University of Tennessee, Knoxville

Trace: Tennessee Research and Creative

Exchange
Masters Theses Graduate School

12-1958

Mutual Coupling Between Loudspeakers

Robert Edward Bodenheimer
University of Tennessee - Knoxville

Recommended Citation
Bodenheimer, Robert Edward, "Mutual Coupling Between Loudspeakers. " Master's Thesis, University of Tennessee, 1958.
http://trace.tennessee.edu/utk_gradthes/2588

This Thesis is brought to you for free and open access by the Graduate School at Trace: Tennessee Research and Creative Exchange. It has been
accepted for inclusion in Masters Theses by an authorized administrator of Trace: Tennessee Research and Creative Exchange. For more information,
please contact trace@utk.edu.
To the Graduate Council:
I am submitting herewith a thesis written by Robert Edward Bodenheimer entitled "Mutual Coupling
Between Loudspeakers." I have examined the final electronic copy of this thesis for form and content and
recommend that it be accepted in partial fulfillment of the requirements for the degree of Master of
Science, with a major in Electrical Engineering.
J. D. Tillman, Major Professor
We have read this thesis and recommend its acceptance:
E. D. Eaves, C. H. Weaver, J. F. Pierce
Accepted for the Council:
Carolyn R. Hodges
Vice Provost and Dean of the Graduate School
(Original signatures are on file with official student records.)
 December 10, 1958

To the Graduate Council:

I am submitting herewith a thesis wr i t t en by Robert

Edward Bodenheimer entitled "Mutual Coupling Between Loud
speakers.n I r e c ommen d that it be a c c e p t 6 d for nine quarter
hours of credit in partial ful fillment of the requirements
for t he degree of Master of Scienee, with a major in Electrical
Engineerinp:.

We have read this thesis

and recommend its acceptance:

, ,ctl8,6!,
..;! '

1!1 q
Uc>--< -4
i '
_L ;..._-----------
__ - ------

-' - (-!-- _!_k?!:::_/ -------- -----

Accepted for the Com1cil:
MUTUAL COUPLING BETWEEN LCUDSPEAKERS

A THESIS

Submitted to
The Graduate Council
of
The University of Tennessee
in
Partial lfillment of the Requirements
for the degree of
Master of Sclence

------

by

Robert Edward Bodenheimer

December,l958
 The author wishes to express his appreciation to Professor

J. D. Till..rnan for sugg e s ti ng the interesting thesis topic <md offering

invaluable assistance durir.g the progress of the thesis; tJ Dr. J. Fo

Pierce for 'is helpful suggestions; to Dr. Eo D. Eaves of the Depart-

ment of Mathematics and to Dr. c. H. 'leaver of the Department of

F.leetricl Engineering for their kind interest and encouragement; and

to Professor P. c. Comll, Head of the D epar tment of Electrical Enein-

eering, for oro vidinP the "'!e<:nB by v..rhich the author co u ld continue his

education.

.) 111 ._()
' .
. ,_
 TABLE OF C:ONTENJ'S

PAGE

I!ITRODLJCTI CN . . * l

DIRECT RADIATOR LOUDSPEAKER , . . . 2

MUTUAL COUPLING BETWEEN DIRECT RADIATOR LOUDSPBAKERS 10

HlJTDAL At:' SELF' IMPEDANCE.::. CF COUPLED LOUDSPEAKliS 16

C Cl C L !J S I 0 S 8 ., " ... ., * 31

REF t Eli C ES . . . .. . . .. .. .. . . . .. . . . . . . . . . . . . "" . .. . ..,

33
 LIST OF FIGURES

FIGURE PAGE

1. A permanent-magnet, direct radiator loudspeaker

3

2. Mechanical airload impedance of a vibrating p iston. 5

3. Electrical equivalent circuit and symbols used for

an analysis of direct radiator loudspeaker....... 7

4. Voice c oil impedance of a direct radiator loud-

speaker. . . . . . . . . . . . . .. . . . . . .. . . . . . 8
5. Piston locati on and geometry f or two pistons mounted

in an infinite baffle . 12
6. Representa tion of a four terminal network 17
t. Block diagram of the circuit used to measure sym

metric and antisy.mme tric impedan ces . 21

8. Self' and mutual impedances for two speaker mounted

in an infinite baffle 23
9. 55 cps directional p attern for speakers mounted in

an infin ite baffle 25

10. 80 cps di recti onal pattern for speaker mou nted in

separate enclosures 26

11. Self and mutual impedan ces for two speakers mounteri

in sep arate enclos ures . 28

12. Hutual impedance for d ifferent speaker separations

when mru nted in 9 . n infinite baffle... 29

INTRODUCTION
-- ,-----

A complete system f or reproduction of s ound in the home

consists of a chain whose first link is ormed by a turntable

pick -up and wh ose final link is made up o f one or more loud-

speakers. Th e loudspeaker is considered am ong the weaker links

of this chain. In fact, the quality of sound from a reprod u c -

ing system can be entirely dependent upon the perf ormance of

ita loudspeaker system.

