Você está na página 1de 9

Applied Acoustics 132 (2018) 109–117

Contents lists available at ScienceDirect

Applied Acoustics
journal homepage: www.elsevier.com/locate/apacoust

Multi-degree-of-freedom low-frequency electroacoustic absorbers through T


coupled resonators

Etienne Rivet , Sami Karkar, Hervé Lissek
Signal Processing Laboratory LTS2, École polytechnique fédérale de Lausanne, 1015 Lausanne, Switzerland

A R T I C L E I N F O A B S T R A C T

Keywords: Electroacoustic absorbers represent an interesting solution for low-frequency sound absorption in rooms. These
Sound absorption systems simply consist of closed-box electrodynamic loudspeakers, whose acoustic impedance at the diaphragm
Acoustic impedance control is judiciously adjusted by connecting a passive or active electrical control circuit. This paper presents a method
Electrodynamic loudspeaker for designing different electroacoustic absorber systems constituted of simple electrical and mechanical com-
Performance optimisation
ponents that are coupled to a primary loudspeaker, resulting in multi-degree-of-freedom resonators. Each system
Electrical shunt
is optimised to maximise the sound absorption performance with respect to different metrics. Experimental
evaluations in an impedance tube confirm the model accuracy and method efficiency for achieving low-fre-
quency sound absorption.

1. Introduction to maximize their sound absorption performance in rooms [18]. When


an appropriate electrical resistance is connected to the transducer
Room modes cause uneven distributions in space and frequency of terminals, an optimal acoustic resistance can be achieved at the dia-
the sound field, thus altering the sound quality [1]. Conventional pas- phragm, but limited to the resonance frequency of the system [19].
sive techniques based on foam-based absorbers and diffusers are often With parallel resistance - capacitance (RC) or resistance - inductance
used to control the reverberation time and early reflections [2]. Since (RL) electrical networks, the peak of sound absorption can be tuned
their size is dictated by the wavelengths in the targeted frequency below or above the transducer resonance frequency respectively, thanks
range, they are not well suited to the low frequencies. Resonant ab- to the reactive electronic components [20,21]. Connecting a series re-
sorbers and Helmholtz resonators are available for low-frequency sound sistance - inductance - capacitance (RLC) electrical network to the
absorption [3–5]. Nevertheless, even if the design of Helmholtz re- transducer terminals becomes a two-DOF resonator, resulting in two
sonators can be optimised for a given room, their high quality factor peaks of sound absorption [21].
causes a narrow frequency band of efficient sound absorption [6]. Thus, Such Helmholtz resonators and shunted transducers can even be
Helmholtz resonators with two and three degrees of freedom (DOF), combined together. A multi-DOF electromechanical Helmholtz re-
which consist of pairs of cylindrical necks and cavities stacked in series, sonator, consisting of a Helmholtz resonator coupled to a shunted
were designed to improve the sound absorption capabilities [7–9]. The piezoelectric at the bottom of the cavity, was developed in Ref. [22].
effect of resonator arrays on the sound field in cavities were evaluated This way, both resonance frequencies were tuned thanks to the shunted
in Refs. [10,11]. These resonators can also be combined with micro- electrical load. The low-frequency sound absorption was also efficiently
perforated panels constituted of very thin perforations backed by a improved using a thin micro-perforated plate coated with a shunted
cavity, which were firstly introduced in Ref. [12], so as to improve the piezoelectric transducer [23]. These different approaches introduce the
sound absorption performance at higher frequencies [13,14]. Recently, idea of using mechanical or mixed electrical/mechanical resonators
an original design constituted of panels arranged with parallel extended coupled to the primary closed-box loudspeaker interacting with the
tubes, was proposed in Ref. [15], resulting in four peaks of absorption sound field, so as to imitate or improve the sound absorption cap-
from 150 Hz to 440 Hz. abilities relative to those obtained with electrical shunts.
Another approach is the active sound absorption with electro- The objective of this paper is to design innovative systems of multi-
acoustic absorbers. These active absorbers are closed-box electro- DOF electroacoustic absorber, which are only constituted of conven-
dynamic loudspeaker systems, whose acoustic impedance at the dia- tional electrical and mechanical components. First, the model of the
phragms is judiciously adjusted with [16] or without sensor [17], so as closed-box electrodynamic loudspeaker is introduced, before presenting


Corresponding author.
E-mail address: etienne.rivet@epfl.ch (E. Rivet).

http://dx.doi.org/10.1016/j.apacoust.2017.10.019
Received 30 March 2017; Received in revised form 3 October 2017; Accepted 10 October 2017
Available online 24 November 2017
0003-682X/ © 2017 Elsevier Ltd. All rights reserved.
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

the sound absorption performance through the definitions of the spe- sound absorption coefficient α (ω) are defined as
cific acoustic impedance, as well as the corresponding sound reflection
Zs (ω)−ρc
coefficient and sound absorption coefficient. Then, different systems of r (ω) = and α (ω) = 1−|r (ω)|2
Zs (ω) + ρc (5)
electrical and mechanical resonators coupled to the primary closed-box
loudspeaker are studied, and the sound absorption performance of each In the following, three different systems of electroacoustic absorber
system is optimised with respect to specific objective functions. Finally, are investigated:
an experimental evaluation of these systems and a discussion on the
measured sound absorption performance are given. • System A: the loudspeaker is coupled to an electrical resonator.
2. Acoustic impedance of multi-degree-of-freedom electroacoustic
• System B: the primary loudspeaker is coupled to a secondary loud-
speaker, where an electrical resistance is connected at its terminals,
absorbers which is denoted the “mechanical resonator”.

