0% found this document useful (0 votes)
445 views5 pagesHorn Loudspeaker Design
Uploaded by
solveaudioCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
445 views5 pagesHorn Loudspeaker Design
Uploaded by
solveaudioCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 5
Horn loudspeaker design
Three articles summarizing the development of design theories and
concluded with two systems for construction
by J. Dinsdale, M.A., M.Sc. Cranfield Unit for Precision Engineering
Reprinted from WIRELESS WORLD, June 1974
The two designs which follow are specific examples d
ived from the d
ign data and tabi
provided in parts one and two of this series. Guidelines for construction are given, although it is
intended that the constructor devise his own v
requirements.
Much has, been written about te best
methods of constructing lodspeaker en
closures, especially regarding rigidity and
the prevention of fesonanocs td leaks and
ts fr asthe hom sconoered these pots
fe equally inportant, The hom encore
tar to sand up to consiierble acouste
fess, and any shortomings in man
fica ore Ubie Wo caume moe sera
tural distress than would be te ease wih
tome ote enonures
Téealy the hom sould becatincicaar
section, bt thistorm of contraction ly
petal wih’ small mode and igh fe
tency hors, The technique best esoped
yc home constructor tocastinplaster
tkParis using 8 pywood mold and ren
fevconests ay netvnry. The elelaed
vote should be se out using plywood
oe held in place by stringer, which
be bared whhin che casting ial as
reinforcer, Ii sso good plan fo
provide a }in panel at the throat end for
frowning tbe loudspeaker, and further
peed serounding te mou which wal
‘help in securing the complete horn assembly
tnd Suing any Secreted
Bass Ror are aloe vaiably oon
sxuced of fat panel ext 502810 approx
mate (othe comer Rare profi Plywood,
Chipboard or boskboard ae stsecery,
either { or fin thick, Composite sand-flled
panels condetng of two tin wall of fin
plywood spaced 3 or lin apart and filled
wan ry "and. provide exterely id
Taonanchliee cndoeues Great cme
Should be ten to prevent any parle of
Sand, stds, Fomenteringthe perch
coil gapof the oudspeaer, recentdemon
Staion jouned by theeuor ast bans
tras pote ty dsoton cased arnt
Jarile of wood i one ofthe ttle load
Speakers, Care most lo be taken tenure
thatthe sand is dry, oberwise the wood
parels wil rot is advable 1 bake be
Sand on shallow trays in an oven toensre
that al motores romoved. After fifeg a
few annus of orgen runic wl lp the
sand’ to eile down rend) for “opping
1¢ wooden panels should be fixed to-
1 using wood-serews and with aliberal
plication of a liquid glue along all mating
faces. This not only adds strength, but also
makes all the joints airtight. Further
sength should be provided by triangular
comer fillets and “glue-blocks" placed at
inervals along the longer joins. In addi-
tion, smaller reflecting plates should be
laced atthe outside ofall sharp corners
case” the wavefronts around bends, and
to help preserve the steady exponential i
crease in horn area, as indicated in Fig. 10.
Manufocuring tolerances should not ex
ceed hin atthe throat but ecrors of fin
at the mouth of the bass horn are unlikely
to have any noticeable effect on the per
formance. It is worth bearing in mind that
the-veloity of sound, on which al design
caleulations are ultimately. based, itself
Wares by at much as 5% at climatic
extremes.
‘A vial detail isto ensure thatthe lous-
speakers can be fed and removed easily,
maintaining an overall airtight construe
tion by means of thin rubber gaskets if
necessary. It should of course be remem
‘ered that the highest pressures occur atthe
Aout, and the greatest effort to ensure
Figilty and absence of leaks should bemade
in this area. As the cross-sectional area of
the hor increases, i is © good plan to ft
longitudinal stiffening panels, made of kin
plywood, across the centre of the horn,
{hereby converting the hora into two sym
metrical adjacent ducts. This reduces air
turbulence effects at bends and makes the
bends themselves less erica in adition
providing. extra cross-bracing between
Danels that might otherwise resonate. It is
‘worth fing longitudinal stifeners forthe
final 25% of each bass horn.
