BRÜEL & KJAER-Influence of Tripods and Mic Clips

Download as pdf or txt
Download as pdf or txt
You are on page 1of 48

PREVIOUSLY ISSUED NUMBERS OF

BRUEL & KJ/ER TECHNICAL REVIEW

3-1985 The Modulation Transfer Function in Room Acoustics


RASTI: A tool for evaluating auditoria
2-1985 Heat Stress
A New Thermal Anemometer Probe for Indoor Air Velocity
Measurements
1-1985 Local Thermal Discomfort
4-1984 Methods for the Calculation of Contrast
Proper Use of Weighting Functions for Impact Testing
Computer Data Acquisition from B & K Digital Frequency Analyz
ers 2131 /2134 using their Memory as a Buffer
3-1984 The Hilbert Transform
Microphone System for Extremely Low Sound Levels
Averaging Times of Level Recorder 2317
2-1984 Dual Channel FFT Analysis (Part II)
1-1984 Dual Channel FFT Analysis (Part I)
4-1983 Sound Level Meters - The Atlantic Divide
Design principles for Integrating Sound Level Meters
3-1983 Fourier Analysis of Surface Roughness
2-1983 System Analysis and Time Delay Spectrometry (Part II)
1-1983 System Analysis and Time Delay Spectrometry (Part I)
4-1982 Sound Intensity (Part II Instrumentation and Applications)
Flutter Compensation of Tape Recorded Signals for Narrow Band
Analysis
3-1982 Sound Intensity (Part I Theory).
2-1982 Thermal Comfort.
1-1982 Human Body Vibration Exposure and its Measurement.
4-1981 Low Frequency Calibration of Acoustical Measurement Systems.
Calibration and Standards. Vibration and Shock Measurements.
3-1981 Cepstrum Analysis.
2-1981 Acoustic Emission Source Location in Theory and in Practice.
1-1981 The Fundamentals of Industrial Balancing Machines and their
Applications.
4-1980 Selection and Use of Microphones for Engine and Aircraft Noise
Measurements.
3-1980 Power Based Measurements of Sound Insulation.
Acoustical Measurement of Auditory Tube Opening.
2-1980 Zoom-FFT.
1-1980 Luminance Contrast Measurement.
(Continued on cover page 3)

TECHNICAL REVIEW
No. 4 1985

Contents
Validity of Intensity Measurements in Partially Diffuse Sound Field
by Svend Gade

Influence of Tripods and Microphone Clips on the Frequency Response


of Microphones
by K. Zaveri
32
News from the Factory

41

VALIDITY OF INTENSITY MEASUREMENTS


IN PARTIALLY DIFFUSE SOUND FIELD
by
Svend Gade, M.Sc.

ABSTRACT
In this article a practical method is proposed and outlined for determining the
Dynamic Capability of Intensity analyzing systems and the Reactivity Index of
Intensity measurements. Furthermore, using this method, the amount of error due
to phase mismatch, the amount of random error, and the useful frequency range
for measuring intensity in different types of sound fields can be determined.
SOMMAIRE
Cet article propose et donne les grandes lignes d'une methode pratique pour
determiner la capacite dynamique des systemes d'analyse d'intensite acoustique
et I'indice de reactivite des mesures d'intensite. De plus, en utilisant cette
methode, la valeur de I'erreur due au dephasage, la valeur de I'erreur aleatoire et
la gamme de frequence utile pour les mesures d'intensite dans les differents
types de champ sonore peuvent etre determinees.
ZUSAMMENFASSUNG
In diesem Artikel wird eine praktische Methode zur Bestimmung der dynamischen
Fahigkeiten und des Reaktivitatsindexes von Intensitatsmessungen vorgeschlagen
und beschrieben. AuBerdem laBt sich mit dieser Methode der Fehler durch
Phasenfehlanpassung, der Zufalligkeitsfehler und der fur verschiedene Schallfelder zulassige Frequenzbereich bestimmen.

Sound Intensity
Sound Intensity is a vector quantity, which describes the amount and the
direction of net flow of acoustic energy at a given position.
It can be shown [1,13] that the intensity vector
direction r can be calculated from
2

1 <9$

'' = - ^ T T T
pc

component

in the

< >

dr

3
r

where d$/dr is the phase gradient of the sound field in the direction, r,
2
p rms is the mean square pressure, pc is the impedance of the medium
and k is the wave number.
Equation (1) shows that intensity calculations using the two-microphone
method involves determination of the mean square sound pressure as
well as the phase difference of the sound field between the two micro
phone positions. The critical point for sound intensity calculations is the
phase measurement. The sound pressure is often taken as being the
mean pressure value between the two microphone signals.

Bias Error, Approximation Error


For a plane (free field) sinusoidal wave, which propagates along the axis
joining the microphones, the two-microphone method assumes that the
free field phase between the two microphone positions is equal to
(k Ar), i.e. proportional to frequency and microphone spacing, while the
measured free field phase is sin (k Ar). Thus the measured intensity, /,
(see Fig.1) is related to the true intensity, /, [1] by:

/ / / = sin (A:- Ar)/(k- Ar)

(2)

For sound intensity measurements in environments with diffuse back


ground noise the approximation error appears to take the same form as
for free field conditions [4,5]. This is because an intensity meter only
responds to the propagating part (active part) of a sound field, and
because the upper frequency limit, where the accuracy is within 1 dB, is
found where the microphone spacing is approximately 6 times smaller
than the apparent wavelength.
If there exists an angle, a, between the direction of propagation and the
direction of probe orientation, r, the approximation error formula
becomes

*
'a ~ 'r

sin {kAr cosa)


~
kAr
sin(/cAr cosa)
kAr cosa

As an example, if the angle, a, between the propagation of sound and


the probe orientation is 60, then the upper frequency limit for intensity
4

Fig. 1. Approximation error, LtAP, at high frequencies in an active sound


field for various spacers
measurements is equal to twice the upper frequency limit as found when
the angle, a, is 0.
Since in a general sound field the direction of flow of acoustic energy
might be different from microphone position to microphone position, any
attempt to correct the measured intensity values by means of equation
(2) would not be advisable. The formula only gives an idea of the upper
frequency limit for a given microphone separation.

