Electro Acoustic free field measurements

in ordinary rooms

using gating techniques

 Electroacoustic free-field measurements in ordinary rooms
— using gating techniques

by Henning Moller and Carsten Thomsen, Brliel & Kjaer

Introduction
By the use of gating techniques, sponse and directional characteris­ System which provides the neces­
many types of acoustic measure­ tics. However, the technique is also sary tone burst test signal and mea­
ment normally confined to ane- applicable to other areas such as sures the peak amplitude of the re­
choic rooms, can now be made in measurement of distortion, early re­ ceived signal, which is gated by an
ordinary reflective rooms. Of pri­ flections, absorption, and phase re­ adjustable time w i n d o w to eliminate
mary importance are the measure­ sponse. These measurements are the influence of reflections and loud­
ments of loudspeaker frequency re­ made possible by means of a Gating speaker transient distortion.

Basic Principles
The Gating Concept
When a tone burst is applied to a
loudspeaker, the acoustic waveform
received at the position of the mea­
suring microphone often bears little
resemblence to the original electri­
cal signal (Fig.1). This is due to a
number of phenomena: (1) over­
shoot of the loudspeaker on the in­
itial part of the burst (2) internal
reflections in the loudspeaker cab­ A Emitted
Tone Burst
inet which most often distort the
first part of the burst (3) overhang
of the loudspeaker at the end of the
burst (4) reflections from the walls, Received
B
Signal
floor, and ceiling of the room,
w h i c h usually arrive after the end
of the burst. All of this information
Measuring
is undesirable w h e n wanting to de­
Gate
termine the free-field sine wave re­
sponse of the loudspeaker, although
it may be useful in describing other
aspects of the loudspeaker's perfor­ Measured
D Signal
s mance. However, somewhere in the
received tone burst, a steady-state
sine wave can be found whose am­
750417
plitude equals the free-field re­
sponse of the loudspeaker. This Fig.1. Principle of the Gating System

1
steady-state information can be
measured by the use of a time w i n ­
dow (gate) which is adjusted to se­
lect only the desired section of the
burst (Fig.2).

•■•■ T- r—
•v-w-l ---^ • .

:- •-+
4133/2619
Power Measuring Amplifier
Gating 2606
Amplifier
System
2706
4440

Level Recorder 2307

L 750735/1

Fig.2. The gate is adjusted to include only Fig.3. Instrument set-up using the Gating System
the steady state portion of the signal

fc_ r

,H
B*xf<?$ 5. K\K-i

_ ^ - _ _ -■.■-■-■-■-■_■ _ _■_■
M ^ - s ^ r i i o G*Jte
F
Ui*f- Width Dfisy ". > I L "

'"^

0» ^s§F^>
S**-Jr> * \ : ^
h ^
x: : i '<i
>{.<<<■ ■-■-'-•1
■. ■ " ■ -
y^

t -H,
<! ' *
-, ! • ~=oo
.; o.-?
^\\ ss^.

I'KjsjerSflfefio

Fig.4. Front and rear view of the Gating System

The measurement of the loud­ Instrumentation ting and receiving section. The sine
speaker's amplitude response using The system for gated measure­ wave from the Sine Generator
this method will be accurate pro­ ments builds on the same basic in­ 1023 is fed into the transmitting
vided that the loudspeaker has fully struments as used for traditional fre­ section where it is converted to a
responded to the input signal as quency response measurements: a tone burst. The duration of the
can be seen on the oscilloscope. If sine generator, measuring micro­ burst is adjustable from 0,1 ms to
the signal has settled to a steady phone, measuring amplifier, and 1 s and its repetition rate may be
value, the use of a longer tone level recorder. To permit gated varied from 0,5 Hz to 1 5 Hz, or the
burst will give no more information. measurements, the Gating System burst may be triggered externally.
Hence this value will equal the and a two channel oscilloscope The tone burst is generated so that
value obtained with a continuous must be added to this basic package it always begins and ends on a posi­
sine wave, which is simply an ex­ (Fig.3). tive-going zero crossing (Fig.5). This
tremely long tone burst. minimizes the generation of other
The Gating System Type 4 4 4 0 frequency components.
(Fig.4) consists of both a transmit­

Fig.5. The tone burst always begins and ends with a positive-going zero crossing

2
 In the receiving section, a measur­ sound in air and to permit selection value until the next tone burst. The
ing gate is adjusted to select only of the desired section of the burst. hold circuit is than reset and holds
the desired portion of the received the new peak value of the next
signal. The w i d t h of this measuring The w i d t h and delay of the mea­ burst. This value is fed as a DC vol­
gate is adjustable over a wide suring gate are adjusted w i t h the tage to a Level Recorder, w h i c h re­
range, from 30/^s to 1 s. In addi­ aid of a t w o channel oscilloscope. A cords the automatically swept fre­
t i o n , this " t i m e w i n d o w " may be positive peak detector gives the max­ quency response. The peak detector
delayed from 0,1 ms to 1 s to com­ imum peak value of the received sig­ operates linearly over a dynamic
pensate for the propagation delay of nal w i t h i n the gate and holds that range of at least 5 0 d B .