2.1. Passive loudspeaker


• System C: the primary loudspeaker is coupled to a secondary loud-
speaker, where an electrical resonator is connected at its terminals,
which is denoted the “electromechanical resonator”.
In the low-frequency approximation, an electrodynamic loud-
speaker can be modeled as a one-DOF oscillator mechanically driven by 2.2. Loudspeaker connected to an electrical resonator
a voice coil within a magnetic field. All forces acting on the transducer,
especially those resulting from the sound pressures Pf and Pr at the front When an appropriate electrical load is connected at the loudspeaker
and rear faces of the diaphragm, are assumed small enough so that the terminals, the active sound power at the diaphragm can efficiently be
governing equations should remain linear. The mechanical part is dissipated through mechanical and electrical losses, thus modifying its
modeled as a simple mass - spring - damper system in the low-frequency sound absorption capability. The electrical load, represented by the
range, that is the mass Mms , the mechanical compliance Cms accounting complex impedance Zl (ω) , is added in series to the blocked electrical
for the surround suspension and spider, and the mechanical resistance impedance Ze (ω) . When the voice coil moves with the velocity V (ω) ,
Rms , respectively. The electrical part is modeled with a DC resistance R e the electromotive force generates the current I (ω) circulating through
and a self inductance Le . We denote the effective piston area by Sd , the the closed electrical circuit, which is expressed as
force factor by Bl, the incoming diaphragm velocity by V, the electrical
Bl
current flowing through the voice coil by I, and the input voltage be- I (ω) = V (ω)
tween the electrical terminals by U. Fig. 1 shows the schematic diagram Ze (ω) + Zl (ω) (6)
of the electrodynamic loudspeaker. The study focuses on the electrical resonator constituted of the
Using the Fourier transform where ω is the angular frequency, the series RLC network, as highlighted in Fig. 2 (in shaded area). The
governing equations of the loudspeaker are expressed as electrical impedance of this shunt network is written as
S2 1
Sd Pf (ω) = ⎜⎛Zms (ω) + d ⎟⎞ V (ω) + BlI (ω) Zl (ω) = jωLl + Rl +
⎝ jωCab ⎠ (1) jωCl (7)

U (ω) = Ze (ω) I (ω)−BlV (ω) (2) The equivalent acoustic circuit of the closed-box loudspeaker con-
nected to this electrical resonator (system A) is illustrated in Fig. 3,
where Ze (ω) = jωLe + R e is the blocked electrical impedance of the where Q (ω) = Sd V (ω) is the volume flow. Combining Eq. (6) with Eqs.
voice coil, Zms (ω) = jωMms + Rms + 1/(jωCms ) is the mechanical im- (1) and (2) gives the expression of the specific acoustic impedance of
pedance of the loudspeaker, Cab = Vb/(ρc 2) is the acoustic compliance of system B as
the enclosure, ρ is the density of the medium and c is the speed of
sound. Zms (ω) Sd ZmeA (ω)
Z sA (ω) = + +
The specific acoustic impedance is defined as the complex ratio of Sd jωCab Sd (8)
the sound pressure Pf (ω) at the front face of the diaphragm to the
where
diaphragm velocity V (ω) . When the loudspeaker is left open circuit,
namely the case where no electrical current I (ω) circulates through the (Bl)2
ZmeA (ω) = 1
voice coil, this quantity is directly derived from Eqs. (1) and (2) as jω (Le + Ll ) + R e + Rl + jωCl (9)
Z (ω) Sd
Zs (ω) = ms + corresponds to the equivalent mechanical impedance of the electrical
Sd jωCab (3)
resonator in series with the electrical impedance of the loudspeaker.
The corresponding resonance frequency is equal to The corresponding resonance frequency of the electrical resonator is
1
f0 =
Cms Cab
2π Mms
Sd2Cms + Cab (4)
Under normal incidence, the sound reflection coefficient r (ω) and Sd Pr
Sd Pf + I
V Rl
U Ll
Cl
-

Fig. 2. System A: the loudspeaker connected to an electrical resonator (series resistance –


Fig. 1. Schematic diagram of the electrodynamic loudspeaker. inductance – capacitance network).

110
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

pressure can be considered uniform in the frequency range of interest.


The sound pressure Pr 1 (ω) at the rear face of the diaphragm #1 and the
sound pressure P f2 (ω) at the front face of the diaphragm #2 are ex-
pressed as
1
Pr 1 (ω) = P f2 (ω) = (Sd V (ω)−Sd2 V2 (ω))
jωCab1 1 1 (13)

Fig. 3. Equivalent acoustic circuit of system A. where Cab1 = Vb1 /(ρc 2) .