Intensity Index Nomogram


It can be shown [1,2,3,13] that in the frequency domain the intensity
function can be calculated from the imaginary part of the cross-spec
trum function, GAB
f{f)

JULGAB

(4)

cop A r
where co is the angular frequency and p is the density of the air.
5

The general formulation of the cross-spectrum is


GAB= E[A* B]
= E[\A\

\B\ (cos <S>BA + ysin $ M ) ]

(5)

where E [ ] denotes expected value of the quantity inside the brackets.


For simplicity we omit this symbol. The star * indicates a complex
conjugate, that is a change of the sign of the imaginary part, which also
corresponds to a change of the sign of the phase.
From (4) and (5) we have
/ cop A r = | A | | B\ sin $AB

(6)

For small angles sin <bAB $AB. We also have that


\A\

|0| ~

2
p rms

Thus equation (6) can be rewritten as


f- kc

or

p Ar= p

/cAr
$AB

rms$AB

2
p rms/pc

(7)

which is the Intensity Index Nomogram relation.


Equation (7) could have been evaluated directly from equation (1) by a
finite difference approximation of the phase gradient, d$/dr - $ A B / A r .
Equation (7) shows the general relation between the free field phase or
relative frequency, (/cAr), wave number or frequency, /c, microphone
2
spacing, Ar, actual phase, $AB (or just $), pressure, p rms, and intensity,
/, for an actual measurement.
When the propagating sound field of interest is contaminated by diffuse
background noise the actual phase $ is smaller than the corresponding
free field phase /cAr.

The Reactivity Index, K is defined as


K = *-'

(8)

or in logarithmic form with pc = 400 and the usual reference quantities


L = - 1 0 l o g K= L,-Lp

(9)

In most situations we have that K ^ 1 and thus LK ^ 0 dB.


The use of LK was introduced by Roland [6]. The Intensity Index Nomogram relation is shown in graphical form in Fig.2. Note that these curves
make no allowance for the finite approximation error at high frequencies.
The nomogram is very useful in evaluating errors due to phase mis
match. For a Reactivity Index of OdB (free field condition, Lp = Lj) it is
seen that when using a 12 mm microphone spacing, the phase of the
sound field between the two microphone positions is 10 at 800 Hz, 1 at
80 Hz and 0,1 at 8 Hz. Here we assume that the probe orientation is the
same as the direction of propagation of sound.
With a Reactivity Index of - 1 0 d B (L, = Lp- 10dB), the measured phase
is 1 at 800 Hz, and with a Reactivity Index of - 2 0 dB the phase is only
0,1.
It is also seen that for a Reactivity Index of OdB and a microphone
spacing of 6 mm, the phase of the sound field between the two micro
phone positions is 5 at 800 Hz, 0,5 at 80 Hz and 0,05 at 8 Hz.
Evidently, a phase mismatch is most critical at low frequencies and for
high reactivity indices of the measurements, and for small spacings
between the two microphones. In fact the measured phase of the sound
field should always be at least 5 times larger than the phase-mismatch of
the system to ensure an accuracy better than 1 dB.
F

Note that in practice the Reactivity Index, LK, is normally negative and
indicates an important character of the sound field as it is measured. LK
is not a direct measure of how reactive the sound field is, or how much
diffuse background noise is present. As an example of a highly reactive
sound field, a standing wave can be used. It can be shown (Refs.[12,13]
or see Appendix A) that in such a sound field the intensity is the
geometrical mean value between the maximum mean square pressure
7

o
CO
<D

c:
o

E
c:
CO
Q>
CO
CO

I
o
CO
CD

Cr
CD

CD
Q

o
CD

u.

and the minimum mean square pressure values normalized with respect
to the impedance of the medium, pc. In other words if the standing wave
ratio is 20 dB, then LK can take any value between -10dB and + 10dB
depending on the observation position in the standing wave.
Also note that LK cannot distinguish between different reasons for the
measured phase. If Lp is 3dB higher than the measured L, the situation
could be that we have
1)

a plane propagating sound field, where the angle between the


direction of propagation of sound and the direction of probe orienta
tion is 60 or

2)

a sound field consisting of two uncorrelated parts of equal strength,


namely a plane propagating sound field and a diffuse sound field. In
this case the direction of propagation of sound and the direction of
probe orientation is assumed to be the same.

In the first case the upper frequency limit for the measuring system will
be two times higher than in the latter case as discussed earlier.
In fact there exists an infinite amount of different combinations of active,
reactive and diffuse sound fields where LK is - 3 d B .
All of the above discussion assumes that the intensity is determined
without error.

Error due to Phase Mismatch


If the measured phase, kAr, is small and if a phase mismatch, <p, exists
between the two measuring channels, the relationship between the
measured intensity, /, and the true intensity, /, for free field conditions
becomes

?/l = sin (kAr cp)/(kAr)


(kAr <p)/(kAr)

(10)

For the more general case, where free field conditions cannot be
assumed, equation (10) becomes
f / / (</> <)/</>

(11)

Using eqs.(8) and (9) the nomogram relation (7) can be written as
(Z

$ = io -*

/10)

(/cAr)

(12)

If a phase mismatch, <p, between the two channels exists and the phase
difference of the sound field is 0, the measured phase becomes
(L /10)

(f) <p = 10 *

(/cAr)

(13)

If we feed the same broadband signal to the two measuring channels, we


simulate a sound field with 0 phase difference between the two mea
surement positions [4,5,23,24]. In this case the difference between the
measured intensity level (the Residual Intensity level, Z_/R, of the analyz
ing system) and the pressure level, LpR, is a measure of the phase
mismatch, <p, between the two measuring channels and is, per definition,
the Residual Intensity Index of the system, LK>0.
LK>0 = LIJR-LPJR

(14)

From Equation (13), the phase mismatch between the two channels is
given by
<p = io^-o/io) .

(15)

{kAr)

Solving equations (13) and (15) for <j> gives


(/

0 = (10(^/10) + io -*'

/10)

) kAr

(16)

Inserting equations (13) and (16) into (11) yields:

f/l = 1 / ( 1 +

1o^o-iK)/^

(17)

or in logarithmic form
L^hase = -10 log [1 + 1 o^o-W/io)]

(18)

Equations (17) and (18) indicate that the error due to phase mismatch
only depends on the level difference between the Residual Intensity
Index of the measuring system and the measured Reactivity Index of the
sound field at the microphone positions.

10

Fig. 3. Error due to phasemismatch

for Intensity

measurements

Equation (18) is derived in Appendix B without the use of the nomogram.


