THE USE OF VOLUME VELOCITY SOURCE IN

TRANSFER MEASUREMENTS

N. Møller, S. Gade and J. Hald

Brüel & Kjær Sound and Vibration Measurements A/S
DK2850 Nærum, Denmark
nbmoller@bksv.com

Abstract
In the automotive industry there is a growing need for measurement of acoustical transfer functions in
connection with transfer path analysis, the main outcome being acoustical source contribution analysis. These
transfer functions are from monopole Volume Velocity at a source location to the resulting sound pressure at a
receiver (listener) position. In most cases it is an advantage to make use of reciprocity, which allows the
monopole source position and the pressure response position to be interchanged. The source to be used for
these measurements must be powerful and omni-directional, and the frequency range of interest is typically 50-
6300 Hz. So the Brüel & Kjær OmniSource™ Type 4295 is in many ways perfectly suited for the application.
This paper will discuss the design criteria for a Volume Velocity Source as well as the verification of the
performance. Also the use of Volume Velocity Source in Transfer Path Analysis often called Source Path
Contribution is described. Practical examples using the Volume Velocity Source will be shown.

Design of the Volume Velocity (VV) Source

The Brüel & Kjær OmniSource™ Type 4295 (see figure 2) has been designed to provide a high level (Lw =
105dB ref. 1pW) of omni-directional sound radiation over a broad frequency range from around 60 Hz to 6.3
kHz. The target application was room acoustics measurements. So the OmniSource is in many ways perfectly
suited for the Volume Velocity application. The only feature missing is a capability to measure the Volume
Velocity output and – in particular – the transfer function from Volume Velocity output to the (pressure)
response at a set of positions. Figure 1 shows a new Volume Velocity measurement adaptor to be fitted on the
Type 4295 OmniSource™. Notice the ¼” intensity microphone pair Type 4178 built into the adaptor. The
cylindrical tube section with a diameter of 4 cm will suppress all non-planar waves at the microphones up to
approximately 5 kHz. The first higher order mode has a single radial node line and it can propagate above 5
kHz. But by measuring the pressure on the tube axis, this mode should in principle not be detected, which fits
with the fact that it does not contribute to the output Volume Velocity. The first higher order mode, which has
non-zero pressure on the axis, can propagate only above 10.4 kHz. At 6 kHz this mode will be attenuated 41
dB over the 3 cm from the opening to the outermost microphone B. So if the two microphones – A and B –
could measure the undisturbed pressure exactly on the axis, then only the propagating plane waves would
contribute significantly to the measurement over the considered frequency range. In practice, the microphones
will disturb the sound field and not measure the pressure exactly on the axis.

Assuming that only plane waves are measured, the plane wave components propagating in the two directions
can be estimated from the two microphone signals. These two plane wave components can be extrapolated to
the opening of the tube, and the Volume Velocity output can be calculated.
 Figure 1: Volume Velocity measurement adaptor for OmniSource™ Type
4295

We measure the pressure at the two microphones: p A and p B . Beyond that we measure the response at a
position R. From these signals we average the Cross spectral matrix:

A B R
A C AA C AB C AR
B C BB C BR
R C RR

We use the following notation

S Cross-section area of the tube.

l See the Figure 1. Typically it is 3 cm.
∆ See the Figure 1. Typically it is 2 cm.
k Wave number (2πf/c)
ρ Density of the air
c Propagation speed of sound

The Auto-spectrum C QQ of the Volume Velocity in the opening of the adaptor and the Cross spectrum
between the Volume Velocity and the Response signal can then be shown to be:

 S  CAA cos2 (kl) + CBB cos2 (k (l + ∆)) − 2 Re{CAB}cos(kl) cos(k (l + ∆))

2
CQQ =   (1)
 ρc  sin2 (k∆)
S C AR cos( kl ) − C BR cos( k ( l + ∆ ))
C QR = (2)
ρc j sin( k ∆ )
Using these two spectra, the Frequency Response Function H QR between this Volume Velocity and the
Response signal can be obtained as:

