A novel technique for measuring the reflection coefficient of

sound absorbing materials

H-E. de Bree1, F.J.M. van der Eerden2, J.W. van Honschoten1

1
Dep. of Electrical Eng. (TT), 2 Dep. of Mechanical Eng. (TMK), University of Twente, The Netherlands
E-mail: [email protected]

Abstract
A new method to measure the acoustic behaviour of sound absorbing material in an impedance tube is
presented. The method makes use of a novel particle velocity sensor, the microflown, and a microphone. The
so-called p·u method is compared to three other methods of which the two microphone technique is well
known. It is shown that the combination of a microphone and a microflown provides direct information on
the acoustic impedance, the sound intensity and the sound energy density. The experimental results are
compared to the results obtained with the conventional impedance tube measurements. To be able to repeat
the measurements in a reliable way a well described test sample with a quarter-wave resonator is used.
Furthermore it is shown that the viscothermal effects on the wave propagation are important, i.e. for the
quarter-wave resonator and to a lesser extent for the impedance tube itself.

The first technique made use of the measured

1. Introduction standing wave ratio (SWR) in the tube for a specific
frequency. Next, in 1980 Chung and Blazer [1]
A number of measurement techniques can be used to presented a technique which is based on the transfer
quantify the sound absorbing behaviour of porous function of two fixed microphones (see Figure 1).
materials. In general one is interested in one of the The standing wave pattern in their case is built up
following properties: the sound absorption from a broadband stationary noise signal. With the
coefficient α, the reflection coefficient R, or the measured transfer function the incident and reflected
surface impedance Z. The techniques can be divided waves are separated mathematically. This leads to
into three categories: the reflection coefficient R of the sample for the
• Reverberant field methods complete frequency band. The normal impedance Zn
• Free field methods and the sound absorption coefficient α can be
derived as well. The transfer function technique,
• Impedance tube methods (Kundt’s tube)
here called the 2p method, has proven to be
The impedance tube technique is the topic of the
efficient and reliable (see ISO 10534-2 1998).
present investigation (see Figure 1).

rigid tube microphones backing plate With the development of the first acoustic particle
1 2 velocity sensor, the microflown, alternative
measurement techniques become possible in the
pB pA impedance tube. These techniques are presented in
section 4:
loudspeaker test sample • The 2u method
Figure 1 Outline of an impedance tube. • The u/p method
• The p·u method
At one end of the impedance tube a loudspeaker is The two latter methods make use of the combination
placed and at the other end a sample of the test of a microphone and a microflown to determine the
material is placed. In the tube a standing wave acoustic behaviour of a sound absorbing sample.
pattern is established as a result of the forward, pB, The p·u method is a new technique and makes use of
and backward, pA, travelling waves. The frequency the measured sound intensity and sound energy
of the sound waves is kept low enough to assure density. By using the p·u method a good indication
plane propagating waves in the tube. of the quality of the measured sound intensity and
energy density is obtained. The first results of the
S1
p·u method are presented and compared to the three
other methods. S2
Furthermore it will be shown that the
combination of a microflown and a microphone, in a
Acoustic wave
slightly different set-up, is useful to calibrate the
microflown with respect to a calibrated microphone

Estimated temperature
(see section 5.1). In section 5.2 the experimental
results of the different methods are mutually
compared. Furthermore they are compared to the
theoretical sound absorption coefficient of a well
described sample with a quarter-wave resonator. In
Acoustic wave present
this way the measurements can be repeated in a Stationary situation
reliable way.
First the microflown is described in section 2. S1 S2
Position
The theory for the viscothermal wave propagation in
the impedance tube and the resonator is given in Figure 3 Photograph of the sensors on top and
section 3. estimated temperature profile.
It follows from Figure 3 that a travelling acoustic
2. The microflown wave causes movements of the air and as a result
heat is transferred from one sensor to the other
The microflown (or µ-flown) was developed by De (harmonically for a single frequency). This causes a
Bree et al. [2]. Instead of pressure the microflown difference in temperature of the two sensors. The
measures the average acoustic particle velocity of a temperature difference between both sensors causes
small finite volume of air. The measurement a differential electrical resistance variation which is
principle of the microflown is based on the measured. To realise a temperature difference that is
temperature difference between two resistive high enough to be able to measure the resistance
sensors which are 40 µm apart as shown in Figure 2 variation the sensors are heated by a DC current up
and Figure 3. The microflown consists of two to about 400-600 degrees Kelvin.
cantilevers of silicon nitride with a platinum pattern The sensitivity of the microflowns shows
on top. The size of the cantilevers is 800×40×1 µm approximately a first order low pass behaviour. The
corner frequency, above which the sensitivity drops
(l×w×h).
6 dB per octave, is between 300 Hz and 1kHz. More
Two sensors on top (40 µm apart) practical characteristics of the microflown are:
• The directional sensitivity varies cosine-like (as
in a figure of eight). Therefore the sensors can
be aligned accurately by hand.
Acoustic
waves
• The absence of a resonance mechanism.
• The low costs due to simplicity and batch size.
• The ability to measure the particle velocity in
the near field where the sound intensity
technique with two closely spaced microphones
breaks down.
• The vulnerability of the two sensors. A
protective package may be used (see also
section 4).
Mass Power supply Die • The microflown has to be positioned outside a
Signal Breaking groove
boundary layer whereas a microphone is usually
Figure 2 Photograph of the microflown. flush-mounted in the impedance tube wall.
A new design of the microflowns is shown in Figure
4. The cantilevers of the two sensors are supported
at both sides for extra stability. Furthermore, the
sensors are smaller to gain extra sensitivity. For the
sake of completeness also a microflown with three
pairs of sensors is depicted in Figure 5 which
measures the three orthogonal velocity components. pB
In this way the measurement of the sound intensity ll
x
vector becomes relatively easy.