Fig.3 shows equation (18) in graphical form. The curves indicate that a
measured Reactivity Index should be at least 7dB higher than the
Residual Intensity Index of the analysing system to ensure an error due
to phase mismatch of less than 1 dB.
Thus we could define the Dynamic Capability of an Intensity Analyzing
System by adding 7dB to the Residual Intensity Index.
It should be noted that for most practical measurements all three
quantities, Dynamic Capability, Reactivity Index and Residual Reactivity
Index will be negative values.
The relationship between sound field and measurement system indica
tors for a given microphone spacing Ar using the two-microphone
method is shown in Fig.4.

11

Fig. 4. Relationship between sound field and measurement system indi


cators where LK is the Reactivity Index and LKt0 is the Residual
Intensity Index

Fig. 5. The useful frequency range for various spacer


configurations,
phase-match and measured Reactivity Index for an accuracy of
1dB

12

Note that increasing the microphone spacing will increase the Dynamic
Capability but decrease both the upper and the lower frequency limit for
the system. Thus the nomogram can be used for calculation of the useful
frequency range as shown in Fig.5.

Random Error
For sound pressure measurements the normalized random error (68%
confidence interval) is inversely proportional to the square root of the BT
product (Bandwidth multiplied by Averaging Time) for BT^ 10.
random [P'rms] =

^]fBf

(19)

For Sound Intensity measurements an equivalent formula exists. From [7]


or from Appendix C we have
* random V] = ^'Wh

/ l - ((1 - y ) / 2y ) f $\x\ $

(20)

where y is the coherence function between the two signals at the


positions of the microphones. A simple model [8,9] is used for calcula
2
tion of y and $ for Sound Intensity measurements, where the sound
source of interest is regarded as a point source uncorrelated with the
diffuse background noise.

Fig. 6. Measurement of sound intensity from a point source situated in


environments with diffuse background noise

13

For simplicity the sound source of interest is situated in the direction


where the microphone probe has maximum sensitivity, see Fig.6. Howev
er, this simplification does not impose any restriction on the orientation
of the probe relative to the source, for measurements to be valid.
The measured Cross-spectrum due to the point source is
GAB = GPP [cos (kAr) + ysin (kAr)]

(21)

where GPP is the Autospectrum (the pressure spectrum).


The measured Cross-spectrum GAB due to the diffuse background noise
is:
GAB=

GDD

[sin (kA r)/(kA r)]

(22)

where GDD is the Autospectrum (the pressure spectrum).


The resulting Cross-spectrum is the sum of the two Cross-spectra and
the resulting Autospectrum is the sum of the two Autospectra. This is
because the point source signal is uncorrelated with the diffuse back
ground noise signal.
In this case the Reactivity Index is
K = Pressure!Intensity

= {GDD + GPP)/GPP

(23)

Note that equation (23) also shows that the Reactivity Index, LK, gives an
indication of the Signal/Noise ratio for the measurements. As an exam
ple, when the propagating part of the sound field contributes 10% and
the diffuse part of the sound field contributes 90% of the total sound
pressure (or total energy density), LK is -10dB. In this case the Signal/Noise ratio is -9,5dB.
Using (21), (22) and (23) and GAA = GBB = GDD + GPP we get
{

y* =

"

AA

BB

~ \GAB\ /(GPP + GDDf


2

= (1//C ) [(cos (kAr) + (K-1)- sin (kAr)/(kAr)f

14

+ sin (/cAr)]

(24)

and

$= tan' [Im GAB/Re GAB]


1

= tan" [sin (/cAr)/(cos (/cAr) + (K-1)- sin (/cAr)/(/cAr))]

(25)

Inserting equations (24) and (25) into (20) shows [8] that the required
averaging time (BT-product) depends on the desired statistical accuracy,
en and the Reactivity Index, K or LK, but not on the relative frequency,
/cAr(the free field phase), in the frequency range of interest as shown in
Fig.5, (see also Ref.[8]),
BT=

M f , y , $) = f2{er, LK)

(26)

The curves for 5%, 10%, 20% and 40% accuracy are shown in Fig.7.
These curves indicate that a change in Reactivity Index of - 5 d B requires
an increase in the averaging time by a factor of approximately 10 to yield
the same statistical accuracy.
Fig.8 shows on linear axes er[l] tfBTas a function of K.

Fig. 7. Normalized random error (68% confidence interval) for Intensity


measurements, diffuse background noise is assumed

15

Fig. 8. Normalized random error (68% confidence interval) for Intensity


measurements
The graph indicates a nearly perfect linear relationship between the two
parameters.
For K>

1 we have
r

er[l]

YBT^

0,42 fP

0,42 (K+

,pc

1)

+ 1)
(27)

Sound Fields
At a given point in a sound field there will be an acoustic pressure, p,
and a particle velocity, u. A natural way of classifying the sound field for
intensity measurements is whether p and u are correlated or not.
The diffuse sound field is an example where p and u are uncorrelated. In
this case there is no amplitude gradient or phase gradient of the sound
field. All the energy is stored in the sound field.

16

Where p and u are correlated a further subdivision is possible. The


situation where p and u are in phase can be defined as an active sound
field. An example of a purely active sound field is a plane wave
propagating in a free field. In this situation a phase gradient exists but
no amplitude gradient. All acoustic energy propagates [1].
The situation where p and u are in quadrature, that is 90 out of phase,
can be defined as a reactive sound field. An example of a purely reactive
sound field is a standing wave. In this situation an amplitude gradient
exists but no phase gradient. All energy fluctuates between the sources
(real as well as imaginary sources) and the medium [1].
Note, that in general a sound field consists of an active, reactive and
diffuse part. Also note that there might exist singular points in a sound
field where the above definitions do not hold. One example is found
where we have pressure maxima in an ideal, standing wave. In these
points u is equal to zero - that is we have a purely pressure field.
Another similar example is where we have velocity maxima in an ideal
standing wave. In these points p is equal to zero - that is we have a
purely velocity field. See A.1, A.2, A.3, see Refs.[25,26] for more
information.
Note that in this paper we have used a mathematical model in which a
point source emitting random noise is placed in a sound field contami
nated with diffuse background noise for evaluation of finite difference
approximation errors, errors due to phase mismatch and random errors
for intensity measurements. Other models have also been used in Ref.[8].
As can be seen from this section, there is still a lot of work to be done
before we have a model using a general sound field for evaluation of
these errors, as well as errors for sound power determination using
sound intensity measurements. Recent developments in this field are
found in Refs. [15-22].