C QR
H QR = (3)
C QQ
Verification measurements in an anechoic chamber

Figure 2: Picture showing the OmniSource™ with VV adaptor

above a turntable

The performance of the Volume Velocity (VV) source was checked in Brüel & Kjær´s anechoic chamber. For
this a Portable PULSE™ Type 3560C with a 3109 Input/Output module was used together with a Power
Amplifier Type 2716 and a single Array Microphone Type 4935 to measure the Response pressure. The
OmniSource™ Type 4295 with a VV adaptor was mounted horizontally approximately 1.2 meter above a
Turntable in such a way that it could be rotated around a vertical line through the output aperture - see Figure 2
below.
The formulae for calculation of the Volume Velocity Auto-spectrum C QQ and the Volume Velocity to Response
transfer function H QR were first programmed in PULSE Language, later they were implemented as External
Functions. Also, the two phase-matched microphones A and B of the VV adaptor were precisely calibrated
using a Sound Intensity calibrator Type 3541.

Measurement of maximum SPL output

Figure 3 shows the 1/3-octave SPL spectrum measured exactly 10 cm in front of the aperture of the VV
adaptor, when the Type 4295 OmniSource™ is driven approximately to its maximum output with Pink noise, the
excitation was frequency band limited to the interval 0-6300 Hz. Figure 4 show the 400 line FFT spectrum at 10
cm distance when the Type 4295 OmniSource™ is driven approximately to its maximum output with White
noise. Notice that loudspeaker distortion is not critical, because we only deal with the radiated sound field –
specifically with a transfer function in that sound field. As seen the VV source has a reasonable flat output as a
function of frequency.

[dB/20.0u Pa] Autospectrum(Response) - Input [dB/20.0u Pa] Autospectrum(Response) - Input

Working : Input : Input : CPB Analyzer Working : Input : Input : FFT Analyzer
100 100

96
90
92

88 80

84
70
80

76
60
72

68 50

64
40
60 50 100 200 500 1k 2k 5k
[Hz]
63 125 250 500 1k 2k 4k
[Hz]

Figure 3: 1/3-octave SPL spectrum at 10 cm Figure 4: FFT based SPL spectrum at 10 cm

distance with maximum pink noise excitation distance with maximum white noise excitation

Measurement of directivity.
At 2 meter distance from the output aperture of the VV adaptor the 1/3-octave SPL spectrum was measured as
a function of the off-axis angle with 10° intervals between the measurement points. Figure 5 shows sequence of
plots of the radiation patterns for the 1/3-octave bands 500, 1000, 2000, 2500, 3150 and 4000Hz. Up to around
3 kHz the SPL does not change more than 5-6 dB over the 360° angle interval. At higher frequencies (not
shown here) the variation over angle goes up to approximately 15 dB.

Figure 5: Directivity patterns from 500Hz to 4kHz

Check of the Transfer Function Measurement

The main application of the Volume Velocity source is measurement of the Frequency Response Function
H QR between this Volume Velocity output and a Response signal. If the VV source radiated as an ideal
monopole point source, then the frequency response function should be given by the following expression:
) e − jkR
H QR = j ρ ck Ideal Monopole Point Source (4)
4π R
So we can compare our measurements with this expression, see Figure 6. At low frequencies the VV source is
highly omni-directional, so the agreement is expected to be good. At higher frequencies, the measured transfer
)
function H QR will tend to deviate from the ideal function H QR in the same way as the radiated pressure
deviates from the omni-directional pressure distribution of an ideal monopole. Because the VV source
concentrates the radiation in the axial direction at high frequencies, the response H QR measured on the axis is
)
expected to be larger than the ideal omni-directional H QR . Looking at the phase (not shown here), the
agreement is best, if the ideal monopole is assumed to be 1.2cm in front of the aperture, i.e. the acoustic center
is located 1.2cm, in front of the aperture.
)
A few comments on the deviations between the ideal omni-directional transfer function H QR and measured
transfer function H QR :
• At 10 cm distance the measured H QR is very smooth and agrees very well with the omni-directional
)
H QR , the only deviation being the high-frequency lift due to the directivity of the source in the direction
of the measurement position. This good agreement shows that the VV adaptor is working very well.
• At larger distances the influence of the environment will introduce further differences between H QR
)
and H QR . The ability to measure these differences is the main benefit of the VV source.
• The influence of the environment (the OmniSource™, its support structure, the measurement
equipment, the floor and the room in general) is seen to be increasing from 10 cm to 1m and again
from 1 m to 2 m distance. At the chosen microphone position the response is mostly decreasing, and
ripples are showing up.
 • In the low/medium frequency areas, these ripples (and the shape of the measured response function in
general) are seen to remain almost constant during the rotation of the source around the aperture.
This indicates that the deviations (the ripples) are not due to parts rotating together with the VV source
– they must be created mostly by the surroundings.

