The audio spotlight: An application of nonlinear interaction of sound waves to a new type of loudspeaker design

Masahide Yoneyama and Jun-ichiroh Fujimoto

Application Products Department, Technology Division, RicohCompany Ltd., 3-6, I-chome,Naka-magome, Ohta-ku,Tokyo143,Japan
Yu Kawamo and Shoichi Sasabe

Research Laboratories, Nippon Columbia Company Ltd., 5-1, Minato-cho, Kawasaki-ku, Kawasaki-shi 210, Japan

(Received12 May 1982;accepted for publication17 January 1983)

This work wasdoneto devise a newtypeof loudspeaker. The theoryfor sound reproduction of thisloudspeaker is based on nonlinear acoustics of soundwaveinteraction in air. A finite amplitude ultrasound wavethat canbeamplitude modulated by anyaudiosignal isradiated from a transducer arrayintoair astheprimarywave.As a result, anaudiosignal isproduced in theair
because of the self-demodulation effectof the AM soundwavedueto the nonlinearityof the air. It

ispossible to geta fiatcharacteristic of reproduced sound pressure by using anequalizer. In some fundamental experiments the characteristic of the reproduced soundpressure is not quiteflat due to an imperfect transducer array. Improvement of the transducer makesit possible to geta flat characteristic. A special featureof this loudspeaker is its very sharpdirectivitypattern,which makes it possible to realizea sound spotlight.
PACS numbers: 43.25.Lj, 43.25.Vt, 43.88.Ja
INTRODUCTION

The nonlinearinteraction of finiteamplitudeultrasonic

waves in theaircan beapplied toa loudspeaker. Thispaper

describes thefundamental concept of a loudspeaker based on
the nonlinear interaction of soundwavesin air. Also, someof

Ps -4-
space.

Ir - r' I ct qr,t -- dr', (3)

CO

wherer is the observation pointposition vector,r' is the sourcepositionvector and v is the nonlinearinteraction
'Whentheprimarywaveconsists of two continuous sinusoidal waves andbothareplanarandwellcollimated, the integralof Eq. (3) is calculated in the samemanneras in

the experimental resultsfrom the operationof a prototype loudspeaker are presented.

When two finite amplitudesound waves (primary

waves),havingdifferentfrequencies, interact with one anotherin a fluid, new soundwaves(secondary waves) whose frequencies correspond to the sumand the difference of the primarywaves may be produced asthe result.

previous papers. :'sWhen thedirectivity ofa circular piston

is takeninto consideration, however, Eq. (3)mustbe used with theexpression of Muir et al.6 A newtypeof loudspeaker hasbeen developed on the
basisof the nonlinear interaction of sound waves mentioned

This phenomenon wasfirstanalyzed by Westervelt 2

and is well known as "nonlinear interaction of sound

waves," or the "scattering of sound by sound. "3 Based on Lighthill's arbitrary fluidmotion equation 4asshown in Eq.
(1), Westerveltderivedan inhomogeneous wave equation whichis satisfied by thesound pressure of secondary waves

above. In thistypeof loudspeaker, ultrasound isamplitude

modulatedby an audio signaland radiatedfrom a transducer arrayasfiniteamplitude waves. Whentheamplitudemodulated ultrasound wave interacts is a nonlinear fashion

produced bythenonlinear interaction [Eq.(2)].

in air, the modulated signal{theaudiosignal) can be demodulated in the air.

a2/ CoVp - xic)x a% , ct 2 j

p: density offluid,To:stress.tensor,

(1)

In thefollowing section, theprinciple underlying this

type of loudspeaker is described.
I. THEORY

1 2p,
(2)

A. Acoustic reproduction by nonlinear interaction of

AM ultrasound in air

When two sinusoidal sound waves are radiated in the

Poo

2 '

In Eq. (2),Psisthesecondary wavesound pressure, p isthe primary wave sound pressure, isthenonlinear fluidparameter,andcoisthesmallsignal sound velocity. The solution for Eq. (21 maybeexpressed by the superposition integralof the Green'sfunctionand the virtual sec-

air, two newwaves with angularfrequencies ofo d- a: arise by nonlinear interaction of the two original sinusoidal waves, whose angularfrequencies are(t)1anda2. Thereforeonemight expectthe secondary wavewhich corresponds to the modulation signal,to appearin the air as
a result of the nonlinear interaction between the carrier ul-

ondsource [right side of Eq.(21] asshown in Eq.(3).