The sound pressure Pr 2 (ω) at the rear face of the
diaphragm #2 is expressed as
equal to Sd2
Pr 2 (ω) = V (ω)
1 jωCab2 2 (14)
f0A =
2π (Le + Ll ) Cl (10) where Cab2 = Vb2 /(ρc 2) .
The coupled equations of motion of the loud-
The poles of the specific acoustic impedance Z sA (ω) are the same as speaker diaphragms can then be written as1
those of the equivalent mechanical impedance ZmeA (ω) , but the zeros Sd1 2 ⎞ Sd1 Sd2

⎜Zms1 (ω) + ⎟V1 (ω)− V (ω) = Sd1 P f1 (ω)−Bl 1 I1 (ω)
cannot be expressed analytically in a simple way. The real and ima- jωC jωCab1 2
⎝ ab1 ⎠ (15)
ginary parts of the specific acoustic impedance Z sA (ω) are derived from
Eqs. (8) and (9) as Sd1 Sd2 Sd2 2 Sd2 2 ⎞
V1 (ω)−⎛⎜Zms2 (ω) + + ⎟V (ω) = Bl 2 I 2 (ω)
Rms (Bl)2 (R e + Rl ) jωCab1 ⎝ jωCab1 jωCab2 ⎠ 2 (16)
Re (Z sA (ω)) = +
Sd 1 2
Sd ⎡ (R e + Rl

)2 (
+ ω (Le + Ll )− ωC ⎤
l )⎦ (11)
The equivalent acoustic circuit of system B is illustrated in Fig. 5,
where Q1 (ω) = Sd1 V1 (ω) and Q2 (ω) = Sd2 V2 (ω) . Combining Eqs. (6),
(13) and (14) with Eqs. (15) and (16) gives the expression of the spe-
Im (Z sA (ω)) =
1
Sd

⎡ωMms− 1
ω ( 1
Cms
+
Sd2
Cab ) ⎤⎦ cific acoustic impedance
1 ⎞
(Bl)2 ⎛ω (Le + Ll) −

ωCl ⎠

Zms1 (ω) Sd1 (Bl 1)2 ZmeB (ω)
− ⎝ Z sB (ω) = + + +
⎡ 1 ⎞ ⎤
2 Sd1 jωCab1 Sd1 (Ze1 (ω) + Rl1 ) Sd1 (17)
Sd ⎢ (R e + Rl)2 + ⎛ω (Le + Ll) −
ωCl ⎠ ⎥
⎜ ⎟

⎣ ⎝ ⎦ (12)
where
For the reactive part, if the values of inductance Ll and capacitance 2
Cl are well chosen, the electrical resonator can partly decrease both
mass and stiffness of the closed-box loudspeaker over a certain fre- ZmeB (ω) = −
( Sd1 Sd2
jωCab1 )
Sd2 2 (Bl 2)2
quency range. Zms2 (ω) + jω ( 1
Cab1
+
1
Cab2 )+ Ze2 (ω) + Rl2 (18)

2.3. Loudspeaker coupled to a mechanical resonator corresponds to the equivalent mechanical impedance of the mechanical
resonator. Just like the force factor Bl for system A, the coupling
The main motivation of the method presented in this paper is to be coefficient Sd1 Sd2/ Cab1 plays a crucial part for the sound absorption
able to mimic, and possibly improve, the sound absorption performance performance of this system. The choice of values for both surface areas
of system A, where the primary loudspeaker is coupled to the electrical and volume of enclosure between both loudspeakers are thus important
resonator, by coupling the primary loudspeaker with a “mechanical for the coupling.
resonator”. Thanks to the resonant behaviour of the loudspeaker, sev-
eral loudspeakers in boxes may be stacked to increase the order of the 2.4. Loudspeaker coupled to an electromechanical resonator
resonant system as well as the sound absorption bandwidth. Shunt re-
sistors are connected to the loudspeakers terminals to adjust the The primary loudspeaker might also be coupled to a combination of
acoustic resistances of the diaphragms. System B is constituted of two two (or more) resonators of different types to improve the sound ab-
loudspeakers, whose parameters are denoted by indices i = [1,2] in sorption bandwidth. Here, the series RLC electrical network substitutes
Fig. 4. Each loudspeaker is connected to a shunt resistor Rli and loaded for the shunt resistor connected to the secondary loudspeaker terminals
in a closed box. In this configuration, the secondary closed-box loud- (that is the mechanical resonator), as illustrated in Fig. 6. This sub-
speaker system (shaded area) is called here the “mechanical resonator”. system denoted “electromechanical resonator” becomes a two-DOF re-
Within each of the two volumes Vb1 and Vb2 of the enclosures, the sound sonator.
The equivalent acoustic circuit of the primary loudspeaker coupled
to the electromechanical resonator (system C) is illustrated in Fig. 7.
Substituting an electrical impedance Zl2 for the electrical resistance Rl2
in Eq. (18) gives the expression of the specific acoustic impedance
Sd1 Pr1 I1 Sd2 Pr2 I2 Zms1 (ω) Sd1 (Bl 1)2 ZmeC (ω)
Sd1 Pf1 + Sd2 Pf2 + Z sC (ω) = + + +
Sd1 jωCab1 Sd1 (Ze1 (ω) + Rl1 ) Sd1 (19)
V1 V2
U 1 Rl1 U 2 Rl where
2

2
- -
ZmeC (ω) = −
( Sd1 Sd2
jωCab1 )
Sd2 2 2
Zms2 (ω) + jω ( 1
Cab1
+
1
Cab2 )+ (Bl 2)
Ze2 (ω) + Zl2 (ω) (20)
corresponds to the equivalent mechanical impedance of the
Fig. 4. System B: the primary loudspeaker is coupled to a secondary loudspeaker. Both
loudspeakers are connected to shunt resistors and loaded in closed boxes. 1
B1 l1 and B2 l2 are noted Bl1 and Bl2 respectively to avoid numerous subscripts.