Conclusion
For intensity measurements in highly diffuse environments one must
always measure the Residual Intensity Index, LK0 (or the Dynamic Capability) of the measuring system as well as the Reactivity Index, LK of the
sound field in the direction given by the microphone orientation at the
point considered.

17

Error due to phase mismatch depends on the difference LK0-LK


random error depends on LK as indicated in Fig.9.

and

Fig. 9. Random error depends on the measured Reactivity Index of the


sound field. Error due to phase mismatch depends on the differ
ence between the measured Reactivity Index of the sound field
and the Residual Intensity Index (Dynamic Capability) of the ana
lyzing system
For the difference approximation error at high frequency it appears that
the underestimation is similar to that under free field conditions.
A practical example where the outlined procedure for determining the
validity of intensity measurements has been used is shown in Ref.[14].
Here measurements were performed inside a highly reactive and diffuse
sound field, namely an empty aircraft in flight.

Acknowledgement
For useful discussions the author wishes to thank Dr. J. Pope and T.G.
Nielsen M.Sc, Bruel&Kjaer.

18

References
[ 1]GADE, S.

"Sound Intensity (Theory, Instrumenta


tion and Applications)", 1982, B&K Tech
nical Reviews, Nos.3 & 4

[ 2] FAHY, F.J.

"Measurements of Acoustic Intensity us


ing the Cross Spectral Density of two
Microphone Signals", 1977, J. Acoust.
Soc. Am. 62 (4), pp.1057-1059

[ 3] CHUNG, J.Y.

"Cross-Spectral Method of Measuring


Acoustic Intensity", 1977, GM Research
Publication, GMR-2617

[ 4] GADE, S., GINN, B.,


ROTH, O. & BROCK, M.

"Sound Power Determination in Highly


Reactive Environments", 1983, Internoise
Proceedings, pp.1047-1050

[ 5] GADE, S., GINN, B.,


ROTH, O. & BROCK, M.

"Sound Power Determination in Highly


Reactive Environments", 1983, B&K Ap
plication Note. Extended version of [4]

[ 6] ROLAND, J.

"What are the Limitations of Intensity


Technique in a Hemi-diffuse Field", 1982,
Internoise Proceedings, pp.715-718

[ 7] ELKO, G.W.

"Frequency Domain Estimation of the


Complex Intensity and Acoustic Energy
Density", 1984, Ph.D. Thesis, Pennsylva
nia State University, pp.77-80, 328-332

[ 8] DYRLUND, O.

"A Note on Statistical Errors in Acoustic


Intensity Measurements", 1983, Journal
of Sound and Vibration, 90, pp.585-589

[ 9] BENDAT, J.S.,
PIERSOL, A.G.

"Engineering Applications of Correlation


and Spectral Analysis", 1980, New York,
Wiley-lnterscience, pp. 167-172

[10] JENKINS, G.M.,


WATTS, D.G.

"Spectral Analysis", 1968, San


co: Holden-Day. Section 9.2.

Francis-

19

[11] SEYBERT, A.F.

"Statistical Errors in Acoustic Intensity


Measurements", 1981, Journal of Sound
and Vibration, 75, pp. 585-589

[12] SCHULTZ, J.,


SMITH Jr. P.W.,
MALME. C.I.

"Measurement of Acoustic Intensity in


Reactive Sound Field", 1975, J.Acoust.
Soc. Am., 67 (6), pp.1263-1268

[13] JACOBSEN, F.

"Measurement of Sound Intensity", 1980,


The Acoustics Laboratory, Technical Uni
versity of Denmark, Report No.28

[14] GADE, S., NIELSEN, T.G.

"Sound Intensity Measurements Inside


Aircraft", 1985, Eleventh European Rotorcraft Forum, Paper no.9, The City Uni
versity, London, England

[15] HUBNER, G.

"Development of Requirements for an In


tensity Measurement Code Determinat
ing Sound Power Level of Machines
Under (Worst) In Situ Conditions", 1984,
Internoise Proceedings, pp. 1093-1098

[16] RASMUSSEN, P.

"Phase Errors in Sound Power Measure


ments", 1985, Dept.13 Publication, Bruei
& Kjaer, Denmark

[17] POPE, J.

"Validity of Sound Power Determination


using Sound Intensity Measurements",
1985, Proceedings of 2nd International
Congress
on
Acoustic
Intensity,
pp.353-360

[18] NICOLAS, J.,


LEMIRE, G.

"Precision of Active Sound Intensity


Measurements in a Progressive and a
Non Progressive Field", 1985, J.Acoust.
Soc. Am, 78, pp.414-422

[19] NICOLAS, J.,


LEMIRE, G.

"A Systematic Study of the PressureIntensity Index", 1985, Internoise Proceedings, pp.1179-1182

20

[20] NICOLAS, J.,


LEMIRE, G.

"Comments on the Validity of the Index


Lt-Lp\
1985, Proceedings of the 2nd In
ternational Congress on Acoustic Inten
sity, pp.337-342

[21] BOCKHOFF, M.

"Sound Power Determination in the Pres


ence of Background Noise", 1985, Pro
ceedings
of
the 2nd
international
Congress
on
Acoustic
Intensity,
pp.275-282

[22] STIRNEMANN, A.,


BOLLETER, U.,
RATHE, E.J.

"Possibilities and Limits of Sound Power


Measurements with a Real-Time Intensity
Analyzer", 1985, Journal of Sound and
Vibration, 98 (3), pp.403-413

[23] FREDERIKSEN, E.,

"Phase Characteristics of Microphones


for Intensity Probes", 1985, Proceedings
of the 2nd International Congress on
Acoustic Intensity, pp.23-30
F

[24] RASMUSSEN, G.,


BROCK, M.

"Transducers for Intensity Measurements", 1983, Proceedings of 11th Inter


national Congress on Acoustics, vol.6,
pp.177-180

[25] TICHY, J.

"Use of the Complex Intensity for Sound


Radiation and Sound Field Studies",
1985, Proceedings of the 2nd Internation
al Congress on Acoustic
Intensity,
pp.113-120

[26] PASCAL, J.C., LU, J.