1532 J. Acoust. Soc.Am. 73 (5), May 1983

trasound and the lowerand uppersideband waves, provided that a finiteamplitudeAM ultrasound waveis radiatedinto
@ 1983 Acoustical Societyof America 1532

0001-4966/83/051532-05500.80

cessed by an equalizerhaving -- 12 dB/oct frequency characteristics beforethe audiosignalis introduced into the AM
modulator.

B. Harmomc

distort,on

In the caseof pure-tone modulation, g(t ) = sintot,the sound pressures arising fromboththesignal secondary wave and the second harmonicdistortionsignalare calculated from Eqs.(6}and (7},respectively,
FIG. 1. Frequency spectra of an AM waveand demodulated wave.

ps(t } = -- (/3pa2mto2/8poCCtr} sin to(t -- r/Co},

(9)

the air. That is, the AM ultrasoundis self-demodulated by

the nonlinear interaction.

pa(t }= lpa2m2to/8poCCtr} cos 2to(t -- r/Co). (10)

From these equations, it is possible to definethe second
harmonic distortion ratio as follows

Figure 1 shows the spectra for both an AM waveand a

demodulated wave. In this case, since the modulation wave

= [ [pa{t )l/[P,(t)[] x 100= mx 100%.

{11}

is reproduced in the air, a new type of loudspeaker can be devisedif the modulationsignalis selected as the program audio signal. Ifa finiteamplitudeultrasound beam,modulated by an audio signalg(t ), is radiatedinto the air from a transducer array,thesound pressurepl of the primarywave(AM wave} at a distance x from the array on axismay be represented by Eq. (4)

Because thesecond harmonic distortion ratioisproportional to m, a gooddistortionratio requiresa very small modulation depth to preventcrossinteractionbetweenthe lower and uppersideband.waves. The signal and distortion sound
wavesare represented by the first and the second term on the right sideof Eci.(5},respectively. The sound pressure of the signalis proportional to m, whilethe distortion is proportionalto m2.In accordance withthisrelation, ifm isselected lessthan 1, the distortion soundpressure will be muchless than the signalsoundpressure. If the equalizer of -- 12dB/oct is used,the modulation depthm varies with thefrequency of themodulation signal,

Pt = Pc[ 1 + mg(t-- X/Co)] e - x sinwo(t-- X/Co},

(4}

wherepcis the initial soundpressure of the ultrasound, m is the parameterindicatingmodulationindex, and a is the absorptioncoefficient of carrier sound. A virtual audio signal sourceoccursin the primary soundbeambecause of the nonlinearityof the acoustic interactionin air. This sound source mayberepresented by Eq. (5) usingEq. {2)and Eq. (4)
q

asexpressed in Eq.{12} rn = too/to 2, moisconstant.

'

poCo

24

--

bt

co Jl
(5)

In the aboveequation,the secondterm on the right side impliesa harmonicdistortioncomponent arisingfrom the interaction between the lower and uppersideband waves.If theprimarysound beamcross section isassumed to becircular with radiusa, then the demodulated audiosoundpressurepsat the point rfrom the array, on axis,can be calculated analyticallyusingEqs. (3}and (5) in the form

a2rn 02 (t--oo r)' p,_lpo 8pocon aa- g

lPo a2m2 02 r

(6}

On the other hand, the soundpressure of a harmonic distortioncomponentmay be expressed as

Pa--16poCgCtrc)-d(t--c-). (7}
TleFourier transform ofEq.{6} can beexpressed as
P{co} = -- (/Ypg a2m/SpoC arko2exp [ - Ar/CokO ] Gs (o},
wherePs(to} istheFourier transform ofps(t }, andG (to} isthe Fourier transformofg{t }. As evidentfrom Eq. (8},Ps{to} is

proportional totoe andthus thefrequency characteristics of

the reproduced sound showa 12dB/oct dependence. Consequently,the audio signal(modulationsignal)must be pro1533 J. Acoust. Soc.Am.,Vol.73, No.5, May 1983
FIG. 2. Front viewof the loudspeaker.