111
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

Fig. 5. Equivalent acoustic circuit of system B.

This objective function maximises the bandwidth BW of efficient


sound absorption.
• Area A over the threshold (in Hz): it corresponds to the area be-
Sd1 Pr1 I1 Sd2 Pr2 I tween the curve of the sound absorption coefficient in Eq. (5) and
Sd1 Pf1 + Sd2 Pf2 + 2 the threshold αth of minimal efficient sound absorption
V1 V2 Rl2 f+
U 1 Rl1 U2 L A= ∫f (α (f )−αth) df
l2 (22)

Cl2
- - It corresponds to the weighted bandwidth wBW minus the band-
width BW of efficient sound absorption weighted by the threshold
αth of minimal efficient sound absorption. Instead of favouring a
large bandwidth BW of efficient sound absorption with a “low”
maximal value of sound absorption coefficient, this objective func-
Fig. 6. System C: two loudspeakers loaded in closed boxes, one connected to a shunt
tion maximises the sound absorption coefficient (that is near a
resistor, the other connected to a series resistance - inductance - capacitance network. perfect sound absorption), over a frequency range as large as pos-
sible. This quantity will be considered for the optimisation process,
but both quantities will be studied.
electromechanical resonator. If the electromechanical resonator is
properly designed, this system behaves as a three-DOF resonator with
In the following, the lower and upper bounds f− and f+ and centre
three distinct peaks of absorption. Note that another electrical re-
frequency fc = f− f+ of each system will be given as additional in-
sonator could be connected to the primary loudspeaker as well, but the
dicators.
present article only investigates the 3-DOF system described above, for
Considering a Visaton AL-170 loudspeaker, the model parameters
the sake of simplicity.
are reported in Table 1. The density of the air at 294 K is ρ = 1.2 kg m−3
and the speed of sound in the air at 1 atm is c = 344 m s−1. The me-
3. Optimisation of the multi-degree-of-freedom electroacoustic chanical parameters were estimated from the measured specific
absorbers acoustic impedance at the diaphragm. The loudspeaker was in open
circuit, mounted at one end of a standing-wave duct in a closed box of
3.1. Strategy volume Vb = 10.1 cm3, and excited by a broadband noise source at the
other end. The experimental setup is detailed in Section 4.1. The
In Section 2, the specific acoustic impedance expressions of systems acoustic radiation impedance, which depends on the environment in
A, B, and C given in Eqs. (8), (17) and (19) respectively show the which the loudspeaker is located (here a waveguide), was taken into
complexity to analyse the effect of each system model parameter. To account in the mechanical impedance Zms (ω) .
maximise the sound absorption performance, parametric optimisations
are proposed through two objective functions using the simplex search 3.2. Choice of the secondary loudspeaker
method developed in [24]. First, a bandwidth BW of efficient sound
absorption is defined as the frequency range over which the total sound The preliminary step is the selection of the secondary loudspeaker
intensity in front of the diaphragm is less than twice the total sound from the optimisation process. Note that, for systems B and C, the
intensity in the ideal case α = 1. This condition is equivalent to mechanical resistance Rms2 of the secondary loudspeaker is not the most
Pf (ω) ⩽ 2 Pi (ω) , where Pi (ω) is the sound pressure of the incident critical parameter, since the total resistance at the denominator of the
wave. This criterion corresponds to a threshold value of minimal effi- equivalent mechanical impedance ZmeB (ω) of the mechanical resonator
cient sound absorption αth = 1−( 2 −1)2 ≃ 0.83. 2 The studied objective
in Eq. (18) can be adjusted thanks to the electrical resistance Rl2 . Same
functions, which are illustrated in Fig. 8, are derived from this criterion
remark applies for the mechanical compliance Cms2 , since the value of
and presented below:
the total compliance at the denominator of the equivalent mechanical
impedance ZmeB (ω) can be increased thanks to the volume Vb2 . As a
• Weighted bandwidth wBW (in Hz): it corresponds to the area de- result, only the effective piston area Sd2 and mass Mms2 may have a
limited by the curve of the sound absorption coefficient between the
significant impact on the performance. Since only one of the three
lower and upper bounds f− and f+ of the bandwidth BW of efficient
parameters of the coupling coefficient Sd1 Sd2/ Cab1 in Eq. (18) is imposed,
sound absorption
the optimisation is processed for every value of volume Vb1. This volume
f+ Vb1 varies from 1 L to 4 L to favour a small absorber. The mass Mms2 , the
wBW = ∫f −
α (f ) df
(21) effective piston area Sd2 , the shunt resistance Rl1, and the volume Vb2 are
free to vary during every optimisation step. The mechanical compliance
Cms2 is arbitrarily fixed and large enough (Cms2 = 1.5 mm N−1) since the
2
The sound pressure level in front of the diaphragm is denoted Li for the ideal case
volume can only increase the stiffness. The term
(α = 1). It is equal to Li + 3 dB for the threshold case (α = 0.83 ), Li + 4.6 dB for α = 0.5 ,
and Li + 6 dB for the worst case (α = 0 ). Thus, the threshold definition corresponds to
Rms2 + (Bl 2)2 /(Ze2 (ω) + Rl2) in Eq. (18) is replaced by a resistance (here
both, a −3 dB criterion in power reduction, relative to perfect sound absorption, and a the inductance Le2 is neglected) and is free to vary during every opti-
−3 dB criteriund on in power absorption relative to the worst case (rigid wall). misation step.