Only in the last few years have s eparate speakers been

emp loyed t o re produ ce a pa rt of the frequency spectrum. In

present day s ound systems tw o or m or e identical speakers are

often installed t o improve the performance over a part of this

spectrum. Since the speakers w ork into t he same elastic medL.un,

there will be c oupling of the s peak era through the mediu.'11. The

literat,1re is alm ost entirely v oi d of i nformation concerning

this c oupling effect. It is the p urpose of this thesis to in-

v estigat e the mutal coupling between identica l loudspeakers

anj to determine whether or n ot this coupling can be neglected

in t he design of multi-unit speaker s yst ems .

DIRECT RADIATOR LOUDSPEAKER

The direct radiator, permanent ma g net lo u dspea ker has

a r a dia t ing element which ie di re c tly co upled to the air. In

its sim pl est form this type of speaker ma y be represented as

in Figure 1. The first elements obser ved in the ust:.al dyna mic

speaker are the basket, or ho ueing, and th e s pecia l ly treated

paper cone, or diaph ragn, mounted on the front of the basket.

The voice coil is wound on a supp ort i ng former which is rig i dly

attached to the d iaphra gm . Motion of the voice coil i mpa rts

motion to the diaphragm; this in turn i mpa rts motion to the

surr oundi n g mediwn, resulting in the propagation of sound waves.

A spider, or centering device, i s prov ide d to c enter

and to r etain the voice coi l in the circular air gap of a per-

mr.t ma g net . Pr oper alignment insures axia l motion of voice

coil in the air gap of the magnetic circ uit, t hus eli mina ting

wobble. The rim of the c o nic a l diaph ragm is mounted on the

supporting b ask et by means of a mounting ga sket, or flexible

annular ring The corrugations al l ow the cone to be elast

ically suspended a t the rim. This i nsu r e s that the cone 11

l
vibrate in a true axial directionQ

Usua lly the co ne is sufficiently st i f f at the low freq

uencies to move as a uni t . For purposes of a na l ysis , it is

mathematically convenient to consider the loudspeaker d ia phra gm

to be a circular piston of diameter equal to the diameter of

the cone and o sc i llat i ng in a large, thin, r i gi d wall ( infinit e

baffle ). This provides a good approximation to l oudspea ker

performance up to frequencies of the order of 1000 cycles.

 3

11ogn(t/ /rarm
ar;d F-b/e _(!!_./ ,

Corrvgaled

Figure 1. A permanent-magnet, direct radiator

loudspeaker.
Above this frequ ency the diaphragm proceeds to "break up" into

several modes of oscillation and ceases to behave a s a rigid

disk.2

An analysis of the m echanical impedance of the air load

upon one side o f a vibrating piston has been presented very

completely in Lord Rayleigh's3aralysis of sound. The air load

mecGanical impedance has resistive and reactive comp oner. ts of

a rather co mplicated character:

( 1)

where

R = radius of piston

p = density of air

c = velo city o f so und

k - - 2.7T
- c. - ---;\"
f = frequency

= wa v eler:gth

and J1 (2kR) and Kl(2kR) are Bessel functi ons o f the first and

second kind. Figure 2 shows the resistive a n reactive co m-

ponents of the me ch an i c al impedance per unit area o f the piston.

