Investigation into the feasibility of using a parametric array control source in an active noise control system

Transcription

1 Poceedings of ACOUSTICS Novembe 005, Busselton, Westen Austalia Investigation into the feasibility of using a paametic aay contol souce in an active noise contol system Laua A. Books, Anthony C. Zande, Colin H. Hansen School of Mechanical Engineeing, The Univesity of Adelaide, Austalia ABSTRACT Conventional Active Noise Contol (ANC) systems that minimise pessue at a point in space have typically used loudspeakes as contol souces, which ae vitually omnidiectional within the low fequency ange of inteest. One obvious disadvantage of this is that locations emote fom the desied contol point may actually expeience an incease in sound pessue level. The paametic aay is capable of poducing a highly diectional beam of low fequency sound via the nonlinea inteaction of emitted ultasonic waves with ai. Although significant eseach and development of the paametic aay fo use in audio systems has been undetaken, the feasibility of using a paametic aay as the contol souce in an ANC system has not yet been fully investigated. Within this pape, the theoy govening the opeation of the paametic aay and the esulting estictions upon the poduction of low fequency sound ae discussed. Expeimental testing of a commecially available paametic aay indicates that although highly diective low fequency noise of sufficient amplitude fo some ANC applications can be poduced, thee ae a numbe of pactical concens. The noise floo of the paametic aay is high at low fequencies, esulting in a poo signal to noise atio. The nonlinea natue of the sound poduction pocess means that the pecise contol in amplitude and phase equied fo ANC cannot, as yet, be achieved. In addition, thee ae concens egading the safety of the high amplitudes of ultasound emitted by the paametic aay. These limitations would all need to be ovecome befoe the paametic aay could be successfully implemented as a contol souce in an ANC system. NOMENCLATURE α attenuation coefficient β coefficient of nonlineaity Γ Goldbeg numbe ρ 0 density of the popagation medium ω modulation angula fequency ω 0 pimay souce angula fequency a effective souce adius c 0 speed of sound within medium E envelope function f c pimay souce fequency HD hamonic distotion k wave numbe l a absoption length m modulation index P o amplitude of pimay caie wave p demodulated fundamental fequency pessue p d demodulated fist hamonic pessue Ra Rayleigh distance axial distance fom the souce plane wave shock fomation distance t time INTRODUCTION Conventional ANC systems which minimise pessue at a point have used one o moe loudspeakes as contol souces, which ae vitually omnidiectional within the low fequency ange of inteest. One obvious disadvantage of this is that locations away fom the desied contol point may actually expeience inceases in sound pessue level. To minimise the effect of an ANC system on locations othe than the desied contol point, a diectional souce may be implemented. This can be done using a lage aay of conventional speakes. Howeve, the amount of hadwae equied and the physically lage size of the aay may be geate than is pactical to implement. It would theefoe be pefeable if a highly diectional beam could be poduced fom a single tansduce. It was known as fa back as the 19th Centuy that as a sound wave popagates it also distots (Ingad and Pidmoe-Bown 1956). Non-linea inteaction between two sound waves, of geate than 130 db sound pessue level (SPL), of angula fequency ω 1 and ω causes demodulation to occu, poducing sound waves at the sum ω 1 +ω and diffeence ω 1 - ω fequencies. Fom the 1930s until the ealy 1960s, both theoetical and expeimental analyses of the poduction of extaneous fequencies as a esult of distotion of popagating sound wee undetaken (Ingad and Pidmoe-Bown 1956, Thuas et al. 1935, Westevelt 1957, Bellin and Beye 1960, and Dean 196). Westevelt (196) pesented a theoetical model of what has come to be temed the paametic aay. The paametic aay emits high amplitude ultasonic waves which demodulate into diectional audible sound due to nonlinea inteaction with the medium though which they popagate. A highly diectional beam is theefoe poduced. A paametic aay is able to poduce a seconday sound beam with simila diectivity to that of the pimay