirely in the reduction gears.

The relation between sound level and speed of a submarine is quite different for surface operation. Figure 3-13, B, shows that the increase in source level of submarine A is gradual, and does not show the abrupt rise due to cavitation that is observed with submerged operation. The higher levels associated with surface operation are attributed to the Diesel engines used for operating on the surface; the electric drive is considerably more quiet. The hump shown in the curve for submarine B, figure 3-13, B, is caused by a singing propeller.

Figure 3-14 gives the spectrum of a submarine running at 6 knots at periscope depth or at 12

Average spectrum of a submarine running at 6 knots at periscope depth or at 12 knots on the surface.
Figure 3-14. -Average spectrum of a submarine running at 6 knots at periscope depth or at 12 knots on the surface.

  Spectra of individual submarines. A, The variation of spectra with speed of submerged submarine; B, effect of increasing depth on the spectra.
Figure 3-15 -Spectra of individual submarines. A, The variation of spectra with speed of submerged submarine; B, effect of increasing depth on the spectra.

knots on the surface. These values are the average of a large number of measurements. It must be borne in mind that there is a great spread in individual measurements, and thus the sounds from a given submarine may deviate decidedly from the values in the figure.

Figure 3-14 shows that the intensity of submarine sounds decreases rapidly with the frequency; the drop in level is about 6 db per octave on the average. In other words, the spectrum level is about 20 db higher at 100 cycles per second than at 1,000 cycles per second and this same proportionate variation continues at least until 30 kilocycles. As a result, the over-all level is largely determined by the lower frequencies.

If the threshold of listening gear were independent of frequency, sounds with such a spectrum would be much more readily detected with sonic than with ultrasonic devices. However, the threshold also decreases with increasing frequency, especially for gear mounted on a moving surface vessel. Until recently this factor has tended to nullify the advantage of sonic listening. On

 
66

sailing vessels, sonic listening retains its advantage, especially if the auxiliaries can be periodically shut down for listening. An effective antisubmarine watch can thus be maintained from such vessels. The same is true of bottom-mounted hydrophones and sonobuoys, both of which use the sonic band.

Sound-level spectra of individual submarines are shown in figure 3-15 and figure 3-16 for various operating conditions. Figure 3-15, A, shows the effect of increasing speed on the sound-level spectrum. A characteristic feature of these curves is a peak at low frequencies, and a tendency for this peak to occur at lower frequencies as the speed increases. This behavior is ascribed to cavitation effects. It is thought that higher propeller speeds produce progressively larger bubbles. The resonant frequency of a bubble is inversely related to its diameter, and thus an increase in speed results in the production of sound of a lower frequency.

The exact position of these peaks also varies from submarine to submarine. Consequently they do not show on the average curve of figure 3-14. Even the peaks of these submarines lie well below the average curve for frequencies of less than 1 kc.

Figure 3-15, B, shows the effect of increasing depth on the sound-level spectrum. The peaks tend to shift toward higher frequencies with increasing depths. The increase in hydrostatic pressure with depth probably reduces the size of the cavities formed at a given speed and thus results in a higher resonant frequency.

Very little is known concerning the location of the particular point, or points, on the ship that can

Variation of spectra of individual submarines
with speed in surface operations.
Figure 3-16 -Variation of spectra of individual submarines with speed in surface operations.

  be considered as the effective source of the radiated sound. There is reason to believe that at periscope depth the engine room is the principal source of sounds at very low speeds, whereas at speeds above 3 knots the propeller is chiefly responsible. however, even at high speeds the engine room may contribute materially to the sound at frequencies below 150 cycles per second. During surface operations the propeller and wake are probably the principal sources of sound at practically all speeds with electric drive. With Diesel drive the engine room is the main source at low speeds and a material contributor at all speeds.

The sounds from submarines are radiated in such a way as to produce approximately a uniform sound field at a distance of several ship lengths from the source. Some observers report a slight decrease in the sound level in the region within 10° or 20° on either bow; at 200 yards this decrease is from 2 to 4 db. A similar shadow astern of the ship has been reported. This shadow is ascribed to the wake.

Surface Ships

The sounds emitted by surface vessels may provide considerable information to an experienced sound operator aboard a submarine. Various forms of underwater mines are detonated by a ship's sound. Ship sounds vary greatly in intensity and spectrum from ship to ship and from one class of ship to another. For a given ship sound intensity and spectrum vary with speed.

