- Email: [email protected]
On the performance of hearing protectors in impulsive noise
Applied Acoustics, Vol. 54, No. 2, pp. 165±181, 1998 #1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain PII: S0003-682X(96)00026-6 0003-682X/98/$19.00+.00
On the Performance of Hearing Protectors in Impulsive Noise D. Smeatham and P. D. Wheeler University of Salford, Department of Applied Acoustics, Salford M5 4WT, UK (Received 4 September 1995; accepted 2 May 1996)
ABSTRACT The purpose of this study was to facilitate the comparison of typical industrial and military impulse noise to that reproducible in a laboratory by means of categorization of the dierent types of impulse noise. This categorization enables us to de®ne laboratory impulse noise sources which can then be used to study the non-linear attenuation characteristics of hearing protectors in a laboratory environment. # 1998 Elsevier Science Ltd. All rights reserved
INTRODUCTION European Community (EC) Directive 86/188/EEC1 sets advisory and mandatory limits, called action levels. These limits apply to both the daily personal noise exposure of employees and the maximum unweighted sound pressure to which employees may be exposed. The daily exposure action levels are currently 85 and 90 dBA and the peak limit is 200 Pa (140 dB SPL). In order to assess the protection provided by hearing protector devices in relation to the peak limit, information is needed which describes the resultant pressure level underneath a hearing protector when subjected to impulse noise. To fully de®ne the acoustic performance of a hearing protector when subjected to a speci®ed impulse requires a detailed knowledge of the dependence of any non-linearity in its response on the peak level, rise time, and duration of the impulse. In other words, a full parametric analysis of the protector's attenuation performance as a function of these variables is needed. It is often impracticable to measure the attenuation of hearing protectors at a working location. Therefore, laboratory impulse sources need to be developed whose characteristics are similar to those found at the workplace to facilitate laboratory attenuation measurements. 165
166
D. Smeatham, P. D. Wheeler
This paper is based on research conducted for the EC-funded IMPRO project which has been studying the possibility of using impulse noise to assess the non-linear performance of hearing protectors. The paper describes the characteristics of impulse noise and gives an indication of the peak levels and durations of many industrial and military impulse sources. The study has been con®ned to sources which are likely to give rise to peak exposures close to, or exceeding, the 200 Pa limit. In order to make modelling in the laboratory simpler, the impulse noises have been divided into four categories which encompass the range of impulse noises in the military and industrial environments. The paper also discusses the performance of various types of hearing protector on the market today and give some indication of their non-linear behaviour. A number of laboratory impulse noise sources are identi®ed with their peak levels and waveform characteristics shown. RESPONSE OF HEARING PROTECTORS TO IMPULSE NOISE Initially it is worth reviewing the manner in which dierent types of hearing protector devices can be expected to respond to impulse noise. If a hearing protector behaves linearly, the peak level eective to the ear can be calculated from the impulse response of the protector. However there is a body of evidence that even conventional passive hearing protectors exhibit non-linear behaviour when subjected to high-level impulse noise. A classic example of this phenomenon is when the cushion of the ear defender lifts o the wearer's head. Additionally an increasing number of electronic or mechanical leveldependent protectors are being introduced into the marketplace, intended to respond selectively to transient and steady-state noise. In seeking to de®ne test noise sources with which to measure the impulse response of hearing protectors, it is important to take account of the ways in which such devices operate. An increasingly wide range of hearing protector types are now available, and are classi®ed below. Passive ear mus The conventional ear mu, when operating in its linear regime, may be regarded as a low-pass ®lter for impulse noise. In very simplistic terms, when the leading edge of the impulse pressure wave is incident on the protector, the ear mu is set into a damped oscillatory motion on the head, at its natural frequency. This resonant frequency, which is typically in the 100±500 Hz range, is determined by the mass of the ear mu and the compliance of the cushion acting on the ¯eshy area of the head surrounding the pinna.2
The performance of hearing protectors
167
If the excursion of the ear mu on the head is large, a gap may be introduced between the cushion and the head, allowing sound energy to penetrate virtually unattenuated. This is more likely to occur when the impulse noise has either a very high peak level or a large low-frequency component. Passive ear plugs Depending on its construction, the conventional ear plug may be modelled as either: (a) a simple acoustic resistance element located in the ear canal, or (b) a mass±spring system comprising the mass of the ear plug itself reacting with the shear compliance of the plug in the ear canal, and including the shear compliance of the ¯esh. Mechano-acoustic non-linear passive ear mus and plugs Hearing protection devices claiming a non-linear attenuation for impulse noise have been in existence for at least 30 years. The principle of operation is that the incidence of a high-level impulse at the protector gives rise to additional attenuation due to either the generation of turbulence in a small ori®ce or the closure of a valve in the sound path. For steady-state noises, such devices in general provide low attenuation, since the sound wave is able to penetrate the protector without impediment. Tests on such devices in the past have found that, in some cases, non-linear attenuation might only be achieved when the incident peak sound pressure exceeds 160 dB. Since the use of these `claimed' non-linear protectors in the workplace (and in leisure pursuits such as shooting) is increasing, often in circumstances where peak levels may lie in the 140±160 dB range, reliable attenuation data covering this range of sound levels are urgently needed. Active electronic non-linear protectors An increasing number of protectors ®tted with a level-dependent `hearingaid' electronic circuit are being used in the workplace in impulse noise situations. External sounds are detected by a microphone placed on the outside of the protector, whose output is fed, via a level-dependent electronic ampli®er, to an earphone mounted in the protector. More usually, these circuits are incorporated in ear mus, although ear plug devices are under development. The level-dependent electronic circuit may be either a simple peak-clipping circuit, restricting the maximum signal delivered to the earphone to that allowed by the ampli®er voltage supply rail, or a more sophisticated automatic gain control (AGC) compressor circuit in which the level of loud
168
D. Smeatham, P. D. Wheeler
sounds is reduced by a control loop which senses the time-averaged rootmean square (RMS) signal level but which allows quiet sounds to be transmitted without attenuation. In either case, the electronic circuits may allow transmission of external sounds to the ear at (well) above natural levels in the unattenuated mode. Devices providing 20 dB ampli®cation of sounds up to 80 dBA external level have been found during our investigations. These protectors may either provide a single (monaural) ampli®er circuit fed to one ear, or a twin-channel stereo system to aid localization of external sounds. In a related series of tests on such protectors, all the devices investigated were found to eectively limit the transmission of impulse sound to that achievable by the protector used in its passive modeÐeither by the setting of an appropriate maximum feed level to the earphone, or by the use of a fastacting AGC circuit. Active (noise cancellation) protectors This class of hearing protector is ®tted with an `antiphase' acoustic noise cancellation electronic circuit whose purpose is to increase the total attenuation provided by the protector. The term active noise reduction (ANR) is sometimes used to describe such protectors, which were originally developed for applications involving exposure to high levels of pseudosteady-state low-frequency noise (military vehicles and aircraft).3 ANR protectors are generally based on high passive attenuation ear mus although ear plug devices are under development. The ANR electronic circuit comprises a microphone mounted inside the protector which detects the internal sound level, a feedback or compensation ®lter, and ampli®er, and an earphone which transmits the `antiphase' signal (and any speech or communications signals needed). The additional attenuation provided by an ANR circuit is typically up to 20 dB in the 100±500 Hz frequency range, with the possibility of an unwanted `in-phase' component which can reduce eective attenuation by typically 3±5 dB in the 2000±4000 Hz range. Usually, no pre-knowledge of the external sound is required, and the device will attempt to follow variation in the instantaneous sound pressure level at the sensing microphone, within the bandwidth±time constraints of the electronic circuit. Although the bandwidth constraints of the feedback circuit will prevent the system `cancelling' an incident impulse, the system is quite capable of responding to the low-pass ®ltered signal under a circumaural hearing protector. Earlier experiments with a prototype ANR earmu showed that some 10 dB reduction in the single event level (SEL) of a
The performance of hearing protectors
169
pistol shot could be achieved. Maximum reproduced sound levels of typically 130±135 dB will be limited by transducer dynamics, supply voltage rails or protection circuits. Passive protectors with communication/relayed sound In some workplaces, employees may use hearing protectors with `piped' music or with speech communication facilities (hopefully) at levels equivalent to less than the 85±90 dBA action levels. In terms of their impulse response these devices may be expected to behave the same as passive mus. CHARACTERISTICS OF IMPULSE NOISE Impulse noise is produced by a number of dierent phenomena giving the impulse its own characteristics. The nature in which the impulse is developed determines the time-history characteristics of the impulse. For example, the impulse produced by an explosion in free air produces a single spike-type overpressure, whereas the impulse of a drop forge has a characteristic ringing time-history due to various complex resonance eects. Because of the diering time histories, dierent descriptors are used to de®ne the impulse characteristics. The types of time histories and their describers are shown below. Classical Friedlander (A-wave) impulse A single explosion or electrical discharge in free air will generally produce the classical Friedlander wave, shown in Fig. 1. The ideal impulse, for which the risetime (t1±t0) is zero, is described by eqn (1) (see Figs 4 and 5). p t pmax
0
ÿ t ÿt t 2 1
1 ÿ t2 ÿt t1 e
ÿ1 < t < t0 t0 t < 1
1
The time taken for the pressure to ®rst return to zero (at t2) is called the A-duration. The tail of the impulse extends for approximately six times the A-duration before the pressure again returns to zeroÐin practice to less than 1% of the peak levelÐat t3 (see Fig. 1). The spectral distribution of the energy in the impulse is controlled by both the A-duration (TD) and the rise time (TR), as shown in Fig. 2. Peak level, rise time, and duration all depend on the type of sourceÐin the case of explosives, the charge weight, design and containment. Gun®re noise (in free space) also resembles these simple Friedlander waves.