caie beam. Fo example, if a 48 khz ultasonic beam is modulated with a 500 Hz beam, the demodulated 500 Hz beam will have diectivity compaable to the 48 khz caie. This diffeence in diectivity between a low fequency audible beam at 500 Hz and a high fequency ultasonic beam at 48 khz is depicted gaphically in Figue 1. The 500 Hz beam is vitually omnidiectional, while the 48 khz beam is highly diectional. Bektay (1965) pesented a quasilinea analysis of the paametic aay. Fom his analysis he was able to pedict the demodulated on-axis fa-field pessue amplitude of a given souce outside the egion of inteaction. Bektay only consideed the possibility of using the paametic aay undewate. Howeve, Bennett and Blackstock (1975) subsequently pesented a successful demonstation of the paametic aay woking in ai. Austalian Acoustical Society 39

2 9-11 Novembe 005, Busselton, Westen Austalia Poceedings of ACOUSTICS 005 Vesions of the paametic aay have been developed and commecialised by both Ameican Technology Copoation (ATC 001) and Holosonic Reseach Labs (HRL 003). Intended applications of these commecial devices include point to point communication ove anges of up to appoximately two hunded metes, non-intusive individual communication to patons/customes in museums and shopping centes, as well as diectional entetainment systems. = ( + α ()) p 1 P0 1 mg t e sin ω whee, () g t = sin ω () c 0 (1) P 0 is the amplitude of the pimay caie fequency at the souce, m is the modulation index, c 0 is the speed of sound within the popagation medium, is the axial distance fom the souce, α is the attenuation coefficient at the souce fequency in Np/m, ω is the modulation fequency and t epesents time. The demodulated seconday fequency pessue is appoximated using Bektay's theoy (Bektay 1965): t 0 0 βp0 a m d p = E ( ) (3) 4 16ρ c α dt (a) whee β is the coefficient of nonlineaity, a is the effective souce adius, ρ 0 is the density of the popagation medium and E is an envelope function. Bektay's appoximation was developed using quasilinea analysis and is only applicable on-axis and in the fa field (outside the nonlinea inteaction egion). An envelope function of the fom: () () E t = 1 + mg t = 1 + msin ω (4) is assumed. Diffeentiating the squae of Eq. (4) twice with espect to time, to enable substitution into Eq. (3), yields: (b) Figue 1. Compaison between diectivity of a) a single souce emitting sound at 500 Hz to b) a souce emitting sound at 48 khz. d dt E () t = mω sin ω + m ω cos ω c 0 (5) The use of paametic aays fo applications such as landmine detection and analysis of undegound stuctues (van Wijk et al. 005), medical focussing (Cahill and Bake 1998) and the detemination of sea bed popeties (Hines 1999) have also been consideed. The feasibility of using a paametic aay as the contol souce in an Active Noise Contol System is yet to be fully consideed. This pape outlines the basic undelying theoy of the paametic aay. A theoetical analysis of the commecially available ATC paametic aay is pesented. Results fom expeimental testing ae also pesented. The diectivity and an on-axis esponse of the souce ae discussed, followed by a discussion on thee main aeas which have the potential to limit the pefomance of the paametic aay in an ANC system: tempoal behaviou of the demodulated signal, hamonic distotion levels and ultasonic levels. BACKGROUND THEORY The pimay caie fequency pessue of a modulated high fequency sound souce is expessed as (Yoneyama et al. 1983): Fom Eq. (5) it can be obseved that components at both the desied fundamental fequency and its fist hamonic ae poduced. Substituting the fundamental fequency component of Eq. (5) back into Eq. (3) yields the demodulated fundamental fequency pessue: βp = 0 a mω p sin ω 4 8ρ 0c0α Fom Eq. (6) it can be obseved that the pessue amplitude of the seconday (audible) fequency, p, is popotional to the squae of the modulating fequency, ω. This esults in a 1 db pe octave oll-off. Hence, at low fequencies, much highe pimay pessue amplitudes ae needed to poduce seconday wave amplitudes compaable to what could be poduced at highe seconday fequencies. Substituting the fist hamonic fequency component of Eq. (5) back into Eq. (3) yields the fist hamonic of the demodulated seconday fequency pessue: βp = 0 a mω p d cos ω 4 8ρ 0c0α (6) (7) 40 Austalian Acoustical Society

