Tube, Solid State,

Loudspeaker Technology

Article prepared for www.audioXpress.com

Horn Theory:
An Introduction, Part 2
By Bjørn Kolbrek

The author continues his look at the various horn types and how they work.

Spherical Wave Horn and the area of the wave-front with not very different from the throat im-
The spherical wave (or Kugelwellen) height h is 2πr0h. Thus for the area to pedance of a tractrix horn.
horn was invented by Klangfilm, the mo- increase exponentially, h must increase
tion picture division of Siemens, in the exponentially:
late 1940s26, 27. It is often mistaken for h = h0emx (20)
being the same as the tractrix horn. It’s where x is the distance of the top of the
not. But it is built on a similar assump- wave-front from the top of the throat
tion: that the wave-fronts are spherical wave-front and
with a constant radius. The wave-front
4πfc
area expansion is exponential. m= .
c
To calculate the spherical wave horn
contour, first decide a cutoff frequency fc Now that you know the area of the wave-
and a throat radius yo (Fig. 20). The con- front, you can find the radius and the
stant radius r0 is given as distance of this radius from the origin.
S = 2πr0h
c
r0 = (18)
πfc FIGURE 21: Assumed wave-fronts in
y=
S
− h2
(21)
spherical wave horns.
The height of the wave-front at the π
throat is
xh = x – h + h0 (22)
h 0 = r0 − r02 − y 20 (19)
The assumed wave-fronts in a spherical
The area of the curved wave-front at the wavehorn are shown in Fig. 21. Notice
throat is that the wave-fronts are not assumed to
S0 = 2πr0h0 be 90° on the horn walls. Another prop-
erty of the spherical wave horn is that it
can fold back on itself (Fig. 22), unlike
the tractrix horn, which is limited to a
90° tangent angle.
The throat impedance of a 100Hz
spherical wave horn—assuming wave-
fronts in the form of flattened spheri-
cal caps and using the radiation imped-
ance of a sphere with radius equal to the FIGURE 22: Spherical wave horn folding
FIGURE 20: Dimensions of a spherical
mouth radius as mouth termination—is back.
wave horn.
shown in Fig. 23. You can see that it is
audioXpress 2008 1
 it reaches F2, the waveguide acts like a 1P horn for a re-
wave-front area stricted frequency range. Above a certain
has expanded, frequency dictated by throat radius and
and to account horn angle, there will be higher order
for this, a small modes that invalidate the 1P assump-
triangular ele- tions.
ment (or, really, a The contour of the oblate spheroidal
sector of a circle) waveguide is shown in Fig. 26. It follows
b1 is added. the coordinate surfaces in the coordinate
The wave- system used, but in ordinary Cartesian
front expansion coordinates, the radius of the horn as a
f rom b1 (line function of x is given as
3-4) continues in
r = rt2 + tan 2 (θ0 )x 2
element a3, and (23)
FIGURE 23: Throat impedance of a spherical wave horn. an element b2 is where
added to account rt is the throat radius, and
for further wave expansion at F3. The θ0 is half the coverage angle.
Le Cléac’h Horn process is repeated, and the wave-front The throat acoustical impedance is not
Jean-Michel Le Cléac’h presented a horn becomes a curved surface, perpendicular given as an analytical function; you must
that does not rely on an assumed wave- to the axis and the walls, but without find it by numerical integration. The
front shape. Rather, it follows a “natural making any assumptions regarding the throat impedance for a waveguide with
expansion.” The principle is shown in shape prior to the calculations. The wave- a throat diameter of 35.7mm and θ0 = 30
Fig. 24. Lines 0-1 show the wave-front fronts are equidistant from each other, is shown in Fig. 27.
surface at the throat (F1). At the point and appear to take the shape of flattened
spherical caps. The resulting contour of
the horn is shown in Fig. 25.
The wave-front expands according to
the Salmon family of hyperbolic horns.
