Closed-Box Loudspeaker Systems

Part I1: Synthesis

RICHARD H. SMALL

School of Electrical Engineering, The University o/ Sydney

Sydney, N.S.W. 2006, Australia

Part I of this paper provides a basic low-frequency analysis of the closed-box loud-
speaker system with emphasis on small-signal and large-signal behavior, basic perfor-
mance limitations, and the determination of important system parameters from voice-coil
impedance measurements. Part II discusses some important implications of the findings
of Part I and introduces the subject of system synthesis: the complete design of loud-
speaker systems to meet specific performance goals. Given a set of physically-realizable
system performance specifications, the analytical results of Part I enable the system
designer to calculate directly the required specifications of the system components.

Editor's Note: Part I of Closed-Box Loudspeaker Systems the modulation distortion [23] produced by a large driver
appeared in the December 1972 issue of the Journal. will be less than that of a small driver delivering the
same acoustic output power.
8. DISCUSSION Thus a large driver has no inherent advantage over
Driver Size a small one so far as small-signal response and efficiency
are concerned. It may in fact have a cost disadvantage.
It has long been an accepted principle that a large bass But where high acoustic output at low distortion is re-
driver is better than a small one. While this attitude quired, the large driver has a definite advantage.
seems to be justified by experience, it has recently been
called into question [22]. The analysis in this paper dem-
onstrates that driver size alone does not determine or Enclosure Size
limit system performance in areas of small-signal re- It is clear from section 4 that an air-suspension system
sponse, efficiency, or displacement-limited power capacity, having a high compliance ratio can duplicate the per-
A large driver inevitably costs more than a small formance of a larger conventional closed-box system
driver having identical small-signal and large-signal pa- having a low compliance ratio. However, once the eom-'
rameters of the kind discussed here. However, it is pliance ratio is made larger than about 4, there is no
physically easier to obtain a large value of VD and hence way to gain a significant reduction in enclosure size
a high acoustic power capacity from a large driver, and without affecting system performance.