3 Poceedings of ACOUSTICS 005 Fom Eq. (6) and Eq. (7) the second hamonic distotion atio can be defined as: pd % HD = 100 = 100m (8) p If the modulation index is set to unity, 100% hamonic distotion occus. Reducing the modulation index educes the pecentage hamonic distotion; howeve the amplitude of the desied seconday fequency is also educed. The hamonic distotion component can also be educed by integating the envelope function twice and taking the squae oot (Kite et al. 1998). This is the pefeed method as it does not advesely affect the amplitude of the desied seconday pessue. THEORETICAL ANALYSIS Peliminay tests Paametic aays wee obtained fom two commecial vendos: Holosonic Reseach Labs (HRL 003) and Ameican Technology Copoation (ATC 001). Peliminay testing of paametic aays fom both of the commecial vendos was undetaken. The paametic aay fom HRL was shown to poduce highe levels of hamonic distotion and theefoe the ATC device was selected fo futhe theoetical and expeimental analysis. The noise floo of both paametic aays was also obseved to be high at low fequencies, esulting in a poo signal to noise atio Novembe 005, Busselton, Westen Austalia 0.15 m and opeates at a pimay fequency of 48 khz. The coesponding absoption length and Rayleigh distance ae 4.3 m and 9.9 m espectively, satisfying the assumption that nonlinea absoption is confined to the nea field. Assuming a 130 db pimay ultasonic souce modulated with a 500 Hz tone, the on-axis sound pessue level (SPL) of the ultasonic caie and the demodulated fundamental fequency wee calculated and ae plotted as a function of ange in Figue 3. The Rayleigh distance and absoption length ae also indicated. A coefficient of nonlineaity of 1., density of 1.1 kg/m 3, and unity modulation ae assumed. Since non-linea inteaction teminates at the Rayleigh distance, Bektay s appoximation is only valid at distances fom the souce geate than this. An ANC system incopoating the paametic aay would most likely have distances of much less than 10 m between the souce and contol locations. Howeve, the pefomance of the aay cannot be accuately modelled within this egion using Bektay s theoy. Outside the egion of non-linea inteaction, the maximum sound pessue level of the demodulated fundamental fequency is only 45 db, well below the contol souce levels desied in an ANC system. Application of Bektay s theoy Demodulation, which occus ove the absoption length, l a, can be visualised conceptually as low fequency sound poduction fom a vitual aay. Non-linea effects cease at a distance temed the Rayleigh distance. A schematic of the poduction of sound fom a paametic aay is included as Figue. Figue 3. On-axis sound pessue level of the ultasonic caie and demodulated fundamental fequency fo a 130 db ultasonic souce modulated with a 500 Hz tone. The quasilinea appoximation upon which Bektay s theoy is based also assumes that the nonlinea effects ae elatively weak, that is, the Goldbeg numbe, Γ, is less than unity (Kim and Spaow 00). The Goldbeg numbe is defined as: Figue. Schematic of the paametic aay. Bektay s theoy assumes that the absoption length is less than the Rayleigh distance at the pimay souce fequency, ensuing that nonlinea absoption is confined to the nea field (Weiguo 003). The absoption length, l a, is simply the invese of the attenuation coefficient: 1 l a = (9) α The Rayleigh distance is defined as: πf a = (10) c Ra c whee f c is the souce fequency. Assuming the ai to be at atmospheic pessue, 0 o C and 0% elative humidity, the themoviscous attenuation coefficient is Np/m (Piecy et al. 1977) and the speed of sound is 343 m/s. The ATC paametic aay has an effective adius of appoximately l Γ = a (11) whee is the plane wave shock fomation distance defined as (Piece 1989): ρc = 0 (1) βkp0 whee k = ω 0 /c and ω 0 is the pimay souce fequency. Substituting Eq. (1) into Eq. (11) yields: ω βp 0 0 Γ = (13) 3 αρ0c0 Peliminay testing indicated that ultasonic levels of between appoximately 130 db and 140 db wee being emitted by the paametic aay egadless of the demodulated fequency and SPL, with coesponding Goldbeg numbes of.0 and 6.4, indicating that the quasilinea appoximation may not be valid fo these cicumstances. In ode to decease the Austalian Acoustical Society 41