There is no simple expansion equa-
tion for the contour of the Le Cléac’h
horn, but you can calculate it with the
help of spreadsheets available at http://
ndaviden.club.fr/pavillon/lecleah.html

Oblate Spheroidal
Waveguide
This horn was first investigated by Free-
hafer28, and later independently by Ged- FIGURE 26: Contour of the oblate sphe-
FIGURE 24: The principle of the Le
Cléac’h expansion. des6, who wanted to develop a horn suit- roidal waveguide.
able for directivity control in which the
sound field both in-
side and outside the
horn could be accu-
rately predicted. To do
this, the horn needed
to be a true 1P-horn.
Geddes investigated
several coordinate
systems, and found
the oblate spheroi-
dal (OS) coordinate
system to admit 1P
waves. Putland7 later
showed that this was
not strictly the case.
More work by Ged- FIGURE 27: Normalized throat impedance of a 60° included
FIGURE 25: Contour of a Le Cléac’h horn. des 29 showed that angle infinite oblate spheroidal waveguide.
the oblate spheroidal
2 audioXpress 2008 www.audioXpress .com
 The OS waveguide does not have a angle, and the contour lies inside that of ity performance of a horn, you need the
sharp cutoff like the exponential or hy- the plane-wave exponential horn, being polar plot for a series of frequencies. But
perbolic horns, but it is useful to be able a little longer and with a slightly small- sometimes you also want an idea of how
to predict at what frequency the throat er mouth flare tangent angle (Fig. 28). the coverage angle of the horn varies
impedance of the waveguide becomes too Unfortunately, the Wilson method only with frequency, or how much amplifica-
low to be useful. If you set this frequency corrects the wave-front areas, not the tion a horn gives. This is the purpose of
at the point where the throat resistance is distance between the successive wave- the directivity factor (Q) and the direc-
0.2 times its asymptotic value30, so that fronts. tivity index (DI)34:
the meaning of the cutoff frequency be- There is not much information avail- Directivity Factor: The directivity factor
comes similar to the meaning of the term able about the Iwata horn32, 33, just a is the ratio of the intensity on a given axis
as used with exponential horns, you get drawing and dimensions, but no descrip- (usually the axis of maximum radiation)
0.2c sin θ0
tion of the concept. It looks like a ra- of the horn (or other radiator) to the
fc = (24) dial horn, and seems to have cylindri- intensity that would be produced at the
π rt
cal wave-fronts expanding in area like same position by a point source radiating
You see that the cutoff of the waveguide a hypex-horn with T = √2. The ratio of the same power as the horn.
depends on both the angle and the throat height to width increases linearly from Directivity Index: The directivity index
radius. For a low cutoff, a larger throat throat to mouth. is defined as: DI(f ) = 10 log10 Q(f ). It
and/or a smaller angle is required. For indicates the number of dB increase in
example, for a 1″ driver and 60° includ- SPL at the observation point when the
ed angle (θ0 = 30), the cutoff is about horn is used compared to a point source.
862Hz. Because intensity is watts per square
The advantages of the OS waveguide meter, it is inversely proportional to area,
are that it offers improved loading over a and you can use a simple ratio of areas35.
conical horn of the same coverage angle, Consider a sound source radiating in all
and has about the same directional prop- directions and observed at a distance r. At
erties. It also offers a very smooth transi- this distance, the sound will fill a sphere
tion from plane to spherical wave-fronts, of radius r. Its area is 4πr2. The ratio of
which is a good thing, because most driv- the area to the area covered by a perfect
ers produce plane wave-fronts. point source is 1, and thus Q = 1. If the
The greatest disadvantage of the OS FIGURE 28: Comparison of the expo- sound source is radiating into a hemi-
waveguide is that it is not suitable for nential horn with the tractrix and the Wil- sphere, the coverage area is cut in half,
son modified exponential horn22.
low-frequency use. Bass and lower mid- but the same sound power is radiated,
range horns based on this horn type will so the sound power per square meter is
run into the same problems as conical doubled. Thus Q = 2. If the hemisphere
horns: the horns become very long and is cut in half, the area is 1/4 the area cov-
narrow for good loading. ered by a point source, and Q = 4.