JANUARY/FEBRUARY
1973,VOLUME
21, NUMBER
1 11
RICHARD H. SMALL

A small air-suspension system, when compared to a can be obtained quite easily from the basic parameters
large air-suspension system, must have a higher cutoff of the system.
frequency, or lower efficiency, or both. As has been A few manufacturers already supply these basic
claimed many times, it is possible to design a small parameters or the directly-related free-field response and
system to have the same response as a large system. But efficiency data. The practice deserves encouragement.
if both are non-wasteful air-suspension designs, then as
shown by (26) or Fig. 8 the efficiency of the small sys- Typical System Performance
tern must be lower than that of the large system in
direct proportion to size. A sampling of closed-boxsystemsof British, American
It is often possible to provide the same maximum and European origin was tested in late 1969 by measur-
acoustic output as well as the same response from the lng the system small-signal parameters as described in
small system, but the lower efficiency of this system section 6. The frequency response and efficiency were
will dicta:te a higher input power rating and therefore then obtained from the relationships of sections 3 and 4.
a driver voice coil capable of dissipating more heat. Resonance frequencies (/c) varied from 40 Hz to 90
Also, it is easily shown that for these conditions the Hz. Total Q (Q'reo) varied from 0.4 to 2.0. Reference
driver of the small system will require a larger magnet efficiencies (*/o) varied from 0.28% to 1.0%. While there
(e.g., a heavier diaphragm of the same size may be was no general pattern of parameter combinations, all
driven through the same displacement, or a smaller but a few of the system's fell into one of two categories:
diaphragm of the same mass may be driven through a 1) Cutoff frequency (fa) below 50 Hz with little or
larger displacement). Thus for this condition the driver no peaking (QTco up to 1.1). Size generally larger
for the small system must be more expensive than that than 40 dma (1.4 ft0).
for the largesystem. 2) Cutofffrequencyabove 50 Hz with definitepeak-
It may be concluded that the pressure to design more ing (QT¢o between 1.4 and 2.0). Size smaller than
and more compact high-quality loudspeaker systems leads 60 dm 3 (2 ft3).
directly ,to systems of reduced efficiency and, in most
cases, reduced acoustic power capacity. If acoustic power One explanation for this situation was spontaneously
capacity is not sacrificed, these compact systems require provided (and demonstrated) by a salesman who sold
expensive drivers and must be used with powerful am- American systems in both categories. Only category 1
plifiers, systemswouldreproducelow organand orchestralfunda-
mentals, while category 2 systems had demonstrably
Performance Specifications stronger bass on popular nmsic. Sales thus tended to be
determined by the musical tastes of the customer. There
Of all the components used in audio recording and is marketing sense in this, and economic sense as well,
reproduction, loudspeaker systems have the least eom- because the same driver which has category 1 per-
plete and least informative performance specifications, formance in a large enclosure has category 2 perfor-
In _he low-frequency range at least, this need not be so. mance--with a higher acoustic power capacity--in a
If a specified voltage is applied to a direct-radiator small enclosure.
loudspeaker system, the output of the system at low
frequencies may be expressed in terms of an acoustic 9. SYSTEM SYNTHESIS
volume velocity which is substantially independent o/
the acoustic load [12], [24]. The "response" of a loud- System-Driver Relationships
speaker system expressed in this way is meaningless to The majority of closed-box systems operate with am-
most loudspeaker users, but a specification of the acous- plifiers having negligible output resistance, have a total
tic power or distant sound pressure delivered into a moving mass no greater than that of the driver on a
standard free-field load by this volume velocity is both baffle, and obtain most of their total damping from elec-
meaningful and useful, tromagnetic coupling and mechanical losses in the driver.
While the sound pressure delivered to a room is dif- For these conditions, (7), (9), (13), (17) and (18) may
ferent from that delivered to a free field, the difference be used to derive
clearly is a property of the room, not of the loudspeaker
system. _f the room performance is very poor, it can Q'rco QEc /c
--_ -- -- (_+ 1)_, (54)
be corrected acoustically or, in some cases, equalized Qa,s Qr_s /s
electronically. This is in no way a deterrent to accurate
specification of the basic loudspeaker system response and thus
by using a standard free-field load. In fact, the findings [_/QTco _ [s/Qrrs, (55)
of Allison and Berkovitz [25] indicate that a 2_r sr free-
field load is a very reasonable approximation to a typi- where QTs is the total Q of the driver at fs for zero
cai room load. sourceresistance[12,eq. (47)], i.e.,
Such a standard-load approach has of course been QTs = QEsQ._m/(QEs+Q_m). (56)
used for years in loudspeaker measurement standards
[18], [26], [27]. If it were applied more universally, it These equations show that for any enclosure-driver
would provide a very useful--and presently unavailable-- combination (i.e., value of a) the system resonance fre-
quantitative means of comparing loudspeaker systems, quency and Q will be in the same ratio as those of the
It is a particularly attractive method for specifying the driver, but individually raised by a factor (_ + 1)%.
low-frequency response of a system, because the nominal This increase is plotted as a function of a in Fig. 13.
free-field low-frequency response and reference efficiency This approximate relationship and the basic response,
12 JOURNAL
OFTHE
AUDIO
ENGINEERING
SOCIETY
 CLOSED-BOX LOUDSPEAKER SYSTEMS PART Ih SYNTHESIS

efficiency and power capacity relationships derived 5

earlier are used below to develop system design pro- [J
I[
cedures for two important cases: that of a fixed driver

given.
design' and that °f °nly the final system specificati°ns f....CC
4 III /
fill
/
OR 3
One difficulty of trying to design an enclosure to "fit" _/] J
I
a given driver is that the driver may be completely un- QC / I I
suitable in the first place. A convenienttest of suitability _ [ I
for closed-box system drivers is provided by (51) and '_ 2 /
(54); the driver parameters must be known, or measured.

Equation
quency must(54) insists
always that the
be lower thandriver resonance
that of fre-
the system.

wastefully large, i.e., he desires an air-suspension system,

1 /"-
_'_ f I

J IJ P
then
If thea designer
must be at least to
wishes 3 and theandriver
avoid resonance
enclosure which fre-is IlJJ
quency must be no more than half the maximum tolerable .5 ] 2 3 5 7 ]0 20
system resonance frequency. 04
Similarly, Qxs must be lower than the highest ac-
Fig. 13. Ratio of closed-box system resonance frequency
ceptable value of QTco, and by approximately the same and Q to driver resonance frequency and Q as a function of
factor which relates fa to the desired or highest ac- the system compliance ratio a.
ceptable value of fo.
Finally, from (51), the value of VAS must be at least for an unfilled enclosure volume of 135 dm a or 4.8 ft a.
several times larger than the enclosure size desired. This would be quite suitable for a floor-standing system.
If the driver parameters appear satisfactory, the de- The a = 9 alignment gives excellent performance in a
sign of the system is carried out by selecting the most volume of only 60 dm a (2.1 fta). The a = 12 alignment
desirable combination of fo and QTco which satisfies (55) could probably be achieved in a 40 dm a (1.4 ft a) en-
and then calculating a from (17). The required en- closure with filling. Qrco would then be lower than
closure size (net internal volume) is then, from (51), shown, probably about unity, giving a cutoff frequency
of about 53 Hz. This would be quite adequate "book-
VB = VAs/a, (57) shelf" performance.
or somewhat smaller if the enclosure is filled. Taking the larger B2-aligned system, the displacement-
The reference efficiency is calculated from (23), and limited acoustic power rating for program material, from
the acoustic power rating from (39) or (42). The elec- (42), is