4 9-11 Novembe 005, Busselton, Westen Austalia Poceedings of ACOUSTICS 005 Goldbeg numbe to below 1, the pimay ultasonic level would need to be deceased to below 14 db. EXPERIMENTAL ANALYSIS All testing was undetaken in the anechoic chambe at the School of Mechanical Engineeing, Univesity of Adelaide, Austalia. In each expeiment, an HP35665a spectum analyse was used to delive a 100 mvpk sinusoidal input to the paametic aay, well below the maximum specified nominal input level of 1.0 V p-p pe channel. Büel & Kjæ (B&K) type 4133 (half inch) and type 4136 (quate inch) micophones wee used to ecod signals within the audible fequency ange and ultasonic fequency ange espectively. Calibation of the micophones was done using a B&K type 40 piston phone. audible demodulated signal and the fist hamonic of the demodulated signal, which wee measued twice and then aveaged, ae depicted in Figues 5-7 as a function of distance. The theoetical ultasonic caie and fundamental demodulated sound pessue levels, as calculated using Bektay s theoy, ae also shown. It has aleady been mentioned that Bektay s theoy is not necessaily applicable within the expeimental ange o fo the expeimental set-up in geneal. Howeve, it was noted that the expeimental esults do follow the geneal tend of Bektay s pedictions. Diectivity A micophone was set-up to otate at a adius of 1.5 m fom the souce on a B&K type 391 tuntable, which was opeated by a B&K type 305 level ecode. A stationay contol micophone was set up at a distance of 1.7 m fom the souce, along the souce axis. A 500 Hz signal was input to the souce and the otating micophone was swept aound an ac fom -90 o to 90 o elative to the souce. The signal measued by the micophone, which was passed though a low pass filte to eliminate the ultasonic caie, was ecoded at 1 o incements using a PICO logge. A diectivity plot of the ecoded SPL, nomalised to the maximum level, is shown in Figue 4. Fom the figue it can be seen that the sound pessue level deceases by 0 db within 10 o eithe side of the speake axis and by 30 db within 0 o, indicating that the paametic aay is highly diectional at this fequency. Levels of moe than 30 db less than the maximum wee not ecoded due to the noise floo of the measuement technique. The signals eceived by the PICO logging equipment wee ecoded as voltages within a specific ange. The voltage ange could have been changed to measue lowe SPLs and the esults convolved with the oiginal esults, loweing the noise floo, howeve this was not done as levels of moe than 30 db below the maximum ae not consideed significant. Figue 5. On-axis sound pessue level of the ultasonic caie ( ), fundamental audible demodulated fequency (---), and the fist hamonic of the demodulated fequency ( ) fo a modulation fequency of 500 Hz. The theoetical appoximations of the ultasonic caie ( ) and demodulated fundamental fequency (- - ) ae also shown fo compaison. The SPL cuves ae not as smooth as would be desied. One eason fo this is that the demodulation pocess is non-linea with fequency. Anothe eason is that the souce level output by the device was seen to vay tempoally, even when the input fequency and amplitude wee held constant. The SPL of the second hamonic is appoximately 10 db less than that of the fundamental signal fo a ange of geate than 1.5 m fo both the 500 Hz and 1 khz cases, indicating significant levels of hamonic distotion. The 3 khz case shows lowe levels of hamonic distotion, howeve at some anges the SPL diffeence between the fundamental and its fist hamonic is less than 0 db. The ultasonic levels poduced by the souce wee in excess of 130 db at some anges fo all thee cases. Figue 4. Diectivity of a 500 Hz signal at a distance of 1.5m fom the paametic aay. The diectivity of the paametic aay was also measued at fequencies of 300 Hz, 1 khz, khz, 3 khz, 4 khz, 5 khz and 10 khz. These esults ae not shown gaphically due to space limitations; howeve, fo each case, the signal was seen to be highly diectional. Axial SPL The on-axis esponse of the paametic aay was measued expeimentally fo a ange of 0.1 m to 3.6 m in 0.1 m incements at fequencies of 500 Hz, 1 khz and 3 khz. The sound pessue level of the ultasonic caie, the fundamental Figue 6. On-axis sound pessue level of the ultasonic caie ( ), fundamental audible demodulated fequency (---), and the fist hamonic of the demodulated fequency ( ) fo a modulation fequency of 1 khz. The theoetical appoximations of the ultasonic caie ( ) and demodulated fundamental fequency (- - ) ae also shown fo compaison. 4 Austalian Acoustical Society