170
D. Smeatham, P. D. Wheeler
Reverberant impact pulse The impulse produced by two metal objects colliding, as in drop-forging and stamping processes, is a far more complex transient decay of energy which is
Fig. 1. Friedlander waveform.
Fig. 2. Spectrum of Friedlander impulse.
Fig. 3. Reverberant impact pulse waveform.
The performance of hearing protectors
171
distributed across the many possible modes of mechanical vibration of the objects in question. It is likely, at its simplest level, to take the form shown in Fig. 3. The decay of the ringing time history is described in terms of its B-durationÐthe time taken for the pressure to fall 20 dB from its peak value. (An alternative term, the D-duration, is sometimes used. This is the time taken to fall 10 dB from the peak level.) The spectral distribution of energy in this impulse is determined both by its overall envelope decay and by the complex 'ringing' infrastructure. It is worth describing brie¯y the dierent mechanisms which produce the ringing time history. Akay4 describes the ®ve fundamental noise generating mechanisms associated with the impact of two bodies. Air ejection When two surfaces approach each other air is ejected from the space between them. When this gap is rapidly narrowed just prior to contact the ¯ow becomes compressible and a pressure gradient is developed in the gaps
Fig. 4. De®nition of C-duration.
Fig. 5. De®nition of B-duration of multiple impulses.
172
D. Smeatham, P. D. Wheeler
between the surfaces. On rebound the newly created gap between the surfaces becomes a low pressure area and air ¯ows inward generating another pressure pulse. The in¯ow also sets the trapped air in the cavity between the surfaces into resonance, causing oscillations in the acoustic ®eld. Rigid body radiation One element in the sound ®eld resulting from the impact of two bodies is due to sudden velocity changes of two bodies on impact. Rigid body radiation is the predominant source of sound when a ball strikes a massive plate. Radiation from the ¯exural vibrations of the thick plate is negligible. Radiation due to rapid surface deformations The deformation of a surface results in the generation of an initial peak sound pressure pulse before the natural mode of radiation of the plate is set up. Pseudo-steady-state radiation Impact processes are generally employed to convert a vast amount of energy into useful work in a very short time. The time span is often too short for the applied energy to be wholly converted to work. The excess energy is absorbed by the mechanical structure. The resulting damped harmonic radiation is referred to as ringing, this generally dominates the total radiation ®eld from forging machinery. This is called pseudo-steady-state radiation in order to distinguish it from harmonically excited steady-state acoustic radiation. Radiation from material fracture The noise resulting from fracture of materials is a consequential source in punch presses and some other impact-forming machines. Eect of environment In the presence of any re¯ecting surface, such as the ground, or in a reverberant space, such as in a factory, there will be additional components in the impulse, produced by re¯ections. If the path lengths involved are short enough, the re¯ections may interfere with the original pulse, producing a complex temporal pattern.The term C-duration is used to describe the total time for which the pulse level exceeds the (peakÐ10 dB) criteriaÐsee Fig. 4. Within a reverberant space such as a factory, a machinery impact noise may now comprise a series of pulses, possibly overlappingÐsee Fig. 5. In such cases, the B-duration relates to the total time for which the level exceeds the peak ÿ20 dB criterion.
The performance of hearing protectors
173
WORKPLACE IMPULSE NOISE Impulse noise is generated in the workplace from sources associated with metal forming and cutting, fabrication and material-dressing, air exhausts and pyrotechnic devices. Although many noise sources to which employees may be exposed in the workplace might be classi®ed as `impulsive' in nature, in the vast majority of cases the peak sound pressure eective to the employee's ear is less than 200 Pa. Such cases need not be considered within the scope of this study. Industrial impulse sources A project undertaken by Bolt Beranek and Newman Inc. and reviewed by Dym5 involved the measurement of a number of impact and impulsive noises commonly found in industry. Information regarding the peak levels of noise are listed in Table 1.5±14 No information on the temporal and spectral characteristics were presented in this report. Other studies of impulse noise are included in the table. The peak levels and duration information provided in Table 1 have been TABLE 1 Source and reference
Duration (mS)
5
Bolt setting gun Hammer on metal plate6 Carpenter's hammer (0.45 kg)5 Piledriver5 Plastic-forming press5 Board hammer (29 000 N)5 Drop forge7 Drop forge8 Drop forge (12-tonne) Drop hammer (11 000 N)5 Pneumatic hammer9 Hand hammering (2 kg on anvil)5 Punch press10, 11, 12 Punch press5 Transistor lead preformer5 Nail shooting13 Shooting6 Stapler14 Impact welder Staple gun5 Power shear5 Pneumatic nailing machine5
B=25
B=60 B=50ÿ100 B=80 130 B=40 128 B=15 B=35 B=10
Peak level (dB) 140 125 147 130 128 145 133.4 160 143 138 140 135 118 140 148 130 155 127 117 148
174
D. Smeatham, P. D. Wheeler
measured at typical operator positions. Table 1 enables us to outline several major sources of impulse noises found in industry. Military impulse sources In general, the military impulse noise environment is characterized by shorter duration, higher peak levels and lower repetition rate than industrial impulse noise environments. The temporal and spectral characteristics of impulses produced by military weapons can be broadly categorized under two dierent classes, that produced by light weapons and that from heavy artillery weapons both exhibiting Friedlander-type waveforms. Light weapons such as pistols and ri¯es have very short rise times (typically 30 s) and short durations (typically 300 s). This sharp time-history produces a spectrum which has a very large bandwidth. Medium and heavy weapon (cannon-type weapons) impulses have longer rise times and duration, therefore producing spectra with large low-frequency components and less energy in the upper frequency ranges. Impulses produced from blasting, quarrying and explosive testing are also characterized in the same manner. TABLE 2 Weapon
Peak (dB)
Howitzer 105 mm
181 167±183 (crew position)
Ri¯e, self-loading RK 7.62 M16 7.62 M14 Bazooka, light Heavy Mortar Antitank Panzerfaust Pistol .38 M/08 9 mm P6 9 mm 9 mm Cannon (130 mm) Machine gun (7.65 mm) M60
175±187 (crew position) 158±165 (5.5 m) 159 (®rer position) 154±158 (2 m) 155 (4.5 m) 155 (®rer position) 155 (®rer position) 165±180 (®rer position) 172±175 166±176 (®rer position) 182 (®rer position) 172 150 (2 m) 152 (®rer position) 159 (®rer position) 176 (2 m behind) 164 (®rer position) 149±161 (1 m)
Duration
Reference
2.5 ms A 2.2±3.5 ms A 8.9±13.3 ms B 2.4±5.6 ms A 2.5 ms A
15 16
0.4 ms A 0.3 ms A 0.3 ms A
2 ms A±10 ms B 2.7 (Hodge) 0.3 ms A
4 ms (Hodge)
Where available the position of the acoustical measurements have been indicated
17 18 19 20 18 21 17 22 22 23 17 24 25 21 21 25 17
The performance of hearing protectors
175
Impulses commonly found in military environments have been studied considerably to try and relate the impulse characteristics to hearing loss. The characteristics of many weapons are described in Table 2. The duration data from Hodge and Soderholm17 is calculated using the time of 20 dB below (before) the peak to the time when the pressure envelope has decreased to 20 dB below the peak. See Table 2.15±25 Categorization of industrial and military impulse noise To ensure the attenuation of hearing protectors in impulse noise environments is calculated correctly, the impulse source used in the laboratory has to resemble the impulse for which the hearing protector is to be used against. To do this one must be able to describe the impulses found at the workplace and relate laboratory impulse sources to common industrial and military impulses. To simplify the discussion, the impulse noises described in this chapter have been categorized into four dierent types. Type 1 These are high amplitude, long B-duration impulses which have most of their acoustic energy distributed in lower frequency ranges (i.e. below 2 kHz). Noises that fall into this category include drop forges, punch presses and stamping. Peak levels range from 125 to 160 dB, with B-durations greater than 30 ms. Type 2 Type two noises are typi®ed by hammer noise, nail shooting, impact welder and staplers and have high peak level, short B-durations and most of the acoustic energy distributed in the higher frequency ranges (i.e. above 2 kHz). Peak levels range from 125 and 150 dB, with B-durations of between 2.5 and 30 ms. Type 3 These are Friedlander-type waves produced by large calibre weapons such as large howitzers, cannons and explosives. Peak levels can reach 190 dB with A-durations between 2 and 6 ms. The majority of the acoustic energy is concentrated into the lower frequencies. Type 4 Again these are Friedlander-type waves produced this time by small-calibre weapons such as pistols and ri¯es. Peak levels range from 150 to 172 dB with A-durations between 0.2 and 2 ms. The acoustic energy is predominantly in the high frequency end of the spectrum.