trical power rating is then, from section 5, PAR = 0.19 W,

PER = PAn/VD. (58) and the corresponding electrical power rating is

Example of Design with a Given Driver PrR = 19 W.

Using a standard baffle and unlined test enclosure, a This is well within the power rating given by the manu-
European-made 12-inch woofer sold for air-suspension facturer, so the system can safely be operated with an
use is found to have the following small-signal parameters: amplifier having a continuous power rating of 20 W.
The "bookshelf" design, because of its higher value of
fs = 19 Hz f:3, has displacement-limited ratings of about 0.5 W acous-
QMs = 3.7 tical and 50 W electrical. This is much higher than the
Qss = 0.35 manufacturer's rating. In the absence of the actual value
Vas = 540 dm a (19 fta). of Premix) on which the manufacturer's rating is based,
Using (56) and (23), it is probably best to limit the amplifier power to 25 W.
The .system can then produce an acoustic output of
QTs = 0.32 0.25 W.
% = 1.02%.

The manufacturer's power rating is 25 W, and the peak Design from Specifications
linear displacement is estimated to be 6 mm (¼ in). The Most engineering products are designed to suit specific
effective diaphragm radius is estimated to be 0.12 m, requirements. Quite commonly, the "requirements" for
giving SD -- 4.5 × 10 -2 m 2 and VD = 2.7 X 10 -4 m a a particular product contain conflicting factors, and the
or 270 cm s.
The values of fa, QTs and Vas for this driver appear Table 1. Expected Performance of the Given Driver
to be quite favorable. The values of/c, Q_co and fa to
a fo, HZ QTco f_, HZ V,B, dm 3
be expected from various suitable values of a are given 4 42.5 0.72 42 135
in Table 1 together with the corresponding enclosure 6 50.3 0.85 44 90
compliance VA_ (volume of an unfilled enclosure). 9 60.0 1.01 47 60
12 68.6 1.15 50 45
The a -- 4 alignment gives almost exactly a B2 response