5 Poceedings of ACOUSTICS Novembe 005, Busselton, Westen Austalia of 50 Hz was input into the speake. Ove a peiod of 30 minutes, the amplitude of the signal poduced by the speake, which was chosen to be appoximately 75 db fo compaability, was obseved to vay by no moe than 0.1 db, indicating that the vaiation was not due to the measuing equipment. The SPL of the fundamental demodulated signal was measued evey 5 seconds until 35 samples wee obtained, fo modulating signals of 50 Hz, 1 khz and 5 khz, to detemine limits on the amount by which the amplitude of the demodulated signal vaies. The SPL of each sample at each fequency is depicted in Figue 9. Figue 7. On-axis sound pessue level of the ultasonic caie ( ), fundamental audible demodulated fequency (---), and the fist hamonic of the demodulated fequency ( ) fo a modulation fequency of 3 khz. The theoetical appoximations of the ultasonic caie ( ) and demodulated fundamental fequency (- - ) ae also shown fo compaison. The tempoal vaiation of the signal poduced by the souce, the high levels of hamonic distotion and the high ultasonic levels wee all highlighted as aeas equiing futhe investigation. Tempoal behaviou of demodulated signal The tempoal behaviou of both the amplitude and phase of the demodulated signal wee investigated. The signal eceived by a micophone at a distance of 1 m fom the souce along the souce axis was passed though a low pass filte and the esulting time tace was obseved on an oscilloscope. A snapshot of the time tace of both a 00 Hz modulation signal input into the souce and the demodulated signal output fom the souce is shown in Figue 8. No diffeence between the phase behaviou of the demodulated signal and that of the input signal was obseved but the amplitude was seen to vay significantly. A vaiation of up to 60% was obseved fom the 1.5 s snapshot; howeve, the amplitude was seen to vay by even geate amounts ove longe peiods of time. Figue 9. Tempoal vaiation in SPL at fequencies of 50 Hz ( ), 1 khz (*) and 5 khz (+). The mean SPL, as well as the ange and standad deviation ae included fo each fequency in Table 1. The amplitude ange and standad deviation ae significant at all thee fequencies, although it can be seen that the vaiabilities ae not as geat in the highe fequency case as they ae at the lowe fequencies. A contol souce expeiencing amplitude vaiations of this level would not pefom well in an ANC system. The contol levels would fluctuate within this ange of vaiability. An adaptive contolle could be used to change the filte weights as the amplitude vaies; howeve, the effects of the filte weights would also be non-linea and subject to amplitude vaiation. It would theefoe still be difficult to eliminate ange fluctuations altogethe. Table 1. Mean SPL, ange and standad deviation of 35 samples measued at 5 second incements, at fequencies of 50 Hz, 1000 Hz and 5000 Hz. fequency (Hz) mean (db) ange (db) standad deviation (db) Figue 8. Time tace of the nomalised input modulation signal ( ) and the nomalised demodulated signal ecoded at a distance of 1 mete fom the souce ( ) fo a 00 Hz modulation signal. In ode to ensue that the amplitude vaiation was due to the paametic aay athe than the associated signal poducing and measuing equipment, the paametic aay was tempoaily eplaced with an odinay loudspeake. A signal It is likely that the most effective means of contolling the souce amplitude lies within the design of the paametic aay itself. It is possible that though bette undestanding of the non-linea phenomenon, paametic aay design may be impoved, amelioating amplitude fluctuations. Hamonic distotion The hamonic distotion poduced by the paametic aay was measued along the axis of the souce at a distance of 1 m fom the souce fo multiple fequencies within the ange Austalian Acoustical Society 43