176
D. Smeatham, P. D. Wheeler
LABORATORY SOURCES A literature survey and measurement programme have been carried out to review methods of producing impulse noise in laboratory conditions. The information from this survey and measurement programme and a survey of military and industrial impulse noises have been consolidated in an attempt to model industrial and military impulse with sources reproducible in laboratories. The literature survey has reviewed a number of dierent methods of producing impulse noise, most of these methods have been developed to facilitate the measurement of hearing loss due to impulse noise. Impulse sources utilise one or more of the following techniques using mechanical, electromechanical or electro-mechano-acoustic methods. The impulse producing methods are described below including their classi®cation. Compressed air release (classi®cationÐtype 3 and 4) In this method a sound wave is generated by rupturing a diaphragm situated between a compression chamber and an expansion chamber. Hamernik et al.26 used two dierent methods of generating impulses using compressed air. The ®rst used a bursting diaphragm and the other a rapid-acting valve to initiate the shock front. Peak levels of 160 dB where achieved from all methods, they were found to be repeatable and adjustable. It was noted that the output from the plate rupturing was similar to a 105 mm Howitzer and the output from the rapid acting valve was similar to an M-16 ri¯e. At PTB Fedtke and Richter27 developed a foil blaster for generating impulsive noise. This device consists of a cylindrical tube ®lled with compressed air and terminated by a thin foil. To shape the pulse, another tube with an open end is mounted in front of the in¯ated one. The foil bursts after having been scratched with a sharp needle. At an air pressure of 2.05 bar, peak sound pressure levels of 170 dB at a distance of 0.5 m were achieved with typical A-durations of 1 ms. The system repeatability was 0.7 dB. Cricket (classi®cationÐtype 4) A piece of concave spring steel is fastened at one end to a holder. When the cricket is squeezed the steel snaps abruptly to its other stable position (i.e. concave or convex) generating a loud click. Ward et al.28 used this method to measure the TTS of humans subjected to impulse noise. Peak
The performance of hearing protectors
177
levels of 145 dB were recorded which could be controlled by varying the distance between source and receiver. The signal's time-domain content was modi®ed by changing the cricket size. Hammer and plate (classi®cationÐtype 2) The impact produced by a hammer falling on a plate can be used to generate an impulse whose peak level and duration can be varied. Tremolieres and Hetu29 used a hammer beating a metal plate and a hollow metal rod in a highly reverberant room. Dierent temporal patterns of acoustic waveforms were obtained either by (a) varying the distance between the location of the shock and the recording microphone, (b) by damping out the oscillations of the metal structures caused by the shock, or (c) by varying the weight of the hammer. Peak levels of 137 dB were produced with `B' durations of 140±470 ms. Loudspeaker (classi®cationÐtype 1±4) Loudspeakers have been used as the `front end' of many dierent impulse sources. Although loudspeakers oer good repeatability and, with the aid of signal processing, more accurate impulse source modelling, the limitations of the loudspeaker with respect to frequency response, directivity, power handling capabilities and the dynamic range required to achieve high level impulse noise severely restricts the use of conventional ampli®er/loudspeakers con®gurations. The driving signals come from a number of dierent sources including synthesized signals,30±33 actual recordings34 and electrical discharge.28 Although loudspeaker systems can be used to investigate the characteristics of hearing protector devices at lower levels, they have limited use for generating impulses in excess of 140 dB. Spark discharge (classi®cationÐtype 4) These devices produce an arc between two electrodes, the resulting pressure pulse is dependent on the voltage and spacing between the electrodes. The discharge system has been developed over the years35±37 but the principle of operation has not altered much. The central part of Brinkmann's device is a 100 mF capacitor that is charged by a voltage of 3.5 kV. Two brass bars are mounted on top of the capacitor and held in a polyethylene block. Two tungsten electrodes are mounted in the brass bars, the distance between the ends being variable from 6 to 12 mm. The system was found to be reproducible and peak levels of 155.5 dB are produced.
178
D. Smeatham, P. D. Wheeler
Pistol (classi®cationÐtype 4) The best method of producing impulses to model military noise is obviously to use the weapon itself. However, this is not always possible due to the practicalities of ®ring the weapons under the correct measurement conditions. Starting pistols have been used as impulse sources to study the eect of impulses on TTSs.38 Peak level variations can be made by moving the subject away from the source. Lamothe and Bradley39 studied the acoustical characteristics of various types of guns with dierent calibres. They conclude that a .38-calibre gun ®ring black powder blank cartridges is a suitable source of acoustical impulse. Measurements undertaken in this project using a 9 mm starting pistol ®ring both black and white powder cartridges have produced peak levels of 158 dB at the ®rer position with very high repeatability (dierences less than 3 dB). It has been found that white powder blank cartridges oer greater repeatability than black powder cartridges. Propane cannon (bird scarer) (classi®cationÐtype 1/3) Impulses provided by a propane cannon were studied to determine if they could be used to model impulse noises found in industry. The propane cannon tested runs o compressed propane gas. The gas ®lls a set of bellows inside the body of the cannon and is then compressed by a mechanical mechanism. Once the gas has been compressed to its set pressure a piezoelectric igniter lights the gas thus causing an explosion which then propagates down a tube and out into free space. Associated with the explosion is the production of exhaust ¯ames therefore all measurements of spectra and time histories are taken from approximately 1 m behind the cannon. Peak levels of 147 dB were recorded using a B and K 6 mm microphone and a FM recorder. Initially the propane cannon produces a `mued' impulse noise, but after approximately four impulses the propane cannon produces louder more crisp impulses which are repeatable (with dierences less than 4 dB) once the device is `warm'. Pyrotechnic devices There are a wide variety of pyrotechnic devices which are capable of producing short duration Friedlander impulses. However the safety issues associated with their use within laboratory conditions make them unsuitable for acceptance as `standard' sources, given that alternatives exist.
