Planar Loudspeaker Technology - Applications in the Real World 3840 (P10.

7)

Christopher 'Topper' Sowden

Senior Consultant
Pelton Marsh Kinsella, Inc.
Dallas, Texas, USA

Frederick _ Ampel
President
Technology Visions S.A.
Overland Park, KS., USA

Presented at A U D I O
the 96th Convention
1994 February 26 - March 01
Amsterdam ®

Thispreprint has been reproducedfrom the author'sadvance

manuscript,withoutediting, correctionsor considerationby
the Review Board. The AES takes no responsibilityfor the
contents.

Additionalpreprintsmay be obtainedby sendingrequest and

remittanceto the Audio Engineering Society,60 East 42nd Bt.,
New York,New York 10165-2520,USA.

Ali rights reserved. Reproductionof this preprint,or any portion

thereof,is not permitted withoutdirectpermission from the
Journal of the Audio EngineeringSociety.

AN AUDIO ENGINEERING SOCIETY PREPRINT

 Planar Loudspeaker Technology - Applications in the Real World
by Christopher "Topper" Sowden
Senior Consultant, Pelton Marsh Kinsella, Inc., Dallas, Texas, USA
and Frederick J. Ampel
President, Technology Visions S.A., Overland Park, KS., USA

This paper provides a report on the development of planar loudspeakers for

professional/commercial sound applications. Included is selected historical
information relative to the development of planar loudspeakers. In addition,
an examination of the construction of a production model of a system designed
to meet specific requirements and applications relative to use in commercial
installation is presented. Detailed engineering data, on this loudspeaker, is
also provided as well as some discussion regarding assets and concerns
relative to its utilization.

1. INTRODUCTION AND DISCUSSION

1.I History

The first historical references to the concepts behind planar transducers extend back over two
centuries. Essentially, those references are concerned with the evolution and technical
developments associated with two specific areas:

· microphone transducers
· reproduction pickups of various types for use with early phonographic systems

However, the basic engineering principles, and physics represented in those primitive devices,
remain as valid today as they were when Edison patented some of them at the close of the
19th century.

While it is far beyond the scope of this paper to discuss those early devices in any detail, the
extensive variety of planar and/or flat panel transducers we know and use today are not totally
new 20th century developments.

The first practical planar loudspeaker made for commercial purposes was produced by the
Automatic Musical Instrument Company of Grand Rapids, MI., in 1930. That electrostatic
device was used in many of the firms early coin operated phonographs, a predecessor to the
modern-day jukebox.

While by today's fidelity standards that loudspeakers' performance would be at best

considered marginal, the fact that it worked reliably, and was used in a commercial product,
validated the principles involved in a way that no amount of research and theory could ever
accomplish.3

One of the authors (Sowden) clearly remembers looking at a Newark Electronics catalog in
the late 1960's that, along with resistors and capacitors, provided information on special "flat"
dipole loudspeakers manufactured by Fisher, KLH and Quad, and noting that the cost of these
dipole loudspeakers was an amount that could not bo rationally spent, by the author, (although
that amount was being spent by others or the devices would not have appeared in tho catalog)
on a consumer audio system.

In tho six decades since that first commercial product, the development of planar or flat panel
loudspeakers, has boon consistently overshadowed by the developments in cone, dome and
compression drivars?

In part, this has been due to the products' final cost and the restrictions placed on widespread
use, by that cost. However when examined in more detail, tho slow and sporadic
dovelopmant of planar transducers is more the result of practical technical and materials
related limitations.

Despite the often slow pace, the continual development of planar loudspeakers remained
appropriate for tho reproduction and reinforcement of music in the eyes of many researchers.
This was due to tho fact that planar transducers could theoretically moro closely emulate tho
directivity characteristics of acoustical instruments because of their omnidirectional radiation
capabilities.

1.2 Development

During tho late 1950's and early 1960's the materials needed to produce the desired large
planar drivers, including various petrochemical plastics and chargeable films, became moro
readily available at realistic prices. Tho accessibility of these materials opened a new chapter
in tho progression of planar transducers. The many new options now feasible caused, not only
a resurgence in the number of companies and individuals pursuing development but, a
rethinking of the basic design methodologies used to achieve tho benefits of
omnidirectionality. This caused tho developmental path for planar transducers to split into
several tangential trails.