From the viewpoint of defense, every ship that is likely to enter water harboring hostile submarines obviously would benefit by an analysis of its own sound output. Such an analysis would disclose the existence of any revealing single-frequency components. These undesirable components are due to causes that can be remedied easily. The analysis also would make possible more accurate estimates of the range at which a ship is apt to be detected by an enemy submarine.

The extreme values of observed over-all source levels range from about 50 db for launches and small auxiliary craft at low speeds to 110 db for battleships at 20 knots. The 110-db value is approximately the source level of a standard sonar projector. The average over-all source levels of submarines range from about 30 to about 75 db.

 
67

Spectra of surface ships.
Figure 3-17 -Spectra of surface ships.

Besides being affected by the speed of the vessel, the over-all source level is a function also of the load or displacement of the ship.

The sources of ship sounds are extremely diversified, and a given source may change its sound output with ship speed. Hence ship sounds are variable and complex and are distributed through the whole range of frequencies. As with

  submarines, the chief sources are the screws, where cavitation produces the sound, and the hull, which transmits the vibrations of the machinery and engines.

Single-frequency components due to propeller singing or to vibrations of the propulsion machinery are common. Ordinarily such sounds occur below 1 kilocycle, but sometimes these single-frequency components are encountered well above this frequency.

Figure 3-17 shows the average spectrum-frequency distribution of sounds from a large number of surface ships. The data on which this figure is based were the average measurements made on 52 ships comprising 12 different types of warships and commercial vessels. The ordinates on the graph are the values of relative spectrum levels-that is, of the spectrum level minus the over-all level (0.1-kc to 10-kc). These differences are averaged for all types of ships in order to obtain the graphs. Because the total spread of the measurements on the individual ships was considerable, due allowance for this spread must

Average spectrum levels for six different classes of ships.
Figure 3-18. -Average spectrum levels for six different classes of ships.
 
68

Effect of varying speed on spectral distribution.
Figure 3-19 -Effect of varying speed on spectral distribution.

be made when using data from this graph and the following graphs.

The level of the sound decreases with increasing frequency at a rate of 7 db per octave. This relation is similar to that shown in figure 3-14 for submarines. Spectra of the different ships vary in average slope from about 5.5 to about 8.6 db per octave. Figure 3-18 shows average spectrum levels for six different classes of ships at normal cruising speeds. The average over-all levels also are indicated.

Figure 3-19 illustrates the effect of varying speed on a ship's spectral distribution. Curve L represents the average spectrum at low speeds, curve H that at high speeds, and curve N that at normal cruising speeds. At very low speeds the chief source of sound is the machinery, and all the machinery contributes materially. Much of the sound from this source is concentrated at the lower frequencies; therefore in this region the spectrum is highly variable, as was previously noted with submarines.

The variability of the spectra in the lower frequency region may be ascribed again to cavitation, which is the chief source of ship sounds at all but the lowest speeds. The sound due to cavitation has a continuous spectrum, whereas machinery sound generally is more likely to consist of many discrete components closely spaced. Above approximately 1 or 2 kc the spectral slope of cavitation sound is very nearly -6 db per octave; but in the region of lower frequencies there is usually a peak (figure 3-15). The frequency at which this peak occurs depends on various factors related to the type and .size of ship and its speed, and thus may provide some information tending toward identification of the vessel.

  At high speeds cavitation may introduce components in the ultrasonic region, as shown by curve H in figure 3-19.

The sound emitted by ships has very little directivity, particularly in the sonic region of frequencies. Average directivity patterns for 15 freighters for the low frequencies (200 to 400 cycles per second) are illustrated in figure 3-20, where sound levels are exhibited as contours-lines joining points of equal intensity. The levels were measured with a bottom-mounted hydrophone.

Contours are somewhat difficult to reconcile with the fact that many ships have two dominant sources of sound, one at the engine room and the other at the screws. In large destroyers these two sources are of equal level at about 12 knots. At 8 knots the engine room is the dominant source, whereas at 16 knots the screws are the dominant

Contours showing the average directivity of ship
sounds.
Figure 3-20 -Contours showing the average directivity of ship sounds. A, Average patterns for 15 freighters for low frequencies (200-400 cycles per second); B, contours of sound levels for a typical freighter at 8 knots. The outline of the ship is indicated by the shaded area.

