JAES_V59_1_2_PG44hirez
-
Upload
ciskowiec2210 -
Category
Documents
-
view
21 -
download
0
Transcript of JAES_V59_1_2_PG44hirez
ENGINEERING REPORTS
Dynamic Measurement of Transducer EffectiveRadiation Area*
WOLFGANG KLIPPEL** AND JOACHIM SCHLECHTER([email protected]) ([email protected])
Klippel GmbH, DE-01309 Dresden, Germany
The effective radiation area SD is one of the most important loudspeaker parameters because
it determines the acoustic output (sound power) and efficiency of the transducer. This
parameter is usually derived from the geometrical size of the radiator considering the diameter
of one-half the surround area. This conventional technique fails for microspeakers and
headphone transducers where the surround geometry is more complicated and the excursion
does not vary linearly with the radius. New methods for measuring SD more precisely are
discussed. The first method uses a laser sensor and a microphone to measure the voice-coil
displacement and the sound pressure generated by the transducer while mounted in a sealed
enclosure. The second method uses only the mechanical vibration and the geometry of the
radiator as measured using a laser triangulation scanner. The reliability and reproducibility of
the conventional and the new methods are verified and the propagation of the measurement
error on the T/S parameters using the test-box perturbation technique and other derived
parameters (sensitivity) is discussed.
0 INTRODUCTION
Electroacoustic transducers can be modeled at low
frequencies by an equivalent circuit comprising a few
lumped parameters. The effective radiation area SD
describes the coupling between mechanical and acous-
tical domains by replacing cone, diaphragm, surround,
and other radiating parts with a rigid piston moving with
the same velocity as the voice coil and having an
equivalent surface generating the same sound pressure in
the far field as the real radiator [1]. The parameter SD
determines the sound pressure on axis (passband
sensitivity) and the total acoustical output power [2]. A
precise value of SD is also a prerequisite for accurate
parameter measurements using the test-box perturbation
technique and for providing reliable T/S parameters
(such as the equivalent air volume of driver stiffness Vas)
for system design [3], [4].
In practice the effective radiation area is usually
derived from the geometry of the radiating surface by
Fig. 1. Estimation of effective radiation area SD from
radiator geometry.
*Manuscript received 2010 June 2; revised 2010 October 28.**Also with University of Technology Dresden, Dresden,
Germany.
44 J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February
reading the diameter d in the center of the surround area,
as shown in Fig. 1. This approach considers the cone
moving as a rigid body and assumes a linear decay of the
displacement,
jx ðx; rÞjjx coilðxÞj
¼1; 0 � r � di
2
do � 2r
do � di
;di
2� r � do
2
8>>><>>>:
ð1Þ
versus the radius r in the surround area. Small showed
that the diameter d referred to the half-width of the
surround is a useful approximation of the effective
radiation area,
SD ¼p4
d2
i þ ðdo � diÞdi þðdo � diÞ2
3
" #’
p4
di þ do
2
� �2
¼ p4
d2 ð2Þ
as long as the surround is not wider than 10% of the outer
diameter do and the assumption in Eq. (1) is valid (see [5,
appendix]).
Significantly higher errors are found in woofers that use
a larger surround designed for realizing a high peak
displacement Xmax and especially in tweeters, head-
phones, horn compression drivers, and microspeakers,
where the outer part of the diaphragm itself is used as the
suspension system.
In those cases the displacement rises from zero at the
outer rim to the maximum value at the voice-coil position
by a function that is usually nonlinear (see solid line in
Fig. 2(b) and [5, fig. 8]) and depends on the particular
clamping condition of the rim, the thickness and curvature
of the diaphragm profile, and the material properties.
Theoretical estimates of the effective radiation area SD for
clamped plates, flat circular membranes, and other simple
geometries are given in [6] but may not be applicable to
real transducers having a much more complicated
geometry. Thus a precise value for the effective radiation
area SD has to be calculated from the total volume
velocity q(x) measured by using mechanical and/or
acoustical sensors [7]–[9].