A number of individuals and companies, seeking to serve tho relatively cost insensitive
audiophile market, chose to stay with tho high-voltage or magnetic suspension approaches,
and proceeded to refine and improve those ideas. These included Martin Logan, Infinity, Stax,
Magnapan, Enigma, Celestion and more recently Apogee, Acoustat and Audio Static, and a
host of others.

Most of these firms planar or dipole loudspeaker assemblies wore developed using variations
on the idea of an electromagnetic or electrostatic field in which a lightweight surface material
moved between two acoustically permeable charged surfaces (plates). Charging tho "plates"
allowed omnidirectional sound from the planar surface. Tho distribution of fields in three
dimensions, around the planar diaphragm, was generally expensive to implement, requiring
linear magnetic structures or high voltage DC power supplies. Additionally the overall SPL
capabilities of such devices were somewhat restricted. This, therefore, made the resultant
loudspeakers too fragile and cost-ineffective for applications other than tho high-end
audiophile market.
Another group of researchers sought to find alternate ways to construct planar diaphragms and
variations on that idea, such as dipolar systems, while simultaneously opening other areas of
application for planar technology.

One example of an alternative approach to dipole loudspeaker technology is manufactured by

Toby Guynn of Fort Worth, Texas, U.S.A. Mr. Guynn has been manufacturing dipole
loudspeakers, using cone drivers, with open backs for the last twenty years and has been
successful in locating the drivers on a panel in a way to make use of the driver's back
radiating characteristics.

Another alternative approach began its development in the late 1960's in Argentina. During
the 1950's Jose Bertagni, a physics professor, had successfully developed a business that
produced both radio and television receivers using local components, for sale within Argentina
under the "Parsons" brand. Because of the success of this venture, he received a license from
the United States based Westinghouse company to manufacture electronic products such as
the Westinghouse brand televisions, radios and high-fidelity equipment for sale and
distribution within Argentina. This was Bertagni's first introduction to the manufacture of
loudspeakers, a technology that was to hold his interest until his death in 1992.

Around 1968, Bertagni began a small research project to study how musical instruments
produced, and more importantly dispersed sound energy. The intent of the research was to
study the radiating effects of musical instruments and model an electro-acoustic transducer
with similar characteristics.

2. RESEARCH AND DEVELOPMENT

2.1 Introduction

A literature search and basic research quickly produced indications that some form of planar
or dipolar device would best model the desired characteristics. The cost of the technology,
required to produce the various existing dipole or planar loudspeaker systems, directed
Bertagni toward the use of a more traditional magnetic motor assembly if he was going to be
able to produce a more cost-effective design. If a planar loudspeaker could be developed using
this technology, construction costs would be noticeably less than the electrostatic or
electromagnetic distributed field approaches described above.

2.2 Diaphragm

The development of the radiating surface, for a conventional motor, required a rigid material
that would produce minimum frequency response variations across the audio frequency band.
Early experiments involved the use of a number of types of plastics, leather and various
thicknesses and densities of wood panels, similar to those used in the construction of acoustic
instruments.

All of these materials were tested with unacceptable results. Bertagni found that panels
constructed of sheet plastics or wood would provide the needed omnidirectional radiation at
certain frequencies. However the directionality was far too inconsistent with frequency and
resulted in an extremely inefficient loudspeaker.

3
It was also found, with the various wood materials tested, that the lighter the panel, the more
efficient the transducer. Tho overall results of this set of experiments clearly indicated that
a balance had to be made between the desired omnidirectionality and realistic efficiency.

A second round of research into diaphragm materials was begun, concentrating on various
plastics. One material tested was a new type of polystyrene bead product. This material was
considered applicable because it could he inexpensively molded into complex acoustical
shapes while maintaining a consistent density over the panel.

After studies of various manufacturers, a product manufactured by BASF was evaluated and
found to maintain density during the molding process. This material was composed of
individual spherical expanded polystyrene "beads" approximately 2mm in diameter. Injection
molding processes, similar to those used with other plastics, could be applied to this product
allowing the forming of a panel containing the various complex shapes that would best meet
the acoustical requirements desired.

After studies of panels produced using this polystyrene product, it was found that the motor
had to bo rigidly supported in a frame adjacent to the panel to ensure stabile and accurate
alignment of the voice-coil's connection to the driving surface.