JANUARY/FEBRUARY
1973, VOLUME21, NUMBER1 13
RICHARD H. SMALL

engineer is called upon to assess the requirements and from (42), as appropriate. This value must be no larger
to adjust them to a condition of physical and economic than a few percent of Va.
realizability. Fig. 8, for example, frustrates the desires The driver is now specified by its most important
of many marketing managers who would be delighted to parameters rs, QEs, VAs, V_ and P_(m_×) as well as its
offer a one cubic foot (28 dma) air-suspension system voice-coil resistance RB which is typically 80% of the
giving flat response to 20 Hz at high efficiency, desired rating impedance. The system designer is faced
The desired response of a closed-box loudspeaker sys- with the problem of obtaining a driver which has the
tern may be based on amplitude, phase, delay or tran- required parameters. If he has a driver factory available,
sient considerations [13], but can always be reduced to he may have the required driver fabricated as described
a specification of/c and QTc. Once the response is speci- in the next section. If he does not possess this luxury, he
fled, either the enclosure volume VB or the reference must find a driver from among those available on the
efficiency ye may be specified independently; the other market.
will then be determined or restricted to a minimum or At present, very few of the loudspeaker drivers offered
maximum value. Finally, the power capacity may be for sale are provided with complete parameter informa-
specified in terms of either Pm_ or PAR' If both PrR and tion, either in the form above or any other. While this
PAR must be fixed independently, this will detemine ye situation will no doubt improve with time, particularly
and thus restrict VR as above, as increasing demands are made on manufacturers to
A typical set of design specifications might start with provide such information, today's system designer must
values of /c, QTc, Va and P^a, together with a rating obtain samples where possible and measure the param-
impedance which fixes RB. Unless a special amplifier is eters as described in [12]. The small-signal parameters
to be used, it can be assumed that QTc = Qrco. Note should be measured with the driver mounted on a stan-
that Va effectively specifies the enclosure; the design dard test baffle having an area of one or two square
problem is then to specify the driver, meters, e.g., [18, section 4.4.1], so that the diaphragm
The design process begins by assigning realistic values air load is approximately that which will apply to the
to Q._m and a. The value of QMe has only a relatively driver in the system enclosure.
minor effect on system performance through k,(0). As
noted in section 7, typical values are 2-5 for systems Example of System Design from Specifications
using filling material and 5-10 for unfilled systems. If
A closed-box air-suspension loudspeaker system to be
no better guide to the expected value of QMc is available, used with a high-damping-factor amplifier is to be de-
assume QMc = 5. The required value of QEc for the sys-
tem is then calculated from (9). signed to meet the following specifications:
If maximum efficiency consistent with the initial speci-
fications is desired, then the air-suspension principle must 1'3 40 Hz
be used. This requires that a be at least 3 or 4, but its
value will otherwise have only a small effect on system Response B2
performance through k,_c) and may be chosen to have Va 2 fta (56.6 dm a)
any value consistent with physical realizability of the
driver. If a is chosen too large, the driver will be found PAa 0.25 W program peaks; expected
to require unrealistically high compliance which, if realiz- peak/average ratio 5 dB.
able at all, may lead to poor mechanical stability of the
suspension. A suitable choice of _ is usually in the The enclosure is to be lined, but not filled. It is assumed
range of 3-10. that the enclosureand driver losses will correspondto
Next, the value of VAB is established. This is equal to Q_c = 5 and that it wilt be physically possible to obtain
Va for unfilled systems, but is increased by the factor a compliance ratio of a = 5.
1.4/yB (typically 1.15 to 1.2) if the enclosure is filled. The first two specifications translate directly into
The required driver small-signal parameters are then,
from (17) and (18), fc = 40Hz
and
fa =/o/(a+ 1)_, (59) Q_e = QTeo = 0.707.
QEs = QEC/( a-_' 1)_, (60) For Q_te = 5, (9) gives
and QEe = 0.824.

Vxs=c_VAB. (51) For a=5, (_+1)_ =¥6=2.45; so from (59) and

(60),
VxT is determined from (49). The reference efficiency
to be expected from the completed system is calculated /s = 16.3 Hz
from (24). Alternatively, kn(Q), k,(o) and k,(o) may be and
evaluated separately and ye determined from (26). The QEs = 0.336.
system electrical power rating P_R is then calculated from Also, for the unfilled enclosure, (51) gives
(58). A comparable or lower value is assigned to Pa(ma×),
depending on the peak-to-average power ratio of the VAs = 10ft a (283 dmZ).
program material with which the system will be used.
Then, from (49),
The required value of VD is calculated directly from
(39) using Fig. 5 or (78) to determine IX(]o01ma_, or VAe = 1.67ft a (47.2dina).
14 JOURNAL
OFTHEAUDIOENGINEERING
SOCIETY
 CLOSED-BOX LOUDSPEAKER SYSTEMS PART Ih SYNTHESIS

From (29), (30)and (31), about x_ax at each end. Also; the voice coil must be
capable of dissipating as heat, without damage, an elec-
k,(0) = 0.858, trical input power PE(lnax). This design problem is familiar
k,(c) = 0.833, to driver manufacturers.
k_(o) = 1.36 X 10 -0. The driver parameter Q_s usually plays a minor role
Thus in systemperformance,but it cannot be neglected en-
tirely. The value of Q_s in practical designs is often af-
k_ = 0.97 X 10 -6 fected by decisions related to performance at higher
and from (26), frequencies. Where the driver diaphragm is required to
be free of strong resonance modes at high frequencies,
VD----0.00351 or 0.35%. the outer edge suspension is usually designed to reflect

The reference efficiency can also be calculated directly a minimum of the vibrational energy travelling outward
from the voice coil through the diaphragm material. This
from (24) because/c, VAT and QEc are known.
The displacement-limited electrical power rating, from means that energy is dissipated in the suspension, and a
(58), is Iow value of Q_s results. The intended use of the
driver or the constructional methods preferred by the
P_n = 71.5 W. manufacturer thus determines the approximate value of