Yonoyama otaL: Audio spotlight

1533

,2o[
lOO 130

120

lO 20

50
f

lOO
( kHz )

, i,

2)o
(v)

FIG. 3. Soundpressure-frequency response characteristics of the transducerarray,for a point4 m fromthetransducer. The inputvoltage is0.5 V.
the transducer.

FIG. 5. Sound pressure versus inputvoltage at 40 kHz, fora point4 m from

In this case,sincethe secondharmonicdistortionratio E is

proportional to 1/co 2, distortion in low-frequency regions

andthedirectivity at 40 kHz (theprimarywave)of the array areshown in Figs.3 and4, respectively. As canbeseen from thesignal frequency characteristics, is dueto the factthat Fig. 3, the frequency response characteristics of the array are Pa (t)isproportional torn 2even throughps (t)isproportional notsymmetrical for 40 kHz. Moreover, therearemanyharto rn. If rn is keptsmallto makedistortion low, the sound monic resonances and antiresonances. The frequencyrepressure ps(t) alsodecreases. sponse characteristics of the secondary sound wavearedisTherefore,either the initial soundpressure Po of the tributed by the resonances and antiresonances. carrierwaveor theradius of theprimarybeam cross section Figure5 shows thesound pressure at 40 kHz, at a point should increase to maintainthe expected Pt (t). 4 m from the array,plottedagainst input voltage. The sound pressure frequency response characteristics II. EXPERIMENT of the secondary waveproduced by the nonlinear self-interA loudspeaker using a finiteamplitude AM ultrasound action of thefiniteamplitude AM ultrasound radiated from radiatedfrom a transducer array wasdeveloped and put to thearray,areshown in Fig. 6. The characteristics weremeaincreases markedly.

Thateisproportional to 1/co 2,inspite oftheflatness of

practical use.This array consisted of 547 PZT bimorph

transducers. The fundamentalresonantfrequencyof each transducer wasabout40 kHz. A front view of the array appearsin Fig. 2.

suredwith modulation depth m = 0.5ata point 4 m fromthe

arrayin an anechoic chamber. The 12dB/oct equalizer was notused. In thefrequency region below1.5kHz, thecharacteristicsalmostfollow the 12 dB/oct curve.The soundpresThe sound pressure frequency response characteristics surecharacteristics of the primarywavehavea flat region within the frequencies of 40 + 1.5 kHz as shown in Fig. 3. When the sideband spectra of the modulated ultrasound deviatesfrom the flat range,the soundpressure of the secondary wave decreases. The peak of the primary soundpressurecurveat 60 kHz produces the peakof the secondary waveat 20 kHz. All of these phenomena canbe predicted

lOO

90

70

60

(Hz)

FIO. 4. Directivity at 40 kHz of the transducer array, for a point4 m from the transducer. The input voltageis 10 V.

FIG. 6. Soundpressure-frequency response characteristics of secondary wave,for a pointof 4 m, rn = 0.5, and inputvoltage of 10 V.

1534

J. Acoust. Soc. Am.,Vol.73, No. 5, May 1983

Yoneyamaotal.: Audiospotlight

1534

92

FIG. 7. Directivity of secondarywave at 1.0 kHz, for a point of 4 m, m = 0.5, and input voltageof 10 V.

FIG. 9. Directivity of secondary waveat 10.0kHz, for a pointof 4 m,

rn = 0.5, and input voltageof 10 V.

from Eq. (9) and the characteristics of the primary wave. The measureddirectivities of the secondarysignal waves at 1.0, 5.0, and 10.0kHz areshownin Figs.7, 8, and9, respectively. To check the relation betweenthe secondarysignal sound pressureps (t I andsecond harmoniccomponent sound pressure Pa(t) of the secondary wave, the secondary wave pickedup by audiomicrophone wasanalyzedby a spectrum analyzerfor variousvaluesof to. Figure 10 showsthe mea-

surement results atf = 5.0 kHz. Theseresults showthat the

relationof the sound pressure levelbetween signal and distortion arepredicted by Eqs.(9}and(10}. For example, if the results ofm = 1.0andrn = 0.5 arecompared, it isclearthat the signallevel(i.e., 5 kHz I decreases 6 dB and the second

] l0dB

10 (a)

15 kHz

10 (b

15 Id-lz

92 6

10

15 kHz

10

15 '.-Iz

(c)

10

15 kHz

FIG. 8. Directivity of secondary wave at 5.0 kHz, for a point of 4 m, m = 0.5, and input voltageof 10 V.