 
69

source. In Liberty ships, however, the two sources are of about equal level at all speeds. The dominance of the propellers as the source of sound for the 15 ships shown in figure 3-20, A, possibly indicates that Liberty ships are not typical of all freighters.

If the source of sound from a ship is concentrated at the screws or over a small part of its hull, the audible sound is independent of direction except for the shadow effect of the hull and wake. This effect is illustrated graphically in figure 3-20, B,

  which shows the contours of pressure levels for a typical freighter cruising at 8 knots. The outline of the ship is shown by the shaded area. The shadow and screening effects are highly variable from ship to ship. These variations and the variable distribution of the sound sources make it difficult to generalize about the sound distribution. It is probable that for large ships the sound-pressure level 400 to 500 feet ahead or astern of the main source of sound is 5 to 10 db below the level at the same distance abeam.
 
Time Patterns and Propeller Beats
 
RHYTHMS AND OTHER TIME PATTERNS

The necessary prerequisite for the detection of a ship or submarine is that its sound have sufficient intensity at the hydrophone to be heard above the background noise. Because the level of background noise usually varies in an irregular manner, a rhythmic sound having a periodic pattern of beats, may be more readily recognized than a nonrhythmic one.

Moreover, intensity alone conveys no information other than that something in the neighborhood is making a noise. Additional information about the source is obtained from the spectrum (high or low pitch) and from any rhythm that is inherent in the sound.

The propeller sounds of a large ship, although produced by cavitation, usually pulsate periodically. In some ships, the beat may be unaccented and occur once per propeller revolution (shaft frequency). Other propeller sounds pulsate several times per revolution; a three-blade propeller gives 3, and a four-blade, 4, beats per revolution (blade frequency). If the beat is unaccented, it is difficult to determine which frequency is involved. However, one blade is often noisier than the others, resulting in an accent repeated at shaft frequency. In favorable cases, therefore, both the number of blades and the propeller rpm can be determined. These items partially identify the class of ships, and certainly differentiate its sound from various intermittent background noises.

Perception of Time Patterns

The manner in which fluctuations in sound level are heard depends on their rate or frequency. Very slow changes in level are not perceived unless

  they are relatively large; they are often called fading. Rhythms are most easily heard and counted when the beats occur two or three times a second. At high rates, counting becomes difficult; with practice, it can be done by counting every third or fourth beat.

When the frequency becomes greater than about 15 or 20 cycles per second, the individual beats are no longer heard. The rhythm is then heard as a "flutter" or "tremolo". Frequencies much above 100 cycles per second are not recognized as periodic, but as a pitch that is inherent in the sound.

SINGLE-FREQUENCY COMPONENTS

Audibility

Previous discussions in this chapter have pointed out that ship sounds in general have continuous spectra-that is, (1) the emitted sound energy is distributed over a wide range of frequencies, and (2) on the average, the distribution of the energy over the frequency range follows a fairly simple pattern-a decrease in the sound level of about 6 db per octave increase in frequency.

Mention has been made, however, of the occurrence in ship sounds of relatively pure tones of audible frequency. On a spectrum plot an absolutely pure tone would be one-dimensional having sound level but no frequency width. A spectrum composed predominantly of such discrete components would be a line spectrum. Actually the so-called single-frequency components comprise a relatively narrow band of frequencies; but if the width of this band is smaller than the width of the band that can be resolved by the ear, the single-frequency components will have a definite

 
70

pitch. It is in this sense that the terms "single-frequency component" and "pure tone" are used.

The ear very readily detects pure tones against a background of complex noise. This detection is possible because the ear is a very efficient analyzer of comparatively high selectivity, and because a pure tone has a distinctive quality that contrasts strongly with random noise, which has no definite pitch. These characteristics make it possible for the ear to detect a pure tone in the audible region even when its sound level is considerably lower (sometimes as much as 20 db) than the over-all level of the background noise.