The performance of hearing protectors
179
CONCLUSIONS The purpose of this study was to facilitate the comparison of typical industrial and military impulse noise to that reproducible in a laboratory by means of categorization of the dierent types of impulse noise. This categorization enables us to de®ne laboratory impulse noise sources which can then be used to study the non-linear attenuation characteristics of hearing protectors in a laboratory environment. A literature survey on the characteristics of impulse noise in the workplace (both industrial and military) has shown that they can be grouped within four categories. These categories are brie¯y described below. Type 1 sources are long B-duration impulses which have most of their acoustic energy distributed in lower frequency ranges (i.e. below 2 kHz). Type 2 sources have short B-durations with most of the acoustic energy distributed in the higher frequency ranges (i.e. above 2 kHz). Type 3 sources are Friedlander type waves with long A-durations between 2 and 6 ms, with the majority of the acoustic energy concentrated in the lower frequencies. Type 4 sources are Friedlander type waves with short A-durations between 0.2 and 2 ms, with the majority of the acoustic energy in the high frequency part of the spectrum. Type 1 and 2 sources are commonly found in industry whereas type 3 and 4 sources are most commonly encountered in a military, mining and quarry environments. A literature survey on methods of producing impulse noise in laboratory environments has been carried out to model impulse noises found in the workplace to facilitate the measurement of hearing protector attenuation against impulse noise. Methods of modelling three of the four types of impulse noise have been found, however methods of modelling type 4 impulses which are produced by large calibre weapons have not been found. The peak levels and the extent of the low frequency energy produced by such weapons is extremely dicult to produce in a laboratory environment. However some success using a membrane device for medium calibre weapons is reported. For testing lower level impulses, loudspeaker systems can be utilized which can produce impulse noise with various characteristics, although at restricted peak levels. REFERENCES 1. Directive 86/188/EEC. On the protection of workers from the risks related to exposure from noise at work. European Commission, Brussels. 2. Shaw, E. A. G., Hearing protector attenuation: a perspective view. Applied Acoustics, 1979, 12, 139±157.
180
D. Smeatham, P. D. Wheeler
3. Wheeler, P. D. and Halliday, S. J., The active noise reduction system for aircrew helmets. NATO AGARD Conference Proceedings, CP-311. 4. Akay, A., A review of impact noise. J. Acoust. Soc. Am., 1978, 64(4), 977±987. 5. Dym, C. L., Sources of industrial impact/impulsive noise. Noise Control Engng, 1977, 8(2), 81±87. 6. Bruel, P. V., Do we measure damaging noise correctly? Noise Control Engng, 1977, 8(2), 52±60. 7. Sulkowski, W. J. and Lipowczan, A., Impulse noise-induced hearing loss in drop forge operators and the energy concept. Noise Control Engng, 1982, 18(1), 24±29. 8. Taylor, W., Noise levels and hearing thresholds in the drop forging industry. J. Acoust. Soc. Am., 1984, 76(3), 807±819. 9. Pekkarinen, J., Exposure to impulse noise, hearing protection and combined risk factors in the development of sensory neural hearing loss. University of Kuopio, 1989. 10. Coleman, R. B., Hodgson, T. H. and Bailey, J. R. Punch press noise signature analysis using digital signal processing techniques. Noise-Con '81, North Carolina State University, Raleigh, NC, 1981, pp. 275±278. 11. Behar, A., Control of impulse noise in punching operations. Noise-Con '81, North Carolina State University, Raleigh, NC, 1981, pp. 271±274. 12. Richards, E. J. and Stimpson, G. J., On the prediction of impact noise, part IX: the noise from punch presses. J. Sound Vibr., 1985, 103(1), 43±81. 13. Passchier-Vermeer, W., Burg, R. V. D. and Leeuw, R., Measurement of impulse noise at workplaces: relation between results from oscilloscope measurements and from measurements with a precision sound level meter. International Symposium on Eects of Impulse Noise on Hearing, Malmo, Sweden, 1980, pp. 85±99. 14. Erdreich, J., A distribution based de®nition of impulse noise. J. Acoust. Soc. Am., 1986, 79(4), 990±998. 15. Forrest, M. R., The eciency of hearing protection to impulse noise. Scand. Audiol. Suppl., 1980, 12, 186±193. 16. Carter, N. L., Eectiveness of foam earplug hearing protection for artillerymen ®ring L118/119 105 mm Howitzers. Military Med., 1989, 154, 473±476. 17. Hodge, D. C. and Soderholm, R. B., A survey of instrumentation for producing impulse noise. US Army Human Engineering Laboratories. Aberdeen Proving Ground, MD, 1964, Technical note 3±64. 18. Price, G. R., Relative hazard of weapons impulses. J. Acoust. Soc. Am., 1983, 73(2), 556±566. 19. Forrest, M. R., Hearing protection against impulsive noise. British Acoustical Society Meeting, 1972, pp. 1±4. 20. Paakkonen, R., Anttonen, H. and Niskanen, J., Noise control on military shooting ranges for ri¯es. Appl. Acoustics, 1991, 32, 49±60. 21. Ylikoski, J., Pekkarinen, J. and Starck, J., The eciency of earmus against impulse noise from ®rearms. Scand. Audiol., 1987, 16, 85±88. 22. Anttonen, H., Sorri, M. & Riihikangas. P., The impulse noise exposure and the eect of protectors in the army. Noise as a Health Problem: Fourth International Congress. Contro Ricerche e Studi Amplifon, Milan, Italy, 1983, Vol. 1, pp. 313±316. 23. Brinkmann, H., Eectiveness of ear protection against impulse noise. Scand. Audiol. Suppl., 1982, 16, 23±39.
The performance of hearing protectors
181
24. Osmundsen, E. and Gjaevenes, K., The noise attenuation of weak shock waves by ear mus. J. Sound Vibr., 1981, 77(3), 375±385. 25. Shenoda, F. B. and Ising, H., Sound attenuation of ear protectors under conditions of impulsive noise: ®eld measurements. Appl. Acoustics, 1988, 23, 297± 307. 26. Hamernik, R. P., Ahroon, W. A. and Hsueh, K. D., The energy spectrum of an impulse: its relation to hearing loss. J. Acoust. Soc. Am., 1991, 90(1), 197±204. 27. Fedtke, Th. and Richter, U., Private communication: Foil blaster for generating impulsive noise, 1992. 28. Ward, W. D., Selters, W. and Glorig, A., Exploratory studies on temporary threshold shift from impulses. Journal, 1961, 33781±784. 29. Tremolieres, C. and Hetu, R., A multi-parametric study of impact noise-induced TTS. J. Acoust. Soc. Am., 1980, 68(6), 1652±1659. 30. Roberto, M., et al., Impact noise and the equal energy hypothesis. J. Acoust. Soc. Am., 1985, 77 1514±1520. 31. Hamernik, R. P., Patterson, J. H. and Salvi, R. J., The eect of impulse intensity and the number of impulses on hearing and cochlear pathology in the chinchilla. J. Acoust. Soc. Am., 1987, 81(4), 1118±1129. 32. Danielson, R., The importance of `temporal pattern' in traumatic impulse noise exposures. J. Acoust. Soc. Am., 1991, 90(1), 209±218. 33. Nicolas, J., Sawan, M., Hetu, R. and Laroche, C., A digitally controlled impulse-noise generation system for the study of the ear response to impulses. Acustica, 1990, 70, 112±126. 34. Walker, J. G. and Behar, A., Problems associated with the reproduction of impulse noise for TTS studies and impulse noise measurements. J. Sound Vibr., 1971, 19(3), 349±354. 35. Loeb, M., Fletcher, J. L. and Benson, R. W., Some preliminary studies of temporary threshold shift with an arc-discharge impulse-noise generator. J. Acoust. Soc. Am., 1965, 37(2), 313±318. 36. Wyber, R. J., The design of a spark discharge acoustic impulse generator. IEEE Trans. Acoust., 1975, ASSP-23 157±162. 37. Brinkmann, H., Impulse noise generator, design and operation. Wehrtechnische dienstelle fur waen und munition, Postfach 1764, D-4470 Meppen, Germany, 1991. 38. Shenoda, F. B., Ising, H. and Fischer, R., Sound attenuation of earmus under conditions of impulse noise-laboratory measurements. Appl. Acoustics, 1987, 21, 295±307. 39. Lamothe, M. J. R. and Bradley, J. S., Acoustical characteristics of guns as impulse sources. Canadian Acoustics, 1985, 13, 16±24.