Selection of Microphone Distance and Pulse Length

W h i l e the response time of the
loudspeaker places a restriction on
the m i n i m u m pulse length, the max­
x
i m u m pulse length w i l l also be lim­ ' i

ited to the time it takes for the / h \

\

first reflection to return to the mea­ /

\

suring microphone. This, of course, /

/
\

Loudspeaker > ^ Microphone

depends on the room dimensions — -^j D

and the positioning of loudspeaker /

and microphone. It is interesting to /

note that the lower frequency limits /

/
obtained are approximately the /
\

same as for anechoic rooms of the \

/

same dimensions. In practice, how­

ever, the gating system may be
used to achieve low frequency per­ 750415

formance as good as or better than Fig.6. The microphone/loudspeaker combination should be centered in the room
any anechoic room, provided that a
large enough room (such as a w a r e ­
house) is available. Optimum Placement This is valid w h e n the length of
In practice, the microphone and the axis of the room along w h i c h
First of all, the distance (d) be­ loudspeaker should be placed along the microphone and loudspeaker
t w e e n loudspeaker and microphone the long axis of the room and are placed is at least 1 5% greater
should be selected. Ideally, the mi­ should be centered between the than h. If this is not the case, see
crophone must be placed in the far walls, and the ceiling and floor. Appendix.
field of the loudspeaker, that is, at (Fig.6). W i t h this set-up, the opti­
least one wavelength from the m u m distance can be calculated The corresponding minimum fre­
speaker at the lowest frequency (See Appendix) w h i c h fulfils the one quency is
(* m i n . ) wavelength criterion and permits
the maximum tone burst length be­ f mm. = 595 (3)
fore the first reflection. If the h
room's smallest dimension is h, and the maximum pulse length (t) is
then the optimum distance (d) be­ 1/fmin. w h i c h at the m i n i m u m fre­
d= M i (D tween loudspeaker and microphone quency includes one period of that
f mm.
is: frequency.

d = 0,58 h (2) Table 1 shows some practical val­

ues for m i n i m u m room dimensions
of 2 to 5 meters.
The size of the loudspeaker must
also be considered in determining
the far field. The microphone h> 2m 3m 4m 5m
should be placed at a distance at
least equal to the largest dimension d 1,15m 1,73m 2,31 m 2,88m
of the loudspeaker. Unfortunately,
due to practical restrictions on room 1 298 Hz 199 Hz 149 Hz 119 Hz
mm.
size, these criteria are often ig­
nored, thus leading to nonreproduci- t 3,3 ms 5,0 ms 6,7 ms 8,4 ms
ble measurements. Certain stand­
ards, of course, also call for fixed Table 1 . Optimum microphone/loudspeaker distances (d), lower limiting frequencies (fmin.) anc
*
distances. pulse lengths (t) for given room heights (h)

3
Pulse Length related to a where M is the shortest reflection self. In this case, the shortest reflec­
fixed distance path, which for a rectangular room tion path from the loudspeaker to
A fixed, possibly arbitrary dis­ is the smaller of L (length) or the reflection object and back to the
2 2
tance between transducers, such as Vh + d , and c is the speed of microphone can simply be mea­
1 m, is often chosen even though it sound. Of course, the minimum fre­ sured and substituted as M in Equa­
ignores near-field considerations at quency again is 1 / t m a x tion (4).
ow frequencies.
Equation (4) may also be used if Complete derivations of these con­
In this case the maximum tone there is some reflecting object pres­ siderations are given in the Appen­
burst length is ent other than the room's bounda­ dix.
ries, such as a cabinet, or perhaps
t _ M — d (4) the measuring instrumentation it­
•■max. -