Figure 1 shows a cross-section of the planar motor and frame assembly in comparison with
a characteristic cone device. During tests of the planar device, it was noted that there were
many frequencies that were noticeably directional, and provided varying spectral response of
the planar driver at different locations. Additionally, it was noted that the maximum sound
pressure level generated was not located on the centerline of the driver. It was found that the
mounting of the radiating surface to the structural frame caused resonances that were related
to the distance from the motor to the frame edge boundary.

The shape of the polystyrene product was modified using mathematical formulae and iterative
methods to develop a shape that would least impact the directionality and frequency response
of the driver assembly.

After intensive studies, and numerous prototypes, it was possible to determine tho optimum
surface and a final shape was developed. Edge boundaries were iteratively designed to trap
frequencies resonating horizontally across the panel causing eventual misalignment of the
driver assembly. The design of this inter-modulation (IM) trap required a thinning of the
polystyrene surface in various locations to minimize motion of the radiating surface.

Figure 2 illustrates tho details of this construction. Some panel designs were also provided
with additional damping weights in specific locations of the diaphragm to reduce motion of
the voice-coil in the magnetic gap and broaden the directional peaks at resonant frequencies.
Excessive horizontal motion of the surface could have caused misalignment of the driver
motor voice-coil.

When completed this structure is composed ora variable thickness diaphragm, clamped rigidly
along its circumference. The variable thickness structure provides a series of frequency
tailoring mechanical compliances as well as mechanical stiffening. This structure presents a
rigid compliance to lower frequencies, while permitting a carefully controlled progressive
decoupling of portions of the diaphragm's surface from the driver at the higher frequencies.

The low frequency voice-coil assembly is finished with a flat surface constructed of 30%
glass-filled Lexan (figure 3). This flat surface, or "hammer" is attached to the radiating surface
via a felt mastic insulating disc. This attachment of the hammer to the insulating disc is made
with a silicone compound allowing separation of the voice-coil from the radiating surface
without damage to the polystyrene material. Figure 4 illustrates the arrangement of a low
frequency motor and its attachment to the radiating surface.

The insulating disc is attached to the diaphragm using an epoxy resin. Extended tests were
run on various materials that could be used in this attachment and the silicone based assembly
was found to provide the most effective transfer of energy from the motor to the radiating
surface.

Further development of the loudspeaker assembly determined that larger radiating surfaces
tended to reduce the high frequency response of the device. To accommodate this physical
condition, two-way bi-directional devices were developed to improve overall high frequency
response. Two motor assemblies, driving significantly different surface areas, were
constructed using the same polystyrene mold. This allowed the mounting of a low frequency
and high frequency motor to a single surface element.

This surface element is subdivided into two separate radiating surfaces through the use of
mechanical and IM traps. The high frequency voice-coil assembly is attached to the radiating
surface directly, using a stiff epoxy resin. Figure 5 illustrates multiple driver configurations
with the right-hand design being the most recent model. This resulting shape provides
isolation of one radiating surface from another, reducing intermodulation distortion between
drivers. A cross-section diagram of a dual-driver unit is provided in figure 6.

The radiating surface acts as a dome radiator that varies in circumference with respect to
frequency. The higher the frequency, the smaller the radiating surface. In essence, the entire
diaphragm radiates energy at lower frequencies, while a controlled, and tailored reduction in
effective diaphragm surface area occurs as the frequency increases as would be predicted by
the physics.

2.2 Magnet Assembly

The horizontal motion of the radiating surface created the requirement for a larger voice-coil
gap to avoid rubbing of the voice-coil on the magnetic pole pieces. The requirement for a
larger voice-coil gap commensurately reduced the efficiency of the driver because of the
reduced magnetic fields adjacent to the voice-coil (gap flux). Experiments in the use of
magnetic fluids showed an increase in the magnetic field near the voice-coil and therefore
improved efficiency. The use of magnetic fluids also increased the power handling capacity
of the driver by allowing more direct heat transfer from the Nomex coil former to the magnet
structure.
2.3 Commercial Applications

Through the years, the foundation of Bertagni's original research and development has
continued to be built upon. One area of concentration has been the utilization of this
technology in commercial sound systems. This application would most benefit from the wide
angular radiation characteristics and convenient mounting conditions associated with planar
or dipole loudspeaker systems.

The basic concept was to produce a commercial panel transducer product that would take
advantage of the technology, and simultaneously offer the consultants and system design
specialists an architecturally appropriate package that would become essentially invisible when
installed.