On the Performance of Hearing Protectors in Impulsive Noise D. Smeatham and P. D. Wheeler University of Salford, Department of Applied Acoustics, Salford M5 4WT, UK (Received 4 September 1995; accepted 2 May 1996)
ABSTRACT The purpose of this study was to facilitate the comparison of typical industrial and military impulse noise to that reproducible in a laboratory by means of categorization of the dierent types of impulse noise. This categorization enables us to de®ne laboratory impulse noise sources which can then be used to study the non-linear attenuation characteristics of hearing protectors in a laboratory environment. # 1998 Elsevier Science Ltd. All rights reserved
INTRODUCTION European Community (EC) Directive 86/188/EEC1 sets advisory and mandatory limits, called action levels. These limits apply to both the daily personal noise exposure of employees and the maximum unweighted sound pressure to which employees may be exposed. The daily exposure action levels are currently 85 and 90 dBA and the peak limit is 200 Pa (140 dB SPL). In order to assess the protection provided by hearing protector devices in relation to the peak limit, information is needed which describes the resultant pressure level underneath a hearing protector when subjected to impulse noise. To fully de®ne the acoustic performance of a hearing protector when subjected to a speci®ed impulse requires a detailed knowledge of the dependence of any non-linearity in its response on the peak level, rise time, and duration of the impulse. In other words, a full parametric analysis of the protector's attenuation performance as a function of these variables is needed. It is often impracticable to measure the attenuation of hearing protectors at a working location. Therefore, laboratory impulse sources need to be developed whose characteristics are similar to those found at the workplace to facilitate laboratory attenuation measurements. 165
166
D. Smeatham, P. D. Wheeler
This paper is based on research conducted for the EC-funded IMPRO project which has been studying the possibility of using impulse noise to assess the non-linear performance of hearing protectors. The paper describes the characteristics of impulse noise and gives an indication of the peak levels and durations of many industrial and military impulse sources. The study has been con®ned to sources which are likely to give rise to peak exposures close to, or exceeding, the 200 Pa limit. In order to make modelling in the laboratory simpler, the impulse noises have been divided into four categories which encompass the range of impulse noises in the military and industrial environments. The paper also discusses the performance of various types of hearing protector on the market today and give some indication of their non-linear behaviour. A number of laboratory impulse noise sources are identi®ed with their peak levels and waveform characteristics shown. RESPONSE OF HEARING PROTECTORS TO IMPULSE NOISE Initially it is worth reviewing the manner in which dierent types of hearing protector devices can be expected to respond to impulse noise. If a hearing protector behaves linearly, the peak level eective to the ear can be calculated from the impulse response of the protector. However there is a body of evidence that even conventional passive hearing protectors exhibit non-linear behaviour when subjected to high-level impulse noise. A classic example of this phenomenon is when the cushion of the ear defender lifts o the wearer's head. Additionally an increasing number of electronic or mechanical leveldependent protectors are being introduced into the marketplace, intended to respond selectively to transient and steady-state noise. In seeking to de®ne test noise sources with which to measure the impulse response of hearing protectors, it is important to take account of the ways in which such devices operate. An increasingly wide range of hearing protector types are now available, and are classi®ed below. Passive ear mus The conventional ear mu, when operating in its linear regime, may be regarded as a low-pass ®lter for impulse noise. In very simplistic terms, when the leading edge of the impulse pressure wave is incident on the protector, the ear mu is set into a damped oscillatory motion on the head, at its natural frequency. This resonant frequency, which is typically in the 100±500 Hz range, is determined by the mass of the ear mu and the compliance of the cushion acting on the ¯eshy area of the head surrounding the pinna.2
The performance of hearing protectors
167
If the excursion of the ear mu on the head is large, a gap may be introduced between the cushion and the head, allowing sound energy to penetrate virtually unattenuated. This is more likely to occur when the impulse noise has either a very high peak level or a large low-frequency component. Passive ear plugs Depending on its construction, the conventional ear plug may be modelled as either: (a) a simple acoustic resistance element located in the ear canal, or (b) a mass±spring system comprising the mass of the ear plug itself reacting with the shear compliance of the plug in the ear canal, and including the shear compliance of the ¯esh. Mechano-acoustic non-linear passive ear mus and plugs Hearing protection devices claiming a non-linear attenuation for impulse noise have been in existence for at least 30 years. The principle of operation is that the incidence of a high-level impulse at the protector gives rise to additional attenuation due to either the generation of turbulence in a small ori®ce or the closure of a valve in the sound path. For steady-state noises, such devices in general provide low attenuation, since the sound wave is able to penetrate the protector without impediment. Tests on such devices in the past have found that, in some cases, non-linear attenuation might only be achieved when the incident peak sound pressure exceeds 160 dB. Since the use of these `claimed' non-linear protectors in the workplace (and in leisure pursuits such as shooting) is increasing, often in circumstances where peak levels may lie in the 140±160 dB range, reliable attenuation data covering this range of sound levels are urgently needed. Active electronic non-linear protectors An increasing number of protectors ®tted with a level-dependent `hearingaid' electronic circuit are being used in the workplace in impulse noise situations. External sounds are detected by a microphone placed on the outside of the protector, whose output is fed, via a level-dependent electronic ampli®er, to an earphone mounted in the protector. More usually, these circuits are incorporated in ear mus, although ear plug devices are under development. The level-dependent electronic circuit may be either a simple peak-clipping circuit, restricting the maximum signal delivered to the earphone to that allowed by the ampli®er voltage supply rail, or a more sophisticated automatic gain control (AGC) compressor circuit in which the level of loud
168
D. Smeatham, P. D. Wheeler
sounds is reduced by a control loop which senses the time-averaged rootmean square (RMS) signal level but which allows quiet sounds to be transmitted without attenuation. In either case, the electronic circuits may allow transmission of external sounds to the ear at (well) above natural levels in the unattenuated mode. Devices providing 20 dB ampli®cation of sounds up to 80 dBA external level have been found during our investigations. These protectors may either provide a single (monaural) ampli®er circuit fed to one ear, or a twin-channel stereo system to aid localization of external sounds. In a related series of tests on such protectors, all the devices investigated were found to eectively limit the transmission of impulse sound to that achievable by the protector used in its passive modeÐeither by the setting of an appropriate maximum feed level to the earphone, or by the use of a fastacting AGC circuit. Active (noise cancellation) protectors This class of hearing protector is ®tted with an `antiphase' acoustic noise cancellation electronic circuit whose purpose is to increase the total attenuation provided by the protector. The term active noise reduction (ANR) is sometimes used to describe such protectors, which were originally developed for applications involving exposure to high levels of pseudosteady-state low-frequency noise (military vehicles and aircraft).3 ANR protectors are generally based on high passive attenuation ear mus although ear plug devices are under development. The ANR electronic circuit comprises a microphone mounted inside the protector which detects the internal sound level, a feedback or compensation ®lter, and ampli®er, and an earphone which transmits the `antiphase' signal (and any speech or communications signals needed). The additional attenuation provided by an ANR circuit is typically up to 20 dB in the 100±500 Hz frequency range, with the possibility of an unwanted `in-phase' component which can reduce eective attenuation by typically 3±5 dB in the 2000±4000 Hz range. Usually, no pre-knowledge of the external sound is required, and the device will attempt to follow variation in the instantaneous sound pressure level at the sensing microphone, within the bandwidth±time constraints of the electronic circuit. Although the bandwidth constraints of the feedback circuit will prevent the system `cancelling' an incident impulse, the system is quite capable of responding to the low-pass ®ltered signal under a circumaural hearing protector. Earlier experiments with a prototype ANR earmu showed that some 10 dB reduction in the single event level (SEL) of a
The performance of hearing protectors
169
pistol shot could be achieved. Maximum reproduced sound levels of typically 130±135 dB will be limited by transducer dynamics, supply voltage rails or protection circuits. Passive protectors with communication/relayed sound In some workplaces, employees may use hearing protectors with `piped' music or with speech communication facilities (hopefully) at levels equivalent to less than the 85±90 dBA action levels. In terms of their impulse response these devices may be expected to behave the same as passive mus. CHARACTERISTICS OF IMPULSE NOISE Impulse noise is produced by a number of dierent phenomena giving the impulse its own characteristics. The nature in which the impulse is developed determines the time-history characteristics of the impulse. For example, the impulse produced by an explosion in free air produces a single spike-type overpressure, whereas the impulse of a drop forge has a characteristic ringing time-history due to various complex resonance eects. Because of the diering time histories, dierent descriptors are used to de®ne the impulse characteristics. The types of time histories and their describers are shown below. Classical Friedlander (A-wave) impulse A single explosion or electrical discharge in free air will generally produce the classical Friedlander wave, shown in Fig. 1. The ideal impulse, for which the risetime (t1±t0) is zero, is described by eqn (1) (see Figs 4 and 5). p t pmax
0
ÿ t ÿt t 2 1
1 ÿ t2 ÿt t1 e
ÿ1 < t < t0 t0 t < 1
1
The time taken for the pressure to ®rst return to zero (at t2) is called the A-duration. The tail of the impulse extends for approximately six times the A-duration before the pressure again returns to zeroÐin practice to less than 1% of the peak levelÐat t3 (see Fig. 1). The spectral distribution of the energy in the impulse is controlled by both the A-duration (TD) and the rise time (TR), as shown in Fig. 2. Peak level, rise time, and duration all depend on the type of sourceÐin the case of explosives, the charge weight, design and containment. Gun®re noise (in free space) also resembles these simple Friedlander waves.