Tests have shown a pure tone can be heard when its level is at least equal to the level of the background noise in a band of a certain width at the frequency of the tone. The width of the band depends on the frequency. These critical bands are from 30 to 50 cycles per second wide for tones of between 100 and 1,000 cycles per second. This

  fact is an indication of the great effectiveness of the ear in discriminating against random noise.

Time Patterns

Pronounced rhythmic time patterns sometimes occur in single-frequency components originating in propeller vibrations. Also, many single-frequency components have their source in reduction gears.

The extreme audibility of single-frequency components, as compared to sounds of continuous spectrum, introduces complications in the techniques of sound measurement. For example, suppose the over-all level of a moored submarine with its motors secured is measured. It has a continuous spectrum of certain over-all level. The motor may produce a pure tone that increases the audibility of the submarine's sound very materially, but may scarcely affect the over-all level.

 
Frequency Considerations in Listening
 
In the over-all problem of detection by listening, two general classes of systems can be distinguished. One class includes those systems with a listening band that falls in the ultrasonic region and an output that is made audible by a heterodyne change of frequency. The other class has its listening band in the audio frequencies and does not need a heterodyne stage to make the output perceptible.

SONIC LISTENING

Sonic listening depends on the sources of sonic sounds. These sources are surface vessels, submarines, torpedoes, explosions of depth charges, and the echo-ranging signals of other vessels. Cavitation sounds have a comparatively continuous spectrum, the level of which falls off about 6 db per octave on the average. They are sufficiently uniform to make it possible to determine the cavitation spectrum of a given class of ship at a given speed by taking a single measurement at some frequency-say 1 kc or 5 kc. Enough measurements on cavitation sounds from various sources have been made to enable the prediction of their level for any class of ship at any speed within about 5 db.

It is not so easy to predict the level of machinery sounds, which are the dominant source of

  low-frequency sound (less than 1 kc) at low speeds. These sounds have very complex and irregular line spectra and differ widely among different ships. They are heard as squeaks, rumbles, groans, and whines.

The spectra of the various types of ambient noise that are encountered in listening have been discussed. Ambient noise is the limiting factor when the listening hydrophone is stationary, provided the sea state is greater than 1 or 2. For a sea state of less than 2, the over-all level of ambient noise drops below 0 db and thus approaches the over-all level of circuit noise, which ranges from -30 to 0 db. In this case, the circuit noise may be limiting. Shrimp noise is usually negligible at lower sonic frequencies.

The data for transmission loss in the frequency range of from 200 to 2,000 cycles per second can be schematically summarized (figure 3-21).

At ranges less than a few hundred yards, the transmission loss, H, is variable because of the interference between direct and surface-reflected sound. This condition is indicated by the double hatching in the figure. Beyond this variable region, the transmission loss increases rapidly out to about 2,000 yards. The frequency is a determining factor in this region. The low frequencies suffer a greater loss than the high frequencies.

 
71

Downward refraction in the upper layers causes this loss to occur at shorter ranges. The single hatching on figure 3-21 shows the region of the rapidly increasing loss.

Beyond this region bottom-reflected sound is dominant, and the transmission loss remains constant out to about 20,000 yards. The magnitude of this loss and the range at which it begins depend on the depth of water. A value of 80 to 85 db appears to be relatively independent of thermal conditions but increases slightly with the hydrophone depth. This value is also subject to irregular fluctuations of considerable magnitude, but they do not appear to bear any systematic relation to the range.

At very long ranges the transmission loss must again increase, but there is very little data to indicate the rate of increase.

The fact that the transmission loss of bottom-reflected sound is nearly independent of range has an important effect on the maximum ranges obtained with sonic gear. If the available signal output is between 60 and 80 db, the maximum range is likely to be less than 1,000 yards and unlikely to be greater than 2,000 yards. Contact is not established until the target becomes audible by way of direct sound. If the available signal output is greater than 80 db, however, the bottom-reflected sound may become useful, and range may suddenly increase to between 10,000 and 20,000 yards.

ULTRASONIC LISTENING

Ultrasonic sound is made audible by heterodyning, so that the loudspeaker of the listening system emits audible sound. The general principles of recognition for heterodyned ultrasonic

Transmission loss H(r) for sonic sound.
Figure 3-21 -Transmission loss H(r) for sonic sound.