One of the authors (Sowden) became aware of the Bertagni electro-acoustic system panels in
1982 when while reading Stereo Review's yearly review of home loudspeaker systems. He
noted that the 1 watt, 1 meter sensitivity of this consumer loudspeaker was published to be
124 dB (!).

He later found that the reviewer had mistakenly printed the maximum sound pressure level
for the sensitivity, but the mistake drew his interest. He notes, "I always like to see firms that
break the laws of physics."

3. TESTING PROCEDURES

3.1 Anechoic Measurements

In December of 1992, a series of measurements were made at the National Research Council
of Canada facilities, located in Ottawa, Canada, using current products based upon the original
concepts. The devices tested comprised samples of the entire product range including the
CTIOBH - a round omnidirectional full-range single driver transducer for commercial
installation and a prototype of the CT73D 0.6lin by 0.61m (24in. by 24in.) full-range dual
driver ceiling tile transducer system.

3,1.1 Frequency Response

Frequency response was measured in an anechoic chamber with Bruel & Kjaer 4133
microphones located at a distance of 2m from the front of the loudspeaker.

Each loudspeaker was mounted vertically on a baffle measuring 1.5m by 1.8m (Sft. by 6ft.).
Input voltage at the loudspeaker was 2.83 volts. Measurements were made at 256 frequencies,
logarithmically spaced between 20 Hz and 20 kHz.

A correction was applied so that the normal anechoic chamber errors that set in below
approximately 200 Hz, for this facility, were substantially removed. The corrected curve is
within 1 dB of true free-field measurements from 30 Hz to 200 Hz, and better above 200 Hz.
The vertical axis in each plot was calibrated to show the sound pressure level at one meter,
though measurements where made a 2m to ensure far-field conditions.
Frequency response data was taken on-axis (0°) and at 15°, 30 °, 45°, 60 °, 75 ° and 90 °
off-axis. Data from for the CT10BH, with 0.1m3 (0.25ftfi) backcan, is provided for on-axis
(0°) and 90 ° measurements in graphic form (figures 7 and 8).

3.1.3 Front Hemisphere Radiation

Mean response in the front hemisphere was calculated from average measurements made at
0°, 15°, 30°, 45°, 60 °, 75 ° and 90°, weighted by the solid angle into which the sound was
assumed to be radiating.

This is an approximation for frequencies above 500Hz, to the response that is measured in
a typical listening room at a distance of about 3m.

Below 500 Hz the room response will fluctuate above and below this curve, depending on the
physical form of the room and the placement of the loudspeaker and listener.

Figure 9 provides the resultant data for the CT10BH.

3.1.4 Impedance

Impedance was plotted as a function of frequency. Data for the CT10BH is shown in Figure
10.

3.2 Reverberant Room Measurements

In January 1993, additional testing was conducted, by one of the authors (Sowdan), in a
typical open-office environment in Dallas, Texas, U.S.A. This space represents a standard
installation environment for the product. It measured 140 square meters (1,500 square feet)
with 2.7 meter (9 foot) ceilings. The space was finished with sheet-rock (1.91cra [0.75 inch]
plasterboard) walls and a standard drop-ceiling configuration using 610mm by 1.22m (2ft. by
4ft.) acoustical ceiling tiles. The floor was covered with industrial grade carpet.

Six (6) Sound Advance Systems CT10BH loudspeakers, with 0. lin 3 (0.25ftfi) backcans, were
installed in two rows of three. Figure 11 displays the loudspeaker layout and associated
measurement positions. All measurements were taken at 1.2m (4ft.) above the finished floor
using an Ivie pink-noise generator, a Crown CTS00 power amplifier, a Larson-Davis model
3100 real-time analyzer (RTA), and a hand-held Ivie 1/3 octave RTA.

Measurements were made to determine the -6dB coverage angle and to ascertain the overall
coverage with one to six devices functioning.

Off-axis radiation characteristics of multiple drivers is illustrated in figure 12.

Sample measurements are provided as follows:

· Figure 13 - Frequency response chart, all speakers operational, measurement position

3.94m (12ft.,l lin.) South of position 16 as shown on fig. 10.
· Figure 14 - Frequency response chart, speakers 1 through 4 operational, measurement
position 1.68m (5ft.,6in.) North, 3.48m (llft.,5in.) West of position 16 as shown on
fig. 10.