112
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

Fig. 7. Equivalent acoustic circuit of system C.

secondary loudspeaker should have an effective piston area much


smaller than that of the primary loudspeaker, at least for this chosen
model. Here, lightweight loudspeakers, whose effective piston area is
between 40 cm2 and 60 cm2, are good candidates for maximising the
sound absorption performance. The Monacor SP-60/8 loudspeaker is
then considered the secondary loudspeaker for systems B and C. The
model parameters, which were estimated from the same experimental
setup as for the primary loudspeaker, are reported in Table 1. The
loudspeaker was also in open circuit and mounted in a closed box of
volume Vb = 10.6 cm3. In the following, the model parameters of the
secondary loudspeaker will thus be fixed and will not be taken into
account for the optimisation of sound absorption performance of sys-
tems B and C.

Fig. 8. Schematic of the objective functions.


3.3. Performance analysis
The optimal values of the effective piston area Sd2 and mass Mms2 are
illustrated in Fig. 9 in function of the volume Vb1. The effective piston For system A, where the closed-box loudspeaker is coupled with the
area Sd2 increases with the volume Vb1, while the mass Mms2 is minimal electrical resonator, the volume Vb1 is fixed and equal to the one iden-
for values around Vb1 = 2.5 L. The area A over the threshold increases tified in the optimisation of system C. The electrical shunt parameters
linearly with the volume Vb1 (from 12.53 Hz to 12.90 Hz). Note that the Rl1, Ll1, and Cl1 may vary without constraint during the optimisation. For
system B, where the primary loudspeaker is coupled with the me-
Table 1 chanical resonator, the volume Vb1 is also fixed and equal to the one
Loudspeaker model parameters of the Visaton AL-170 and Monacor SP-60/8. found in the optimisation of system C. The shunt resistances Rl1 and Rl2
and the volume Vb2 are free to vary during the optimisation. For system
Visaton AL-170 Monacor SP-60/8 Unit
C, where the primary loudspeaker is coupled with the electro-
Notation Value Notation Value mechanical resonator, the secondary loudspeaker is the same as the one
used for system B. The electrical shunt components Rl1, Rl2, Ll2, Cl2 , and
Sd1 133 Sd2 55 cm2 the volumes Vb1 and Vb2 may vary without constraint during the opti-
Mms1 13.2 Mms2 6.5 g misation. This way, the performance of the three systems can be com-
Rms1 0.63 Rms2 0.97 N s m−1
pared for a same volume Vb1 relative to the basic configuration, that is
Cms1 574 Cms2 852 μ m N−1
when the closed-box loudspeaker system is in open circuit and is not
Bl1 6.9 Bl2 4.8 N A−1
Le1 0.9 Le2 0.3 mH coupled to any additional resonator.
R e1 5.8 R e2 6.8 Ω The parameter values resulting from the optimisation for systems A,
fs1 58 fs2 68 (Hz) B, and C are summarised in Table 2. The fixed and optimised

Table 2
Values of the fixed and optimised parameters and corresponding sound absorption per-
100 8 formance indicators according to the objective function A .

90 System A System B System C


7

80 Fix. Vb1 3
(dm ) 2.14 2.14 –
6
Optimised Parameters Vb1 (dm3) – – 2.14
Sd2 (cm2)

Mms 2 (g)

70 Rl1 (Ω ) −0.1 ∞ ∞
5 Ll1 (mH) 9.0 – –
60 Cl1 ( μ F) 99.3 – –
4 Vb2 (dm3) – 1.2 1.3
50
Rl2 (Ω ) – 0.9 −0.1
3 Ll2 (mH) – – 11.5
40
Cl2 ( μ F) – – 109.2

30 2 Perf. Indicators fc (Hz) 160.1 142.6 139.9


1 1.5 2 2.5 3 3.5 4
f− (Hz) 117.6 105.4 95.1
Vb (L)
1 f+ (Hz) 219.8 198.9 208.0
Fig. 9. Optimal values of the effective piston area Sd2 and mass Mms2 displayed in wBW (Hz) 97.06 89.76 107.52
A (Hz) 12.41 12.27 14.00
function of the volume Vb1.

113
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

as low as possible to maximise the sound absorption performance.