A Practical Measurement Procedure

A suggested measuring procedure tone burst. If the tone burst is tor of the Gating System Type
is given as follows: too long, the received signal will 4 4 4 0 only responds to positive
appear as in Fig.7, where the di­ peaks. In addition, measuring on
1. Place the loudspeaker and mea­ rect and reflected signals over­ the second half of the single pe­
suring microphone along the lon­ lap. This should correspond w i t h riod will give a more accurate re­
gest axis of the room. Center the the calculated values from sec­ sult since the system has had
microphone/loudspeaker combi­ tion "Selection of Microphone more time to respond and hence
nation w i t h respect to all three Distance and Pulse Length". the waveform is less distorted.
axes of the room. Choose a suit­
able distance between the trans­ Reflection Now sweep the
tne generator
ducers as indicated in Section through the higher frequencies
"Optimum Placement". and if necessary trim the adjust­
ment of the measuring gate so
2. Connect the instrumentation as that it only measures the steady-
indicated in the 4 4 4 0 Gating Sys­ state portion near the end of the
tem Instruction Manual. Note tone burst.
that there are two GATE OUT­ Direct Sound
750734
PUTS on the rear panel of the 7. The rate at which the sweep can
4 4 4 0 , one for the transmitting Fig.7. Too long a burst results in overlap of be made depends on the tone
and one for the measuring sec­ the direct and reflected sound
burst Repetition Rate, and the de­
tion. The transmitting section sired resolution at the maximum
GATE OUTPUT must be con­ 5. Use a relatively low Repetition frequency of interest. It is import­
nected to the Ext. Trigger input Rate and note the distance (in ant to note that the frequency re­
of the oscilloscope. The measur­ time), along the X-axis of the os­ sponse measurements obtained
ing section GATE OUTPUT con­ cilloscope, that it takes the ampli­ using the Gating System are
nects to the Channel 2 input of tude of the reflections to drop at made at discrete frequencies
the scope. least a factor of 10. The Repeti­ since the gate is only open a
tion Rate must be adjusted so short time during each repetition
3. The received tone burst can be that the reflections have decayed period. Hence details of the
displayed on Channel 1 of the os­ at least 2 0 dB before a new curve may be lost if too high a
cilloscope. This signal is avail­ burst is transmitted. The more re­ sweep rate is used. Sweep rates
able at the output of the Measur­ verberant the room, the slower given by Level Recorder Paper
ing Amplifier. (The signal can the required Repetition Rate. Speeds of 1 or 3 m m / s are typi­
also be taken from the AC OUT­ cal. The exact relationship that
PUT of the Gating System but 6. Decrease the frequency of the os­ gives the Paper Speed (P) is:
must then be inverted 1 8 0 ° to cillator until only one period is ob­
display it properly on the oscillo­ served on the oscilloscope. At
scope.) this frequency, adjust the width 22 B R (5)
P*S
and delay of the measuring gate f max.

4. Begin w i t h a relatively short tone to measure the peak value of the

burst (about 3 ms at about 1 kHz) second half of the period. If this where B is the frequency resolu­
and observe the received wave­ peak is negative-going, the loud­ tion (in Hz) at the maximum fre­
form on the oscilloscope. Note speaker leads must be reversed quency f max. and R is the Repeti­
the point of the first reflections to change the polarity. This is ne­ tion Rate. This formula is derived
and increase the duration of the cessary because the peak detec­ in the Appendix.

4
Practical Examples
On Axis Response details of the t w o curves differ. The crophone/loudspeaker position and
Ideally, a measurement in an ordi­ difference below 2 0 0 Hz is due to distance in the t w o different rooms.
nary room using the Gating System the low frequency limits already dis­ Experience shows that even a smal
should yield the same results as a cussed. angular displacement of the micro­
measurement in an anechoic cham­ phone in an anechoic room will
ber. A practical comparison is The difference in detail between yield different results.
shown in Figs.8 and 9. It will be the t w o curves is due to the diffi­
noted that above 2 0 0 Hz only the culty in exactly duplicating the mi­

■*-*.
Bruel & K)3?r Bruel & K|3?r Bruel ft Kjser ™ Bruel & Kjwr Srue! & Kjgjr Bruel & Kja&r
ZJ D D n ~ ~ r ~ i " i - ■: 11 n n L ^ n n n ^ - i n n n n n r i L i - i i j o c c D T r i Z Z D U u z i D n c D zi D n ~ D c C c n a D zi •z r n u u a a c z : a c a n c c z ) a c z : n z j G n c z i D Z ] r : G C a D i J D C z : D n D D c n c n c t
B r u e l & K j a e r Potentiometer Range 50 dB Rectifier R M S L o w e r L i m Freq 20 H : Wr Speed 80 m m sec Paper S p e e d 3 m r n - sec B r u e l & K j c e r P o t e n t i o m e t e r R a n g e ^ 0 d B R e c t i f i e r D C . . L o w e r L i m F r e q ^ 0 0 Hz W r S p e e d . ? P ° m m s e c P ap*?r S p e e d J - m m ^ s e c .
:i: I !
Copenhagen 50r25r k
t
l~f f---f- - -^ - T
^z^:jr_^rz^0
< : ■ J
7h C o p e n h a g e n 5 0
^
-I - T - T T T1 .... r T T
^r^ w u w ■ m * '■ ' _I m * 1 V ■ _p m ' 1 m r
m ri _E
i - L — - T 1
1
^ - ■ J "T.

14-
3 1 _]_. - 1■ " " 4 - .
—fa * - ■ fa fa M
_i
— ■ - _ F * - ^ t1
I " ""
' 1
'
1
1 | _ -
1

dBdB T^TZdBdB dBdB' 1 J L ■ n

- .. 1. . ' TZTI JdB_ d B
^Frequency Response of loudspeaker"" d . J

- i — 1 » F r p f i u p n r v Rpcrvm^P n f l o u H s r w ^ k p r
1 F ™ a
■

. _ _ _i
■

_ _ .
ri ■

1
J
1 r 1 1-
-ILHttpk-
^measured with pure sine in anechoic cham ber ■ ■*-■ ■ ~ r rts.*,f, ,r^A ../.*¥* n.*.*\~w Ctr-teif* A A AC\ - - _ _ .■ _ _ —
- ri
1 , 1
i !
I J
.
,
' r -
— I H C 0 5 U F CVJ W l L M V J O U I W O V J L d l l "TfTJ T ^ ' ■ " -
4020- - — - 8 '60 40 2 0 1
Y 1
60
r ;in a nrHinarv/
' 1
r n n n i 1 . "-
I— •-
Measuring Ob| - 1 - . T
M e a s u r i n g O b | ,. '_ ^ |— . ■ ■ — * —1 — — 1 —1