· Figure 15 - Frequency response chart, speaker 4 operational, measurement position

1.68m (5ft.,6in.) North of position 16 as shown on fig. 10.

· Figure 16 - Frequency response chart, speakers 2, 4 and 6 operational, measurement

position 1.75m (5ft.,9in.) South of position 16 as shown on fig. 10.

· Figure 17 - Frequency response chart, speaker 2, 4 and 6 operational, measurement

position 3.05m (10ft.,0in.) South of position 16 as shown on fig. 10.

4. RESULTS OF TESTS

It was noted, from the obtained data, that tho directionality of the planar loudspeaker varies
less, but more randomly than conventional 8-inch loudspeakers used in the same application.

A May 15, 1989 report from Mark Engebretson of Summit Laboratories provided
measurements of the average beamwidth of a Sound Advance Systems CT72D device and
compared it to the average beam width of a typical 8-inch loudspeaker.

Figures 18 and 19 are presented to demonstrate tho comparatives for these two devices. Thoro
appears to be a directional peak of the planar loudspeaker at approximately 1000 Hz that is
not apparent in the traditional 8-inch loudspeaker type.

This I kHz in directivity increases the mid-frequency response of the device directly below
the transducer, creating a noticeable mid-band presence directly under the device. Noting the
extension of high frequency beam width produced by the Sound Advance Systems device, the
typical 8-inch loudspeaker produces less energy off-axis in the upper sibilance range, making
it less intelligible than the planar loudspeaker device at extreme off-axis angles.

The tests, performed for this paper, showed an improvement to this beamwidth response
indicating a more consistent non-directional mid-frequency distribution of the planar
loudspeaker allowing it to be used at a greater spacing in a distributed ceiling system with
reduced loss of intelligibility. The reverberant room tests indicated that the planar
loudspeakers may be located approximately 2.7 times the distance measured from the
loudspeaker position to the average listener height in horizontal spacing across tho room.
Interaction between drivers at this spacing provided reasonably consistent frequency response
and fidelity for playback of music with acceptable intelligibility results.

5. CONCLUSIONS

It was found that the planar dipole loudspeaker has application in areas where fidelity is
required, but spacing and position of the loudspeakers is not optimum. The loudspeakers may
be painted or surfaced with thin wall surface materials with only a slight degradation of the
overall response of the transducer. The radiating characteristics of the loudspeaker provide
increased spacing for use in paging and sound reinforcement in a ceiling surface. This
loudspeaker technology is cost-effective for use in commercial applications.

Footnotes:

tread, Oliver, and Welch, From TinFoil to Stereo, secondedition,page 316.HowardW. Sams,1976.

2Although many variations on the dipole loudspeaker have been developed through the years,
only a few technical papers have been written describing the technology. In 1982, a paper
published by Celestion offered various computer models of a dipole, and the varying polar
response characteristics of their SL600 loudspeaker. Peter Walker, of Acoustical
Manufacturing Co. Ltd., presented a paper at the 63rd AES Convention in Los Angeles,
supplying a simplified analysis of the dipole transducer, and discussing the bandpass
characteristics caused by the dimensions of the panel.

Bibliography

[1] Engebretson, M., "Laboratory Measurements of Directional Radiation

Characteristics B.E.S.T. CT12E and CT72D Transducer Systems A Report", (May 15,
1989).
[2] Ampel, Frederick J., "Planar loudspeakers: an alternative technology",
Sound & Video Contractor, v. 11 n. 3, pp. 135-143 (1993 March).
 A SUSPENSION(BEST)/ SURROUND(CONE)
B FRAME
C DIAPHRAGM
D HAMMER (BEST)/ COLLAR (CONE)
E VOICECOIL
F POLE PIECE
G REAR PLATE
H MAGNET
I FRONT PLATE
J SPIDER

Figure 1

A FRAME
B SUSPENSION
C DIAPHRAGM

Figure 2

A COIL
B 30% GLASS FILLEDLEXAN HAMMER
C TINSELS
Figure 3
A Rear Plate Pole Piece
B VoiceCoil CenteringHoles
C Ceramic Magnet
D Front Plate
E VoiceCoil
F Hammer
G Silicone
H InsulatingDisc
___-------- I Epoxy
J CouplingDisc/Diaphragm
Figure 4