This engineering report tackles the problem and
searches for simple, robust, and convenient measurement
techniques giving more reliable values of SD. At the
beginning the effective radiation area SD will be defined
by using the sound pressure output predicted by the
Rayleigh equation. In the second part the pros and cons of
the traditional techniques are discussed. A new technique
is presented which calculates SD from vibration and
radiator geometry using the results of modern laser
scanning techniques. The accuracy of the new method is
investigated and compared with traditional methods.
0.1 Symbols
ra far-field observation point
rc point on radiator surface
r0 distance between radiator center and observation
point ra
x angular frequency
x0 angular resonance frequency
p(x,ra) complex sound pressure at far-field observation
point
pbox(x) complex sound pressure in sealed test enclosure
x(x,rc) complex displacement at radiator surface
xcoil(x) complex voice-coil displacement
SD(x) complex value of effective radiation area de-
pending on frequency
SD effective radiation area describing piston mode at
resonance
q0 density of air
Sc total radiator surface
k wavenumber
ravg average radius (usually at voice-coil position)
rr outer radiator radius
q(x) total volume velocity generated by radiator
Cab acoustic compliance of air volume in sealed
enclosure
DVbox variation of air volume in test enclosure
Vbox total air volume of test enclosure
p0 static air pressure
c speed of sound
xcoilþDV(x0) voice-coil displacement at resonance fre-
quency measured in box with enlarged air
volume
pboxþDV(x0) sound pressure at resonance frequency
measured in box with enlarged air volume
Fig. 2. Variation of displacement with radius. (a) Geometry
of diaphragm and voice coil and graphically enhanced peak
and bottom displacement. (b) Amplitude of headphone
diaphragm displacement (—) measured at 297.4 Hz by laser
scanning compared to linear approximation (- - -) as used in
geometrical method for predicting SD.
J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February 45
ENGINEERING REPORTS MEASUREMENT OF EFFECTIVE RADIATION AREA
di inner diameter of surround area
do outer diameter of surround area
d diameter referred to center of surround area
1 DEFINITION OF SD
A radiator mounted in an infinite baffle generates a
sound pressure
p ðx; raÞ ¼ �x2q0
2p
ZSC
e�jkjra�rcj
jra � rcjx ðx; rcÞ dS ð3Þ
at a listening point ra in the far field using the Rayleigh
integral of complex displacement x(x,rc) over all points rc
on the surface Sc of the radiator, as illustrated in Fig. 3.
This approximation implies that the radiating surface has
a relatively flat shape and all dimensions are much
smaller than the wavelength of the sound considered for
calculating SD.
If the listening point ra is on axis, far away from the
radiator, and much larger than the diameter 2rr of the
radiator, then the distance jra – rcj ’ r0 may be assumed
constant and Eq. (3) simplifies to
p ðx; raÞ ¼ �x2q0
2pe�jkr0
r0
ZSC
x ðx; rcÞ dS: ð4Þ
The same sound pressure
p ðx; raÞ ¼jxq0
2pq ðxÞ e
�jkr0
r0
ð5Þ
is generated by a volume velocity
q ðxÞ ¼ jxZSC
x ðx; rcÞ dS ð6Þ
emitted by a point source in the center of the radiator. An
equivalent volume velocity
q ðxÞ ¼ jxx coilðxÞS DðxÞ ð7Þ
can be generated by a piston vibrating with the reference
displacement xcoil(x) and an effective radiation area
SD(x).
Combining Eqs. (6) and (7) gives the effective
radiation area,
S DðxÞ ¼q ðxÞ
jxx coilðxÞ¼
ZSC
x ðx; rcÞ dS
x coilðxÞð8Þ
which is a complex quantity and depends on frequency.
At low frequencies, where all parts of the radiator
vibrate in phase, the complex function SD(x) becomes a
real value, and the effective radiation area can be
defined as
SD ¼ jS Dðx0Þj ð9Þ
using the fundamental resonance frequency x0 as a
useful reference.
2 MEASUREMENT TECHNIQUES
The following dynamic measurement techniques make
no assumption regarding the displacement x(x,rc) on the
radiator surface.