An amplifier of this power rating must be used to obtain Q_m. In a completed closed-box system, the value of
Q._s and the enclosure and filling material losses deter-
the specified acoustic output. For the expected peak/av- mine Q._m and therefore the value of k_(o) for the system.
erage power ratio, the thermal rating PE(,_×_ of the
driver must be at least 22.5 W.
Using (42) for the program power rating, Drivers for Air-Suspension Systems
VD = 3.4 X 10 -4 m 3 or 340 cm 0. It was stated earlier that the compliance ratio of an
air-suspension system is not very important so long as
This is only 0.6% of V,, so linearity of the air corn- ,it is greater than about 3 or 4. This means that the exact
pliance is no problem, values of driver compliance, resonance frequency and Q
are not of critical importance. It is in fact the moving
10. DRIVER DESIGN mass M_is and the electromagnetic damping Bele/R_ that
are of greatest importance. These can be calculated
General Method directly from the system parameters alone. Substituting
The process of system design leads to specification of (16), (17) and (18) into (61), (62) and (64), or using
the required driver in terms of basic parameters. These (3), (6), (8) and (25),
parameters are used to carry out the physical design M_ s = SD2MAc = poC2SD2/(4_r2fG2VAT), (66)
of the driver.
First, Vi) must be divided into acceptable values of and

SD and x ..... . The choice of S_) may have to be a compro- BelS//RB = 2rr/cM_m/QEc. (67)
mise among cost, distortion, and available mounting area.
The required mechanical compliance of the diaphragm The exact value of mechanical compliance is not
suspension is then critically important so long as it is high enough to give
approximately the desired compliance ratio. This is an
C_m = CAs/Sn 2 = VAS/(poC2SD_), (61) advantage of the air-suspension design principle, because
and the required total mechanical moving mass is mechanical compliance is one of the more difficult driver
parameters to control in production.
M_ s = 1/[ (2rrJs)eC_m]. (62)

This total moving mass includes any mass added by Example of Driver Design
filling material, as well as the air loads M3t_ and M_m on The driver required for the example in the previous
front and rear of the diaphragm. The latter can be evalu- section has the following parameter specifications:
ated from [I, pp. 216-217]. The mechanical mass of the
diaphragm and voice-coil assembly is then is = 16.3 Hz
: QEs = 0.336
M_iD = M_m- (M_l + M._m), (63) V^s = 283 dm s
less any allowance for mass added by filling material. V D = 340 cm a
The magnet and voice coil must provide electromag- PE(max) = 22.5 W

netic damping given by The driver size will probably have to be at least 12
B212/RB = 2rr/sM_s/Qr.s , (64) inches to meet the specifications of Vz) and Pr(_,x_. This
is checked by assuming a typical diaphragm radius of
or, for the value of RE specified, a BI product given by 0.12 m for the 12-inch driver, giving

Bl = (2rr/sREM_s/Q_s)_. (65) So = 4.5 X 10 -2 me.

This B1 product, together with the mechanical compli- For the required displacement volume of 340 cm s, the
ance, must be maintained with good linearity for a
peak linear displacement must be
diaphragm displacement of --Xma x. This effectively means
that the voice-coil overhang outside the gap must be x_ = VD/SD = 7.5 X 10 -a m = 7.5 mm (0.3 in).