pA

Figure 6 A prismatic tube.

S2
S1 p( x) = p A e Γ k x + p B e − Γ k x (1)

u ( x) =
G
ρ 0 c0
(
p A e Γ k x − p B e −Γ k x ) (2)

Figure 4 A bridge-type microflown. where u(x) is the averaged velocity over the cross-
section so that a one-dimensional model arises. The
viscothermal effects are present in the wave
S1 propagation coefficient Γ.
The solution of Zwikker and Kosten was
S2 rewritten in a dimensionless form by Tijdeman. In
the so-called ‘low reduced frequency model’ four
S1 S2 dimensionless parameters, s , kr, σ and γ characterise
the wave propagation (see Appendix: List of
Symbols). The shear wave number s is a measure for
the ratio between the inertial and viscous forces and
can be seen as an unsteady Reynolds number.
S1S2
ρ 0ω
s=l (3)
µ
Figure 5 A microflown for three dimensions.
For small values of s the viscous effects are
dominant. In that case a tube is called ‘narrow’. For
3. Viscothermal wave propagation s » 1 the tube is called ‘wide’. Beltman has
demonstrated that for most acoustic problems the
3.1 The wave propagation coefficient low reduced frequency model is sufficient.
Compared to other models, like the simplified
In small tubes the wave propagation is affected by Navier Stokes model, the low reduced frequency
the viscosity and the thermal conductivity of the model is a relatively simple one.
fluid. There is a variety of literature on the The wave propagation coefficient Γ is a complex
viscothermal wave propagation. Tijdeman [3] and number, i.e. Γ = Re(Γ ) + i Im(Γ ), where c0 / Im(Γ )
Beltman [4] give a complete overview of the represents the effective phase velocity and Re(Γ )
different analytical solutions. It is shown there that represents the attenuation of a propagating wave. It
the approach of Zwikker and Kosten leads to an is noticed that Γ is a function of the shear wave
efficient and accurate solution. They assumed a number s and therefore a function of the frequency.
constant pressure across the tube cross-section and In the present investigation s » 1 applies for the
included the effects of inertia, compressibility, tubes. As a result a first order approximation of the
viscosity and thermal conductivity of the fluid. For a low reduced frequency solution can be used. For
prismatic tube as shown in Figure 6 the harmonic large values of the shear wave number s, i.e. for
pressure perturbation p(x) and the harmonic velocity ‘wide’ tubes, the propagation coefficient Γ can be
perturbation u(x) are given by (1) and (2). estimated according to Kirchhoff:
i +1  γ −1+ σ 
Γ =i+   (4)
2  sσ 
The accompanying velocity coefficient G is: where α is the fraction of the incident energy which
is dissipated.
i
G=− (5) With known amplitudes pA and pB the impedance
Γ
ζ(x) can be determined and with the use of the
From (4) it can be seen that both viscous and definition of the reflection coefficient R(x) one can
thermal effects are included. These effects are taken derive:
into account for the impedance tube and the quarter- ζ ( x) − 1
R( x) = (9)
wave resonator. ζ ( x) + 1

3.2 The reflection coefficient

3.3 A quarter-wave resonator
The amplitudes pB and pA of the forward and
backward travelling waves in the tube are governed A sample with a single tube, a quarter-wave
by the boundary conditions at both ends of the tube. resonator, is placed at the end of an impedance tube
In Figure 7 the possible boundary conditions are with a square cross-section. The sample is depicted
shown. in Figure 8.