*[Pa/(m³/s)]* PLF - Hqr Monopole (Magnitude) *[Pa/(m³/s)]* PLF - Hqr Monopole (Magnitude)

100k 10k

30k 3k

10k 1k

3k 300

1k 100

300 30
b. 1m distance, 0° off-axis
100
a. 10cm distance, 0° off- 10

50 100 200 500 1k 2k 5k 50 100 200 500 1k 2k 5k

[Hz] [Hz]

*[Pa/(m³/s)]* PLF - Hqr Monopole (Magnitude) *[Pa/(m³/s)]* PLF - Hqr Monopole (Magnitude)

3k 3k

1k 1k

300 300

100 100

30 30

c. 2m distance, 30° off-

10 10
d. 2m distance, 90° off-
50 100 200 500 1k 2k 5k
[Hz]
50 100 200 500 1k 2k 5k
[Hz]

Figure 6: Ideal (blue) and measured (green) transfer function H QR for a

response position on and off-axis at various distances from the aperture plane.

The Frequency Response Function H QR of Figure 6a - measured at 10 cm distance (upper left) - has some
ripples around 5.5 kHz. These ripples are probably associated with corresponding ripples in the phase of the
cross spectrum between microphones A and B – (not shown here). Probably the phase ripple is caused by the
interaction of the sound field in the tube with the two microphones and their extensions across the tube. This is
a small problem to be addressed in the future.

Measurement Example
Some measurements were carried out on a Volvo60 passenger car using the VV source, see Figure 7, the
intentions was to measure the transfer paths for both airborne noise as well as structure borne noise. This
series of measurements were unfortunately not finished, when this paper was written, so only an example is
shown here. The VV source was placed in the drivers seat (right hand steering wheel car), a ½” microphone
Type 4190 was placed under the oil sump and an accelerometer was placed on the right front wheel
suspension.
Measurement were performed using both FFT (Fast Fourier Transform) as well as SSR (Steady State
Response) techniques. For the FFT techniques 400, 1600 and 6400 lines were used in order to see the
influence of resolution and both Random Signal (with Hanning Weighting) as well as Pseudo Random Signal
(with Uniform Weighting) was used in order to see the influence of excitation signals. SSR technique uses
stepped sine excitation, here the step frequency interval was chosen so it corresponded to the 400 line FFT
resolution.
Figure 8 shows some measurement results. The functions used were the following: Autospectra, Coherence,
Cross-correlation, Volume Velocity and Frequency Response Function (Volume Velocity to Pressure) and for
SSR also Measurement Time and Standard deviations as a function of frequency were displayed. Autospectra
of the two microphone positions A & B in the adaptor should not differ too much (Figure 8a). Coherence
between the between microphone B and the response position, R should not be too low due to background
noise.
 Cross-correlation was used to identify the major time delays between
excitation point (microphone B was used) and the response point, R at he oil
sump (Figure 8b), Major delays at 6, 14 and 21ms. The spectrum of the
Volume Velocity output (inverse proportional to frequency) was used to verify there
was not too much acoustical influence from the surroundings, i.e. lack of free field
(Figure 8c). Finally the Volume Velocity Spectrum was calculated (Figure 8d).

Figure 7: VV source in a
Volvo60 passenger car

b. Croscorrelation (AP)

a. Ch. A & Ch .B pressure

d. Frequency Response
c. Volume Velocity

Figure 8: Measurement results from a Volvo60 passenger car

Conclusion
In this paper we have introduced the design and use of a Volume Velocity Source based on a powerful OmniSource™
Type 4295 fitted with an adaptor, which includes a pair of phase-matched microphones in order to measure the VV-output
as well as the VV to pressure Frequency Response Functions.