To sum up, the OS waveguide pro-
vides excellent directivity control and
fairly good loading at frequencies above
about 1kHz.
FIGURE 29: Wave-fronts in the Western
OTHER HORNS Electric type exponential horn17.
Three other horn types assuming curved
wave-fronts that are worth mentioning
are: the Western Electric horns, the Wil- DIRECTIVITY CONTROL
son modified exponential, and the Iwata Control of directivity is an important
horn. What these horns have in common aspect of horn design. An exponential
is that they do not assume curved wave- horn can provide the driver with uniform
fronts of constant radius. loading, but at high frequencies, it starts
The Western Electric type horn17 uses to beam. It will therefore have a cover-
wave-fronts of constantly increasing ra- age angle that decreases with frequency,
dius, all being centered around a vertex a which is undesirable in many circum-
certain distance from the throat (Fig. 29). stances. Often you want the horn to radi-
In the Wilson modified exponential ate into a defined area, spilling as little
horn31, the waves start out at the throat sound energy as possible in other areas.
and become more and more spherical. Many horn types have been designed to
The horn radius is corrected in an it- achieve this. FIGURE 30: Contour of the Iwata horn32.
erative process based on the wall tangent For the real picture of the directiv-
audioXpress 2008 3
 For a horn with coverage angles α and where beam widths of a typical multicellular
β as shown in Fig. 31, you can compute fI is the intercept frequency in Hz where horn are shown in Fig. 33. The fingering
Q as the horn loses directivity control, at high frequencies is shown in Fig. 34.
180
Q= (25) x is the size of the horn mouth in mm in The beam width of a multicellular
 α β
sin −1  sin sin  the plane of coverage, and horn with different number of cells is
 2 2
Most constant directivity horns try to θ is the desired coverage angle in degrees shown in Fig. 3534. The narrowing in
act as a segment of a sphere. A sphere in that plane. beam width where the dimensions of the
will emit sound uniformly in all direc- You thus need a large horn to control di- horn are comparable to the wavelength is
tions, and a segment of a sphere will emit rectivity down to low frequencies. evident.
sound uniformly in the angle it defines, Most methods of directivity control
provided its dimensions are large com- rely on simulating a segment of a sphere.
pared to the wavelength11. But when the The following different methods are list-
wavelength is comparable to the dimen- ed in historical order.
sions of the spherical segment, the beam
width narrows to 40-50% of its initial MULTICELLULAR HORNS
value. Dividing the horn into many conduits
A spherical segment can control direc- is an old idea. Both Hanna36 and Slepi-
tivity down to a frequency given as an37 have patented multicellular designs,
25106 with the conduits extending all the way
fI = (26) back to the source. The source consists
x θ
of either multiple drivers or one driver
with multiple outlets, where each horn
is driven from a separate point on the
diaphragm.
FIGURE 33: -6dB beam widths of Elec-
The patent for the traditional mul- tro-Voice model M253 2 by 5 cell horn42.
ticellular horn belongs to Edward C.
Wente38. It was born from the need to
accurately control directivity, and at the
same time provide the driver with proper
loading, and was produced for use in the
Bell Labs experiment of transmitting the
sound of a symphonic orchestra from
one concert hall to another39.
A cut view of the multicellular horn, as
FIGURE 31: Radiation into a solid cone patented by Wente, is shown in Fig. 32.
of space defined by angles α and β.