It may be seen fro m Figure 2 that when the wavelength o f t he

radiated srn1nd wave is greater than the circumference of the

pisto n ( Z7TR < 1), the radiation resistance decreases as the

square of the frequency. Above the f requency where ZnR:: 1,
A
 !OO.O j

--l
Figure 2. Mechanical air load resistance

1
I
and reactance per unit area of
. ibra ting in an infinite
a pis t on v
I baffle.
i

I -
. so.o - - -- -- -- --
pc =+I ohms
!
I
i

I
I


200! I
{ I
I

1:
I i
100

"' ,X,
'

5.0 I

IJ I
I

.,

"

\) I

20L_
\J
i
I
I
I
0.02 o.os 0./ 0.2 o.s /.0 2.0 so /0.0

I(R= 2nRh_ 2l7rR/c

==
 '

the radiation resistance is constant. To h ave a constant rad-

iation resistance at 100 cyc l e s would require a cone with a

diamet er of approximately 3 fe et . Since t h er e is a limit to

the physica l size of t he co n e , it is a ppar en t that t he rad-

iation resistance will n ot be constant at low frequencies.

The e quiva l e nt e le ct r ic al eircu i t

4 of the direct rad-

iator l o udspe aker is illust rat ed in Figure 3. Use of the

e qu i v al en t circuit an d e leme n t ar y electromagnet ic theory leads

to an e xpress i o n for t he voice coil imp edance S of t he l oud-

speaker. This imp edce may be written symbolically a s

Zv.c.. = Zc. + Z,.,.o

( 2)

Rc. +-fw Lc:. +

RA + RM +1wM + -.1--r
1 W '-- M

where c. is the elect rical im pedance of the v oice coil in the

absence of motion , an d MOis t h e motional elec t r i c im pe danc e

due t o the movement of the tne c ha nic al system. The motional

impedance is d11e solely to the mo ti on of the voice cotl in

the ma gnetic field and the emf ind11ced in he coil. The ef

fect of this vibrating mechanical system is to intro duc e a

resonant condit ion in th e <'oice coil impedance curve as shown

in Figure 4. At the extreme high frequencies the motional

impe da nc e app roa ch e s zerJ, and the i mpedan ce curve is deter-

mined primarily by the reactance of the vo i ce coil induct ance.

From the e l ec t ri c a l e qu i val en t ci r cui t , the sound power

o utpu t S is given by
 7

Rc = voi ce coil re sis tan c e.

Lc = voice coil inductance.

Zc = voice coil impedance.

M = mass cf \ibrat iq:; system.

= mass or cone and coil + ma&s of air load.

Cm --
con e c omplia.nce of suspensions.

Rm = mecha n i c al resistance of system.

Zm = mechanical impdance,

Ra = air loa d or radiation resistance,

B = air gap maf!Det :ic flux density.

e = length of the voice coil.

F = f'orceexerted en diaphraF,.:m due to a cur

rent in the voice coil.

F. = voltat;e a}:plied to loudspeaker terminals.

v = velocity of the cone.

Figure 3. Electrical equivalent circuit and symbols

for an analytical study of the direct
radiator loudspeaker.


--- -
----

Figure 4 Voice coil impedan ce for

a direc t radiator loud

l
speaker as a function of
frequency.

')

sr
..._
I

>J
4 L '\
.;;:

@-

------ - --

\j

t..,)
<

20 50 100 200 500 /000 2000 5000

Cll)
Fre,juency ;n C.P.S
 9

( 3)
where

r +1wM +- 1
t'wC"'
Low sound power output at th low frequencies is primarily due

to the small radiation mechanical resistt:.nce. The 011tp'tt at

the hirh frequencies is lir.lited by the mechanical mass rea.:t-

ance of the vibrating system. To impro;e the lot- frequency

response, a large diaphragm is required. 'rhis inherently means

a large rna ss, thu.:S pro due in a poor high frequ::;nce response.

or these reasons most systems built for rer!'o:-mance over a

wide frequency ra": e consist.. of two o:- mo 1e 1 ou,"lspeakers, a

large lou.-.l.speaker (a large radtatiol ress:ane) to reproduce

the 1 ow freq1enc les, and a small spea}: er (a small ma s s react-

ance of v i\)rating system ) to reproduce the high frequencieJ *

The principal ,_!:sadvan+.n;es of the direct radiator lotld-

speaker are low efficiency arrl a narrow cirect:tvity pattern

at high frequenc les. The a dvan ta1;es of this t,Tpe speaker,

which lead to its universal acceptance9 are simplicity of

construction, small space requirements, and relatively uni

form response characteristics.S

 In maey l ou dsp eake r ap p l ic a t ions two or more ident leal

loudspeakers ar e used. Ch aracteristics of these mul t i - uni t

yst ems are greater acoustic power out put and wider angle of

coverage. The one quantity l..rhich i s common to both loudspeak-

ers is t he medium. Si nce a radiating loud s p ea ke r whose Stlr-

face is in c on t ac t l..Ti ih an elastic mer'lium will radiate to all

polnts of i
t he me d um , it follows that t here will be mutual coup-

l ing of the loudspeakers throug h the medium. This matual couplinv

will appear to the loudspeakers a s a change in the mechanical

air load impedance.