"Advantage of the vectorial nature of


Acoustic Intensity to describe Sound
Fields" 1984, Internoise
Proceedings,
pp.1111-1114

21

APPENDIX A

Reactivity Index in a Standing Wave


A standing wave consists of two plane waves travelling in opposite
directions.
Thus using complex notation the instantaneous pressure as a function of
time, t, and position, x, is
p

AeJ(o>t-kx)

Bej[ut+kx)

The instantaneous particle velocity can be calculated from eq. (A.2),


Refs.[1,13]
u

- - - L f JE- m

(A.2)

p J dx
J(a3t kx)

which in this case gives

u = A
cop
The mean square pressure is

ikx

^2 = - L p p* = J - (A e~
2

= (A
v
2

2Jkx

+ B + AB e~

+Be

-B e
cop

+ikx

) (A e

2Jkx

+ BA e

2
p max

= r- (A + Bf

+jkx

i{wt+

+ B e~

and

2
p min

= {A-Bf

(A.3)

Jkx

= {A + B + 2 A B cos 2/ex)
This means that

kx)

(A.4)

at antinode, kx = 0 + rnr (A.5)

at node, kx = *fe + mr

(A.6)

The mean square particle velocity is


| (A e ~
pc )
22

jkx

Jkx

-Be )

(A e

Jkx

kx

- B e~' )

= - y - ( I (A + B - AB e'
= I

j2kx

+J2kx

- AB e

) (A + B - 2 AB cos 2/ex)

(A.7)

This means that


2

= J

u mjn

and

2
u max

J . - (A - Bf

at antinode, kx = 0 + mr

{A + B)

at node, kx = TT/2 +

(A.8)

rnr

(A.9)

The pressure has maxima where velocity has minima and vica versa.
The time averaged intensity is
l=Re[

p u*

= Re f {A e~

\-pc
= Re f Lpc

jkx

+ Be

Jkx

) (A e

J
2

= Re f (A - B + 2ABjs\n

=
pc

- Be*)'

2
J
2
2
J2kx
J2kx
]-{A -B - AB e+ AB e )
2

Lpc

Jkx

2kx)
J

1- (A - B ) = - 1 - (A + 8) (A - B)
v
y
2
'
pc
2

(A.10)
'

Eq. (A.10) shows that the intensity is constant along the standing wave.
Now we define the standing wave ratio, R, as

2
P min

(A-Bf

Using eqs. (A.5) and (A.10) we get

P max''pc = A + B =
/
A-B

and using eqs. (A.6) and (A.10) we get

= (
min/pc

'

I = R ( A 13)
{
A-B>

Combining eqs. (A.12) and (A.13) we get


/ = ]/ (Pmln/pC)

(P~ max/pC)

(A.14)

23

that the intensity is the geometrical mean value between the maximum
and minimum mean square pressures normalized with respect to pc.
Thus on a logarithmic (dB) scale the intensity level is the arithmetical
mean value between the maximum and minimum sound pressure level.
See also Refs. [12,13].
L
L, =

+ L
(A.15)

Fig.A.1 shows pressure, intensity and velocity levels for the second
mode in a tube where the standing wave ratio is 25 dB.
Fig.A.2 shows the phase difference between pressure and velocity for
the same case. There is a maximum phase difference between the 2
quantities when their levels in dB are the same. For a standing wave
ratio of 25dB the phase difference is 83. At these points the sound field
is highly reactive.

Fig. A.1. The level of pressure, Intensity and velocity for a standing wave,
where the standing wave ratio is 25dB. One wavelength is
shown
24

Fig. A.2. The phase difference between pressure and velocity for a
standing wave, where the standing wave ratio is 25dB. One
wavelength is shown
Fig.A.3 shows pressure, intensity and velocity levels for the second
mode in a tube where the standing wave ratio is 100dB.
Where the pressure and the velocity have their maximum or minimum
values the phase difference is 0. At these points the sound field is
purely active.
Figs.A.1 and A.3. also reveal that there are more positions in a standing
wave where the pressure level is higher than the intensity level (LK is
negative) than where the intensity level is higher than the pressure level
(LK is positive).
Fig.A.4 shows the relative amount of positions in a standing wave as a
function of standing wave ratio, where we have LK positive.
In this Appendix reference values of p0 = 20n Pa, u0 = 50nm/s, and
2
l0 = 1 pW/m have been used.
25

Fig. A.3. The level of pressure, Intensity and velocity for a standing wave,
where the standing wave ratio is 100dB. One wavelength is
shown

Fig. AA. The relative amount of positions in a standing wave, where the
Intensity level is higher than the pressure level, that is LK > OdB
26

APPENDIX B

Error due to Phase Mismatch


If we consider the Residual Intensity level as being the system noise for
our measurements, and that the measured intensity level as being the
signal contaminated by the system noise, we have
L,: Signal Noise

(B.1)

LhR: Noise

(B.2)
2

where /_, and LIR are measured quantities. If instead of using 1 pW/m
and 20 fi Pa (/_, = OdB and Lp = OdB) as our references use the pressure
levels LpR and Lp as indicated in Fig.B.1 we have
LK : Signal Noise

(B.3)

LKS>: Noise

(B.4)

The relative amount of error due to system noise can be defined as


/

-in i~~

\'measured]

/r,

L f p = 10 log

(B.5)

'true

Fig. B.1. Signal to system noise ratio for Intensity

,-*

measurements
27

Combining eqs. (B.3), (B.4) and (B.5) we get


L /1

/
L*,P -

1 0 l0

= 10

10

9 (10LK/IO

' (l

10 L K ,O/IOJ

+10(^0-^10)
(Z

L /10

= - 10 log (l + io -^- ^

(B.6)

Equation eq. (B.6) is shown in Fig.B.2 for reactivity indices larger as well
as smaller than the Residual Intensity index.
The upper right hand curve shows the case when

Sig

Lf

measured

S l

Lf

true

= Sign

LlResiduai

When LKtQ-LK = - 3 d B , we have that LIJrue = LitResiduah so that the mea


sured intensity level, Llmeasured, will be 3dB higher than the true intensity
level. The error is a 3dB overestimation.

Fig. B.2. Error due to phase mismatch as a function of LKi0-LK


28

The lower curve shows the case when


s

'9

^/, measured

S l

Lf true +

S l

Li; Residual-

In this case the intensity will always be underestimated.