170
D. Smeatham, P. D. Wheeler
Reverberant impact pulse The impulse produced by two metal objects colliding, as in drop-forging and stamping processes, is a far more complex transient decay of energy which is
Fig. 1. Friedlander waveform.
Fig. 2. Spectrum of Friedlander impulse.
Fig. 3. Reverberant impact pulse waveform.
The performance of hearing protectors
171
distributed across the many possible modes of mechanical vibration of the objects in question. It is likely, at its simplest level, to take the form shown in Fig. 3. The decay of the ringing time history is described in terms of its B-durationÐthe time taken for the pressure to fall 20 dB from its peak value. (An alternative term, the D-duration, is sometimes used. This is the time taken to fall 10 dB from the peak level.) The spectral distribution of energy in this impulse is determined both by its overall envelope decay and by the complex 'ringing' infrastructure. It is worth describing brie¯y the dierent mechanisms which produce the ringing time history. Akay4 describes the ®ve fundamental noise generating mechanisms associated with the impact of two bodies. Air ejection When two surfaces approach each other air is ejected from the space between them. When this gap is rapidly narrowed just prior to contact the ¯ow becomes compressible and a pressure gradient is developed in the gaps
Fig. 4. De®nition of C-duration.
Fig. 5. De®nition of B-duration of multiple impulses.
172
D. Smeatham, P. D. Wheeler
between the surfaces. On rebound the newly created gap between the surfaces becomes a low pressure area and air ¯ows inward generating another pressure pulse. The in¯ow also sets the trapped air in the cavity between the surfaces into resonance, causing oscillations in the acoustic ®eld. Rigid body radiation One element in the sound ®eld resulting from the impact of two bodies is due to sudden velocity changes of two bodies on impact. Rigid body radiation is the predominant source of sound when a ball strikes a massive plate. Radiation from the ¯exural vibrations of the thick plate is negligible. Radiation due to rapid surface deformations The deformation of a surface results in the generation of an initial peak sound pressure pulse before the natural mode of radiation of the plate is set up. Pseudo-steady-state radiation Impact processes are generally employed to convert a vast amount of energy into useful work in a very short time. The time span is often too short for the applied energy to be wholly converted to work. The excess energy is absorbed by the mechanical structure. The resulting damped harmonic radiation is referred to as ringing, this generally dominates the total radiation ®eld from forging machinery. This is called pseudo-steady-state radiation in order to distinguish it from harmonically excited steady-state acoustic radiation. Radiation from material fracture The noise resulting from fracture of materials is a consequential source in punch presses and some other impact-forming machines. Eect of environment In the presence of any re¯ecting surface, such as the ground, or in a reverberant space, such as in a factory, there will be additional components in the impulse, produced by re¯ections. If the path lengths involved are short enough, the re¯ections may interfere with the original pulse, producing a complex temporal pattern.The term C-duration is used to describe the total time for which the pulse level exceeds the (peakÐ10 dB) criteriaÐsee Fig. 4. Within a reverberant space such as a factory, a machinery impact noise may now comprise a series of pulses, possibly overlappingÐsee Fig. 5. In such cases, the B-duration relates to the total time for which the level exceeds the peak ÿ20 dB criterion.
The performance of hearing protectors
173
WORKPLACE IMPULSE NOISE Impulse noise is generated in the workplace from sources associated with metal forming and cutting, fabrication and material-dressing, air exhausts and pyrotechnic devices. Although many noise sources to which employees may be exposed in the workplace might be classi®ed as `impulsive' in nature, in the vast majority of cases the peak sound pressure eective to the employee's ear is less than 200 Pa. Such cases need not be considered within the scope of this study. Industrial impulse sources A project undertaken by Bolt Beranek and Newman Inc. and reviewed by Dym5 involved the measurement of a number of impact and impulsive noises commonly found in industry. Information regarding the peak levels of noise are listed in Table 1.5±14 No information on the temporal and spectral characteristics were presented in this report. Other studies of impulse noise are included in the table. The peak levels and duration information provided in Table 1 have been TABLE 1 Source and reference
Duration (mS)
5
Bolt setting gun Hammer on metal plate6 Carpenter's hammer (0.45 kg)5 Piledriver5 Plastic-forming press5 Board hammer (29 000 N)5 Drop forge7 Drop forge8 Drop forge (12-tonne) Drop hammer (11 000 N)5 Pneumatic hammer9 Hand hammering (2 kg on anvil)5 Punch press10, 11, 12 Punch press5 Transistor lead preformer5 Nail shooting13 Shooting6 Stapler14 Impact welder Staple gun5 Power shear5 Pneumatic nailing machine5
B=25
B=60 B=50ÿ100 B=80 130 B=40 128 B=15 B=35 B=10
Peak level (dB) 140 125 147 130 128 145 133.4 160 143 138 140 135 118 140 148 130 155 127 117 148
174
D. Smeatham, P. D. Wheeler
measured at typical operator positions. Table 1 enables us to outline several major sources of impulse noises found in industry. Military impulse sources In general, the military impulse noise environment is characterized by shorter duration, higher peak levels and lower repetition rate than industrial impulse noise environments. The temporal and spectral characteristics of impulses produced by military weapons can be broadly categorized under two dierent classes, that produced by light weapons and that from heavy artillery weapons both exhibiting Friedlander-type waveforms. Light weapons such as pistols and ri¯es have very short rise times (typically 30 s) and short durations (typically 300 s). This sharp time-history produces a spectrum which has a very large bandwidth. Medium and heavy weapon (cannon-type weapons) impulses have longer rise times and duration, therefore producing spectra with large low-frequency components and less energy in the upper frequency ranges. Impulses produced from blasting, quarrying and explosive testing are also characterized in the same manner. TABLE 2 Weapon
Peak (dB)
Howitzer 105 mm
181 167±183 (crew position)
Ri¯e, self-loading RK 7.62 M16 7.62 M14 Bazooka, light Heavy Mortar Antitank Panzerfaust Pistol .38 M/08 9 mm P6 9 mm 9 mm Cannon (130 mm) Machine gun (7.65 mm) M60
175±187 (crew position) 158±165 (5.5 m) 159 (®rer position) 154±158 (2 m) 155 (4.5 m) 155 (®rer position) 155 (®rer position) 165±180 (®rer position) 172±175 166±176 (®rer position) 182 (®rer position) 172 150 (2 m) 152 (®rer position) 159 (®rer position) 176 (2 m behind) 164 (®rer position) 149±161 (1 m)
Duration
Reference
2.5 ms A 2.2±3.5 ms A 8.9±13.3 ms B 2.4±5.6 ms A 2.5 ms A
15 16
0.4 ms A 0.3 ms A 0.3 ms A
2 ms A±10 ms B 2.7 (Hodge) 0.3 ms A
4 ms (Hodge)
Where available the position of the acoustical measurements have been indicated
17 18 19 20 18 21 17 22 22 23 17 24 25 21 21 25 17
The performance of hearing protectors
175
Impulses commonly found in military environments have been studied considerably to try and relate the impulse characteristics to hearing loss. The characteristics of many weapons are described in Table 2. The duration data from Hodge and Soderholm17 is calculated using the time of 20 dB below (before) the peak to the time when the pressure envelope has decreased to 20 dB below the peak. See Table 2.15±25 Categorization of industrial and military impulse noise To ensure the attenuation of hearing protectors in impulse noise environments is calculated correctly, the impulse source used in the laboratory has to resemble the impulse for which the hearing protector is to be used against. To do this one must be able to describe the impulses found at the workplace and relate laboratory impulse sources to common industrial and military impulses. To simplify the discussion, the impulse noises described in this chapter have been categorized into four dierent types. Type 1 These are high amplitude, long B-duration impulses which have most of their acoustic energy distributed in lower frequency ranges (i.e. below 2 kHz). Noises that fall into this category include drop forges, punch presses and stamping. Peak levels range from 125 to 160 dB, with B-durations greater than 30 ms. Type 2 Type two noises are typi®ed by hammer noise, nail shooting, impact welder and staplers and have high peak level, short B-durations and most of the acoustic energy distributed in the higher frequency ranges (i.e. above 2 kHz). Peak levels range from 125 and 150 dB, with B-durations of between 2.5 and 30 ms. Type 3 These are Friedlander-type waves produced by large calibre weapons such as large howitzers, cannons and explosives. Peak levels can reach 190 dB with A-durations between 2 and 6 ms. The majority of the acoustic energy is concentrated into the lower frequencies. Type 4 Again these are Friedlander-type waves produced this time by small-calibre weapons such as pistols and ri¯es. Peak levels range from 150 to 172 dB with A-durations between 0.2 and 2 ms. The acoustic energy is predominantly in the high frequency end of the spectrum.