‘Unlike the majority of loudspeaker en
closures, there is no need to provide any
Sound. absorbent material within the en-
Closure itself (except within the compression
chamber, i fied, which may be lined with
acoustic wadding, long hair woo, ete, to
absorb high frequeney sound). Th interior
of the hora should have all sharp edges
removed with sandpaper and all internal
Gormer filed with putty or a simula setting,
plastic compound and smoothed down by
tneans ofa finger. This practice, whichis not
mandatory, also has the effet of sealing
any remaining sit leaks. The wholeinteror
surface should be treated witha thick coat
of gloss paint
Design ofa “mini-horn”
The intention of the minihorn is to pro-
vide a many a8 possible of the benefis
of horn-loading within an enclosure which
is sufcieny small for use in a smal ving
oom, where the overall size is ofcourse
specially important when a quadraphonie
fof even a stereophonie installation is under
onsideration, ‘The room for which this
tions, using the d
ign data, to suit his particular
particular mini-hora was originally designed
{imposed limitations of 20in asthemaximum
intrusion into the room, and an overall
height of 4; fortunately, corner position:
was acceptable.
Tt was clear that only one loudspeaker
could be used, and after some though
the Eagle FR63 was chosen. This is aco
axial twin-cone loudspeaker in which the
inner (tweeter) cone is itself shaped as a
small horn. This subsidiary cone will handle
the extreme top of the feequency range and
‘beam it out axially through the treble horn,
The loudspetker has a nominal diameter
of 65in, a frequency range of 35 10
18000H2 and power handling capacity
of 10 watts. It is clear that, since top
frequencies wil be dealt with by the vweeter
‘cone, the bass horn need only cover 3
‘octaves, ie. from TOHz to S60Hz, and the
‘middlefrequency horn can take over
Gay) S0OHz. This middle-frequency horn
‘be most efficient for 4 octaves, ie. upto
about 8kHz, at which point the tweeter
‘cone will already be taking over. The
‘complete frequency spectrum wil therefore
be:
Bass horn ‘TOHzt0 $S0Hz
Middle hora 500Hz to 8000H2
Top hom
(tweeter cone) _8kHz upwards
‘The other design consideration at this
stage is the power handling capacity. A
bass power of 0-3 watts ata distortion level
cof up to 1% was decided.
Bass horn
In order to derive the greatest benefit from
ccomer-positioning, the mouth of the bass
hhora should be at floor level and should
stretch horizontally feom wall to wall. A
‘mouth consisting of a quadrant of a circle
fof 19in radius was considered, giving a
horizontal arc of 2-48. Examination of
Table 3 (Part 2) shows a minimum mouth
area of 256 sq.ft for a horn capable of
reproducing down to 70Hz, and dimen-
sions of 248 x 1-038 (29-Tin x 12-Sin)
Were therefore chosen for the bass horn.
Table 7 suggests a throst area of 0-048
sqft (ie. 45 sq.em) and Fig. 9 indicates
that for 196 distortion at 7 times the cut
off frequency (ie, 490H2) the power at the
throat will be 0.007 watts/sa. em, giving
0-3 watts total, which isthe specified value.
Table 9 shows that the bass hom
‘will have length of 6:18ft (exponential
contour) or 5-24f (tractrix contour). hwas
decided to adopt the tractrix 50 a5 10 give ‘a shorter overall length, and the complete
‘contour has been constructed in Fig. 12.
‘Treble horn
‘The treble horn will load the front of the
loudspeaker, commencing at the nominal
aphragm area of 23 59.n, which is thus
the throat area for this hom. Thé lowest
frequency to be handled is SOOHz, and
from Table 4 the mouth area is 130 sq. in,
‘which may conveniently be realized. as
IO-lin x 12.9in, Table 10 now gives the
horn length as-4-42in,
It is also possible to adopt a circular
format for this hom, in which case the
mouth diameter will be 12-9in or altern-
atively some degree of horizontal
directivity may be introduced by adopting
fan aspect ratio of 25:1 (larger dimension
horizontal). In this event the mouth dimen-
sions become 18-08in x 7-23in, and the
flare contours should be arranged to give
the appropriate expansion (see Fig. 13).
Integration and complete design
Radiation at the throats of the mid
frequency and bass horns will be in anti
phase, since these hors load the front and
‘back respectively of the single loudspeaker.
Since the mouths of both horns will be inthe
‘same vertical plane, the combined length
should be an add number of half-wave-
lengths at the crossover frequency to ensure
‘that radiation from both mouths isin phase
at this frequency. The total length is
‘5-240 plus O37, ie. 5.61. At SOOHZ
this length corresponds closely to five half
wavelengths as shown by Table 11, This
‘design thus includes a satisfactory combin-
of horns.