— 1 — *""!-- _ . ._ . ._L._t_ ' ' '" "

—. 1 . 1
■ J ' J
1 1 ■ -
- 1 11 11 1 1 1i _
! 1
- - .■
■ 1
1
■
1
1 r—
■1 _i •
L \ •
1 — ' — ■
1

1 1
J
! ' 1 1 ^
-30'15 - 3 0 15 1 1 1

_ _ _ . L ^ ^ - ^ ^ T 1-
15
■
———
"T 1 1 " __ . L. fi

1
,■ Li l KJ■ J L
■ 1 1 ■ ■ r -
1 r * ' ■ T — 1
T 1 1
___ _ T *" r
— 1
—. — I ' ■ r
1 1
.■ ■
1
L " 1 ." "
1 — 1

1 —
1 —

—
1 1 1 i_
1

■ -— - ■ — . _1
1 \jJJFm
- ^ ^ — ^ m - —

_. -■ 1 . 1 1
I
■ """" _JLJ^H L 1 — Y
■ 1 1 -

20i10 ™
10 —
L b_. ..
r
1 J-L r
-
.

1i " i T4-
1

_ _ _
L

1
- 1 - — -
.■
fa. _ L
L
_ _ —
— ' — T l ' r^ 30
■ m " ■ ■ " ""

■ - 1-f- . 1 -
- I I U" L _ | 1
■ ■ ■ 1■ r h
: ^ - : ~ - -t *- T 1 1 "
i — - - ■
"A~r . 1 ■ — 1 — ■ h ___ _ - -r
4 -
- --ri--
f■ ■
1
' 1
J

j : 1 : ;
r - -—j 1
fa
F
j 4 - r ■ r
{ n x - x —

-10-b A L "
10 -6
J 1 h ,
! [
-
■
1 1 1 T r-2 15

XT'-
■ -
" 1 r - - + , " ■ | 1 .
1
_ V — - - L
h - -3-- ' T
J ! . . . . . - 1

■ - . r 1
l__ J L 4- 4 -m- -
T T ^ _
Rec N o - - t__ __
Rec Ng ■ ^ ■ ^ — t : . ' '
— -- - , ,--—-- i -
D a t e 17-4-7b Date t a ' 4 75 r — ■ ri ■
1
. - H 1 1 J ■ —

| 1
Sign H V Q>_Q
0
Sian H.M. 3
rt _ T_H
\J
n L 1 1 J 3 I 1 ■
1
'

■
1

1
1

1 SK

10 20 ffr 50 100 500 1000 2000 5000 100O0 20O0OU0OO0D A B C lin ^ 20 Hz 50 100 200 500 10OO 2O0O 5000 10000 20000 4O0O0O A B C l.nLru
10
75CWCr9
^_ f « ^ l _ - l_ "?„_.-. I -. ^E
QP 1 1 2 4 M u l t i p l y F r e q u e n c y S c a l e bV Zero Level 5 0 dB 1612/2112 A B CLm i«-k. #-s ^ ^ J>I ■ iL J . . t + . « . | . y
PVIUli
t * ■ - . « . . « « ^ i i O . ^ ^ ^ ^ A t*
c
j "
r PCLJIj e n v y O L d l t 1U 1 rit il 1 1 Z
1 1
C
QP 11 2 4 iply uy eru L
e vet "- ^ 751)408

Fig.8. Frequency response of loudspeaker measured w i t h pure sine in Fig.9. Frequency response of t h e same loudspeaker measured w i t h t h e
anechoic chamber Gating S y s t e m

To verify this, measurements the microphone position was identi­ these curves it can be seen that
were made in an anechoic room cal in both measurements and the this error is negligible above the
both w i t h (Fig. 10) and without only source of error will be the influ­ lower limiting frequency of about
(Fig. 1 1) the Gating System. Thus ence of the Gating System. From 3 0 0 Hz.

m
™ B r u e l & K 3 a?r Bruel & Kjicr Brufl & Kj.xr Bruel & K j ^ r Bruel & Kjav Bruel & K|ser
D n " n n i _ . L _ ._. ~ z: ZJ :z c :.J I J LJ ! J C ^ D LI [J u U LJ H 3 r— 1—1 —\ f
] I C G C a C C C D a D D : C U ' J L j D C D Z I D D D n C E
U
BruelSiKjaBr Potentiomoler FrO ' RMS .Lo w e r
B r u e l r i K j c e r' Potentiometer Range — dB Rectifier ^ -Lower Lun Freq - Hz W r Speed m m 'air. Pspe r Speed - m m st^r Ranqe - - d B Rectifiei Lim Freq .- ^ Hz W r Speed _I2V.. m m -'sec P a p ^ r S p e e d ^ mm-'sec.