Systems A and B present nearly the same value A , which is larger for
system C. System B presents a smaller weighted bandwidth wBW than
system A, but it is larger for system C. If the frequency ranges of effi-
cient sound absorption between the systems are not taken into account
for the comparison, system C provides the best sound absorption per-
formance, in terms of area A over the threshold and weighted band-
width wBW, thanks to the three peaks of resonances.
To understand the effects of the additional resonators on the sound
absorption capabilities, the real and imaginary parts of the specific
acoustic impedance at the primary loudspeaker diaphragm of each
system, according to the results of the objective function A , are given
in Fig. 11. For system A illustrated in Fig. 11a, the real part of the
specific acoustic impedance ZmeA (ω)/ Sd1 of the electrical resonator in
Eq. (11) prevails over that of the closed-box loudspeaker at low fre-
quencies. On the other hand, the imaginary part of the additional re-
sonator has an influence on the system impedance only around the
resonance frequency. Beyond this range, the closed-box loudspeaker
reactance prevails. Note that the resonance frequency of the closed-box
loudspeaker f0 and the one of the electrical resonator f0A are equal to
maximise the sound absorption capabilities. As can be seen in Eqs. (11)
and (12), the expression of the equivalent mechanical impedance
ZmeA (ω) of the electrical resonator is equivalent to the inverse of the
mechanical impedance of the closed-box loudspeaker. Their behaviours
are reversed over the frequency band of interest. This way, the ima-
ginary part of the overall specific acoustic impedance Z sA (ω) of the
system is closer to zero over a greater frequency range than in the case
of the basic configuration (open circuit), thus increasing the sound
absorption capabilities around the resonance frequency. For system B
illustrated in Fig. 11b, the real and imaginary parts of the specific
acoustic impedance ZmeB (ω)/ Sd1 of the mechanical resonator are not
symmetrical because of the double pole involving the coupling coeffi-
cient Sd1 Sd2/ Cab1 in Eqs. (18) and (20). Nevertheless, the behaviours of
the primary loudspeaker and secondary loudspeaker are reversed as
well, leading to sound absorption capabilities similar to system A. Fi-
nally, for system C illustrated in Fig. 11c, the specific acoustic im-
pedance Z sC (ω) is a “mix” of those obtained with systems A and B. This
Fig. 10. (a) Bode plot of the specific acoustic impedance and (b) corresponding sound
system is optimised so as to exhibit three resonances, providing the best
absorption coefficient of the electroacoustic absorber computed in the case of systems A, sound absorption performance of the three systems.
B, and C after optimisation, and for the basic configuration.
4. Experimental results and discussion
parameters, and corresponding sound absorption performance in-
dicators, namely the centre frequency fc , the bounds f − and f+ of 4.1. Experimental setup
bandwidth BW of efficient sound absorption, the weighted bandwidth
wBW, and the area A over the threshold, are given. The Bode plot of To experimentally validate the results found in Section 3, a wave-
the specific acoustic impedance computed in the three cases is dis- guide was designed (length L = 1.97 m and internal diameter
played in Fig. 10a, and the expected performance in terms of sound ∅ = 150 mm) as depicted in Fig. 12. The duct was closed by each system
absorption coefficient is given in Fig. 10b. one after another at one end and by a closed-box electrodynamic
For the basic configuration when the closed-box loudspeaker system loudspeaker, delivering a band-limited pink noise of bandwidth [2 Hz
is in open circuit, the corresponding sound absorption coefficient is to 2 kHz] at the other end. The specific acoustic impedance and cor-
below the threshold αth of minimal efficient sound absorption defined in responding sound absorption coefficient were evaluated according to
Section 3.1. Thanks to the additional resonators presented in the dif- ISO 10534-2 standard [25]. Three 1/2” PCB 378B02 microphones were
ferent systems, the sound absorption performance increase in terms of wall-mounted at positions x1 = 1.02 m, x2 = 1.51 m, and x3 = 1.62 m
weighted bandwidth wBW and area A over the threshold. For system from the sound source, measuring the sound pressures
A, as discussed in Section 2.2, connecting an electrical resonator to the p1 = p (x1,t ), p2 = p (x2,t ) , and p3 = p (x3,t ) . The frequency response
loudspeaker terminals causes the stiffness and mass of the primary functions H13 = p3 / p1 and H23 = p3 / p2 were processed through a Brüel
loudspeaker to be reduced over a relatively narrow frequency band. The and kJær Pulse multichannel analyser (type 3160). With this setup, the
sound absorption performance increases around the resonance fre- performance is evaluated for plane waves under normal incidence over
quency. For systems B and C, as the total volume Vb1 + Vb2 is higher than a frequency range up to 1340 Hz. Nevertheless, the displayed frequency
for system A, the centre frequency f0′ is lower. Note that the volume Vb1 range is reduced from 63 Hz to 400 Hz to focus the analysis on the
between both loudspeakers is slightly larger than the volume Vb2 for this bandwidth over which the sound absorption is supposed to be efficient.
combination of loudspeakers. Also, the connection of a resistance to the For the experimental evaluation, the three systems were made and
terminals of the loudspeaker #1 is useless, since the results of Table 2 measured, according to the optimal parameters found from the opti-
shows that any additional acoustic resistance is not required for this misation. Since the target parameter values given in Table 2 were not
loudspeaker. This result even suggests that the value Rms1/ Sd1 should be exactly realised in practice, the measured values of the actual para-
meters of the experimental systems are given in Table 3. The setting

114
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

Fig. 11. Real and imaginary parts of the specific acoustic impedance of the electroacoustic absorber computed in the case of (a) system A, (b) system B, and (c) system C after
optimisation.

configurations used for the measurements were the same as those not changed for system B. The volumes taken by the loudspeakers in the
presented in Section 3.3. First, the primary loudspeaker was connected boxes were taken into account in the design of the prototype, to get
to a series electrical RLC network in a closed box of volume equivalent volumes very close to the values reported in Table 2. For the
Vb = 2.17 dm3. Then the primary loudspeaker was coupled to a sec- series RLC networks used for systems A and C, as the resistive parts of
ondary loudspeaker, in a layout of closed boxes of volumes inductances and capacitances have non-zero values, a current inversion
Vb1 = 2.17 dm3 and Vb2 = 1.34 dm3, which almost correspond to the op- negative impedance converter, including two resistors R and a resistor
timal values for system C presented in Section 3. The volume Vb2 was Rn as illustrated in Fig. 12, was used. The resistive value of the networks

Fig. 12. Schematic of the experimental setup for system C when the primary loudspeaker is coupled to the electromechanical resonator.