‘The cavity which couples the rear of the
loudspeaker. diaphragm with the throat
of the bass horn should now be designed
to cut off radiation from the bass horn at
S50Hz.
S/n
where
f
3
1
{
necaontal ai 1
Fu
T
I
: 1
= ‘ciccular | |
gs] ot a
a + 4
senses |
1
\
‘ T
1 {
i 1
1 wom ueworn cnt
sectien! dares v=,
Fig. 13. Flare contour for the treble section
Of the mint-horn.
Secuvol frequency, whence Y=
27:23 eu.in. The volume taken up by the
‘magnet, etc, of the loudspeaker is approxi
mately 21" cu.in (obtained by direct
measurement) and thus the overall cavity
volume will be some 48 cuin. The para
meters of the mini-horn have been
summatized in Table 12
Finally, the bass horn should be folded
into a suitable shape, and the two horns
integrated together. In view of the limited
lable and the desire for simplicity,
257 sah
Cosseate
Saat
4 eatin ncucing 21 cuir
loudepenker wale)
Treble hom
Froqueney range SOOMz upwards
Driverunt eagle FRCS.
Mout eres 13007 sain
Trrostares— Z3-in
Contour Exponential
rath 8am
a design with only one major fold may be
opted, shown in Fig. 14) (overpage)
‘The mouth of the bass. orn is at the
bottom of the enclosure between the {40
walls, making contact with the floor. The
mouth of the mig-hom is placed imme:
diately “above this, inthe same plane,
leading back to” the loudspeaker ‘sell
The loudspeaker is mounted ons small
baffle board. which supports the middle-
frequeney horn atthe front and the evity
coupling tothe throat of the base horn
a the rear. The bass horn bends vertcally
Sipwards almost immediately afer the
throat to « point some aft high at =f
doubles back on itself down the core:
the room to form the mouth. The er.
sectional area may conveniently be made
trapezoidal, but the design shown will not
preserve “plane ness” of wavefronts around
the bends. Fig. 14(b) (overpage) ilustrates
the general arrangement only, as readers
‘may well wish to make modifications for
personal reasons. The material used
should be Zin chipboard, ete, except the
side and front panels of $in plywood.
201 rin pass trequeney 70 He
cavott trequeney 2 =s0He
JY
z a
g x
Hom LENGTH Undo © 7 6 ry rd % fa %
‘Yq exponential area 698 G08 1107 1554 2041 966 3500 4572 5990 Bate 1026 1343 57 2304 3007 3951 S146
Ma teacitin aren 690 GSTS
Woe 35 4036 730
Fig. 12. Flare contour for the bass section of the mint-horn,
ore a7? R
van Y= Design of a no-compromise horn
Following the many qualifications already
stated in this article, it must be clear that
“no-compromise” is in itself a misnomer
hor design consists largely of making
the most effective compromises between
conficting requirements. However, this
design is aimed at the situation where the
best possible practice can be followed with
‘out being unduly hampered by limitations
of either size or position. Nevertheless, it
would be somewhat pointless (o design
‘an enclosure which cannot be built without
professional tools, facilities and materials,
so the design has been conducted with a
large living room (or small hall) in mind,
‘and directed towards the competent do-
it-yourself enthusiast.