^ "25
r- ¥ T
V W
& l r w ■ ■ * ■ -■ W J ■ P .■ _» _ • ■ 31 1 V T T V lTLn/D
Copenhagen ■ f ■r ■ T L"
f i - H 4"^. L~L F~H P"I -^i K i j"n r i l —
^-^ ' ^ "'-' L

_... . " _ ^_ _ _ __ i^opcnrtayeri r r

.l.ii-: z!
H _ ^ _ ■*■ t - 4- - L !
. ___ _14__. -I- - -- ' ■

, -J + 1 -1 ■+ -
■
■ !
4
i
i ■4- r

r
1

r r
r
1 —
__ L' J '" 1
4
—
^
— ~^--
- -
-
■ "' : 1' 1 "J
r , i i_ 1 ■ - 1
, * 1 , rC^
dB _rJB ^
— p - ■ ~ " '
1
1
""t^
. . . j n
d B rlR
_i F-fa
dBrdBl 1

— — \ u .
■ T
i_
L y _ _
: —7*v - — , 5~~H ' 1 1 jGDidti
V +
1

1
_. _ -M _ ± 1 4- . ■
7I T ■ h ,_ ri -- 1r ■ 1
- m~ - m , - . / ' 1
1 1 1
T 1 ■ - 4 _j 1 r L

_.:
- 4 ■ r
■
x - —
1 V I 1 ' ' 1 —
■ ~^~ ■ r ■ ■
T H
I ~ -J 1 ! ' ■ 1 _
" j ^ -p
1 1 L 1

40
- - - _ j

20
H
20 . i ' |_jf
i . . , _B-60 40 ! ' 1 a - 60
41
L _L ±
- - ■
- L ■ T n
L L J ^ L
r ■ !
' \ 1
j 1■ i - - \— r^ - - - "T"*" " "^ L -L—1— f
_. ': -_J - - f — I L .-.: T
i ~I" ^ - 1 1 1 ■ \—

"~T"
_ —-pF" -J —L ■ ■ .—I. ■
L J j
I ! I .
Measuring Obj Measuring Ohj _ _■ 7 1 j ' r
1 1 -
. - - __ _ 1
. - .._._ .. _,_ a J - 1 - « - 1 ' M— H

y ~~" - ■X " - 4 - ■ -
. . . +
1 1
^ - - - r - - — - x - r - . - - - _ _
rj " ■ ■ "

A~\-
■

f 1
J 1 t|\ ■ " "T 1 ! '
1
^— . — 1
.
1

^-_ T._ _ I l T 1
- - 1 — r
—
_ k ^ _ ^ ^ H
. J / L^~^ iff ; 1 . _ . 1 . _ ^ -
_ __ h- — ■ J '
" T — —1 W -U- x -fa 14U J L L . J 1

.
-""-£=. I
1
1
1
! Ai
**
* ^ » 1
P " 1 1 U
r ■ j . - i - , r ■
—
^^^ 1
L— 1 L_
. , |¥
■""fa J"X J ^~
t^ r 1 "^ "^ "i "r" m -4"

L
m
m

' .
^

r h J ri-
■
1 * 1
"^ ™ X I L^ ■J ■ t f
* 1 u1 -
- 30 1 5r
p
r '
" 1 . 1 ■—: 1 ■ : — X -1 i 30'15 ^ - —— —
p
| i ■ ! r 1 r fi- 45
__ r - x
1
I * r - ■ i 1 # ■
1 '"- ^ r" 1
h
\ f ^ ^ F
■ ■
■ I 1
1
r -
L

■ / ■ ■ * - - - - 1 1 1
- r 1 ; ■ i- —
1 — ■ — ; — ■ — r
-Xf - —u 1 - 1 . :.
- 4--i
1 M M x ^ - — h a
T r L ,

1 . 1
■
| |
"T "t . 1 T J
_ 1^ 1 ' i r~ "■
■ 1 1 ■

f - " —" " —' - + ri 4 V r T ,

\ 1 1
- B— j 1 I i ^ ■
1
t - ^ ^ ^ - 1 - - F ^ _ L l_
-■jP v ■. - ■1 L . ■ 1 -, 4-
- - - -— l _ ^r. : 1 : i _ _ 1 l_ _
JL 1 | _ 1 L — -r -¥ -r -
1 -L-4 _ _ h - - -
J r■
X 1 , L ■ - ■
r - n -
. - ...i : x ■ \ 1 _,
. __ J _ __ ^ ■ - , ^ 1
r
1
- 1 ■ r h n ■ h - 1 , 1 1 fa . . _L _ . fa ■