In this first kind of multicellular horn,
the individual horns started almost paral-
lel at the throat, but later designs often
used straight horn cells to simplify man-
ufacture of these complex horns. As you
can see, the multicellular horn is a cluster FIGURE 34: High frequency fingering of
of smaller exponential horns, each with a EV M253 horn at 10kHz42.
mouth small enough to avoid beaming in
a large frequency range, but together they
form a sector of a sphere large enough to RADIAL HORNS
control directivity down to fairly low fre- The radial or sectoral horn is a much
quencies. The cluster acts as one big horn simpler concept than the multicellular
at low frequencies. At higher frequencies, horn. The horizontal and vertical views
the individual horns start to beam, but of a radial horn are shown in Fig. 36.
because they are distributed on an arc, The horizontal expansion is conical, and
coverage will still be quite uniform. defines the horizontal coverage angle of
The multicellular horn has two prob- the horn. The vertical expansion is de-
lems, however. First, it has the same lower signed to keep an exponential expansion
midrange narrowing as the ideal sphere of the wave-front, which is assumed to
segment, and, second, the polar pattern be curved in the horizontal plane. Direc-
shows considerably “fingering” at high tivity control in the horizontal plane is
FIGURE 32: Multicellular horn38. frequencies. This may not be as serious fairly good, but has the same midrange
as has been thought, however. The -6dB narrowing as the multicellular horn. In
4 audioXpress 2008 www.audioXpress .com
 vertical direction quency, kc = 2πfc .
only, then the di- c
rection of expan- The problem of midrange narrowing
sion is changed. was solved by having a more rapid flare
The wave-f ront close to the mouth of the conical part of
expansion is re- the horn. Good results were obtained by
stricted vertically, doubling the included angle in the last
and is released third of the conical part. This decreases
horizontally. The the acoustical source size in the frequen-
result is that the cy range of midrange narrowing, causing
horizontal pres- the beam width to widen, and removing
sure that builds up the narrowing. The result is a horn with
in the first part of good directivity control down to the fre-
the horn causes
the wave-front to
FIGURE 35: Beam width of a multicellular horn constructed as expand more as it
shown in the insert34. reaches the sec-
ond part. That it
is restricted in the vertical plane helps
further.
Because the wave-front expansion is
to be exponential all the way, the dis-
continuity at the flare reversal point
(where the expansion changes direction)
is small. In addition, the change of cur-
vature at the flare reversal point is made
smoother in practical horns than what is
shown in the figure.

CE HORNS
In the early 1970s, Keele, then working
for Electro-Voice, supplied an answer to
the problems associated with multicel-
lular and radial horns by introducing a
FIGURE 36: Profile of a radial horn42. completely new class of horns that pro-
vided both good loading for the driver FIGURE 37: Wave-front expansion in
addition, there is almost no directivi- and excellent directivity control42. reversed flare horns41.
ty control in the vertical plane, and the The principle is based on joining an
beam width is constantly narrowing with exponential or hyperbolic throat seg-
increasing frequency. ment for driver loading with two conical
mouth segments for directivity control.
REVERSED FLARE HORNS The exponential and conical segments
The reversed flare horn can be con- are joined at a point where the conical
sidered to be a “soft diffraction horn,” horn of the chosen solid angle is an op-
contrary to Manta-Ray horns and other timum termination for the exponential
modern constant directivity designs that horn. Keele defines this as the point
rely on hard diffraction for directivity where the radius of the exponential horn
control. This class of horns was patented is
for directivity control by Sidney E. Levy 0.95sin θ (27)
r=
and Abraham B. Cohen at University kc
Loudspeakers in the early 1950s40, 41. where
The same geometry appeared in many r is the radius at the junction point,
Western Electric horns back in the early θ is the half angle of the cone with solid
1920s, but the purpose does not seem to angle Ω,
be that of directivity control17.
Ω FIGURE 38: Example of the Electro-
The principle for a horn with good θ = cos-1 (1 - ), and Voice CE constant directivity horns. This
2π
horizontal dispersion is illustrated in Fig. horn covers 40° by 20°42.