  Transmission loss H(r) at 24 kc for various thermal
conditions.
Figure 3-22 -Transmission loss H(r) at 24 kc for various thermal conditions.

sound are thus identical with those applying to audible sound. However, several quantitative differences exist.

In the first place, ultrasonic receivers usually have pass bands not more than 1 kc wide. The spectrum of the heterodyne output may thus be confined to the range of from 300 to 1,330 cycles per second, as compared with a range of 10,000 cycles per second in sonic listening.

In the second place, a 1-kc band of one ultrasonic spectrum is very similar to a 1-kc band of another. There are no single-frequency peaks, and although most spectra slope 5 to 9 db per octave, the change in spectrum level over a 1-kc band is negligible for many purposes. This principle applies to background noise as well as to the sound output of ships.

Thus, there usually is no one frequency of the heterodyned sound that is more audible than another. There is no tonal quality to distinguish the signal from the background.

In general, the recognition differential for ultrasonic listening is zero. The ultrasonic sound from a ship's screw, however, is usually rhythmically modulated in intensity. Recognition occurs when the maximum level of a rhythmic signal is equal to, or possibly a few decibels less than, the average level of nonrhythmic background. The maximum level of screw sounds is usually about 3 db above the average level. Because most measurements yield average values, they must be increased by about 3 db in calculating the available signal. This increase is sometimes loosely called a "recognition differential."

 
72

An exception to these statements occurs when the target vessel is echo ranging. The pings are heard as tonal pulses of sound which have a high recognition differential, as well as a high source level.

These considerations introduce some simplification into the calculation of ranges. The spectra of the signal and the background noise need not be considered in detail; it is sufficient to state the spectrum levels at the midpoint of the listening band.

The situation with regard to background noise is similar to that of sonic listening. That is, if the listening vessel is quiet, ambient noise predominates; whereas if the listening vessel is noisy, the noise of the listening vessel predominates. In ultrasonic listening, however, when ambient noise is limiting, shrimp crackle becomes important. The ordinary levels of ultrasonic ambient noise range from -78 to -53 db depending on sea state. If shrimp are present, however, the ambient noise levels may be -49 to -39 db.

When used at ultrasonic frequencies, listening gear discriminates against ambient noise. A directivity index, D, of -23 is common among standard echo-ranging transducers.

The graphs of figure 3-22 should be compared with figure 3-21 to contrast the transmission loss of the ultrasonic frequencies.

The curves in figure 3-22 are based on the anomaly of figure 3-23, and the same numbering is used. Because there is no horizontal portion of

  Average transmission anomaly under various
oceanographic conditions.
Figure 3-23 -Average transmission anomaly under various oceanographic conditions.

these curves (figure 3-22) the ultrasonic ranges should show less variation than do the sonic ranges. Because of this fact, also, there seems less probability of achieving great improvement in the performance of ultrasonic systems by a reduction in self-noise.

The values of the transmission anomaly in figure 3-23 have been determined by experiment. In this figure D2 is depth for which the change in temperature is 0.3° F, which is the smallest temperature change that can be detected by the present bathythermograph. Note that D2 has the following values:

Curve 1-0 ft<D2<5 ft.
Curve 2-5 ft<D2<20 ft.
Curve 3-20 ft<D2<40 ft.
Curve 4-40 ft<D2<80 ft.
Curve 5-80 ft<D2<300 ft.
 
Sonar Listening Systems
 

SONIC LISTENING SYSTEM

Many of the problems that affect underwater detection by receiving and analyzing both sonic and ultrasonic sound energy have been discussed in this chapter. Block diagrams and a brief description of the function of the various components of both types of listening equipment will now be given. The sonic listening equipment consists of a hydrophone, training unit, receiver-amplifier, and headphones or speaker. This system is shown in figure 3-24.

Hydrophone

The hydrophone used in listening equipment is primarily of the magnetostriction type. Some

  equipments use the crystal type. The primary purpose of the hydrophone is to convert sound energy in the water into electric energy that can be amplified and heard from the loudspeaker. The hydrophone must have a directional characteristic. This directional characteristic is used in two ways. First, it allows the operator to discriminate against unwanted sound, and, second, it enables the operator to determine the direction from which the desired sound is coming.

The discussion thus far applies to both sonic and ultrasonic hydrophones. There is little difference between them. The sonic hydrophone must have larger dimensions than the ultrasonic hydrophone for the same directivity index.