p x p1 - pressure perturbation R = 0.0046 m

0
u0 L = 0.07 m
u1 - velocity perturbation A
L d = 0.0028 m
z 0 z 1- scaled acoustic impedance A = (0.04)2 m2
Figure 7 One-dimensional tube and possible
2R L
boundary conditions.
d
For a tube terminated with a sound absorbing
material an impedance condition can be used. The Figure 8 Sample with a quarter-wave resonator.
specific acoustic impedance is denoted by Za and Due to inlet effects the effective length Leff of the
relates the pressure, at a specific position or at a resonator is longer than the actual length L. The end
surface, to the normal velocity. In the present correction d for a tube centrally located in another
investigation the impedance is scaled to the tube with radius R2 is [5]:
characteristic impedance of a travelling plane wave.
The characteristic impedance of a freely plane 8R  
d= 1 − 1.25 R R  ; with R < 0.6 (10)
propagating wave is ρ0c0. However, in a duct where 3π  2 R2
the viscothermal wave propagation is included the For the square cross-section the radius R2 is
characteristic impedance is –ρ0c0/G. Therefore the estimated by R2 = 0.04m.
scaled (dimensionless) impedance ζ becomes: With the use of (1), (2) and (6) the impedance at
−G −G p( x) the entrance of the resonator is deduced:
ζ ( x) = Z a ( x) = (6)
ρ 0c0 ρ 0c0 u ( x) cosh(Γ k Leff )
ζ ( x = 0) = (11)
More convenient quantities for sound absorbing sinh( Γ k Leff )
materials are the reflection R and the sound
The impedance of the sample itself is calculated by
absorption coefficient α. The reflection coefficient
using a reference frame across the surface of the
is the ratio between the reflected and the incident
sample. The pressure and the mass flow on the left-
wave and by using equation (1) gives:
and right-hand side of the reference frame are equal,
p A eΓ k x so one has:
R( x) = (7) −1
p B e −Γ k x  π R2 
ζ sample =  ⋅ ζ ( x = 0) (12)
 A 
The acoustic energy in a plane wave is proportional  
to the squared amplitude of the pressure. Therefore
|R|2 is the ratio of the reflected energy to the incident
energy.
The sound absorption coefficient α is defined as
2
α = 1− R (8)
4. The impedance tube 4.3 The 2u method

4.1 Experimental set-up The measured transfer function is:

u Su u
The impedance tube is sketched in Figure 9. The H 2u = 2 = 2 1 (15)
u1 S u1u1
four methods are shortly discussed in the next
sections: the 2p, 2u, p/u and the p·u method. With H2u and the theory presented in section 3 the
reflection coefficient R becomes:
s = 0.045 m
L = 0.643 m H 2u − e − Γ k s eΓ k L
p R ( x = L) = ⋅ (16)
H 2u − e Γ k s e −Γ k L

p2 4.4 The p/u method

p1
u
L The transfer function Hp/u is measured in the
s
frequency domain according to:
p S pu
Figure 9 Impedance tube and sensor positions. H p/u = = (17)
u Su u
The microphone positions for the 2p method are
indicated by p1 and p2. The same positions are used If the microphone and the microflown are located at
for the 2u method but then two microflowns are x = 0 then the impedance at the sample at x = L, is:
used. For the p/u and the p·u method the positions of H p / u cosh(Γ k L) − ρ 0 c0 sinh( Γ k L)
the microphone and the microflown are indicated by ζ ( x = L) = (18)
ρ 0 c0 cosh(Γ k L) + H p / u sinh( Γ k L)
p and u, respectively. Notice that for the latter two
methods there is no separation distance s and that Equation (18) is obtained by using the theory for the
the distance to the sample L is different one dimensional wave propagation.
(L = 0.407 m). The microphones for the 2p method Acoustic impedance measurements with the p/u
are ¼ inch Kulites. For the p/u and p·u method a ½ method were first performed by Schurer in 1996
inch B&K microphone and a ½ inch ICP µ-flown directly in the throat of a horn [6].
Technologies is used.
4.5 The p·u method
4.2 The 2p method
By measuring the sound intensity and the sound
The measured transfer function is: energy density (at x = 0) it is possible as well to
p2 S p2 p1 derive the amplitude reflection coefficient |R|. Fahy
H 2p = = (13) [7] shows that in a tube the mean speed of the
p1 S p1 p1
energy transport ce is the ratio between the mean
where p2 and p1 are the pressures in the frequency intensity and the mean energy density:
domain, Sp2p1 is the cross-spectrum and Sp1p1 is the
I 1− | R |2
auto-spectrum. The transfer function H2p can be ce = = c0 (19)
measured directly with a two channel FFT analyser. E 1+ | R |2
With the theory for the one dimensional wave So the reflection coefficient |R| at the surface of a
propagation as presented in section 3 one can derive sample (at x = L, see also equation (7)), and by taken
the reflection coefficient at the surface of the into account the viscothermal losses in the
sample, i.e. at x = L: impedance tube, is:
H 2p − e − Γ k s eΓ k L 2
R ( x = L) =
Γ ks
⋅
−Γ k L
(14) E c0 − I e Γ k L
− H 2p + e e R =
2
⋅ (20)
E c0 + I e − Γ k L