2.1 Far-Field Measurement
Using Eqs. (5) and (7) the frequency-dependent
function becomes
S DðxÞ ¼ �2p
x2q0
r0
e�jkr0
p ðx; raÞx coilðxÞ
ð10Þ
which in turn becomes the effective radiation area,
SD ¼ jS Dðx0Þj ¼2pr0
x20q0
jp ðx0; raÞjjx coilðx0Þj
ð11Þ
at resonance frequency, according to Eq. (9). This
measurement technique requires a calibrated microphone
for measuring the sound pressure p(x,ra) on axis at a
distance ra from the radiator mounted in a sufficiently
large baffle and providing anechoic conditions. The
measurement of microspeakers, tweeters, and other
transducers having a high resonance frequency can be
accomplished by using a baffle according to IEC 60268-5
and medium-size anechoic rooms. This method is less
practical for measuring woofers in anechoic rooms, where
standing waves affect the accuracy of sound pressure
measurements below 100 Hz.
2.2 Near-Field Measurement in Baffle
Moreno et al. [5] suggested a measurement of the
sound pressure p(x,ra ’ rc) close to the surface of the
radiator to relax the requirements put on the acoustic
environment and calculate SD using the equation
SD ¼ jS Dðx0Þj ¼p4
2
x20q0
jp ðx0; raÞjjx coilðx0Þj
� �2
: ð12Þ
However, Moreno et al. [5] observed a significant
influence of the radiator geometry and measured a
Fig. 3. Sound pressure generated on axis at observation point
ra by radiator mounted in an infinite baffle.
46 J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February
KLIPPEL AND SCHLECHTER ENGINEERING REPORTS
difference of more than 10% in the SD value for flat and
conical pistons of the same size.
2.3 Box Method
A radiator mounted on a small enclosure, as shown in
Fig. 4, generates a sound pressure inside the enclosure
pboxðxÞ ¼
q ðxÞjxCab
ð13Þ
which at low frequencies is almost independent of the
microphone position [9]. Calculating the acoustic com-
pliance of the enclosed air,
Cab ¼Vbox
q0c2ð14Þ
by using the air volume Vbox, the air density q0, and the
speed of sound c gives the equivalent radiation area,
SD ¼ jS Dðx0Þj ¼Vbox
q0c2
jpboxðx0Þj
jx coilðx0Þj: ð15Þ
The volume Vbox of the air in the test enclosure has to
be measured carefully considering the free volume of the
enclosure and the air displaced by the loudspeaker parts.
In practice determination of Vbox limits the accuracy of
the measurement technique.
2.4 Differential-Box Method
Performing two measurements with different air
volumes, as illustrated in Fig. 5, dispenses with the
measurement of the static air volume displaced by the
transducers. Combining Eq. (15) of the first measurement
with
SD ¼Vbox þ DVbox
q0c2
jpboxþDV
ðx0Þjjx coilþDVðx0Þj
ð16Þ
of the second measurement and substituting Vbox in both
equations gives
SD ¼1
q0c2
DVbox
jx coilþDVðx0Þjjp
boxþDVðx0Þj
� jx coilðx0Þjjp
boxðx0Þj
: ð17Þ
A medical syringe, as shown in Fig. 6, may be used as
test enclosure for microspeakers, headphones, and tweet-
ers, which allows precise variations of the box volume
DVbox. Although the differential box puts the lowest
requirement on the acoustic environment, the major
disadvantage of this technique is that a minor leakage
may cause significant error in the measurement. Some-
Fig. 4. Measurement of internal pressure generated in a
sealed enclosure.
Fig. 5. Performing two measurements while using two test
enclosures having precise differences in air volume DVbox.
(a) First measurement. (b) Second measurement.
Fig. 6. Differential measurement of effective radiation area
of a headphone by using a syringe.
J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February 47
ENGINEERING REPORTS MEASUREMENT OF EFFECTIVE RADIATION AREA
times there are also minor leakages in the material of the
diaphragm and in the connection to the surround and dust
cap, which affect measurement accuracy.
2.5 Laser Technique
A different approach to determine SD can be based on
the relations shown by Eqs. (8) and (9). There SD is
defined as the integrated displacement of the radiator
surface divided by the voice-coil excursion xcoil(x).