10
 ,,-=3_EI

Figure 5
LOW FREQUENCY DRIVER
HIGHFREQUENCY
DRIVER

COUPLING
DISC(DENSITY
'A') _,
DIAPHRAGM
(DENSITY
'B')
Figure 6
110 . i i i iHi i i i illl i i i i,ll i

,°°

_,o /__4/
= 8°[

, iiii i i . II.I I I , II,I I"

8% 5 1 2 IO 5( 11 5 lC 20 ( 1FTc'--
FREQUENCY (HZ) mgure 7

11oi ..................... '

100

80
_ / _,, , ,

'°_
r _/
60:
,]i-,:,H,
1t 0
!
2( O
i ,
5
,111 i
2
i '_,I,_
i
51
ii1,
10
i:
2( ;

FREQUENCY (HZ) Figure 8

11
 110_..' ' ' ,,,, , , , ,,H i , , i1,, ,
5'

_loo. J

90
=: 1L/ _ II_k/VAl
80
=: P'i , v_F, Vl_ h,l,,

i iiH B i i iiii i i i llll i';

60 9 5 1(9 6' 0
FREQUENCY (HZ)
1 2 1( 2t Figure 9

110 i I I III I I I I III I I I I III I ,

100

_ O0

_80 _.:
_ L
70
o3

I,,, i
6o , , ,,_ltO '2_o P ,5 2 ' ,51"_ '2( ;
FREQUENCY(HZ) Figure 10

X_ X_ X, X_ 0_

_C_T_N_p) . X_ X,i xl,ls._x X, X3

a sR<a

Figure 11

i 15'-."_
eo. , _ dBA1o.PK
iao.(lO,.O-)
165.
(12,.11.) ·
Figure 12

12
 Figure 13

Figure
14

Figure
15

Figure 16

13
 Figure 17

................
IC_Hz Ik_ F_*_, 10kin _k _ Figure18

14
Figure Identification

Figure 1 Cross-section diagram comparing components of planar transducers and cone

transducer
Figure 2 Cross-section diagram of planar transducer diaphragm illustrating inter-
modulation (IM) trap
Figure 3 Low-frequency voice-coil assembly, planar transducer
Figure 4 Cross-section diagram of planar transducer motor assembly showing voice~coil
attachment to the diaphragm
Figure 5 Planar transducer multiple driver configuration example
Figure 6 Dual-driver planar transducer cross-section
Figure 7 Frequency response curve, Sound Advance Systems, Inc. CT10BH transducer
mounted in 0.1m 3 (0.25ft?) backcan, on-axis
Figure 8 Frequency response curve, Sound Advance Systems, Inc. CT10BH transducer
mounted in 0. lm 3 (0.25ft. 3) backcan, 90 o off-axis
Figure 9 Front hemisphere radiation frequency response curve, Sound Advance Systems,
Inc. CT10BH transducer mounted in 0.1m3 (0.25ft. 3) backcan
Figure 10 Impedance plot, Sound Advance Systems, Inc. CT10BH transducer mounted in
0.1m3 (0.25ftfi) backcan
Figure 11 Floorplan of facility used for reverberant room measurements
Figure 12 Off-axis radiation characteristics of mulitple CT10BH transducers mounted in
0.1m 3 (0.25ft. 3) backcan.
Figure 13 Frequency response chart, all speakers operational, measurement position 3.941n
(12ft.,l lin.) South of position 16 as shown on fig. 10
Figure 14 Frequency response chart, speakers 1 through 4 operational, measurement
position 1.68m (5ft.,6in.) North, 3.48m (llft.,5in.) West of position 16 as
shown on fig. 10
Figure 15 Frequency response chart, speaker 4 operational, measurement position 1.68m
(5ft.,6in.) North of position 16 as shown on fig. 10
Figure 16 Frequency response chart, speakers 2, 4 and 6 operational, measurement
position 1.75m (5ft.,9in.) South of position 16 as shown on fig. 10
Figure 17 Frequency response chart, speaker 2, 4 and 6 operational, measurement position
3.05m (10ft.,0in.) South of position 16 as shown on fig. 10
Figure 18 Average beamwidth vs. frequency chart for Sound Advance Systems, Inc.
CT72D transducer
Figure 19 Average beamwidth vs. frequency chart for typical 8-inch speaker

15