176
D. Smeatham, P. D. Wheeler
LABORATORY SOURCES A literature survey and measurement programme have been carried out to review methods of producing impulse noise in laboratory conditions. The information from this survey and measurement programme and a survey of military and industrial impulse noises have been consolidated in an attempt to model industrial and military impulse with sources reproducible in laboratories. The literature survey has reviewed a number of dierent methods of producing impulse noise, most of these methods have been developed to facilitate the measurement of hearing loss due to impulse noise. Impulse sources utilise one or more of the following techniques using mechanical, electromechanical or electro-mechano-acoustic methods. The impulse producing methods are described below including their classi®cation. Compressed air release (classi®cationÐtype 3 and 4) In this method a sound wave is generated by rupturing a diaphragm situated between a compression chamber and an expansion chamber. Hamernik et al.26 used two dierent methods of generating impulses using compressed air. The ®rst used a bursting diaphragm and the other a rapid-acting valve to initiate the shock front. Peak levels of 160 dB where achieved from all methods, they were found to be repeatable and adjustable. It was noted that the output from the plate rupturing was similar to a 105 mm Howitzer and the output from the rapid acting valve was similar to an M-16 ri¯e. At PTB Fedtke and Richter27 developed a foil blaster for generating impulsive noise. This device consists of a cylindrical tube ®lled with compressed air and terminated by a thin foil. To shape the pulse, another tube with an open end is mounted in front of the in¯ated one. The foil bursts after having been scratched with a sharp needle. At an air pressure of 2.05 bar, peak sound pressure levels of 170 dB at a distance of 0.5 m were achieved with typical A-durations of 1 ms. The system repeatability was 0.7 dB. Cricket (classi®cationÐtype 4) A piece of concave spring steel is fastened at one end to a holder. When the cricket is squeezed the steel snaps abruptly to its other stable position (i.e. concave or convex) generating a loud click. Ward et al.28 used this method to measure the TTS of humans subjected to impulse noise. Peak
The performance of hearing protectors
177
levels of 145 dB were recorded which could be controlled by varying the distance between source and receiver. The signal's time-domain content was modi®ed by changing the cricket size. Hammer and plate (classi®cationÐtype 2) The impact produced by a hammer falling on a plate can be used to generate an impulse whose peak level and duration can be varied. Tremolieres and Hetu29 used a hammer beating a metal plate and a hollow metal rod in a highly reverberant room. Dierent temporal patterns of acoustic waveforms were obtained either by (a) varying the distance between the location of the shock and the recording microphone, (b) by damping out the oscillations of the metal structures caused by the shock, or (c) by varying the weight of the hammer. Peak levels of 137 dB were produced with `B' durations of 140±470 ms. Loudspeaker (classi®cationÐtype 1±4) Loudspeakers have been used as the `front end' of many dierent impulse sources. Although loudspeakers oer good repeatability and, with the aid of signal processing, more accurate impulse source modelling, the limitations of the loudspeaker with respect to frequency response, directivity, power handling capabilities and the dynamic range required to achieve high level impulse noise severely restricts the use of conventional ampli®er/loudspeakers con®gurations. The driving signals come from a number of dierent sources including synthesized signals,30±33 actual recordings34 and electrical discharge.28 Although loudspeaker systems can be used to investigate the characteristics of hearing protector devices at lower levels, they have limited use for generating impulses in excess of 140 dB. Spark discharge (classi®cationÐtype 4) These devices produce an arc between two electrodes, the resulting pressure pulse is dependent on the voltage and spacing between the electrodes. The discharge system has been developed over the years35±37 but the principle of operation has not altered much. The central part of Brinkmann's device is a 100 mF capacitor that is charged by a voltage of 3.5 kV. Two brass bars are mounted on top of the capacitor and held in a polyethylene block. Two tungsten electrodes are mounted in the brass bars, the distance between the ends being variable from 6 to 12 mm. The system was found to be reproducible and peak levels of 155.5 dB are produced.
178
D. Smeatham, P. D. Wheeler
Pistol (classi®cationÐtype 4) The best method of producing impulses to model military noise is obviously to use the weapon itself. However, this is not always possible due to the practicalities of ®ring the weapons under the correct measurement conditions. Starting pistols have been used as impulse sources to study the eect of impulses on TTSs.38 Peak level variations can be made by moving the subject away from the source. Lamothe and Bradley39 studied the acoustical characteristics of various types of guns with dierent calibres. They conclude that a .38-calibre gun ®ring black powder blank cartridges is a suitable source of acoustical impulse. Measurements undertaken in this project using a 9 mm starting pistol ®ring both black and white powder cartridges have produced peak levels of 158 dB at the ®rer position with very high repeatability (dierences less than 3 dB). It has been found that white powder blank cartridges oer greater repeatability than black powder cartridges. Propane cannon (bird scarer) (classi®cationÐtype 1/3) Impulses provided by a propane cannon were studied to determine if they could be used to model impulse noises found in industry. The propane cannon tested runs o compressed propane gas. The gas ®lls a set of bellows inside the body of the cannon and is then compressed by a mechanical mechanism. Once the gas has been compressed to its set pressure a piezoelectric igniter lights the gas thus causing an explosion which then propagates down a tube and out into free space. Associated with the explosion is the production of exhaust ¯ames therefore all measurements of spectra and time histories are taken from approximately 1 m behind the cannon. Peak levels of 147 dB were recorded using a B and K 6 mm microphone and a FM recorder. Initially the propane cannon produces a `mued' impulse noise, but after approximately four impulses the propane cannon produces louder more crisp impulses which are repeatable (with dierences less than 4 dB) once the device is `warm'. Pyrotechnic devices There are a wide variety of pyrotechnic devices which are capable of producing short duration Friedlander impulses. However the safety issues associated with their use within laboratory conditions make them unsuitable for acceptance as `standard' sources, given that alternatives exist.