' —h r-
I
r\ j ~~ 1
i 1
10
-
^ n T 1

- ^\J -10
1 1
1 J . ^ ■-^^» JT " ^ I fa' —L _ ■ J ^ 1 _ -L . J—■ J-1.
^ ^ ^ ™
zU l ■ ■
/ 1 ^ ■1 - * ■ L L . : 1 . . , 4 - J O 1 ■ f — ^ ■ ! " "
I , . ~ ■ 1 ■ x - J 1 P ' . . 4 -30
J F ^ J ■ - t fa ■ _ 1 1 r^
TT i :. i . : . ; L -
■ T ■
/ - 1 - - 1

FT" . I TT " T J 1 r~ —
I
X
ri
1
■ _j
r
r
■
r _ J
L B

- U -4-
—
. _ . , — i_ 1 •f ^ _

1
1
1

1
■ |

J
r -
1

l1 _— ,
■
i
i
,
1 - x
L
-
- — "
. ^ i :_. X

_ _ l_
1 '
J 1
'
_ _ j 1
--^-~r- ■+ * t
■
1
. \ , . . _^ — - 4- ^__ _
r-t-L
■ L J ■ , ■ ■ , -1 .■
j -
■ + r
" " T
X _■ _ J - ^
1 ■
■ 1 ■

_ ~ 1_ * _ . -- t ' J — 1 1 1 ~'
_ ^ L _
t- - J ■ - ■ - 1 — J — * — « - — * - _
,_ _ ___ ' . . ___
t T-
. 1 1 ■

H J
10 5 ■ 1 ■
J J r ■ . . _L ___ J - J 4 1—
■ t -+
r

j ■ r- 2 ■ 1 5 — — 10f5 . — 1 — ■ — 1 — 1 - 2 -
1 ■■ ■■

-A 1 J . . + v n T- L — -- — *— - 1
1 5
___ _ L __ _ 1 1 1
1 1 1 1
- -f
j
— 1 — 1 — 1 — 1 —
l
— " - 1 " ' ■ L 4 1 -—. ■ — L .■ ■
- - ' ' j i i

u _ 1 — 1 — 1 — 1 — 1 —
| . ._
1 r J _ L . 1 ■ i k— "

. :_:
1 1 j 1 i _ _

Rec No ;_ -r
1
1
i - ■ ^
+ 1- - - . -
1
4 .
h —
-
r
.
|
_ ,
Rec No - 1
r F
ri — r - - 1 ~* H
*
1 1 1 1 * T ■ 1— "
1 . . . -—• J. —. — - + t T " - 7 " - T"
Dale i _
-- ■ ■
^ : Dale ■1 i -- _ " \ 1 . _ _^
i . .: - :"l-J„
■ -t - J - F . - f a — ■

-— ■ ■ ■ 1 1 1 * 1

S ign . ny- J ^ | | | Siqn n .rv 1 — r __ ^ : : 1 _ .* fc_ ,

* * n ar
1 1
r— h

10 -2G—fW 50 100 200 500 1000 2000 5000 10OO0 20000 40t>00& A b C im" " 10 — 2lo H? 50 100 200 500 1000 2000 5000 10000 20000 40t)0QO A B C liniru
7?>Q7b4 75G76G
H I . .1+ _ 1
QP 1 1 2 4 Multiply Frequency Scale by Zero Level 1612 2112 A B C Lin rvi uiT i f P i y r r e q r U t I I I .-V J L dHe u £_
er u
1
LI
;v tl - C U 1 .£.' £ 1 1 -£ M D L- L I U
QP 11 24 r

Fig.10. Frequency response of t h e loudspeaker in Fig. 11 measured Fig.11. Frequency response of a loudspeaker (not t h e same as in
w i t h Gating S y s t e m in anechoic r o o m . The difference be­ Figs.8 and 9} using pure sine in anechoic r o o m
t w e e n these t w o curves indicates t h e influence of t h e Gating
System

Bruel & Kjfer B r u e l & Kja&r Bruel & Kj*r

Zi r r, r. r. n r. n n n ™ n - n n n r. u :J .. L. : I n n n a ii u zi u u .' : i L: ~ ^ "^ r n n r u zi i? L_ L: □ r -. J —

5 0 R M 2 , 6 0
B r u e l & K j e e r Poteritmmetf-r Range dB Rcrtificr ^ Lowpr Lim Freri 9. H/ Wr Speed . m m sec Paper S p e e d ' m m ' b
f - n j . > \ F ~ . 1 H B B H ■ _ B U U J ■&■ — _ i _ ■ ■! r~fa
U 1
b b ■ ¥ ; ! J ■ i i i■ ■ :■ f p p f j v 75
Copenhagen ° ^ ■ ± _i i .. — i . —
■_! j
.: _- rr™i i , T - j
4 "

■ J * - > ■ * -i
i
4
-1■ 1 4 x —L H
4 - j -

' - *■ -r ■ x
- — -:

_i n -■ ■rb
L3D
X

i i . _i_ —
4 -■

— —
-r -
ii
, 1
- 4- . T - - T --L- - ^ , _,

dB
I ■ j X -- v 4 —¥ —f*\ — -x

. IllVI'" — L\ ■ - ■+--{ ""

jk
i 1 ' 1

_ ^
L

/
1L_- ■.
-
^W t£A ■4BBjH'Bt
W "I T
i i . :
The usefulness of the Gating Sys­ 40 20 ■ s—
-M X J
■ r r ■ ■ l l l f l l l l l 11 p J
60
— 4 - — fflll 1 7 1 1 . * - - > - ■ 9

Measuring Obj
-
— __
. 1
- 1 -
-i—
1 — ■ LI ifalM*" i'.ifc. 1 "U E w ' ■.■■ ft 1
1
1
_L
"
1 ^ "
-l
|
X 1
■ A iy. -■ B i n i ' I I I f
X ^ . _ 4

. _ __ !|4rBVIvJjT "HLBJBJ^^B'

tem is seen by comparison to L:

l

1
L _ ^ ___ 1 1 ■
. j

■ ■ i i i i i MiU
ESli'ij!
^jp" n ^fa.
— 1 —
1
1 B^BF "^BTBTT* "+ B¥BT ~ ^— —

1
1-
_i
- ■

-- —f ■ S i 1 B | 1 ■ T il lif lnT rn S HB^V ~B^B1 T

~^M —1 .
- A _, ., t ,. ,

J J
1 1 . T
III 1 L ' L ^ '

30M5 ' T—' ■ribaflvaiii r f j — ^ ~ ^ — T W *' ■" ^m~ J

; , . 6 i4 5
Fig. 12. Here a frequency response zt ri
1 w i ■ *T| * r f
- | 1 ,
r 1 1—■
■ ■ ■■
J ^m ■ ♦ ' , ' : J

--L±
i 1 ^ ^fj -
H ■" - * -■■;■■ "■ ". : " T i
I
1
__ - - —.
""/"
— -- - ^ - I 4 ^_ ^
4 -
W
1- -- + - , , , ,

measurement was made in a reflec­ T / ^

10
_ —
■ J H- —
"

_1
^
y * -
■—1■ — 1 — i
- * —¥ — _ 1
+—
ri . i .
r
-^- . —

^ 1 r r
^u ■ i . r L L . 1 - - ; * '3 0
-- __ _ i■ i "*
i

- --f-
tive room using a continuous sine f\^ i ^— 1 — 1 ■ - + — r -, - m
F T
i l 1l ' X 4i fa m
U. - : - J J.
■ "

±
I
m - -
n L
4
i
■:-
1
-
" 1
J _
-1—
1
—
1

S\"
^" ■ 4 - h

f
■
4 ■ ■
1 T - * I ■ : 4
■ ■ ■ '

signal. 1 rt
5 1
r^
i

■
--
1
L
L

j .
-
^ I
L
■
" t
L
r
■ ■
ff
IT
t:
^
■ '
^ . .
1

!
1 i
i
|
I" — "
^ ^ ^
1

,
i

;
,

, .
,
- " ■
,_

j 2- 1 5
"
j
i '

\tt ;
~ ~ m ■

L
*
A
1
1
1
1 — |
■
1 —
, ^ x
1 1
- -^ ■
1
1 "- i . i - — — ■ -
- -1
II 1 ■ 1
1 ■
. _ i 1 — i -
"■ - + - - - -
i i . .
Rec No 4
L ^ l_ i __ ■ 1 ■ h - -■ i

1
■
■ " i -fa
Date 1 * - t 1
! ■
....!-__ "'"! — 1
~l I
i i "■ } i X 1 _
H
Sian - n ■ T 1 "" ■ 1 1 -.
'0
0 J B H J 50 100 200 500 1000 2000 5000 10000 20000 40000D A B ChnU
?507t 55
OP 1 1 2 4 Multiply Frequency Sc^Le bv Zero Level - - 1612/211? A B C Lin

Fig. 1 2 . Frequency response of loudspeaker measured in a r o o m w i t h reflecting surfaces

5
Directional Characteristics
Figures 1 3 to 16 show the influ­
ence of the Gating System on the
measurement of the directional
characteristics of a loudspeaker.
Here w e see a good correlation be­
---■ o 0 ODDODOOo 0

tween the two curves. However, it

is seen that at the higher frequen­
cies the Gating System levels out
some of the very sharp notches.
One reason for this can best be ex­
plained by describing the cause of
these sharp dips. The sound arriv­
ing at a given point to the side or
rear of a loudspeaker travels by
more than one path due to reflec­
tion and diffraction phenomena in
the loudspeaker. Since the path
lengths will differ, the sound sig­
nals will reinforce or cancel each
B
i ^aao/aoaaa^m ^^t^l^sS^
other depending on their relative Fig. 13. Directional characteristics of loud­ Fig.14. Directional characteristics of the
phase. This results in the lobes speaker measured with pure sine in same loudspeaker measured with
w i t h the deep notches seen in direc­ anechoic chamber the Gating System in an ordinary
room. Note the 10 kHz curve
tional characteristics measured w i t h
shows the background noise level
pure sine. (It is the same kind of of the room
standing wave phenomena that
causes the sharp dips in Fig. 1 2.)
'0 0 o o Q a D o o a 0 „ g „OOOOO[](]DDO 0 O -;