JANUARY/FEBRUARY
1973, VOLUME21, NUMBER1 15
RICHARD H. SMALL
The total "throw" required is then 15 mm (0.6 in)which ment in an anechoic environment or by an indirect
is realizable in a 12-inch driver. By comparison, the same method [24].
displacement volume requires a throw of 22 mm (0.9
in) for a 10-inch driver, or 9.6 mm (0.38 in) for a 12. CONCLUSION
15-inch driver. The quantitative relationships presented in this paper
Continuing with the 12-inch design, make possible the low-frequency design of closed-box
So2 = 2.0 X 10 -am 4. systems by direct synthesis from specifications and clearly
show whether it is physically possible to realize a de-
The required mechanical compliance and mass are then, sired set of specifications. They should be useful to loud-
from (61)and (62), speaker systemdesignerswho wish to obtain the best
CMs = 9.9 × 10 -4 m/N, possible combination of small-signal and large-signal per-
M_s ----97 g. formance within the constraints imposed by a particu-
lar design problem.
M_ts is the total moving mass including air loads. As- These relationships should also be useful to driver
suming that the front air load is equivalent to that for manufacturers, because they indicate the range of basic
an infinite baffle and that the driver diaphragm occupies driver parameters needed for modern closed-box system
one-third of the area of the front of the enclosure, the design and the extent to which costly magnetic material
mass of the voice coil and diaphragm alone is must be allocated to satisfy both the small-signal and
MMv = M_s -- (3.14a a + 0.65_rpo
aa) -- 87 g. large-signal requirements of the system.
Because the low-frequency performance of a completed
The magnetic damping must be, from (64), system depends on a small number of easily-measured
Bel2/Rr = 30 N ' s/m (MKS mechanical ohms), system parameters, it is always possible to specify--and
verify--the low-frequency small-signal performance for
For an "8a" rating impedance, RB is typically about 6.5 standard free-field conditions. This information is of
fl. The required Bl product for the driver is then much greater value to users of loudspeakers than fre-
quency limits quoted without decibel tolerances and
B1 = 14T.m
without specification of the acoustic environment.
which must be maintained with good linearity over the It is sincerely hoped that the quantitative relationships
voice-coil throw of 15 mm (0.6 in). The voice coil must and physical limitations presented here--and in later
also be able to dissipate 22.5 W nominal input power papers .for other types of direct-radiator systems--will
[12, eq. (6)] without damage, not only be useful to system designers but will also con-
Further examples of driver synthesis based on system tribute eventually to more uniform, realistic and accurate
small-signal requirements are contained in [28]; the product specifications.
method used is based on the same approach taken above
but is arranged for automatic processing by time-shared 13. ACKNOWLEDGMENTS
digital computer. (The Thiele basic efficiency [17] used
This paper is part of the result of a program of post-
in this reference is based on a 4_r sr free-field load and
graduate research into the low-frequency performance of
gives one-half the value of the reference efficiency direct-radiator electrodynamic loudspeaker systems. I am
used here.) indebted to the School of Electrical Engineering of The
University of Sydney for providing research facilities,
11. DESIGN VERIFICATION supervision and assistance, and to the Australian Com-
The suitability of a prototype driver designed in ac- monweatth Department of Education and Science for
cordance with the above methods may be checked by financial support.
measuring the driver parameters as described in [12].x I am particularly grateful to J. E. Benson, R. F. Allison
For an air-suspension driver, it is not necessary that )'s, and R. H. Frater for reviewing early manuscripts and
QEs, and VAs have exactly the specified values. What is making valuable suggestions for improvement.
important is that the quantities/s2VAs and/s/Qr_s, which
together indicate the effective moving mass and electro- 14. APPENDIX--SECOND-ORDER
magnetic coupling, should check with the same corn- FILTER FUNCTIONS
binations of the specified parameters. Then, if VAs is General Expressions
large enough to give a satisfactory value of a for the
system, the driver design is satisfactory. Tables of filter functions normally give on'ly the de-
Similarly, the completed system may be checked by tails of a low-pass prototype function. The corresponding
measuring its parameters as described in section 6 and high-pass or band-pass forms are obtained by suitable
comparing these to the initial specifications. 1 The actual transformations. The general form of a prototype low-
system performance may also be verified by measure- pass second-order filter function, G_,(s), normalized to
unity in the passband, is
1
1A recent paper by Benson contains an improved method GL(s ) ---- (68)
of Q measurement which compensates for errors intro- s2To2+alsTo q- 1'
duced by large voice-coil inductance [32, Appendix 2]. The
compensation is achieved by replacing fo in eq. (45) of where TO is the nominal filter time constant, and the
Part I of this
expression paper-- measured
¥ [_[2.The and fs in [12, of
values eq.fo(17)I--with
and /s are the
un- coefficient a_ determines the actual filter characteristic.
changed, and no other equations are affected. The corresponding high-pass filter function, Gsl(s), which
16 JOURNAL OF THE AUDIO ENGINEERING SOCIETY
 CLOSED-BOX LOUDSPEAKER SYSTEMS PART Ih SYNTHESIS

preserves the same nominal time constant, is obtained Types of Responses

by the transformation The range of system alignments which may be ob-
Gu(BTu) = G_(1/STo).' (69) tained by varying QTo are thoroughly described in [13].
Particular alignments of interest, with brief character-
This gives the general high-pass expression istics, are:

s2To2 Butterworth maximally-flat-amplitude response (B2)

GH(s) ----- . (70)
s2To 2 -'}-alsT o + 1 [13], [29]

Equations (68) and (70) have exactly the same form Q_c = 1/¥ 2 = 0.707, la/fo = 1.000
as (20) and (19) for the displacement and response Bessel maxim_lly-flat-delay response (BL2) [13], [29],
functions of the closed-box system. The two sets of [30]
equations are equivalent for Q_o = 1/¥'-_= 0.577, /a/Jo = 1.272