In order to cover a wide frequency range,
‘a three-horn design has been adopted, using
three separate loudspeaker units,
Bass Hom 40Fz to44oH
Mid-frequency Horn 400H2 03.82
Treble Horn 3.5H2t0 202
‘A total acoustic power handling capacity
fof 1W at a distortion level up to 2% has
‘been chosen as being more than adequate
for the situation envisaged. A wal:
‘mounting design is adopted, using a tractrix
‘bass hor, to simplify the eventual position:
ing of the system. (Of course, there is no
reason why a design intended for wall
‘mounting should not be placed in a corner
bbut it is mot recommended that corner
designs be placed against a wall)
The choice of loudspeaker drive units
is not straightforward; there are many com-
‘mercial units worth using in horns, although
individual designers inevitably have their
‘own favourites. For this design it was
decided to use:
Hom Unit PowertWifrequencisphiagm
Range Area
(He) Wominal
sai
Bess KEFBIS9 30 20-1k 42
Middle KEF8II0 1520-5 13
Treble KEFTZ7 6 3k-30k 15
Bass horn
‘The mouth area for 2 minimum frequency
of 40H: is given in Table 2 as 15-73 sq,
realized as 4Tin by 48in, The diaphragm
area of the B139 loudspeaker is approx
{mately 42 sqin (the diaphragm is oval
in shape) which corresponds closely with
the Sin speaker of Table 7. This table
sives a recommended throat area for the
bass horn of 0.073 sq.ft (68 sq.em). The
highest frequency to be handled by this
hhorn will be 440Hz, Le. 11 times the cute
off frequency. Fig. 9 shows that for 2%
distortion at 11 times the cut-off frequency,
the throat will handle 001 watts/sqem,
which for a mouth area of 68 sqem
ives 0-68 watts total power. The length
of the bass horn, from Table &, is 13-1
using the tractrix contour, and this curve
may be constructed to a suitable scale
im the same way as that produced for the
mini-horn. The form of the tractrix should
‘commence with a throat radius equivalent
to four times the chosen throat area (eight
times in the ease of the corner horn) and
terminate at a mouth radius giving ® peri
meter equal to the cut-off wavelength. The
area at & series of points along the horn
(eg. every Gin) may be obtained by
reading the radius from the graph and
taking one quarter of the corresponding
Middle horn
‘Attention should now be directed to the
mid-frequency horn. The cutoff frequency
‘of 400H2, together with the area of the
chosen loudspeaker (13 sq.in) result in a
‘mouth area of 203-6 sq, a throat area of
13.75 sq. and a length of 8-78in exponen-
tial contour). Again the contour should be
‘constructed {0 a suitable scale (which may
be 1:1 for the mid-frequency and treble
horns). Since the throat areas of the
‘middle and treble hors are made equal
to their respective loudspeaker diaphragm
areas, there is no problem regarding air
overload distortion—one could do litle
ifthere were.
Treble horn
In view of the throat area of 1-5 sqin for
this horn, itis suggested that a mouth area
‘of (say) 30 sqin is adopted, giving an
‘exponential length of I-lin.
However, at these frequencies itis quite
acceptable "to mount the loudspeaker
‘ireclly onto a flat bale board without
any horn.
Integration and complete design
The three loudspeaker drive units should
rive their respective horns in phase.
Initially it will be assumed that, whereas
the middle and treble horns-must load the
front of the loudspeakers (to avoid dif-
faction effects caused by the frame and
‘magnet assembly atthe rer), the bass hon
will n fact load the rear of the loudspeaker.
‘This implies thatthe bass loudspeaker must
be connected in antiphase to the middle
and treble loudspeakers. If examination
Of the behaviour of the bass and middle
hhorns. at their mutual crossover point
reveals that the radiation is in anti-phase,
Frequency range 40H: 10 440He
Briverunt ” KEFei99
Mouthores S73 sae
Twowtaes 0673 sah
Coneowr Sra
anath vain
Cry volume 5 cuin act at front of
foudsoeaker
Mista nom
Frequency reage 400He to 3:84
Drierunt "KETO
203.8 ain
1375 se
Exponent
878m
‘kite upwarse
Kerra?
the bass horn can be arranged to load the
front of its loudspeaker, which will then
have to be connected in phase with the other
two. In ct the total length ofthe bass a
sed fh and al 1 hel
that this is nearly equivalent to. an
number of half-wavelenaths at 400H2, the
crossover frequency. The bess loudspeaker
may therefore be reversed so thatthe horn
Toads the front of the diaphragm; this wil
also make the design of the acoustic cavity
‘much simpler.
‘The cavity between the bass loudspeaker
and its hora should be designed to give a
‘cutoff frequency of 440Hz. Applying the
sforementioned formula gives a volume of
SI cuin. There is no real need to employ
4 similar cavity at the crossover between
the middle and treble horns; the fact that
these horns are not folded, together with
their large throats, reduces distortion at
high frequencies to negligible proportions.
Finally, the three horns must be com-
bined into composite enclosure. As with
the min-horn there are many ways of
achieving this, and it would be invidious
to specify a particular design to the exclu:
sion of all others. However, certain basic
rules apply, and the following suggestions
may be of value.
‘The rectangular mouth of the bass horn
should be placed at floor level, with the
mouths of the middle and teble hors
placed above
back-to-ront
ture is ta premium, the middle and t
horns may be mounted on top ofthe.
plete folded bass horn, giving a very high
‘cabinet. If, however, height is at a premium,
then the bass horn may be folded behind
the middle and top horns, thus minimizing
the overall height but increasing the width
This later approach is shown in Fig. 14(3),
and the complete design of the no-compro-
mise hom is summarized in Table 13.