As s1unin g the loudspeaker can be simulated by a piston

vibrating in an infinit e baffle, the mechanical air load im-

pedance for th is p is t o n can be s o l ved by first s ol v in g the

wave eq uat i on for the velocity p otential p5. The velocity po

tent i al is subject to kn own boundary condition which are the

maximum v e l o c ity of the p is ton , the p ist on radiu s , the freq-

uency of o sci llation of t he pist on, and the v el o c it y of the

baffle. The pressure can be o b t a i n ed in t erms of the velocit y

potential. This pressure is not c cnst a.nt o ver the surfnce of

the piston. Therefore, t he total force of the air load o n t he

pis t on st be fcund by integrating the pressure over the sur-

face afea of t he pist on. The mec han i cal air load impedance
'

A is t hen the rat io of the total reactive force on the pis

t on to the velocity of the p iston. The solution for r_A is

presented in the pr eceding se c t i on

 II

There is particular interest in the derivation o f the

mutual air load impedance between two pistons of ra d ius A

mounted in an infinite b affle . Assume the pistons are sep

arat ed a distance Las shown in Figu re 5(a}. If there is no

mot ion of piston 2, a velocity pot en t i a l

7
al wi ll be set up

at any point in the meditun by the motion of an elementary area

dS of Piston 1. The veloc ity p oten t ial is given by

.f(w"t- k..-)
d.;- u;, ciS
2rrr
f {5)

U0 = velocity amplitude of element ary area dS

= velocity amplitude of Piston 1.

r = distance from e lementa ry area to the point

k =

E
1(wt-_..) = fr eel y traveling, simple harmonic hemispher ical

wave.

Since only thP velo city potential on th e surface of

Piston 2 is of interest in determinjnt:S the mutual air load im-

p edance, the geometry wil l be confined to th e plane of the

baffle and, in par t icu la r , to a point on t he s urface of Piston

2. The p ressure is given in terms of the velocity potential

and densit y of the medium to be

 l
L
l"""""'"- "-----------
i
!
i
P/sloa #/
P;slo/7 #2
E--::-_:- ------=3------ --i-- - - -- ---=t------

tu
------

t u.
'

Figure 5(a).
y

- p

' '7fJ
R 1,_.--:-_-_

___
m
c A -
- Parallel chord- X

i Figure 5 (b)

Figure ) . Two pistons of ra dius a vibrating in an infinite baffle.

N
 13
Using this fundamental expression for pressure, the pressure

at a point on the surface of Piston 2 caused by the sou rc e ele-

ment dS is

.j {wt- kr) (7)

Uo dS

From Figure 5( b), r and dS are defined in terms of the plane

geometry of the pistons to be

dS= a.. da.. d ( 8)

,_z.= R 2. + aw1. - Z I? a.. cas ( - e)

If the assumpti.on is nade that the pressure will be

constant along the chords of Piston 2 rarallel to the y-axis,

th e expression for can br

?
r'-

r
2.
= R 2.+ a..2. - zo rJ a.. cos Jt .

This is truA s1ncB the pressure at po1nt p and the intersecti:,n

of the chord paasinc through point r parallel to they-axis

and the x-axis will be the same. In,;ration over thn surfa.ce

area of Piston 1 loads to the pressure at any point on the

x-axis passine; throneh ?iston 2. The total press'Ire is

j(wt-lc)
(10)
ada.. d'f.
r
 1/.f

The exp ression may be simplified by examin ing the quan-

1
....
tity -
, where
r
I I
r
(11)
LrR +a;, - .2./f'a.. J1.]z
z.
co.s

Factoring out R2 leaves

I I[ 2. ]- !
2 (12)
- I + Rz.
-2
a, ?1.
-
r
=
f( cos r .

I
The expression f or - now becomes
r-

Ff'[
I (2. )]-.!.. (13)
;: :::: ' + ( h - .z h M.JJ . .

If t he above expression i s expanded by the binomial theorem

and t he series rearranged in ascending power of h, t he co-

efficients of h take on a mat hemat ical form known as the

8 I
Legendre Polynomials may now be written as
r

00
I I
r R 2 (M.)F\ { 14}

'l=o

Subst itut ing this into the expression for pressure yields
There is no available means of integratlng the above expression.