Replacing the integral by a sum over discrete elements
gives a good approximation of the effective radiation
area
S DðxÞ ¼X
x ðx; rc;iÞ � DSc;i
x coilðxÞð18Þ
where x(x,rc,i) is the displacement in the z direction and
DSc,i is the respective surface area projected on the r, uplane, as illustrated in Figs. 7 and 8. The vibration and
geometry of the radiator surface can be measured with
high accuracy by using a triangulation sensor and an
automatic scanning technique [10].
For the measurement of the coil excursion xcoil(x) it is
advisable not only to measure a single point but rather to
conduct spatial averaging by measuring multiple points,
for example, on a circle close to the voice coil,
x coilðxÞ ¼
Z2p
0
x ðx; ravg;uÞ du
2p: ð19Þ
The averaged radius ravg should be selected close to the
voice coil to be representative of the coil motion in the
piston mode. An optimal value for postprocessing of the
scanned data can be found using the graphical user
interface depicted in Fig. 9.
2.5.1 Frequency Dependence of SD(x)
By applying Eq. (18) a frequency-dependent effective
radiation area SD(x) is obtained. At low frequencies
below the breakup of the cone the amplitude of SD(x) is
constant and represents the effective radiation area of the
piston motion, which is equal to the T/S parameter SD
[compare Eq. (9)].
The graphs of jSD(x)j are shown in Fig. 10 for a set of
different ravg values in the center of an oval microspeaker.
Over a broad frequency range of 10 Hz to 1 kHz the
derived SD is nearly constant. Averaging in that frequency
range will further improve the accuracy of the nominal SD
of the driver. At higher frequencies SD(x) is no longer
constant but depends on the specific vibration properties
at the averaged radius ravg in relation to the total volume
flow q(x).
As long as the ravg used is close to the voice coil of the
driver, the respective SD(x) represents a linear transfer
function between a given coil displacement xcoil(x) and
the far-field sound pressure on axis. See Eq. (20), which
can be derived from Eq. (10),
p ðx; r0Þ ¼ �x2q0
2pe�jkr0
r0
S DðxÞx coilðxÞ: ð20Þ
2.5.2 Average Radius ravg
If ravg is too far from the voice coil then the coil
excursion xcoil(x) will contain an error, which leads to an
incorrect SD. The measured values of xcoil and SD are
shown in Fig. 11 for various locations of the averaged
radius ravg and for a frequency of 100 Hz. It can be seen
that the amplitude of the coil displacement is nearly
constant in the center of the driver up to a radius of 7 mm
and gives a valid value of SD ¼ 3.4 cm2. If the average
radius ravg is larger than the displacement, xcoil drops to
zero until the outer driver edge and the derived SD goes to
infinity.
Fig. 7. Laser scanning technique for measuring vibration and
geometry of radiator surface.
Fig. 8. Cone surface elements DSc,i.
48 J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February
KLIPPEL AND SCHLECHTER ENGINEERING REPORTS
2.5.3 Measurement Grid Resolution
The resolution of the scanning grid for measuring the
displacement x(x,rc,i) is important for the accuracy of SD.
An investigation of the effect of reducing the number of
radial and angular measurement points was conducted for
the two microspeakers shown in Fig. 12. In both cases a
full, detailed scan of 3201 points was coarsened by re-
moving the data from single measurement radii or angles.
The average radius has been set to 6 mm for the circular
speaker and to 10 mm for the oval speaker.
The nominal value of SD remains very accurate for the
circular speaker, even when only three radial lines or four
angular lines were kept. Even a combination of three
radial and four angular steps will result in SD¼3.375 cm2,
which is only 1% less than the full resolution result of SD
¼ 3.408 cm2 (see Fig. 13). This means that in this case,
with x(x,rc,i) measured at only 9 points, it is possible to
determine the SD value of the driver with acceptable
accuracy. But it should be stated that this result cannot be
applied to all circular speakers. In general at least a higher
radial resolution is necessary to cover the decay of the
displacement amplitude toward larger radii properly. As
long as the driver shape is axis symmetrical, a higher
angular resolution will only improve the signal-to-noise
ratio of the measurement and can help with evaluating the
influence of rocking modes correctly.
For the oval speaker the error becomes larger if too few
radial or angular steps are considered. In particular a
Fig. 9. User interface of SD calculation tool required to specify voice-coil position.