115
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

Table 3 acoustic impedance at the primary loudspeaker diaphragm and corre-


Measured values of the parameters and corresponding sound absorption performance sponding sound absorption coefficient is shown in Fig. 13. The mea-
indicators.
sured sound absorption performance indicators are given in Table 3.
System A System B System C Thanks to the coupling, both the acoustic resistance and reactance of
each system are changed to reach the desired effect according to the
Parameters Vb1 (dm3) 2.17 2.17 2.17 optimisation results (see Section 3.3). The dynamics of the primary
Rl1 (Ω ) −3.0 – – loudspeaker are noticeably modified, and two (respectively three) dis-
Ll1 (mH) 9.0 – –
tinct resonances are visible, when it is coupled to a resonator with one
Cl1 ( μ F) 94.6 – –
(respectively two) DOF(s). For system A, the experimental results
Vb2 (dm3) – 1.34 1.34
Rl2 – 3.4 −1.9
follow closely the simulations (see Table 2). The sound absorption is
(Ω )
Ll2 (mH) – – 11.4 perfect (α = 1) at both resonance peaks at 140 Hz and 191 Hz. For
Cl2 ( μ F) – – 94.6 system B, even though the volume Vb2 is slightly different from the one
found with the optimisation, the area A over the threshold is as ex-
Perf. Indic. fc (Hz) 169.6 156.6 156.0
pected. Nevertheless, the measured weighted bandwidth wBW is larger
f− (Hz) 123.7 106.5 96.2
f+ (Hz) 215.5 206.7 215.7
than the value found in Table 2 in this configuration. System C presents
wBW (Hz) 87.84 95.01 111.51 the most interesting performance indicators of the three systems, thanks
A (Hz) 11.77 12.04 12.46 to the three resonance peaks. The weighted bandwidth wBW is larger
than that obtained by simulation and the value of area A over the
threshold is almost as expected. It provides at least one octave of the
bandwidth BW of efficient sound absorption, notably in the low-fre-
quency range.

4.3. Discussion

For the three systems, the measurements are consistent with the
corresponding simulations. Slight differences are observed on the spe-
cific acoustic impedance and sound absorption coefficient. These dif-
ferences can be attributed to imperfections in the model parameter of
both loudspeakers and to the approximation of the volumes of the boxes
and the volumes taken by the loudspeakers. The general trend is con-
firmed and these slight discrepancies do not contradict the results
presented in this paper.
For system A, when the closed-box loudspeaker is coupled to the
electrical resonator, the sound absorption performance is improved
relative to the basic configuration, when the system is in open circuit.
Note that the performance indicators remain the same whatever the
value of the volume, providing that the value of the inductance Ll (or
capacitance Cl ) is tuned. Even though the measured values of in-
ductances and capacitances are different from the ones indicated by the
manufacturers, the resonance frequency f0A (Eq. (10)) can easily be
matched to that of the closed-box loudspeaker f0 (Eq. (4)).
Systems B and C are more difficult to tune without an optimisation
process, because of the strong interaction effects of the parameters on
the specific acoustic impedance of both systems as seen in Eqs.
(17)–(20) respectively. When one parameter value is modified, the
sound absorption performance of systems B and C is less predictable
than for system A. The coupling coefficient Sd1 Sd2/ Cab1 in Eqs. (18) and
(20), depending on both effective diaphragm surface areas and above
all the volume Vb1 between the loudspeakers when these are fixed, plays
a crucial part for the sound absorption capabilities. Here, the volume Vb1
is chosen to be the same for the three systems, and consequently, the
centre frequency fc is lower for systems B and C relative to system A.
For systems A and C, a negative impedance converter was used to
compensate the resistive part of inductances and capacitances. Because
of the negative value of the resistance Rl1 in the experimental realisation
Fig. 13. (a) Bode plot of the measured specific acoustic impedance and (b) measured of systems A and C, these systems are not considered as purely passive,
corresponding sound absorption coefficient of the experimental realisation of systems A, since an active electrical power is supplied to the system. Thus, the
B, and C, based on the optimisation results with the objective function A . main advantage of the design of system B is to imitate the sound ab-
sorption performance of system A, without requiring any additional
was adjusted to those of the optimal resistances Rl1 and Rl2 given in electrical power. In response, the total volume is larger.
Table 2. The main criterion of both objective functions is based on the
threshold value of minimal efficient sound absorption, the ideal case
being the matching of the specific acoustic resistance at the primary
4.2. Acoustic impedance and sound absorption measurements diaphragm with that of the air. Note that the bandwidth of efficient
sound absorption of such a system is inversely proportional to its
The performance of each system in terms of measured specific acoustic resistance.