FIO. 10.Relations of secondary signal sound pressurep, andsecond harmonic sound pressurepa, (a}m = 1.0,(b}ra = 0.7,(c}m = 0.5,(d}m = 0.3,
and {e}rn = 0.1.

1535

J. Acoust. Soc.Am.,VoL73, No.5, May 1983

Yoneyama eta/.: Audio spotlight

1535

co 2characteristics, anequalizer isrequired forfiatresponse.

Usually, it isquite difficult to produce low-frequency sound
because of distortion.

5oce skjnaZ
g(t)

FIO. 11.Construction of theloudspeaker.

Onespecial feature of thisloudspeaker isitsverysharp direefivitypattern.This loudspeaker canbeused asa sound spotlight. Since an acoustic spotlight hasneverexisted in an audible sound region,various uses for thisloudspeaker may be anticipated. For example, the sharpdirecfivity would make it possible to speakto one groupof peoplewithout disturbance to neighboring groups.In a museumor an exhibit, expensive soundbarriersbetween exhibitswould be
unnecessary.

harmonic distortion level(10 kHz) decreases 12dB. Accord-

ingly,thesignal sound pressure isproportional to m andthat

ACKNOWLEDGMENTS

of thedistortion isproportional to m2.

III. DISGUSSION

The authors wishto express theirsincere appreciation

to all the members of the NonlinearAcousticSociety of

panfortheirhelpful comments. In particular, special thanks

are due to Dr. A. Nakamura and Dr. T. Kamakura for their

An entirelynew type of loudspeaker hasbeendeveloped.This research is based on the phenomenon of the nonlinear interaction of sound waves. That is, the self-modula-

tion effect of finite amplitude AM ultrasoundby the nonlinearity of theair hasbeen applied in theconstruction of the loudspeaker. This loudspeaker consists of an ultrasound transducer array,a drivingamplifier for the array,an AM modulator,a pure-tone oscillator for the carrierfrequency andequalizer asshownin Fig. 11. The soundpressure obtained from the loudspeaker is proportional to thedepthm of themodulation. However, m should beassmallaspossible because thesecond harmonic distortion ratio E is equalto m. The sound pressure of the secondary wave isalso proportional to thesquare of theinitial sound pressurePo of thecarriersound andthesquare of the beamradiusa. Thesevalues mustbeaslargeaspossible to obtainadequate sound pressure for practicaluse. Since thefrequency response ofthe secondary wave has

generous discussion. Finally, the authorswishto acknowledgeDr. C. Schueler for hishelpin revising the English of this manuscript.

M. Yoneyama, Y. Kawamo, J.Fujimoro, andS.Sasabe, "Anapplication ofnonlinear parametric interaction toloudspeaker," Meeting ofInstitute.
of Electronics and Communication Engineers of Japan, PaperEA81-65
(19821.

2p.j. Westervelt, "Parametric Acoustic Array,"J. Acoust. SOc. Am. 35,

535-537 0963}.

3R.T. Beyer, "Nonlinear Acoustics," Navel Ship Command {1974). 4M.J. Lighthill, "Onsound generated aerodynamically, I," Proc. R. Soc.
LondonA211, 564-587 {1952).

H. O. Berktay, "ossible exploitation ofnonlinear acoustics inunderwater

transmitting applications," J. Sound Vib. 2, 435-461{1965}. 'T.13.Muir andJ. 121. Willere, "Parametric acoustic transmitting arrays,"
J. Acoust.SOc. Am. $2, 1481-1486(1972).

1536

J. Acoust.Soc. Am., Vol. 73, No. 5, May 1983

Yonoyamapt a/.: Audiospotlight

1536