 
73

Training Unit

The method of training the hydrophone may be either manual or power. The more modern equipments use an electrically operated drive called a servo-mechanism. This drive is usually an amplidyne system. The operator can train the hydrophone on any bearing relative to the ship. A synchro repeater system is used to give bearing indications to the operator in both relative and true bearings. The two functions of the training device are therefore to allow the operator to train the hydrophone and to provide him with visual indication of the bearing to which the hydrophone is trained. The training unit is the same for both sonic and ultrasonic listening. In some installations it is used for both.

Receiver-Amplifier

The receiver-amplifier is simply an audio amplifier with wide-frequency response. The electrical signals from the hydrophone enter the receiver where they are amplified until their intensity is sufficient to drive a loudspeaker or headphones, as the need may be. There is no necessity for any frequency conversion because the signals entering the receiver in sonic listening are already in the audible frequency range-unlike

  those entering the receiver in ultrasonic listening in which heterodyning is necessary. This frequency conversion is the principle difference between the sonic and ultrasonic systems. It may be desirable under some conditions to limit the frequency response of the receiver in sonic listening. Band-pass filters are usually included for, this purpose. If the principal signal desired by the operator were about 500 cycles per second it would be possible to increase the signal-to-noise ratio by cutting down the band pass of the receiver just enough to include this frequency. Any noise falling outside the band pass of the receiver would not be heard.

Headphones and Speaker

The choice of headphones or speaker is dictated by the airborne noise of the surroundings. The purpose of these devices is to convert the electric signals from the receiver into sound impulses that can be heard by the operator. The cycle is completed by this conversion. In the water the signal is first sound energy, and when it falls on the hydrophone, electric impulses are generated. These electric impulses are amplified and reconverted into sound by loudspeakers or headphones.

Block diagram of a sonic listening equipment.
Figure 3-24. -Block diagram of a sonic listening equipment.
 
74

Block diagram of the ultrasonic converter. From left to right:
Hydrophone
Pre-Amplifier
71 KC Low Pass Filter
Variable Attenuator
First Mixer with input from First Oscillator 102-154 kc
88-94 kc band-pass filter
Second Mixer with input from Second Oscillator 94 kc
5 kc Low-Pass Filter

Input of Amplifier at Point Marked A on Fig 3-24
Figure 3-25 -Block diagram of the ultrasonic converter.
ULTRASONIC LISTENING SYSTEM

Ultrasonic listening equipment will be discussed by comparing it to sonic listening equipment. The ultrasonic hydrophone is the same as the sonic hydrophone in most cases. A small ultrasonic hydrophone, however, gives directivity similar to that of a large sonic hydrophone. Thus the directivity index of the same hydrophone used for both sonic and ultrasonic listening is greater for ultrasonic listening. An ultrasonic hydrophone, therefore, gives sharper bearing indication.

The training equipment of ultrasonic systems is identical to that of sonic systems.

The principal difference in sonic and ultrasonic listening is in the receiver-amplifier. When the sound to be heard is in the ultrasonic frequency range some method must be used to bring it into the audible, or sonic range. Heterodyning in the receiver accomplishes this change in frequency.

  Note the signal path in the block diagram of the ultrasonic converter shown in figure 3-25.

Usually there is a broad-band amplifier stage at the receiver input. This stage is followed by a filter system and an attenuator. The signal is then fed into the first mixer, where it is mixed with the output of a variable-frequency oscillator. The tuning of this oscillator provides for the adjustment of the receiver to various frequency inputs. The signal from the mixed stage is amplified through an intermediate-frequency amplifier similar to that of any superheterodyne radio receiver. This intermediate frequency is usually above the frequency of the ultrasonic signal and is the sum of the ultrasonic signal and the output of the oscillator. The intermediate frequency is then fed into a second mixer where it beats with a second oscillator to give an output in the audible-frequency range. This converting system is in addition to the regular audio amplification of the receiver, which drives the speaker.

 
75

Previous Chapter
Previous Chapter
Sonar Home Page
Sonar Home Page
Next Chapter
Next Chapter

 

Copyright © 2013, Maritime Park Association.
All Rights Reserved.
Legal Notices and Privacy Policy
Version 3.01