When LK0-LK = OdB and LK and LK0 have opposite signs we have that
Li, true = Lfi measured + 3dB. This means that the intensity is underestimated
by3dB.
LK0-LK=
+ oodB occurs when the true intensity level L, true is equal to
the Residual Intensity level liiR but with opposite sign. In this case the
measured intensity level L,measured = -oodB and thus we have an infinitely
high underestimation of the intensity.
The upper left hand curve shows the case when

'9

'9

Lit measured +

'9

Lf

true

=4= S i g n Lf

Resjduai

or
s

Lf

true

S i g n Lf measured

'9

L( Residual

Where LK0-LK = + 3dB we have the case that the measured level
L
is
i, measured
equal to the true level of the signal L{ true but having the
wrong sign.
When measuring in a sound field where the intensity is equal to zero, the
two upper curves go towards infinity.
In this case the measured intensity will be the Residual Intensity of the
analyzer LfR. Thus LK0 - LK= OdB and we have an infinitely high overestimat'ion of the true intensity.
These curves are of course only of academic interest and should never
be used for correction of measured intensity levels. On the other hand
they indicate that the measured intensity level should always be at least
7dB above the Residual Intensity level to ensure a correct sign and an
accuracy in estimating the true intensity level better than 1 dB.

29

APPENDIX C

Random Error
From Refs.[7,10] we have that the variance of the imaginary part of the
Cross-spectrum can be expressed as
2

var [Im GAB] = a [im GAB]


[GAA

G B B +

= YBT

lm G^e~Re GAB]

(C.1)

where a is the standard deviation of the quantity inside the brackets.


The intensity is calculated from the imaginary part of the cross-spectrum
as shown in equation (C.2).
/= - ^

(C.2)

pcoAr
Now we have

a [I] = a [Im GAB]/pooAr

and

a [I] = (<r[lm GAB]/puArf

(C.3)
= ^ ^ f f

(C.4)

The relative random error is defined as

e, ['] =

~f

(C5)

Combining eqs. (C.1), (C.2), (C.4), and (C.5) gives


2

2rn

- [

|m

AB]

Im GAB
1
=

\&AA

2BT I
30

'

GBB

lm

GAB

lm G e

Re

AB~\

-I

/Q

g\

Inserting GAA GBB = (Re G^ e +


2

where y

\m GAB)/y

is the coherence function, we get


2

2
6

1 rRe GAB + lm GAB


2
2
2BT \y lm GAB

rcot *

. - 1 - L[1

2er

lm GAB
2
lm GAB

_ Re GAB
2
lm GAB

- i

J-2

< - ?Vt']
2
J

,0.7,

sin
$
2
using cot $ =
=
we have
sin^$
rl

2BT
1
=

er

7 sin $
2

(1 _ 7 )

r
L

1 +

2 7 sin $

1
(C8)

The normalized random error is a function of the BT product, the


calculated coherence and the calculated phase difference between the
two channels including signals and measuring chains. It should be noted
that Seybert [11] has evaluated an equivalent formula
r

r 1

(1 - y ) cot

$1

under the assumption that the phase angle $ is small.


The formula eq. (C.8) is a more general formula since it is not evaluated
under this assumption.

31

INFLUENCE OF TRIPODS AND MICROPHONE CLIPS


ON THE FREQUENCY RESPONSE OF MICROPHONES
by
K. Zaveri, M.Phil.

ABSTRACT
Use of microphone clips and tripods to support microphones causes disturbance
of the sound field and thus causes errors in sound level measurements. This
article illustrates the amount of errors introduced for different mounting configu
rations, and shows how these errors can be kept to a minimum.
SOMMAIRE
Les pinces et les trepieds utilises pour maintenir les microphones provoquent des
perturbations du champ sonore qui peuvent fausser les mesures du niveau
sonore. Get article illustre I'amplitude des erreurs provoquees par les differents
types de montage, et montre comment ramener ces erreurs a un minimum.
ZUSAMMENFASSUNG
Mikrofonhalter und Stative verursachen Storungen im Schallfeld und beeinflussen
somit die Schallmessung. In diesem Artikel wird der sich fur verschiedene
Aufbauten ergebende Fehler diskutiert und gezeigt, wie sich diese Fehler auf ein
Minimum reduzieren lassen.

Introduction
A free-field equalized microphone measures the sound pressure existing
at the position of the diaphragm in the absence of the microphone, thus
compensating for the interference created in the sound field due to the
microphone itself. If a microphone clip, extension rods and tripod are
used to support the microphone, however, disturbance of the sound field
can be caused, especially if the latter is mounted close to the micro
phone [1]. To investigate the amount of error introduced, measurements
were carried out in an Anechoic Chamber, using a small loudspeaker,
and a 1/2" free field Microphone Type 4133 suspended by thin strings
from metal wires. Their relative positions in the Anechoic Chamber are
shown in the sketch of Fig.1.
32

Fig.1 Microphone, Tripod and loudspeaker positions in the Anechoic


Chamber
The record length in the Narrow Band Analyzer 2031 corresponding
to full scale frequency range of 20 kHz is 20 ms. As the "after
trigger recording" was set to 0,4, reflected signals from objects at
3
a distance of (344xO,4x20x10~ )/2 = 1,38m would be recorded.
Thus signals reflected from the tripod wil be recorded whilst most
of those reflected from the walls will be subdued

Fig.2. Measuring instrumentation

set-up
33

Measurement Procedure
The instrumentation set-up used is shown in Fig.2. As a test signal an
impulse was transmitted via the loudspeaker, and analysed using a
Narrow Band Spectrum Analyzer Type 2031. The frequency response
obtained is shown in Fig.3a. The rather heavy ripples in the frequency
range above 6 kHz, as seen in the figure, were found to persist when the
analysis was repeated. It was found to be caused by sound reflections
from the wire mesh constituting the floor. Covering the wire mesh with
rockwool and repeating the measurements gave a response curve as
illustrated in Fig.3b. The ripples can be seen to be significantly reduced.
This spectrum was stored in the memory of the analyzer as a reference.
As there is not enough energy in the impulse below 500 Hz, measurement
results in this article are not valid below this frequency.
w

A microphone clip UA0588 with its swing arm in the horizontal position
was now mounted on the microphone. To examine the influence of
sound reflections from the clip, the impulse signal was again analyzed
and the reference spectrum was subtracted from it. The difference in dB
is shown in Fig.4a. A similar curve was also obtained with the swing arm
in the vertical position, and is illustrated in Fig.4b. As expected, the
disturbance caused by the swing arm in the vertical position is higher