Fig. 10. Effective radiation area jSD(x)j versus frequency
using eight different values of ravg for averaging voice-coil
displacement.
Fig. 11. (a) Cross-sectional view of radiator with enhanced
peak and bottom displacement. (b) Mean value of displace-
ment measured at a circle of radius ravg on diaphragm. (c)
Effective radiation area SD versus averaged radius ravg at
100 Hz.
J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February 49
ENGINEERING REPORTS MEASUREMENT OF EFFECTIVE RADIATION AREA
higher angular resolution is necessary here to measure the
displacement distribution on an asymmetrical speaker
with sufficient accuracy (see Fig. 14).
The measured cone displacement data should cover the
complete driver area to give a proper value for SD by
applying Eq. (18). But sometimes this is not possible as
when certain cone regions are not accessible by laser
because they are hidden behind a cover. Then the
integrated surface displacement will be too low, which
leads to a too low SD value.
Furthermore the triangulation laser can produce
errors when measuring the vibration at sharp edges or
poorly reflective spots on the cone. These erroneous
measurement points will also affect the calculation of
SD. The scanning software used for the present
examples can recognize erroneous points by correlat-
ing two subsequent measurements from different
distances. These points will be removed from all
further analyses by setting the measured displacement
to zero. Again the integrated surface displacement will
be lower depending on the number and location of the
erroneous points.
Fig. 12. Intensity plot of cone vibration and sectional view. (a) Circular microspeaker. (b) Oval microspeaker.
Fig. 13. Calculated values of effective radiation area SD for
circular speaker. (a) Radial grid resolution. (b) Angular
resolution.
Fig. 14. Calculated values of effective radiation area SD for
oval speaker. (a) Radial grid resolution. (b) Angular
resolution.
50 J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February
KLIPPEL AND SCHLECHTER ENGINEERING REPORTS
3 DISCUSSION
The laser technique proposed in Section 2.5 performs a
vibration measurement of the complete cone. Measuring
the displacement at several points instead of only one
point is also advisable for the single-box technique
(Section 2.3) and the differential-box technique (Section
2.4). The reason is that especially small speakers without
properly centering suspension are susceptible to asym-
metrical vibrations and rocking modes, as shown in Fig.
15.
These vibration modes usually occur at low frequencies
where the speaker is supposed to perform a piston motion.
The displacement measured at only one point will not be
representative of the motion of the complete cone in the
presence of asymmetrical vibration modes. Therefore it is
necessary to perform a set of vibration measurements at
selected points on the cone and average the measured
data.
4 CONCLUSION
An accurate measurement of the effective radiation
area SD is the basis for a reliable prediction of acoustical
output. A 12% error in SD causes an error of 1 dB in
passband sensitivity.
Exploiting the geometry of the radiator by reading
only the diameter in the center of the surround is the
simplest technique, which may be applicable to woofers
where the width of the surround is small compared to
the total diameter of the radiator and the total volume
velocity is generated by the cone vibrating as a rigid
body. This technique is not reliable for headphones,
microspeakers, compression drives, and other transduc-
ers where the vibrations vary significantly with the
radiator surface. Here the total volume velocity has to be
measured using acoustical and/or optical sensors. The
far-field measurement suggested in this engineering
report is a simple technique giving more accurate results
than near-field measurements but it requires a large
baffle and an anechoic environment. The differential-
box technique puts lower requirements on the test
equipment but requires careful sealing of the test
enclosure.
Measurement of the voice-coil displacement requires a
mechanical measurement, which can be accomplished by
optical laser triangulation sensors. The measurement of
multiple points is required to consider non-axis symmet-
rical modes, as found on most transducers having no
spider.
If the radiator surface is completely accessible to an
optical laser scanner no acoustical measurements are
required and a precise value of the effective radiation area
can be calculated from the measured vibration and
geometry.
5 REFERENCES
[1] L. Beranek, Acoustics (Acoustical Society of
America, New York, 1996), p. 231.
[2] R. H. Small, ‘‘Direct-Radiator Loudspeaker System
Analysis,’’ J. Audio Eng. Soc., vol. 20, pp. 383–395 (1972
June).