6 9-11 Novembe 005, Busselton, Westen Austalia Poceedings of ACOUSTICS 005 of 100 Hz to 10 khz. The measued SPLs of the fundamental demodulated signal and the fist hamonic ae depicted in Figue 10. The fist hamonic level is not plotted at fequencies above 6 khz as it was not obsevable above the noise floo. Figue 11. Fequency esponse of the paametic aay. Figue 10. SPLs of the fundamental demodulated signal and its fist hamonic as a function of fequency. At 100 Hz the hamonic distotion is geate than 100%. At fequencies highe than this and up to 1.3 khz, the fist hamonic is, on aveage, slightly moe than 0 db less than the fundamental signal. It is only at fequencies highe than this that the hamonic distotion dops off appeciably. The highe levels of hamonic distotion at low fequencies ae due to the 1 db oll-off pe octave inheent in the pocess. To maintain low fequency SPLs compaable to those in the mid to high audible fequency ange, the modulation index can be inceased (since the demodulated fundamental fequency pessue is popotional to the modulation index). Howeve this compomises the hamonic distotion. It is possible that these levels of hamonic distotion may have little effect upon the peceived sound when the paametic aay is used to poduce music o speech; howeve the effects upon an ANC system, which is geneally used to contol noise within this low fequency ange, would be moe significant. The poduction of additional enegy at fequencies othe than those at which noise contol is being implemented simply add to the total enegy within the egion. Even if noise at the taget fequencies is minimised thee is the potential that the incease in noise level at othe fequencies would be poblematic. Ultasonic levels The fequency esponse of the souce was ecoded at a distance of 1.5 m fom the souce along the souce axis fo thee cases: sound souce tuned on with no input signal; sound souce tuned on with a 500 Hz input signal; and sound souce tuned on with a 3 khz input signal. The fequency esponses ae depicted gaphically in Figue 11 on the one set of axes. The noise floo is also indicated. The SPLs ae simila fo both the 500 Hz and 3 khz modulation envelopes. The esponse fo the 3 khz input signal case shows a geate numbe of distinct peaks. These can be explained by the pe-pocessing technique which is applied within the device to the modulating signal to decease hamonic distotion. A squae oot opeato is applied to the signal befoe it is utilised to modulate the caie in ode to coect fo hamonic distotion. This intoduces multiple sidelobes and theefoe vastly inceases the bandwidth of the emitted signal. The highe the modulating fequency, the geate this bandwidth becomes. Accoding to Howad et al. (005), fo health and safety easons, the level of ultasound to which a peson is exposed should be maintained at levels below 110 db, egadless of the duation of exposue. Fom Figue 11, the ultasonic levels can be seen to peak at appoximately 130 db, fa geate than the ecommended maximum. Even when no modulating signal was deliveed to the souce, the ultasonic caie was still poduced at levels of almost 10 db. This is of paticula concen, as any peson standing in the main beam poduced by the souce will only hea a low level hum fom the souce, mainly emanating fom the cooling fan, and hence they may be completely unawae that they ae being subjected to high ultasonic SPLs and consequently may not take the heaing potection cautions which would be taken if such high SPLs wee being emitted within the audible fequency ange. CONCLUSIONS A theoetical analysis of a paametic aay, which is commecially available fom ATC, has shown that, although the nonlinea absoption is confined to the nea field, the equiement of the quasilinea appoximation that the nonlinea effects being elatively weak ae not met. This indicates that Bektay s theoy may not accuately model the pefomance of the device. Because of the limited size of the anechoic chambe used fo tests, expeimental investigations of the device wee undetaken within anges of less than fou metes of the paametic aay. This distance is well within the egion of nonlinea inteaction, placing futhe doubt on the accuacy of using Bektay s theoy to model the esults. As had been anticipated, the paametic aay was shown to be highly diective, which is an ideal souce chaacteistic in a localised ANC system. Howeve thee exist a numbe of factos which may advesely affect the pefomance of the paametic aay in such a system. The noise floo of the paametic aay was seen to be high at low fequencies, esulting in a poo signal to noise atio. The vaiations in the amplitude of the demodulated signal output by the souce as a function of time wee shown to be significant, esulting in a potential degadation of the pefomance of an ANC system. 44 Austalian Acoustical Society