Fig.3. Frequency response of impulse


a. with bare wire mesh
b. with wire mesh covered with rockwool
34

Fig.4. Errors caused by microphone clip with swing arm


a. in horizontal position
b. in vertical position
c. "Control Meas"
and of the order of 0,5 dB. To ensure that this was caused exclusively
by the microphone clip, the experiment was repeated without the clip,
and the spectrum subtracted from the reference spectrum. The result is
shown in Fig.4c, and the difference is seen to be less than 0,2dB. In
the following, such a control measurement was taken after each of the
interfering objects was removed, and the curves are indicated by "Con
trol Meas" on the figures.
To investigate the effects of connecting the extension rods of the tripod
UA0587 to the microphone clip, similar measurements were carried out.
Figs.5a, b, c, and d, show the interference caused by the microphone
clip and when the extension rods are connected at 0, 30, 60 and 90
from the horizontal. Again the largest interference is caused for the 90
case, however, the error is not greater than 0,5dB. In Fig.5d the peaks
and valleys are shifted in frequency relative to those in Fig.4b, however,
the magnitude of the error is still of the order of 0,5 dB. Fig.5e
illustrates the "Control Meas".
Oftentimes in literature, the microphone is seen to be mounted directly
on top of the ball-joint of the microphone tripod. The errors caused by
35

Fig.5. Errors caused by microphone


connected at
a. 0 from the horizontal
b. 30 from the horizontal
c. 60 from the horizontal
d. 90 from the horizontal
e. "Control Meas."

clip and two-piece

extension

rod

this form of mounting are illustrated in Fig.6a, whereas Fig.6b shows


results for the same conditions but with the tripod as well. The difference
between Fig.6a and Fig.6b is not significant, indicating the influence of
the microphone tripod to be negligible. However, the error due to the
ball-joint is increased to 1,3dB.
As one would expect, this interference is removed when the microphone
is mounted on the two-piece extension rod fixed to the tripod, as can be
seen in Fig.6c. The errors caused here are less than 0,5dB, similar to
those of Figs.5a, b, and c, as long as the angles between the horizontal
microphone and the extension rod is 60 or less.
To further improve the results of Fig.5a, (i.e. to minimize the reflections
from the microphone clip being mounted so close to the microphone),
the flexible extension rod UA0196 was made use of. The results are
shown in Figs.7a, b, and c for 0, 60, and 90 between the microphone
36

Fig.6. Errors caused by microphone clip when


a. mounted directly on ball-joint
b. mounted directly on ball-joint together with the tripod
c. mounted to tripod with the two-piece extension rod
d. "Control Meas"

and the two-piece extension rod. It can be seen that the errors are
reduced to 0,2dB for angles less than 60 and to 0,5dB for 90.
The use of the flexible extension rod is probably even more imperative in
conjunction with the pressure microphone, when it is used in a free field,
and has therefore to be mounted at 90 incidence. Fig.8b shows the
reduction in error achievable with the flexible extension rod compared
with Fig.8a where the microphone clip is mounted close to the
microphone.
Finally, Fig.9 shows the results obtained with an 1 " Microphone Type
4145 with the microphone clip UA0802 and the two-piece extension rod
mounted at 90 from the horizontal microphone. Compared to the
results of Fig.5d for the 1/2" microphone, the error is significantly lower,
approximately 0,3dB. This is because the 1 " microphone is consider
ably less sensitive than 1/2" microphones to reflections coming from
behind the microphone as can be seen from Figs.10a and b.

37

Fig.7. Errors caused using a flexible extension rod and the two-piece
extension rod at
a. 0 from the horizontal
b. 60 from the horizontal
c. 90 from the horizontal
d. "Control Meas"

Fig.8. Errors caused for 90 sound incidence by the microphone


when it is used
a. without the flexible extension rod
b. with the flexible extension rod
38

cup

Fig.9.a Errors caused by microphone clip for a V' microphone with the
two-piece extension rod at 90 from the horizontal
b. "Control Meas"
Conclusions
From the results it is obvious that mounting of the microphone directly
on the tripod should be avoided. To keep errors within 0,5dB, the
two-piece extension rod should be made use of, and should be mounted
preferably less than 60 from the horizontal. For the same configuration,
the error can be reduced further down to 0,2 dB by inserting the
flexible extension rod between the microphone and the preamplifiers on
which the microphone clip is mounted. One inch microphones are
considerably less sensitive to clips and tripods than 1/2" microphones.
It should be noted that results obtained in this article using narrow band
analysis can be considered to be the worst cases, such as obtained
when dealing with pure tones or very narrow bands of noise. In practice,
1
where broad band noise is emitted and measurements carried out in /3
octaves, which is very often the case, considerably lower errors will
occur on account of averaging in the relatively broader bandwidths.

Acknowledgements
The author wishes to thank Peter Moller for his assistance with the
measurements.

39

References
[1] ZOLLNER, M.

"EinfluB von Stativen und Halterungen


auf den Mikrofonfrequenzgang, Acous
tics, Vol.51 (1982), pp.268-272

Fig.10. Directional Characteristics


a. 1" microphones
b. 1/2" microphones
40

for

News from the factory


Portable Machine Vibration Analyzer Type 2515

Bruel&Kjaer's Type 2515 is a portable battery-powered FFT analyzer


designed for the requirements of everyday machine monitoring. The
solidly built single-channel analyzer has waterproof and dust-proof char
acteristics better than IP44 in accordance with IEC529. A large non
volatile memory holds up to 100 constant percentage bandwidth spectra
or 50 constant bandwidth spectra, cepstra or time records, along with
the pushbutton settings used in the recording. When a recording is
recalled and displayed, the original pushbutton settings are also
reactivated.
With its clearly laid-out front panel, the Type 2515 is easy to operate and
makes day-to-day monitoring a straightforward matter. Newly measured
spectra are easily compared with reference spectra. The built-in charge
preamplifier allows accelerometers to be connected directly, and the
2515 also incorporates Bruel&Kjaer's unique speed compensation
technique.
The Analyzer incorporates a wide range of facilities for troubleshooting
vibration problems, including cepstrum, non-destructive zoom, harmonic
cursor with fine tuning, reference cursor, phase read out, computation of
overall vibration levels, advanced triggering facilities, external sampling,
scan analysis, exponential averaging, time averaging, expanded time

41

function, and a choice of weightings for continuous or transient signals.