Since the pressure at any point on the x-axis through

Piston 2 is knor.n, the force on a chord segment of Piston 2 can

be found. Integration over t he area of Piston 2 gives the,total

reactive force of Piston 2 by Piston 1. The ratio of t he re

active force to the velocity of Piston 2 gives the mutual load

impedance on Piston 2. Note that as s oon as there is motion

of Piston 2, the solution for p is no longer valid since Pis

1
ton 1 is no longer in an absolute infini te baffle. Ho\vever,

as a first approximation, the ass1ption tha t p 1 is unaffected

by the m otion of Piston 2 could be made.

P1 could be in tegrated if it is assumed that t he dis

tance between pistons is large compared with the radius of the

pistons. Since in teret in the mutual coupling is confined

to t he case of close coupling, t hat i, the spacing bEtween

speakers is small, this derivatlon is not pursued='

It should now be apparent that some other scheme ot

determining the mutual coupling m ust be used since the inte

gration for p is very complicated, if not impossible . The

1
method to be used is to measure the self and mutual impedances

at the voice coil terminals of one speaker due to the coupling

or change of the air load impedance by the s econd speaker.

/
 (l
A four terrntn:;J. 11nnnr l j lateral network , no matter

how complex, may be represented by Figure 6. The e;enera l

vol t;a.ge loop equations for this network may be v..rri t ten as

(16)

(17)

14and zz are the self impedances of lo op s 1 an d 2 respet-

lvely. The self impedances are defined as

;!.11 = '
...1,
with loop two open ci rcui t

and Z:.zz, = Vz. w i th loop one open circu.i t.

I.z.
is the impedance reflected i nto loop one by a current

in loop two, and r_2J is the impedance in loop two by a cur

r ent in loop one. P_c cording to thereciprocity theorem, C,z.= ru

if the impedances of the network are linear and bi lateral.

The self andmutual impednnces for most l i near bilateral

circuits are easily determned ty open c irc'lla ting the proper

loop and making the a:rpropriate measurements. It woulrl be

very d i f fi cult to me n s u re the self and mu tual impedances for

a loudspeaker system in this manner sine the lodspeaker must

be ope n c ircnited. To open circuit a. loudspeaker w ould require

the diaphragm to be motionless, that is, the ve locity of the

d i aphragm be equal to zero. For the velocity to be zero, it

would be necessar'Y fo r the speaker to work in to an infinite

air load imp e dan ce .

 17

.-
T

<
l

--- ------- -- - - --- - - - -- - --

Figure 6. Representation of a four

terminal network.
 li

The metbod for determ!ntng the self and nru.tual im-

pedances of coupled loudspeakers is to measuPe the symmetric

10' 11
and antisyrnmetric impedances T:) explain this rnethod,

(18)

(19)

where Vs anc Va are the syrmnetric an: anti sy:unetric vol tagos

respectively. In the same manner, defirAe

-r ::::.
..J. 5
- :z; +Zi,
2 (20)

{21)

Hhere Is and Ia are the s ymme tric an1 antisynmetric currents.

Solving ec;_uations (18) and (19) simultaneously for the applied

voltages and ( 20) an (21) for the loop currnnts yielrl

v, = -t- Va...
Vz. Vs - VQ...
(22)
=

I, = Is +Zo-
Iz. =Is- IQ,
For ident:tcal speakers 11 i equal to :Czz., qnd r,z is equal

to .2';u s i
nc e the loudspeaker ls a linear bila tera1 net\orork.

Therefore, substituting equations { 22) into equations (16) and

( 17) and regrouping the t e rms give

 19

(23)

(2lt)

Addition of (23) and (24) y ield

V.s = I5 ( :c,, + ,z.)

from w hich

.r:
v.
:: rs = r,11 + IZ
(25)

Zs is the syoo;etric impedance and is the input impedance at

e it'l.er o f the loudspeaker term l na. ls when e qual vol t ages are

appl ied in phase a cross the terminals of both loudspeakers.

Hence,zs 1s simply the input impedance at one loudspea ker

when both are driven in phase.