116
E. Rivet et al. Applied Acoustics 132 (2018) 109–117

5. Conclusion small room acoustics. In: Audio engineering society convention, vol. 94. Audio
Engineering Society; 1993.
[4] Frommhold W, Fuchs H, Sheng S. Acoustic performance of membrane absorbers. J
The method presented in this paper aims at designing efficient low- Sound Vib 1994;170(5):621–36.
frequency multi-degree-of-freedom electroacoustic absorbers through [5] Yu G, Li D, Cheng L. Effect of internal resistance of a helmholtz resonator on
acoustic energy reduction in enclosures. J Acoust Soc Am 2008;124(6):3534–43.
coupled resonators. First, an appropriate electrical shunt, such as the [6] Fahy F, Schofield C. A note on the interaction between a helmholtz resonator and an
series resistance - inductance - capacitance network, can be connected acoustic mode of an enclosure. J Sound Vib 1980;72(3):365–78.
to the loudspeaker terminals, resulting in a two-DOF resonator. [7] Doria A. Control of acoustic vibrations of an enclosure by means of multiple re-
sonators. J Sound Vib 1995;181(4):673–85.
Stacking loudspeakers in closed boxes points out that the order of the [8] Xu M, Selamet A, Kim H. Dual helmholtz resonator. Appl Acoust 2010;71(9):822–9.
system increases in the same manner with an equivalent sound ab- [9] Guan C, Jiao Z. Modeling and optimal design of 3 degrees of freedom helmholtz
sorption performance without requiring any additional electrical resonator in hydraulic system. Chin J Aeronaut 2012;25(5):776–83.
[10] Cummings A. The effects of a resonator array on the sound field in a cavity. J Sound
power. The primary loudspeaker may also be coupled to a combination
Vib 1992;154(1):25–44.
of the two previous resonators, resulting in a three-DOF resonator. [11] Li D, Cheng L. Acoustically coupled model of an enclosure and a helmholtz re-
Parametric optimisations were carried out by maximising the sound sonator array. J Sound Vib 2007;305(1):272–88.
absorption coefficient over a frequency range as large as possible. [12] Maa D-Y. Potential of microperforated panel absorber. J Acoust Soc Am
1998;104(5):2861–6.
Experimental evaluations confirmed the performance improvement of [13] Gai X-L, Xing T, Li X-H, Zhang B, Wang W-J. Sound absorption of microperforated
the three systems in the low-frequency range by using conventional panel mounted with helmholtz resonators. Appl Acoust 2016;114:260–5.
components. [14] Zhao X-D, Yu Y-J, Wu Y-J. Improving low-frequency sound absorption of micro-
perforated panel absorbers by using mechanical impedance plate combined with
The performance of the proposed systems could be improved by helmholtz resonators. Appl Acoust 2016;114(9):92–8.
adding perforated panels or multi-layered elements, so as to absorb at [15] Li D, Chang D, Liu B. Enhancing the low frequency sound absorption of a perforated
mid and high frequencies as well. In this study, commercial loudspea- panel by parallel-arranged extended tubes. Appl Acoust 2016;102:126–32.
[16] Rivet E, Karkar S, Lissek H. Broadband low-frequency electroacoustic absorbers
kers were chosen for the sake of simplicity of the optimisation process. through hybrid sensor-/shunt-based impedance control. IEEE Trans Control Syst
As their model parameters are limited by physical and technological Technol 2017;25(1):63–72.
constraints, and are initially intended for sound generation, the study [17] Boulandet R, Rivet E, Lissek H. Sensorless electroacoustic absorbers through syn-
thesized impedance control for damping low-frequency modes in cavities. Acta
was only focused on the optimisation of the closed-box volumes and
Acust United Acust 2016;102(4):696–704. 9.
electrical shunt networks. A further stage of optimisation could consist [18] Rivet E, Karkar S, Lissek H, Thorsen TN, Adam V. Experimental assessment of low-
in designing specific transducers adapted for sound absorption appli- frequency electroacoustic absorbers for modal equalization in actual listening
rooms. In: Audio engineering society convention, vol. 140; 2016.
cations.
[19] Boulandet R, Lissek H. Optimization of electroacoustic absorbers by means of de-
signed experiments. Appl Acoust 2010;71(9):830–42.
Acknowledgment [20] Fleming AJ, et al. Control of resonant acoustic sound fields by electrical shunting of
a loudspeaker. IEEE Trans Control Syst Technol 2007;15(4):689–703.
[21] Lissek H, Boulandet R, Rivet E. Optimization of electric shunt resonant circuits for
This research was supported by the Swiss Commission for electroacoustic absorbers. In: Acoustics 2012, Nantes, France; 2012.
Technology and Innovation (CTI) under grant agreement no 14220.1 [22] Liu F, Horowitz S, Nishida T, Cattafesta L, Sheplak M. A multiple degree of freedom
PFNM-NM. electromechanical helmholtz resonator a. J Acoust Soc Am 2007;122(1):291–301.
[23] Chang D, Liu B, Li X. An electromechanical low frequency panel sound absorber. J
Acoust Soc Am 2010;128(2):639–45.
References [24] Lagarias JC, Reeds JA, Wright MH, Wright PE. Convergence properties of the nel-
der–mead simplex method in low dimensions. SIAM J Optim 1998;9(1):112–47.
[25] I. 10534-2-1998. Acoustics - determination of sound absorption coefficient and
[1] Kuttruff H. Room acoustics. 5th ed. Spon Press; 2009.
impedance in impedance tubes - part 2: Transfer-function method. Tech rep.
[2] Cox TJ, D’Antonio P. Acoustic absorbers and diffusers: theory, design and appli-
Internation Standard Organization, Geneva, Switzerland; 1998.
cation. 2nd ed. Taylor and Francis; 2009.
[3] Voetmann J, Klinkby J. Review of the low-frequency absorber and its application to

117

Você também pode gostar