The fully annotated display axes can be set to log. or lin. scales, and the
operator can select metric, British/American or engineering units, or
dB's.
The IEEE-488 interface allows the analyzer to be connected to further
post-processing instruments for trend analysis, data management and
storage/retrieval of reference spectra. Hard copy can also be obtained
using either the analogue or video outputs. The Type 2515 can be
supplied in a reinforced leather case with shoulder strap and comes
complete with a mains supply adaptor/battery charger.

Field Balancing Set Type 3537

The portable Field Balancing Set, Type 3537 a development of the earlier
Type 9537, is a handy, battery operated system which combines all the
measuring instruments needed for both single- and two-plane balancing
of rigid rotors without dismounting them from their own bearings.
Measurement of the unbalance vibration is made via two Type 4370
Delta Shear piezoelectric Accelerometers. Signal conditioning is car
ried out in a Charge Preamplifier, Type 2635 and the vibration level
displayed on an Indicator Unit, Type 2433 which has a thermometer-type
logarithmic display. The unbalance phase is displayed with a resolution
of 1 on a Phase Indicator, Type 2976 with liquid crystal display. The
phase reference is provided by an infra-red tachometer probe, which
can operate at up to 800 mm from the rotor. Signal filtering is provided
by a Tracking Filter, Type 1626 which tracks the rotation frequency
42

without prior tuning. Furthermore, Type 1626 has a sweep mode which
enables frequency analyses of machine vibration to be performed.
The 3537 comes combined in a light-weight carrying-case together with
battery chargers. Weighing only 10 kg, the Type 3537 is truly portable
and ideal for field balancing applications.

Indicator Unit Type 2433

A new battery-operated AC voltmeter, the Indicator Unit Type 2433 is a


development of an earlier model, the Type 5743, which has been avail
able for some time on special order.
The Type 2433 uses a light emitting diode (LED) thermometer-type
display composed of 41 LEDs enabling a 100 mm high-resolution scale
to be accommodated in an instrument only 34,5 mm wide. Overrange,
underrange, or a signal with too high a crest factor are shown by a
flashing indicator. The Indicator Unit has three full-scale outputs of 1; 0,3
and 0,1 V, matching the outputs of a number of B & K instruments, and
uses a logarithmic RMS detector that can accommodate the high crest
factors often found in mechanical vibration signals. Two selectable timeconstants, of 1 s and 10 s, enable measurement to be made of either
deterministic or random signals with frequency components as low as
1 Hz.
The Type 2433 occupies 1/12 of the B & K standard rack width and can
be incorporated with other B & K instruments in a variety of portable
measuring systems built into a carrying-case. If the Indicator Unit is
combined with a charge amplifier and any suitable piezoelectric accelerometer a versatile vibration meter results. Adding a tracking filter and
phase indicator produces a portable balancing set.

43

Tracking Filter Type 1626

The battery-operated Tracking Filter Type 1626 (a development of the


earlier Type 5856 which has been available on special order for some
years) is specifically designed for field balancing of rigid rotors and
vibration analysis when used with other suitable instruments from B&K.
The Type 1626 contains a highly selective band-pass filter which auto
matically tracks the rotation frequency of the rotor to be balanced,
preventing unwanted vibrations from interfering with the measurements.
The Tracking Filter has three fixed bandwidths of 0,1; 1 and 10 Hz,
automatically selected as a function of the tracking frequency. The
automatic bandwidth selection Can be partly or fully disabled, giving
down to 0,1 Hz bandwidth over the full frequency range. Display of the
rotation frequency is given in either Hz or RPM on a large liquid-crystal
display. For frequency analysis of machine vibration the Type 1626 has a
sweep mode which can sweep the filters through a frequency range of
2 Hz to 2 kHz.
The other instruments normally required for field balancing together with
the Type 1626, namely two accelerometers, a preamplifier, a voltmeter, a
phase indicator and a tachometer probe, are all available from B & K and
together can be installed in a carrying case to form a versatile portable
balancing set and vibration analyzer. For frequency analysis the Type
1626 can tune a level recorder to synchronize the motion of the preprint
ed paper to give a hard-copy of the frequency spectrum.

44

PREVIOUSLY ISSUED NUMBERS OF


BRUEL & KJ/ER TECHNICAL REVIEW
(Continued from cover page 2)
4-1979 Prepolarized
Condenser
Microphones
for
Measurement
Purposes.
Impulse Analysis using a Real-Time Digital Filter Analyzer.
3-1979 The Rationale of Dynamic Balancing by Vibration Measurements.
Interfacing Level Recorder Type 2306 to a Digital Computer.
2-1979 Acoustic Emission.
1-1979 The Discrete Fourier Transform and FFT Analyzers.
4-1978 Reverberation Process at Low Frequencies.
3-1978 The Enigma of Sound Power Measurements at Low Frequencies.
2-1978 The Application of the Narrow Band Spectrum Analyzer Type
2031 to the Analysis of Transient and Cyclic Phenomena.
Measurement of Effective Bandwidth of Filters.
1-1978 Digital Filters and FFT Technique in Real-time Analysis.
4-1977 General Accuracy of Sound Level Meter Measurements.
Low Impedance Microphone Calibrator and its Advantages.
3-1977 Condenser Microphones used as Sound Sources.
2-1977 Automated Measurements of Reverberation Time using the Digi
tal Frequency Analyzer Type 2131.
Measurement of Elastic Modulus and Loss Factor of PVC at High
Frequencies.
SPECIAL TECHNICAL LITERATURE
As shown on the back cover page, Bruel & Kjaer publish a variety of
technical literature which can be obtained from your local B & K
representative.
The following literature is presently available:
Mechanical Vibration and Shock Measurements
(English), 2nd edition
Acoustic Noise Measurements (English), 3rd edition
Architectural Acoustics (English)
Strain Measurements (English, German)
Frequency Analysis (English)
Electroacoustic Measurements (English, German, French, Spanish)
Catalogs (several languages)
Product Data Sheets (English, German, French, Russian)
Furthermore, back copies of the Technical Review can be supplied as
shown in the list above. Older issues may be obtained provided they are
still in stock.

Printed in Denmark by Naerum Offset

BV 0022-11

You might also like