The time averaged energy density is defined as:

 1 p p* 1
E= ρ0 u u* + (21) 10
4 4 ρ 0 c0 2 Measurements
0 Theory
10
and the active intensity (also time averaged) is:
1 −1

| [−]
I= Re{ p u *} (22) 10
2

u/p
−2

|H
where p and u are complex amplitudes in the 10
frequency domain.
−3
Using a two channel FFT analyser one has: 10
1 1
E= ρ 0 Su u + Spp (23) 10
−4
2 2 ρ 0 c0 2 0 1000 2000 3000 4000
Frequency [Hz]
I active = Re{ S pu } (24) Figure 11 Transfer function Hu/p (magnitude).
With (23), (24) and (20) the value of |R| can be From the measured and theoretical values (theory:
determined. without gain) a correction function is obtained as a
The results of the new p·u method are compared function of the frequency. The difference for the
to the results of the other methods in section 5.2. amplitude and the phase is given in the Figure 12
and 13. As a first approach a third order polynomial
5. Experimental results is used for the final interpolated correction.

Before the reflection coefficient R (or the sound −40

absorption coefficient α) of a sample with a quarter-
wave resonator is measured the sensors are
−50
calibrated. The procedure and the calibration results
Difference [dB]

for the combination of a microphone and a

microflown are described in the following section. −60

5.1 Calibration results

−70
For the calibration of the combination of the Measured
microphone and the microflown the position of both Interpolated
sensors is indicated in Figure 10. −80
0 1000 2000 3000 4000
Frequency [Hz]
L
Figure 12 Amplitude difference in dB.

microphone 300
microflown
Difference [degrees]

Figure 10 Set-up for the calibration measurements. 250

The measured and theoretical transfer function Hu/p
are given in Figure 11. A minimum occurs when the 200
velocity has a minimum value at the location of the
microflown. The right-hand side of the impedance
tube is approximately acoustically hard (ζ(L) = 150
1000) so that there is always a pressure maximum. Measured
Interpolated
−G  cosh(Γ k L) 
Hu / p =  sinh( Γ k L) +  (25) 100
ρ 0c0  ζ ( x=L)  0 1000 2000 3000 4000
Frequency [Hz]
Figure 13 Phase difference in degrees.
For the present tests the measured transfer function
Hu/p is corrected in the following way: 1
~
20 log H u / p → 20 log H u / p + Correction [dB] (26)
0.8
~
arg( H u / p ) → arg( H u / p ) + Correction

|R| [−]
[deg] (27) 0.6

0.4
5.2 A sample with a quarter-wave resonator
0.2 Measurements
The measurement set-up is in Figure 9 and the Theory
results are presented in the Figures 14, 15 and 16. 0
For the results with the 2u method reference is made 500 1000 1500 2000
to [8] where it is shown that the 2u method provides Frequency [Hz]
the same results as the 2p method. Figure 16 Reflection coefficient (magnitude) using
the p·u method.
The figures show that the three methods provide
1
identical results for the frequency range where the
quarter-wave resonator is effective, i.e. from 1000 to
0.8
1400 Hz. For the lower frequency range the u/p and
the p·u method are less accurate. This may be
|R| [−]

0.6
improved with a more accurate calibration of the
transfer function Hu/p. It is noted that the phase-
0.4
←α
calibration can be performed with the microphone
flush mounted. In this way the microphone shows
0.2 Measurements the same phase-response behaviour because the
Theory incident sound field is the same as for the
0 experiments with a sound absorbing sample. In this
500 1000 1500 2000
Frequency [Hz] respect it has to be remarked that the calculation of
R for an acoustic hard wall is very sensitive to
Figure 14 Reflection coefficient (magnitude) using measurement inaccuracies.
the 2p method.
The figures also reveal that the theoretical model
of section 3 predicts the acoustic behaviour very
accurately for the complete frequency range. This
1 indicates that the viscothermal wave propagation is
correctly modelled.
0.8
It has been shown that the p·u method is a practical
technique to measure the reflection coefficient, the
|R| [−]