The performance of hearing protectors
179
CONCLUSIONS The purpose of this study was to facilitate the comparison of typical industrial and military impulse noise to that reproducible in a laboratory by means of categorization of the dierent types of impulse noise. This categorization enables us to de®ne laboratory impulse noise sources which can then be used to study the non-linear attenuation characteristics of hearing protectors in a laboratory environment. A literature survey on the characteristics of impulse noise in the workplace (both industrial and military) has shown that they can be grouped within four categories. These categories are brie¯y described below. Type 1 sources are long B-duration impulses which have most of their acoustic energy distributed in lower frequency ranges (i.e. below 2 kHz). Type 2 sources have short B-durations with most of the acoustic energy distributed in the higher frequency ranges (i.e. above 2 kHz). Type 3 sources are Friedlander type waves with long A-durations between 2 and 6 ms, with the majority of the acoustic energy concentrated in the lower frequencies. Type 4 sources are Friedlander type waves with short A-durations between 0.2 and 2 ms, with the majority of the acoustic energy in the high frequency part of the spectrum. Type 1 and 2 sources are commonly found in industry whereas type 3 and 4 sources are most commonly encountered in a military, mining and quarry environments. A literature survey on methods of producing impulse noise in laboratory environments has been carried out to model impulse noises found in the workplace to facilitate the measurement of hearing protector attenuation against impulse noise. Methods of modelling three of the four types of impulse noise have been found, however methods of modelling type 4 impulses which are produced by large calibre weapons have not been found. The peak levels and the extent of the low frequency energy produced by such weapons is extremely dicult to produce in a laboratory environment. However some success using a membrane device for medium calibre weapons is reported. For testing lower level impulses, loudspeaker systems can be utilized which can produce impulse noise with various characteristics, although at restricted peak levels. REFERENCES 1. Directive 86/188/EEC. On the protection of workers from the risks related to exposure from noise at work. European Commission, Brussels. 2. Shaw, E. A. G., Hearing protector attenuation: a perspective view. Applied Acoustics, 1979, 12, 139±157.
180
D. Smeatham, P. D. Wheeler
3. Wheeler, P. D. and Halliday, S. J., The active noise reduction system for aircrew helmets. NATO AGARD Conference Proceedings, CP-311. 4. Akay, A., A review of impact noise. J. Acoust. Soc. Am., 1978, 64(4), 977±987. 5. Dym, C. L., Sources of industrial impact/impulsive noise. Noise Control Engng, 1977, 8(2), 81±87. 6. Bruel, P. V., Do we measure damaging noise correctly? Noise Control Engng, 1977, 8(2), 52±60. 7. Sulkowski, W. J. and Lipowczan, A., Impulse noise-induced hearing loss in drop forge operators and the energy concept. Noise Control Engng, 1982, 18(1), 24±29. 8. Taylor, W., Noise levels and hearing thresholds in the drop forging industry. J. Acoust. Soc. Am., 1984, 76(3), 807±819. 9. Pekkarinen, J., Exposure to impulse noise, hearing protection and combined risk factors in the development of sensory neural hearing loss. University of Kuopio, 1989. 10. Coleman, R. B., Hodgson, T. H. and Bailey, J. R. Punch press noise signature analysis using digital signal processing techniques. Noise-Con '81, North Carolina State University, Raleigh, NC, 1981, pp. 275±278. 11. Behar, A., Control of impulse noise in punching operations. Noise-Con '81, North Carolina State University, Raleigh, NC, 1981, pp. 271±274. 12. Richards, E. J. and Stimpson, G. J., On the prediction of impact noise, part IX: the noise from punch presses. J. Sound Vibr., 1985, 103(1), 43±81. 13. Passchier-Vermeer, W., Burg, R. V. D. and Leeuw, R., Measurement of impulse noise at workplaces: relation between results from oscilloscope measurements and from measurements with a precision sound level meter. International Symposium on Eects of Impulse Noise on Hearing, Malmo, Sweden, 1980, pp. 85±99. 14. Erdreich, J., A distribution based de®nition of impulse noise. J. Acoust. Soc. Am., 1986, 79(4), 990±998. 15. Forrest, M. R., The eciency of hearing protection to impulse noise. Scand. Audiol. Suppl., 1980, 12, 186±193. 16. Carter, N. L., Eectiveness of foam earplug hearing protection for artillerymen ®ring L118/119 105 mm Howitzers. Military Med., 1989, 154, 473±476. 17. Hodge, D. C. and Soderholm, R. B., A survey of instrumentation for producing impulse noise. US Army Human Engineering Laboratories. Aberdeen Proving Ground, MD, 1964, Technical note 3±64. 18. Price, G. R., Relative hazard of weapons impulses. J. Acoust. Soc. Am., 1983, 73(2), 556±566. 19. Forrest, M. R., Hearing protection against impulsive noise. British Acoustical Society Meeting, 1972, pp. 1±4. 20. Paakkonen, R., Anttonen, H. and Niskanen, J., Noise control on military shooting ranges for ri¯es. Appl. Acoustics, 1991, 32, 49±60. 21. Ylikoski, J., Pekkarinen, J. and Starck, J., The eciency of earmus against impulse noise from ®rearms. Scand. Audiol., 1987, 16, 85±88. 22. Anttonen, H., Sorri, M. & Riihikangas. P., The impulse noise exposure and the eect of protectors in the army. Noise as a Health Problem: Fourth International Congress. Contro Ricerche e Studi Amplifon, Milan, Italy, 1983, Vol. 1, pp. 313±316. 23. Brinkmann, H., Eectiveness of ear protection against impulse noise. Scand. Audiol. Suppl., 1982, 16, 23±39.
The performance of hearing protectors
181
24. Osmundsen, E. and Gjaevenes, K., The noise attenuation of weak shock waves by ear mus. J. Sound Vibr., 1981, 77(3), 375±385. 25. Shenoda, F. B. and Ising, H., Sound attenuation of ear protectors under conditions of impulsive noise: ®eld measurements. Appl. Acoustics, 1988, 23, 297± 307. 26. Hamernik, R. P., Ahroon, W. A. and Hsueh, K. D., The energy spectrum of an impulse: its relation to hearing loss. J. Acoust. Soc. Am., 1991, 90(1), 197±204. 27. Fedtke, Th. and Richter, U., Private communication: Foil blaster for generating impulsive noise, 1992. 28. Ward, W. D., Selters, W. and Glorig, A., Exploratory studies on temporary threshold shift from impulses. Journal, 1961, 33781±784. 29. Tremolieres, C. and Hetu, R., A multi-parametric study of impact noise-induced TTS. J. Acoust. Soc. Am., 1980, 68(6), 1652±1659. 30. Roberto, M., et al., Impact noise and the equal energy hypothesis. J. Acoust. Soc. Am., 1985, 77 1514±1520. 31. Hamernik, R. P., Patterson, J. H. and Salvi, R. J., The eect of impulse intensity and the number of impulses on hearing and cochlear pathology in the chinchilla. J. Acoust. Soc. Am., 1987, 81(4), 1118±1129. 32. Danielson, R., The importance of `temporal pattern' in traumatic impulse noise exposures. J. Acoust. Soc. Am., 1991, 90(1), 209±218. 33. Nicolas, J., Sawan, M., Hetu, R. and Laroche, C., A digitally controlled impulse-noise generation system for the study of the ear response to impulses. Acustica, 1990, 70, 112±126. 34. Walker, J. G. and Behar, A., Problems associated with the reproduction of impulse noise for TTS studies and impulse noise measurements. J. Sound Vibr., 1971, 19(3), 349±354. 35. Loeb, M., Fletcher, J. L. and Benson, R. W., Some preliminary studies of temporary threshold shift with an arc-discharge impulse-noise generator. J. Acoust. Soc. Am., 1965, 37(2), 313±318. 36. Wyber, R. J., The design of a spark discharge acoustic impulse generator. IEEE Trans. Acoust., 1975, ASSP-23 157±162. 37. Brinkmann, H., Impulse noise generator, design and operation. Wehrtechnische dienstelle fur waen und munition, Postfach 1764, D-4470 Meppen, Germany, 1991. 38. Shenoda, F. B., Ising, H. and Fischer, R., Sound attenuation of earmus under conditions of impulse noise-laboratory measurements. Appl. Acoustics, 1987, 21, 295±307. 39. Lamothe, M. J. R. and Bradley, J. S., Acoustical characteristics of guns as impulse sources. Canadian Acoustics, 1985, 13, 16±24.