Material used for construction is tin block
board, plywood, ete, and all joints should
be screwed and glued to make them air
tight.
‘When converting from a basic parameter
design, as described in this section, t0
complete working drawings, the temptation
is often to press on rapidly and’ adopt
certain compromises. Unfortunately, the
final construction is a “once only" event,
and horn structures cannot easily be modi:
fied if major audio deficiencies (eg, reson
ances or “holes in the frequeney spectrum)
‘become apparent during listening tests. It
Table 16
section details forthe mini bass,
Bec. Length Aves Reid
lish Gein it
er
8 ots te
€ 3 i382
5 bet
2 3s 8
f &
6 sb ce
Ho 8 up
J $261 zoe 10mm
% 880 FBS Tablet
Cross section details forthe no-
compromise bass horn
‘See. tangih Area Realited
Gy ine)
| a 105 08
% 10 4
© 30 278
D 4% 380 Se
& > So be
Fo 8980S
G98 1° tae
Ho 8 25023
Se 58 13.8% 29 igh
k 13 26138 SS
L120 4017 x Ss nom
M18 307 x 37 gh
Nt 580 2h Span
P ¥39 1128 2ex sone
R183 2268 49x aah
is therefore strongly recommended thatthe
final design takes place over an extended
Period, with several alternative approaches
being worked on simultaneously until one
fof them emerges as the right solution for
the parameters and overall concept in mind
‘The three loudspeaker units must be con:
nected via suitable filers so. that each
handles frequencies only within its
appropriate pass-band. The simplest way
of achieving this is by means of passive
crossover networks at the output of the
power amplifier. However, this method
Feduces the beneficial effects of electro-
‘magnetic damping of the loudspeaker
movement afforded by the low output
impedance of the amplifier, and a better
hod is to use three separate power
flrs: whose inputs are fed via active
and low pass filters, as outlined in
Part2.
It is wellworth experimenting over an
‘extended listening period unt the optimum
bandwidth and sensitivity of each horn has
been realized, paying particular atention
tothe crossover points.
Conclusions
This article has taken the form of a critical
review of work which took place largely
between 20 and 50 years ago. The author
of such reviews benefits from hindsight,
bot inevitably loses much ofthe excitement
and impact of the original work. I have
teen in contact with many individuals who
were personaly involved with the develop-
ment of horns, in both amateur and pro-
fessional capacities) 1 thank them all for
their advice and comments, and hope that
Thavedone juste to thelr suggestion.
in spite of the obvious disadvantages of
large Sze and hgh cost, and the ifculties
o realizing an adequate design, the expon-
ential horn loudspeaker stil has many
enthusiastic users, the present author
fmong them. The clear advantages com-
ferred by the hora in terms of presence,
clear bass, low distortion and sheer realism,
mibine to make horn enthusiasts redouble
k efforts to design a beter horn rather
toadoptanaterativetypectenctosure.
1 will be clear fo readers of this aril
that, wth the possible exception of staigh
horns’ of eieslar seston constructed in
very stil materia, the simple horns de
ieribed here ean only appr imate tothe
{deal performance oer by this seme of
reproducer. Although the pioneer develop-
ment work was conducted between 50 and
70 years ago, engineers are continuously
designing new horns and investigating
different aspects oftheir performance, often
with the ald of computers to construct a
‘mathematical model for the analysis of
‘conditions in a practical horn (38, 39). It
‘must be emphasized again that first-class
results may be obtained by following the
basic design data and constructional advice
given in this article. Loudspeakers in
‘general, and horns in particular, are contro-
versial subjects, and T have no doubt that
‘many will wish to challenge some of the
statements I have made. T hope that this
anticle, together with any discussion, will
stimulate many audio enthusiasts to design
‘and make their own horns, and to write
about the results s0 that all may benefit
from thee findings.
Finally, 1 acknowledge with thanks the
helpful advice given by Mr Gilbert Telfer,
whose experience of the design and manu:
facture of horns has been a constant
encouragement.
JACK DINSDALE was educated at
Trinity College, Cambridge and later at
Cranfield. After » craft apprenticeship in
mechanical engineering. he joined the
lilt Autornaion Group where fr nine
years he was concerned with mise gud
ance, digital computer design and lately
with’ online applications of computers
{o industrial and mica contrl systems
Mr Dinsdale is now Principal Research
Engineer with the Cranfield Unit Tor
Precision “Engineering, & commercial