TO = Tc and al -- 1/Q_e. (71) "Critically-damped" response [13]

Study of the steady-state magnitude-rs-frequency be- QTo = 0.500, ]s/lo = 1.554
havior of filter functions for sinusoidal excitation is
facilitated by using the magnitude-squared forms Chebyshev equal-ripple__ response (C2) [13], [31]
Q_o > 1/¥ 2, other properties given by (75)-

[GL(]co) 12 = 1 (72) (78). A very popular alignment of this type is

co4To4 -t- AltO2To2 + 1 QTo = 1.000, la/Jo = 0.786,
and lG(jrO)
Jmax= lxci_)lm_x= 1.155 or 1.25 dB.
oj4Io 4
IG_.4ico)J, = (73) REFERENCES
co4To4+ Al_2To 2 + 1 '
[1] L. L. Beranek, Acoustics (McGraw-Hill, New
where York, 1954).
A 1 = al 2- 2. (74) [2] H. F. Olson and J. Preston, "Loudspeaker Dia-
phragm Support Comprising Plural Compliant Members,"
U.S. Patent 2,490,460. Application July 19, 1944; pat-
Cutoff Frequency ented December 6, 1949.
The half-power frequency cea = 2=[a of the high-pass [3J H. F. Olson, "Analysis of the Effects of Nonlinear
function is obtained by setting (73) equal to _5 ,and Elements Upon the Performance of a Back-enclosed, Di-
rect Radiator Loudspeaker Mechanism," J. Audio Eng.
solving for co. Using (71) and (74), the normalized half- Soc., vol. 10, no. 2, p. 156 (April 1962).
power frequency of the closed-box system is given by [4] E. M. ViUchur, "Revolutionary Loudspeaker and
Enclosure," Audio, vol. 38, no. 10, p. 25 (Oct. 1954).
fa/fo = [5] E. M. Villchur, "Commercial Acoustic Suspension

· (75) [6/ E. M. Villchur, "Problems of Bass Reproduction

(1/QTc -- 2) + ¥ 2(1/QTc 2 -- 2) 2 + 4 -]
._] '_ Speaker," Audio, vol.
in Loudspeakers," J. 39,
Audiono. Eng.
7, p. 18
Soc.,(July
vol. 1955).
5, no. 3, p.
122 (July 1957).
· Frequencies of Maximum Amplitude [7] E. M. Villchur, "Loudspeaker Damping," Audio,
vol. 41, no. 10, p. 24 (Oct. 1957).
The frequency of maximum amplitude of either fre- [8] R. C. Avedon, W. Kooy and J. E. Burchfield, "De-
quency response or diaphragm displacement is found by sign of the Wide-Range Ultra-Compact Regal Speaker
taking the derivative of (72) or (73) with respect to System," Audio, vol. 43, no. 3, p. 22 (March 1959).
[9] E. M. Villchur, "Another Look at Acoustic Suspen-
frequency and setting this equal to zero. This yields for sion," Audio, vol. 44, no. 1, p. 24 (Jan. 1960).
the normalized frequency of maximum response [10] R. C. Avedon, "More on the Air Spring and the
Ultra-Compact Loudspeaker," Audio, vol. 44, no. 6, p. 22
1
fa ..... /lo = (76) (June 1960).
[1- 1/(2Qrc2)]¼ [11] R. F. Allison, "Low Frequency Response and Ef-
ficiency Relationships in Direct Radiator Loudspeaker
for Q'rc > 1/¥ 2. For QTe _- 1/¥ 2, fom×/]c is infinite. Systems," J. Audio Eng. Soc., vol. 13, no. 1, p. 62 (Jan.
The normalized frequency of maximum diaphragm 1965).
[12] R. H. Small, "Direct-Radiator Loudspeaker Sys-
displacement is tem Analysis," IEEE Trans. Audio and Electroacoustics,
Jx .... /Jo = [1 -- 1/(2Qre2)]¼ (77) vol. AU-19, no. 4, p. 269 (Dec. 1971); also J. Audio Eng.
Soc., vol. 20, no. 5, p. 383 (June 1972).
for QTc > 1/¥ 2. For Qu?c-_ 1/\I 2, fXma×/fc is zero. [13] J. E. BonBon, "Theory and Design of Loudspeaker
Enclosures, Part 2--Response Relationships for Infinite-
Baffle and Closed-Box Systems," A.W.A. Tech. Rev., vol.
Amplitude Maxima 14, no. 3, p. 225 (1971).
Substituting the above values of frequency into the ex- [14] J. D. Finegan, "The Inter-Relationship of Cabinet
Volume, Low Frequency Resonance, and Efficiency for
pressions for JG(jco)[2 and IX(]co)j2 corresponding to Acoustic Suspension Systems," presented at the 38th Con-
(72) and (73), the amplitude maxima are found to be vention of the Audio Engineering Society, May 1970.