Subtracting eqnat ions (23) and (2l.t-} give the antisym-

metric imp e dance to be

y'a..
Ia..
EE -:ea.,= II - j!-12. {26)
where Za is the antisymmetr1c tmpedance. Za is t he

input impedance at either of the loudspeaker terminals when

equal volta;7es are apylied out of phn to b0th l o u dspeaker

terminals. Thus Za is oota ined by drt ving the speakers out

of phase. Once the symmetric and a nti symme tr ic impedances

are determined, the self and mutual impedances may be obt ained

by solving equations (10) and (11) simultaneously. The re-

sults are
 2P

2!s +-ea.. (27)

.z/1 -

Zs- ;Go.. (28)

;g,z
z
=

The s ymme t ric and antisymmetric im pedan c es may be mea

sured by use of the ele ctr i cal circuit shown in Figure 7(a).

Th e two s peak e r s work into an inf inite medium and are mounted

i n a 4' by 8 sheet of plywood in sach a mariller that the dis

t a nc e between th e speakers is adjusta bl e. This plywo o d baffle

is su ff i ci en t ly large t o approximate an infinite baffle and

to elimi na t e do,lblet a ction. When the ef f ec ti ve pathlength

from the rear t o the front of t he spenker di aphr a gm becomes

smal l compare d with the wavelength of the propagating s ound

wave, the a ppr oximat i o n is no longer v al i d. The problem is

even more complex d ue to the fact t he re are two s pe ak ers mounted

in th e baffle. However, sinc e th e paths existing from the

rear, around the baffle, to the front of t he sp eak er are o f

different lengths, the doublet action will not arrive i n ph as e

even at extremely low audio frequences.

Since the impedances to be me asur ed are complex, it is

not o nly necessary t o measure the magnitude of t his impedance

but also it s phase angle in o r der to determine the resistive

and reactive co mpo nents of the self and m u tu al impedan ces.

This may be a ccomplished with use of a VAW meter, Which will

measure the voltae, current, and power delivered dir ectly

to t he sp eaker input termi nal s . The fr equenc y range over

which the meas ur em en t s were rr.ade is well within the frequency

jl;ewlerl /..J;;Irard I flr.talhlr/f .--Heoll?lrt/1 John Flu/(
1 /'1ode/
IW.de Range
200CO
Osci/lolor
---

1
Node/ wA-P2
Preamp/;(ter
W5M)25 wall
Amplifier ___j Node/ 10 3
VAW Neier

(a) 2 -Jensen Co/icerl

Jl
Senes
11

P/'1. Speakers; Type No. P/2- T

12 9
"'
- wart.

fal/c ----, lllew/eff Paclrard

JT-30C /'1odel 400D
1-------1
Cryslal Vac<.um Tube
j f\1/crophone Vollmel-er

(b)

Figure 7 o (a) Electrical circuit for measuring the setric and

antisymmet ric impedan ces.
(b) Additional circuit required for measuring the
directivity pattern.

 22.

range o f all the t es t e quipment . Me ters e mp loye d in the VAW

meter a re w i thin plus or minus 3% of full scale for any read

ing wi thin t hera ted f reque ncy range (20-20,000 cy cles ) of the

meter.

The self and mu tual i mpedances may be determined as a

funct i on of fre que ncy for a given se parat i on of th e speakers

by apply i ng e q ua t i ons (27) and (28). The res is t ive and react

i ve compone nts of the self and mu tual impe da nc es for th e small

es t speaker sepa ra t io n are shown in Figure 8. From Figre 8

it a p J e ar s that there is a deviation in theresistive component

of the self impedance as c ompa r ed wi th that of an isolated speak

er. T ests show tha t, al thoccgh the spe ake r s are the same model

and type, there is a differe nce in the i ndividual resonant freq

uencies.

The cone velocity qf a s i n gle speaker will be contr olled

by the nmss rea c t a nce above the res ona nt frequency and by the

compliant re ac tan ce bela theresonant fre que ncy . If two s pea k

ers of the sa'!le type are used in the s arne s ys te m, and th ei r

resonant frequencies differ, the n for frequencies between the

resonances, the cone v eloci t y of the speaker with the lower

resonant frequency wlll be rr:a ss c ontrolled. The cone ve l o c i t y

of the speaker with the higher resonant frequency will be con

tr o l le d by the me chani cal compl ia nce. In o ther words, a.l though

the speakers are b eing driven in ph ase , the c i aphracrn of on.

speaker '..till lag t ha t of the o ther. As the t\vO pres sure waves

are not in phase, there w i ll he a Pedution in the pressure

 2() 30 50 /00 200 300 500 /000
6 --
5
I
4

._ ---
If)

3 -R"'.L.-
" "
---

\)
2

0 )(

Figure 8. Resistive

and reactive
of the s e lf I
for
components
I
and mutual impedances
1611 center
to cente
spac ing of the lod
!0 speakers in an nfnitej
1
baffle.