7 Poceedings of ACOUSTICS 005 It is possible that though bette undestanding of the nonlinea phenomena and the electonic implementation used in the commecially available device, the paametic aay design may be impoved fo ANC system applications, amelioating this poblem. The paametic aay poduces high levels of hamonic distotion within the fequency ange of inteest, which have the potential to advesely affect the pefomance of an ANC system by inceasing the noise level at fequencies othe than the taget contol fequencies. Again it is possible that in the futue, hamonic distotion may be educed without compomising amplitude though impoved paametic aay undestanding and design. The high levels of ultasound poduced by the device, which may potentially have an advese health effect upon exposed individuals, wee also noted. These high levels ae of paticula concen as they ae intinsic in the govening physical sound geneation mechanism and theefoe cannot be educed significantly though impoved paametic aay design. ACKNOWLEDGEMENT Suppot fo this eseach fom the Austalian Reseach Council is gatefully acknowledged. REFERENCES Ameican Technology Copoation, (001) HSS: Technology Intoduction, Technical Repot, Ameican Technology Copoation, USA. Bellin, J. L. S. and Beye, R. T. (1960) Scatteing of Sound by Sound, Jounal of the Acoustical Society of Ameica, vol. 3, no. 3, pp Bennett, M. B. and Blackstock, D. T. (1975) Paametic aay in ai, Jounal of the Acoustical Society of Ameica, vol. 57, no. 3, pp Bektay, H. (1965) Possible exploitation of non-linea acoustics in undewate tansmitting aays, Jounal of Sound and Vibation, vol., no. 4, pp Cahill, M. D. and Bake, A. C. (Septembe 1998) Numeical simulation of the acoustic field of a phased-aay medical ultasound scanne, Jounal of the Acoustical Society of Ameica, vol. 104, no. 3, pp Dean, L. W. (196) Inteactions between Sound Waves, Jounal of the Acoustical Society of Ameica, vol. 34, no. 8, pp Hines, P. C. (Decembe 1999) Quantifying seabed popeties in shelf wates using a paametic sona, Measuement Science and Technology, vol. 10, no. 1, pp Holosonic Reseach Labs (003) Audio spotlight: Use s Manual, Instuction manual. Howad, C. Q., Hansen, C. H. and Zande, A. C. (June 005) A eview of cuent ultasound exposue limits, Jounal of Occupational Health and Safety - Austalia and New Zealand, vol. 1, no. 3. Ingad, U. and Pidmoe-Bown, D. C. (May 1956) Scatteing of Sound by Sound, Jounal of the Acoustical Society of Ameica, vol. 8, no. 3, pp Kim, W. and Spaow, V. W. (00) Audio application of the paametic aay implementation though a numeical model, Poceedings of the Audio Engineeing Society 113 th Convention, Octobe 5-8 Los Angeles, Califonia USA. Kite, T. D., Post, J. T. and Hamilton, M. F. (June 1998) Paametic aay in ai: Distotion eduction by pepocessing, ICA/ASA Poceedings, vol., pp , Seattle WA Novembe 005, Busselton, Westen Austalia Piece, A. D. (1989) Acoustics An Intoduction to its Physical Pinciples and Applications, Acoustical Society of Ameica, New Yok. Piecy, J. E., Embleton, T. F. W. and Sutheland, L. C. (June 1977) Review of noise popagation in the atmosphee, Jounal of the Acoustical Society of Ameica, vol. 61, no. 6, pp Thuas, A. L., Jenkins, R. T. and O Neil, H. T. (Januay 1935) Extaneous Fequencies Geneated in Ai Caying Intense Sound Waves, Jounal of the Acoustical Society of Ameica, vol. 6, no. 3, pp van Wijk, K., Scales, J. A., Mikesell, T. D. and Peacock, J. R. (Januay 005) Towad noncontacting seismology, Geophysical Reseach Lettes, vol. 3. Weiguo, D. (003) Self-demodulation of acoustic waves in ai, Tenth Intenational Congess on Sound and Vibation, July 7-10 Stockholm, Sweden. Westevelt, P. J. (1957a) Scatteing of Sound by Sound, Jounal of the Acoustical Society of Ameica, vol. 9, no., pp Westevelt, P. J. (1957b) Scatteing of Sound by Sound, Jounal of the Acoustical Society of Ameica, vol. 9, no. 8, pp Westevelt, P. J. (196) Paametic Acoustic Aay, Jounal of the Acoustical Society of Ameica, vol. 35, no. 4, pp Yoneyama, M., Fujimoto, J., Kawamo, Y. and Sasabe, S. (May 1983) The audio spotlight: An application of nonlinea inteaction of sound waves to a new type of loudspeake design, Jounal of the Acoustical Society of Ameica, vol. 73, no. 5, pp Austalian Acoustical Society 45