37. The wave is allowed to expand in the kc is the wave number at the cutoff fre-
audioXpress 2008 5
quency dictated by the mouth size. The Manta-Ray horn incorporates ture of the distortion from the adiabatic
For a horn with different horizontal the same rapid mouth flaring as the CE curve for air (Fig. 40). The undisturbed
and vertical coverage angles, the width horns to avoid midrange narrowing, pressure and specific volume of air ( 1 )
ρ
and height of the mouth will not be but does not use radial expansion of the
equal. The aspect ratio of the mouth will walls. The reason for this is that radial is indicated in the point P0V0. Devia-
be given as walls produced a “waist-banding” effect, tion from the tangent of the curve at
θ in which the horn lost much energy out this point will result in the generation
sin H
XH 2
R= = , (28) to the sides in the upper midrange. This of unwanted frequencies, the peak of
X V sin θH
2 effect cannot be seen in the polar plots the wave being stretched and the trough
or, if θH and θV are limited to 120°, for the CE horns, which suggests that compressed.
“waist-banding” can be a result of the The speed of sound is given as
θ
R≈ H . (29) Manta-Ray geometry, and not solely of p
θV c= γ
radial wall contours. ρ (31)
The lower frequency of directivity con- where
trol will also be dictated by the mouth NEW METHODS γ is the adiabatic constant of air, γ =
aspect ratio. Substituting equation 26 Most newer constant directivity designs 1.403.
into equation 29 and solving for the ratio have been based on either the conical You can see that the speed of sound in-
of intercept frequencies, you get horn, some sort of radial horn (includ- creases with increasing pressure. So for
fIH  θH 
2 ing the JBL Biradial design), or diffrac- the high pressure at the peaks of the
≈ . (30) tion methods such as the Manta-Ray wave-front, the speed of sound is higher
fIV  θ V 
design. The only notable exception is than at the troughs. The result is that as
For a 40° by 20° (H-V) horn, the verti- the oblate spheroidal waveguide (cov- the wave propagates, the peaks will gain
cal intercept frequency will be four times ered previously) introduced by Geddes. on the troughs, altering the shape of the
higher than the horizontal intercept fre- The general trend in horns designed waveform and introducing harmonics
quency. for directivity control has been to focus (Fig. 41).
on the control issue, because it is al-
Manta-Ray Horns ways possible to correct the frequency
The Altec Manta-Ray horn sought to response. A flat frequency response does
solve the problems of the CE horns, not, however, guarantee a perfect im-
mainly the inability to independently pulse response, especially not in the pres-
specify the horizontal and vertical inter- ence of reflections. Reflected waves in
cept frequencies43. To achieve directiv- the horn at the high levels in question
ity down to a lower frequency in the will also cause the resulting horn/driver
vertical plane, the vertical dimension of combination to produce higher distor-
the mouth must be increased. Because tion than necessary, because the driver is
the dispersion angle is smaller, the ex- presented with a nonlinear and resonant
pansion must start further back, behind load. (See next section.)
where the horizontal expansion starts.
The result is the unique geometry shown DISTORTION
in Fig. 39 (although it’s not so unique As mentioned, the horn equation is de- FIGURE 40: Adiabatic curve for air.
anymore). rived assuming that the pressure varia-
tions are infinitesimal. For the intensities
appearing at the throat of horns, this as-
sumption does not hold. Poisson showed
in 1808 that, generally, sound waves can-
not be propagated in air without change
in form, resulting in the generation of
distortion, such as harmonics and inter-
modulation products. The distortion is
caused by the inherent nonlinearity of
air.
If equal positive and negative incre- FIGURE 41: Distorted waveform due to
FIGURE 39: The Manta-Ray geometry43. ments of pressure are impressed on a non-constant velocity of sound5.