05

-;;:
<::5

x,..
0
--- -
_

,. , _

""" " """" - --- -

--

R,..
' ' /
____...
---
-0.5 ..._/

t1
-10
rrequrznC!f it? CPs
corresponding to the :Hrection of grnatest phase difference.

Since the spacing of th e speakers is small, this effect will

be observed mostly along the axis of the speak ers. Verifi

cation of this fact is shown in t he directivity pattern of

Figure 9. The pattern is meas'.lred at 55 cps which was be

tween the r esonant frequencies of 51 cps and 60 cp s for the

individual speakers. A crystal microphone and vacuum tube

voltmeter, shown in Figure 7(b) are used in addition to t he

equipment described previously, t o determine the directional

pattern. The response of the microphone is not critical since

the measurements are made for constant freq'Jency.

The investigation of this e ffect is extended by placing

the t wo speakers in separate enclosures, Where the spacing of

the speakers rernains unch anged. Since the enclosure co mplete

ly separates t heont and back of the speaker, the do ublet

action is now eliminated. The added series compliance, due

to an eight cubic foot enclosure, raises t heresonant frequen

cies of t he speakers to 75 cps and 80 cps. The speaker with

the l ower resonant frequency is t hen tnned by decreasing its

enclos ure volume until its resonant frequency corresronds to

that of the speaker with the higher resonant freq'.lency.

The directivity pattern is then obtained at the resonant

frequency (BO cps ) of the system and is shown in Fig1e 10 .

The reduction in pressure, a s measured along the axis of the

loudspeakers when t here is a difference in the resonant freq-

11encles, is eliminated by tuning the speakers to the same resonant

 /Odh

-6(/'.

I
I
I /
-900 -----------'-! ----=:--
-----+ ---- : 900
---"---'---

Figure 9. 55 cps directional pattern for 16" spacing

of speakers located in an infinite baffle.
 /Odb

I
-90 -- \._
___ - ------- - --- --

Figure 10. 80 cps directional pattern for

lb" spacing of speakers located
in separate enclosures.
 Z.1
frequenc y . The compone. nt s of th e self and mutual impeda nces

are shown in Fi gure 11 for the case wh ere t he s peakers are

pl ac ed in s e para t e enclosures. The irregular v ariat ion of

the r e sis ti v e and reactive components of the s elf and mutual

impedances in the region of the resonant fre quen cy is now

elimina te d.

There is also interest in the \'ariatlon of the mutual

impedance as a function of sp eaker s ep aration for a fixed

frequency. The re si sti ve and rea c tive components of the mut-

1.lal impeda nce are shown in Figure 12 for thr e e different

frequencies. The periodicity of the impe dan ce variation with

distance is roughl y sinusoid al. As the separation varies up

to a half-wavelength (60 inc he s ) at 101.: cps, the mutual imp ed

ance comp onents vary thr o ugh a half-cycle. At 200 cps the

separat ion varies up to one wavelength (60 inches ) . The rnut

llal imp e d a nc e para."Tleters have a variation of one c omp le te cycle.

It ap p ears then t hat t he sign of the Inutual imp e danc e for any

fr equency depends upon t:1e s pacing of the spe ake rs in wav e

lengths for that p a rt icula r frequency.

When two pistons ar e driven in phase, the pressure at

the f a ce of each piston is in phase. Therefore, if the spac

ing of the sp e a ker s is small compar ed with the wavel ength of

the s ound wave, there will only be a slight phase difference

in the pressures at any point. However, if t here is a speaker

spac ing of a half-wavelength for that s ound wave, t he p r essure

arriving at Piston 1 due to P isto n 2 will be out of phase with

 5

')

3

2 --

- (\____
I

v
--

0
..

0.5 - "--

'?
R
\-::: 0 ,/
/
"

......
' /
...
'

-0.5 ' '- /

20 30 50 /00 200

Frequ2nc(j ;n CPS.
Figure 11. Resistive and reactive com onents of the mutual p,
and self impedances for 16' spacing of the
speakers in separate enclosures when both
speakers are tuned to eo cps resonant freqllenc ies.
 !1t.lu11/ Couplmg
02 2ll

0/ /00 CPS.