[3] D. Clark, ‘‘Precision Measurement of Loudspeaker
Parameters,’’ J. Audio Eng. Soc., vol. 45, pp. 129–141
(1997 Mar).
[4] W. Klippel and U. Seidel, ‘‘Fast and Accurate
Measurement of Linear Transducer Parameters,’’ present-
ed at the 110th Convention of the Audio Engineering
Society, J. Audio Eng. Soc. (Abstracts), vol. 49, p. 526
(2001 June), convention paper 5308.
[5] J. N. Moreno, R. A. Moscoso, and S. Jonsson,
‘‘Measurement of the Effective Radiating Surface Area of
a Loudspeaker Using a Laser Velocity Transducer and a
Microphone,’’ presented at the 96th Convention of the
Audio Engineering Society, J. Audio Eng. Soc. (Ab-
stracts), vol. 42, p. 405 (1994 May), preprint 3861.
[6] R. Ballas, G. Pfeifer, and R. Werthschutzky,
Eletromechanische Systeme der Mikrotechnik und Me-
chatronik (Springer, Berlin Heidelberg, 2009).
[7] S. Jonsson, J. N. Moreno, and H. Bog, ‘‘Measure-
ment of Closed Box Loudspeaker System Parameters
Using a Laser Velocity Transducer and an Audio
Analyzer,’’ presented at the 92nd Convention of the
Audio Engineering Society, J. Audio Eng. Soc. (Ab-
stracts), vol. 40, p. 433 (1992 May), preprint 3325.
[8] J. N. Moreno, ‘‘Measurement of Loudspeaker
Parameters Using a Laser Velocity Transducer and
Two-Channel FFT Analysis,’’ J. Audio Eng. Soc., vol.
39, pp. 243–249 (1991 Apr).
[9] D. K Anthony and S. J. Elliot, ‘‘A Comparison of
Three Methods of Measuring the Volume Velocity of an
Acoustic Source,’’ J. Audio Eng. Soc., vol. 39, pp. 355–
366 (1991 May).
Fig. 15. Irregular vibration of headphone diaphragm. (a)
Asymmetric motion. (b) Rocking motion.
J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February 51
ENGINEERING REPORTS MEASUREMENT OF EFFECTIVE RADIATION AREA
[10] W. Klippel and J. Schlechter, ‘‘Measurement
and Visualization of Loudspeaker Cone Vibration,’’presented at the 121st Convention of the Audio
Engineering Society, J. Audio Eng. Soc. (Abstracts),
vol. 54, pp. 1251, 1252 (2006 Dec.), convention paper
6882.
THE AUTHORS
W. Klippel J. Schlechter
Wolfgang Klippel studied electrical engineering at the
University of Technology, Dresden, Germany. After
graduating in speech recognition, he joined VEB
Nachrichtenelektronik in Leipzig, Germany, where he
was engaged in the research of transducer modeling,
acoustic measurements, and psychoacoustics. In 1987 he
received a Ph.D. degree in technical acoustics. After
spending a postdoctoral year at the Audio Research
Group in Waterloo, Ont., Canada, and working at Harman
International, Northridge, CA, he returned to Dresden in
1997 and founded Klippel GmbH, which develops novel
kinds of control and measurement systems dedicated to
loudspeakers and other transducers. In 2007 he became
professor of electroacoustics at the University of
Dresden.
Joachim Schlechter was born in Oschatz, Germany, in
1981. He studied computer and media science with a
focus on acoustics and signal processing at the University
of Technology, Dresden, Germany. He attended a master
program for sound and vibration at the Chalmers
Technical University, Gothenburg, Sweden, from which
he received a Master’s degree in the field of active noise
control in 2005. He also received a Diploma degree in
computer science, Dresden, Germany, in 2006. The
supervision of his thesis, ‘‘Visualization of Vibrations of
Loudspeaker Membranes,’’ was shared between the
University of Technology of Dresden and Klippel GmbH.
After graduating he joined Klippel GmbH, where he is
currently engaged in the research and development of
loudspeaker control and measurement systems.
52 J. Audio Eng. Soc., Vol. 59, No. 1/2, 2011 January/February
KLIPPEL AND SCHLECHTER ENGINEERING REPORTS