mass of air, the changes in volume of
At the point where the horizontal ex- that mass will not be equal. The volume There are thus two kinds of distortion
pansion starts, the wave is diffracted to change for positive pressure will be less of a sound wave: one because of the un-
fill the width of the horn, and dispersion than that for the equal negative pres- equal alteration of volume, and another
is controlled by the horn walls. sure44. You can get an idea of the na- because of the propagation itself. This
6 audioXpress 2008 www.audioXpress .com
last kind of distortion is most noticeable One point I need to mention is the
f
in a plane wave and in waves that expand D2 %  = 1.7310 −2 It (33) importance of reducing the amount of
fc
slowly, as in horns, where distortion in- reflection to reduce distortion. At the
creases with the length propagated. Both where high levels involved, the reflected wave
kinds of distortion generate mainly a It is the intensity at the throat, in watts from diffraction slots or from the mouth
second harmonic component. per square meter. will not combine with the forward prop-
Fortunately, as the horn expands, the Holland et al.45 have investigated the agating wave in a linear manner. The
pressure is reduced, and the propaga- distortion generated by horns both with result will be higher distortion, and a
tion distortion reaches an asymptotic the use of a computer model and by nonlinear load for the driver. A driv-
value, which can be found for the horn measurements. The model considered er working into a nonlinear load will
in question, considering how it expands. the harmonics required at the throat to not perform at its best, but will produce
It will be higher for a horn that expands generate a pure sine wave at the mouth higher distortion levels than it would
slowly near the throat than for one that (backward modeling), and also took re- under optimum loading conditions45.
expands rapidly. For example, a hyper- flections from the mouth into account. Directivity of the horn also plays a role
bolic-exponential horn with a low value For a horn with a 400Hz cutoff and in the total distortion performance46. If
for T will have higher distortion than a 4″throat, and a mouth SPL of 150dB, the horn does not have constant direc-
conical horn. For an exponential horn, the distribution of harmonics is shown in tivity, the harmonics, because they are
the pressure ratio of second harmonic to Fig. 42. The peak at the cutoff frequency higher in frequency, will be concentrated
fundamental is given as44 is due to the very high level required at toward the axis, while the fundamen-
the throat to generate the required SPL tal spreads out more. This means that
p2 γ + 1 p1t ω −e− mx / 2
= (32) at the mouth. distortion will be higher on-axis than
p1 2 2 γp0 c m / 2
Figure 43 shows the level of the off-axis.
where harmonics at the throat at 1kHz for a
p1t is the RMS pressure of the funda- given SPL at the mouth. Measurements HIGHER ORDER MODES
mental at the throat, showed that the prediction of the second At low frequencies, you can consider
p1 is the RMS pressure of the funda- harmonic level was quite accurate, but wave transmission in most horns as one-
mental at x, measured levels of the higher harmon- dimensional (1P waves). When the wave-
p2 is the RMS pressure of the second ics were higher than predicted. This was length of sound becomes comparable
harmonic at x, recognized as being due to nonlinearities to the dimensions of the horn, however,
p0 is the static pressure of air, and in the driver. cross reflections can occur. The mode
m is the flare rate of the exponential As you can see from the results, the of propagation changes from the sim-
horn. level of harmonics is quite low at the ple fundamental mode to what is called
You can see that distortion increases levels usually encountered in the home higher order modes. The behavior of
with frequency relative to the cutoff fre- listening environment, but can be quite these modes can be predicted for the uni-
quency. This is easier to see in the sim- considerable in the case of high-level form pipe and the conical horn47, 48, 49,
plification for an infinite exponential public address and sound reinforcement and it is found that they have cutoff fre-
horn given by Beranek34: systems. quencies below which they do not occur.

FIGURE 42: Level of harmonics at the throat for a sinu-

audioXpress 2008 7
 In 1925, Hoersch conducted a theo- an Infinite Hyperbolic Horn,” JASA vol. 11, April
retical study of higher order modes in a 1940, pp. 467-476.
conical horn, and calculated the equipres- 29. E.R. Geddes, “Acoustic Waveguide The-
sure contours for two kinds of modes. ory Revisited,” JAES Vol. 41, No. 6, June 1993,
The results (Fig. 44) show the equipres- pp.452-461.
sure contours including both the radial 30. D. Mapes-Riordan, “Horn Modeling with
and non-radial vibrations. The left part of Conical and Cylindrical Transmission-Line Ele-
the figure shows a pattern that resembles ments,” JAES Vol. 41, No. 6, June 1993, pp. 471-484.
what Hall measured in a conical horn 31. P. Wilson and G. Webb: “Modern Gramo-
(Fig. 16). For a flaring horn such as the phones and Electrical Reproducers,” Cassell and
exponential, however, the higher order Company Ltd, 1929.
modes will occur at different frequencies 32. N. Iwata, Jap. patent no 54-59129, 1979.
at different places in the horn8. 33. J. Hiraga, Les Haut-Parleurs, 3rd edition,
Dunod 2000.