"')
0
(
-01

x,..
-02

02

01 i - -- 200 CP5_
....
t.;:
.....
0
(
-01 1 - -- -L -

-02

02
r-
01 400 CP5
II)
0
C
-01

-02
o ;o" 20 30 40' so'' 6o"

Spealter Spacing in inchs

Figare 12. The resistive and reactive components of
the mutual impedance as a function of speaker
spacJ.ng for frequencies o f H:0, 200 and 400 cps.,
Speakers are located in an infinite baffleQ
 jo

the pre ss u r e of Piston 1. Thus it seems that the variation

of the mut'lal coupling as a function of spacing wi 11 be s::>me

type o f damped sinusoid. This correlates the resCJ.lts as meas

ur e d in Figu re 12, wr;ich shows the variation in mutual impedance

as a functi::m of sreaker spacing measured at t he speaker term

inals. In all the above cases the magnttude of the mutual

impedance is small compared with the magnitude of the self

impedance of the loudspeaker.

CONCLU3I ONS
__..,_______ .__

A multi-unit speaker system may be d esign ed by c o ns i der

ing the pe rforman c e of each speaker independently if their resonant

frequencies are the same. The e ff e c t of the mutual coupling be-

tween tl.ro identical s pea k ers can be ent:lrely neglected sincA

the mutual !mpe rnr.ce is small compar0t1 wi. th the self imp eda nc e

of t speakers. If there is a difference in the mechanical

resonances of the individual speHkers, design precautions must

be taJ..en to avo:id p re s s u re reductions in the d ir e c t ivity pat-

terns for frequenci es between the resonan.ces. This effect can

be completely eliminated by mountin the speakers in sepa.r&te

enclcsuren and tun i ng them to t he same resonant frequenc:;'. It

is nece s sary that this effect be e liminated in the stereophonic

reproduction of sound since the phase relationstlip ezistirg be-

tween the two channels is very important.

REFERENCES
 REFERENCES

1. HI-FI LOUDSP EA K ER S & EN CL O SURES (book), ABRAHAM B. COHEN.

John F. Fider Publ13her, Inc., New York-y--:-;- I95'b;-p-p.12-27.

2. PRINCIPLES OF LOUDSPEA K E R DESIGN AND OPERATION, JOSEPH

CHERN OF. I .R .F.
Tre.nsactiohs on Audio,
-
vol . AU-5, no0,
Septemter::ocCob_e_r--1957, p.li1r:---

3. THE THEORY OF SOUND (book), LORD RAYLEIGH. Dover Publications,

Inc., New York, N. Y., vol.Ir;-19;-pp:-162-65.
4. PRA C TICA L ELECTROACOUS'l'ICS (book), MICHAEL RE'M'I NGER.
Chemical Publishing Co., Inc., New - ----
York, N. ----
Y.,- -r=:
19,, p. 63.

5. ACOUSTICAL ENGINEERING (book), HARRY F. OL.SCN. D. Van Nostrand

Company, Inc.> Princeton, N J., 1957, PP 124-31, 134.

6. ELEMENTS OF SOUND R ECORDIN G (book), h_Q...!'B!YN _li.l;{_OL'E. _

John Wiley and Sons, Inc., New York, N. Y., 1950, p. uO'B.

7. P.COUSTIC3 (book ), J OSE PH L. H1JN'T'f?R. Prentice-Hall, Inc. 1

Englevsood Cliffs, N-J-.-; 1956,pi):" 152-62.

8. FLECTROHAGNETIC WA VES AND RADIATING SYSTEHS (book), E.r. JORDAN.

Prentice-Hall, Inc., Englewood Cliffs, N .J., 19SS, pr-l)'fi.:.'K6:-
.

9. LINES ANL FIELDS ( book), JOHN D. RYDER.Prentice-Ha.ll,

P 2-13.-----------
NETWORKS,
Inc New York, N. Y., 1953, P

10. TRANSMISSION-LINE THEORY (book), R ONA LD W. P. K ING. McGraw-

Hill Book Company, Inc., New 'tork,N;Y., 1955, pp-; 293-91;.

11. THE T H EO RY OF LINEAR A N1'E NNA S (book) SON ALI: W. P. K ING.

Harvard University Press, Cambridge, Mass.f9"5b:;- pp; - -)46-47.