0.6
sound intensity and the sound energy density. Next,
0.4 the transmission loss in an impedance tube and the
sound intensity vector and the energy density
0.2 outside a pipe can be measured.
Measurements
Theory
0 6. Conclusions
500 1000 1500 2000
Frequency [Hz]
Four methods to measure the acoustic behaviour of
Figure 15 Reflection coefficient (magnitude) using sound absorbing material in an impedance tube have
the p/u method. been presented: the 2p, 2u, p/u and the p·u method
and the results have been mutually compared Three
methods make use of a novel particle velocity
sensor, the microflown. It has been shown that the
microflown can be an attractive alternative to
microphones. Furthermore, the combination of a
microphone and a microflown provides direct E Time averaged sound energy density per unit volume
information on the acoustic impedance, the sound G Parameter for the velocity u(x)
intensity and the sound energy density. Besides it H2p Transfer function of two microphones
has been demonstrated that for narrow tubes the i = −1 Imaginary number
viscothermal effects on the wave propagation are
I Time averaged or mean sound intensity [J/(m2s)]
important, i.e. for a quarter-wave resonator and to a
k = ω c0 Wave number [m-1]
lesser extent for the impedance tube itself.
k r = l ω c0 Reduced frequency
l=R Half the characteristic length of the cross-section [m]
Acknowledgements
L Length of a tube [m]
The authors would like to thank W.F. Druyvesteyn p Pressure perturbation [Pa]
and G.J.M. Krijnen for their fruitful discussions. R Reflection coefficient
The comments of H. Tijdeman, R.M.E.J. Spiering, R Radius [m]
P.J.M. van der Hoogt and T.G.H. Basten are s = l ω ρ 0 µ Shear wave number
gratefully acknowledged. This research is supported s Distance between two microphones [m]
by the Dutch Technology Foundation (STW). Spu Cross-spectrum
Spp Auto-spectrum
References t Time [s]
T Kinetic energy per unit volume [J/m3]
1. Chung, J.Y., Blaser, D.A. Transfer function method u Particle velocity perturbation [m/s]
of measuring in-duct acoustic properties. I. Theory, J.
U Potential energy per unit volume [J/m3]
of the Acoust. Society of America, 68(3), 907-913
(1980) x Cartesian co-ordinate [m]
2. Bree, H-E. de, Leussink, P.J., Korthorst, M.T., Za Acoustic impedance [kg/(m2s)]
Jansen, H.V., Lammerink, T., Elwenspoek, M. The
Microflown: a novel device measuring acoustical α Sound absorption coefficient
flows, Sensors and Actuators A, 54, 552-557 (1996) γ = C p CV Ratio of specific heats
3. Tijdeman, H. On the propagation of sound waves in Γ Viscothermal wave propagation coefficient
cylindrical tubes, J. of Sound and Vibration, 39(1), 1-
λ Thermal conductivity [J/msK]
33 (1975)
4. Beltman, W.M. Viscothermal wave propagation µ Dynamic viscosity [Pa s]
including acousto-elastic interaction, part I: theory, J. ρ0 Mean density [kg/m3]
of Sound and Vibration 227(3), 555-586 (1999) σ = µ C p λ Square root of the Prandtl number
5. Bies, D.A., Hansen, C.H. Engineering Noise Control, ω Angular frequency [rad/s]
theory and practice, E&FN SPON, London (1996) Ω Porosity of a surface
6. Schurer, H., Annema, P., de Bree, H-E., Slump, C.H., ζ Acoustic dimensionless impedance
Herman, O. Comparison of two methods for
measurements of horn input impedance, J. Audio Eng.
Re{ } Real part
Society, 44, 649 (1996)
7. Fahy, F.J. Sound intensity, E&FN SPON, London Im{ } Imaginary part
(1995) | | Amplitude
8. Eerden, F.J.M., de Bree, H-E., Tijdeman, H. * Complex conjugate
Experiments with a new acoustic particle velocity
sensor in an impedance tube, Sensors and Actuators
A, 69, 126-133 (1998)

Appendix: List of Symbols

A Surface [m2]
c0 Undisturbed speed of sound [m/s]
ce Mean speed of energy transport [m/s]
Cp Specific heat at constant pressure [J/kgK]
CV Specific heat at constant volume [J/kgK]
d Length increment due to inlet effects [m]