34. L. Beranek, Acoustics, McGraw-Hill, 1954.
35. “Notes on Loudspeaker Directivity,” Altec
Technical Letter no. 211.
36. C.R. Hanna, “Multiple Inlet Horn,” US Pat-
ent no. 1 715 706, 1925/1929.
37. J. Slepian, “Sound-Generating Device,” US
FIGURE 44: Higher order modes in a
Patent no. 1 684 975, 1926/1928.
conical horn47.
38. E.C. Wente, “Acoustic Device,” US Patent
no. 1 992 268, 1933/1935.
Higher order modes will also be gen- 39. E.C. Wente and A.L. Thuras, “Loud Speak-
erated by rapid changes in flare, such ers and Microphones,” BSTJ April 1934, pp.
as discontinuities, so the slower and 259-277.
smoother the horn curvature changes, 40. S.E. Levy and A.B. Cohen, “Acoustic De-
the less the chance for generating higher vice,” US Patent no. 2 690 231, 1950/1954.
order modes. 41. A.B. Cohen, “Wide Angle Dispersion of
The effect of the higher order modes High-Frequency Sound,” Audio Engineering, Dec.
is to disturb the shape of the pressure 1952, pp. 24-25, 57-59.
wave-front, so that directivity will be 42. D.B. Keele, “What’s so Sacred About Expo-
unpredictable in the range where the nential Horns?,” AES preprint no. 1038, 1975.
modes occur. According to Geddes, they 43. C.A. Henricksen and M.S. Ureda, “The
may also have a substantial impact on the Manta-Ray Horns,” JAES Vol. 26, No. 9, Sept.
perceived sound quality of horns50. 1978, pp. 629-634.
44. A.L. Thuras, R.T. Jenkins, H.T. O’Neil, “Ex-
CLOSING REMARKS traneous Frequencies Generated in Air Carry-
In this article, I have tried to present ing Intense Sound Waves,” BSTJ Jan. 1935, pp.
both classical and modern horn theory in 159-172.
a comprehensive way. A short article like 45. K. Holland and C. L. Morfey, “A Model of
this can never cover all aspects of horns. Nonlinear Propagation in Horns,” JAES Vol. 44,
But I hope it has provided useful infor- No. 7/8, Jul/Aug 1996, pp. 569-580.
mation about how horns work, maybe 46. T. Kikkawa, A. Yukiyoshi, and N. Sakamoto,
also shedding light on some lesser known “A New Horn Loudspeaker Design yields low
aspects and research. Distortion and wide Dispersion,” AES preprint no.
Finally, I would like to thank Thomas 1151, 1976.
Dunker and David McBean for proof- 47. V.A. Hoersch, “Non-radial Vibrations With-
reading, discussion, and suggestions. in a Conical Horn,” Physical Review, Feb. 1925, pp.
aX 218-224.
48. P.M. Morse and K.U. Ingard, “Theoretical
REFERENCES Acoustics,” 1968.
26. Klangfilm GmbH., “Lautsprecher mit Expo- 49. W.P. Mason, “Electromechanical Transduc-
nentialtrichter,” Swiss Patent no. 279947, 1948/1951. ers and Wave Filters,” Van Nostrand, 1942.
27. H. Schmidt, “Ü ber eine neuartige 50. L.W. Lee and E.R. Geddes, “Audibil-
Lautsprecherkombination,” Funk und Ton no. 5, ity of Linear Distortion with Variations in
1950, pp. 226-232. Sound Pressure Level and Group Delay,” AES
28. J.E. Freehafer, “The Acoustical Impedance of ConventionPaper no. 6888, Oct. 2006.

8 audioXpress 2008 www.audioXpress .com