Is(j_)l .... = [x(k0lm,_._=[ Q_c 4

Qe_).25 1, _ (78) [15] T.
Stabilized
sponse Matzuk,
in Small "Improvement
Negative-Spring
Loudspeaker of Low-Frequency
Principle,"
Systems byJ. Means Re-
Acous. of Soc.
the
Amer., vol. 49, no. 5 (part I), p. 1362 (May 1971).
for Qxc > 1/¥ 2, and unity otherwise. [16] W. H. Pierce, "The Use of Pole-Zero Concepts in

JANUARY/FEBRUARY
1973. VOLUME21. NUMBER1 17
RICHARD H. SMALL

Loudspeaker Feedback Compensation," IRE Trans. in Home Listening Rooms," J. Audio Eng. Soc., vol. 20,
Audio, vol. AU-8, no. 6, p. 229 (Nov./Dec. 1960). no. 6, p. 459 (July/Aug. 1972).
[17] A. N. Thiele, "Loudspeakers in Vented Boxes," [26] American standard recommended practices for
Proc. IREE (Australia), vol. 22, no. 8, p. 487 (Aug. 1961). loudspeaker measurements, ASA Standard S1.5-1963,
Also, J. Audio Eng. Soc., vol. 19, no. 5, p. 382; no. 6, p. New York, 1963.
471 (May, June 1971). [27] British standard recommendations for ascertaining
[18] IES Recommendation, Methods of Measurement and expressing the performance oJ loudspeakers by ob-
for Loudspeakers, IEC Publ. 200, Geneva (1966). jective measurements, British Standards Institution Stan-
[19] J. King, "Loudspeaker Voice Coils," J. Audio dard B.S. 2498, London, 1954.
Eng. Soc., vol. 18, no. 1, p. 34 (Feb. 1970). [28] J. R. Ashley, "Efficiency Does Not Depend on
[20] J. R. Ashley and T. A. Saponas, "Wisdom and Cone Area," J. Audio Eng. Soc., vol. 19, no. 10, p. 863
Witchcraft of Old Wives Tales About Woofer Baffles," J. (November 1971).
Audio Eng. Soc., vol. 18, no. 5, p. 524 (Oct. 1970). [29] L. Weinberg, Network Analysis and Synthesis,
[21] J. L. Grauer, "Acoustic Resistance Damping for Chapter 11 (McGraw-Hill, New York 1972).
Loudspeakers," Audio, vol. 49, no. 3, p. 22 (March [30] A. N. Thiele, "Techniques of Delay Equalisation,"
1965). Proc. IREE (Australia), vol. 21, no. 4, p. 225 (April
[22] V. Brociner, "Speaker Size and Performance in 1960).
Small Cabinets," Audio, vol. 54, no. 3, p. 20 (March [31] A. N. Thiele, "Filters With Variable Cut-Off Fre-
I970). quencies," Proc. 1REE (Australia), vol. 26, no. 9, p. 284
[23] P. W. Klipsch, "Modulation Distortion in Loud- (Sept. 1965).
speakers," J. Audio Eng. Soc., vol. 17, no. 2, p. 194 [32] J. E. Benson, "Theory and Design of Loudspeaker
(April 1969); Part 2: vol. 18, no. 1, p. 29 (Feb. 1970). Enclosures Part 3--Introduction to Synthesis of Vented
[24] R. H. Small, "Simplified Loudspeaker Measure- Systems," A.W.A. Tech. Rev., vol. 14, no. 4, p. 369
ments at Low Frequencies," Proc. IREE (Australia), vol. (November 1972).
32, no. 8, p. 299 (Aug. 1971); also J. Audio Eng. Soc.,
vol. 20, no. 1, p. 28 (Jan./Feb. 1972). Note: Dr. Small's biography appeared in the December
[25] R. F. Allison and R. Berkovitz, "The Sound Field 1972 issue of the Journal.

18 JOURNALOF THEAUDIO ENGINEERINGSOCIETY