With the relatively short tone

bursts used w i t h the gating system,
all the standing wave phenomena
do not have time to build-up and
hence the sharp dips are not seen.
It can be debated w h i c h of the two
curves is more correct, but this is
probably fruitless since the exact
shape of the directional characteris­
tic to the rear is of no pratical sig­
nificance and the numerous lobes
and dips will be obliterated by the
standing wave patterns of the room
in w h i c h the speaker is used. """"^t^ii^^m
Fig. 15. Directional characteristics of the Fig. 16. Directional characteristics of the
same loudspeaker measured with same loudspeaker measured with
A second reason for the differ­
the Gating System in an anechoic pure tones in ordinary room
ence in the curves lies in the nature chamber. The sharp dips and peaks
of the tone bursts themselves as de­ are not as pronounced as for pure
scribed in the next section. sine

Tone Burst Characteristics

A tone burst contains not only the cies arise due to the square wave frequency components and the click
frequency of the sine wave con­ by which the sine signal is gated. arise at the abrupt transition from a
tained in the burst but also a band These components can be heard as flat line to sine wave at the begin­
of frequencies centered around the a "click" at the beginning and end ning and end of the burst.
sine wave frequency. These frequen­ of the tone burst. Physically, these

6
 5 Hz 1 ms 15 Hz
The frequency spectra of various
tone bursts in Fig. 17 give us a
clearer view of the these pheno­
mena. The spectra are recorded on
a Real-Time Narrow Band Analyzer
(Type 3 3 4 8 ) and are displayed on a
linear frequency scale from 0 to
2 kHz. The frequency of the gated
sine wave in the burst is 1 kHz and , -£X OlO X_X^. y£ . y j $ X^X* to" ^ tfffl V ' ffO > >X O O O ^ ^ ^ S t t ^ ^ W
v^xtfott!?*#^xtfft:_W

l^o^^^^S*>SKi^^

occurs in the middle of the main frxoraiSfltfE-^^

_oxtt*;¥^t_^o^:ra

lobe in the middle of the screen.

The width of this lobe is 2 / t where
t is the duration of the tone burst. 3 ms
Thus a shorter burst gives a broader
frequency spectrum. In the limiting
case of an infinitely short burst,
(Dirac unit impulse) the frequency
spectrum will be infinitely wide,
i.e. include all frequencies. This
case is approached by t h e 1 ms i - u j ■--:ro_:.!t' *+ . ■.

w - c w ^ "> ^ ^ _ o ^ M - -.opo

burst at the top of Fig. 17. A t the - < - -- o^TO-OiWx *&■ < ^ £ WV
c!-- ^ o x= x t f r x - x o i s ^ K > -co x t t

--- « E W S P - ' / W m ™ > ^ K W f p } * * r t X -CM

-o^-o*^- ^ o M ^ X K ^ > \ £ : o f r . ^ ->

.Jp.-^-^-. ■■■_0.-!-,-,-vv(Or-" - S ^ - \ ^
■■ ■■ > pioo .-x --ri-; o_-_+tf -y«-^o^H-J^-xtfr"< ■■ *
TOfrioVri- ^xtt^^^SxWS^oitt:^^ —- <+**!(

bottom of the figure w e see that the _■ x ^ ^

^ o ^ ^ J K t t W ^ ^ x r a S

i ^ ! ^ x o _ o _ t t = : £ _ : _ > > > ^ ^ -co- x = f i r

SWWfcKW
«o«*tt- "«J*—i^
xtf^Kv^ >££:»&-o-
^wwrao^- - x ^ ^ ^ v S f t w ^ ^ o ; ^ -^^OVP
S_tfG-1 xtt*_*£_#-x#r* tftfTOWM
o_- ^ ^ M M . V ^ ' -. ■: o '■ X^."-XK O ^ - - \ S S » X - - > - ;
- *******

xt7W^J>-J

1 0 0 ms burst approaches the single ^a-P*x«_:_*o:tf^^^

- i * ^ - t o x ^ > ^ ^ ; top- x-. ■: ^ - O > X E ^ K ■^xrx^vi-xix^+r.-' > soo>^ -->^^«^^^vo.'^-$4«W^-x^^^^^,wK

^^^sw:>^^i?x^

line spectrum of a continuous sine

wave.
10 ms

Of particular interest is the

—3 dB bandwidth of the center lobe
which equals
eauals 1 i //tt.. Thus a tone
i nus a
burst can be regarded to have a
spectrum similar to narrow band
noise which is often used as a loud­
speaker test signal. (However, the
tone burst has numerous side lobes
( 1 / t wide) which the narrow band
noise does not have.)
30 ms
This raises the obvious question:
" H o w can a tone burst w i t h a rela­
tively broad spectrum give as good
frequency resolution as a continu­
ous tone w h i c h has an infinitely nar­
row spectrum?" The answer, which H^^*>+---T
_
'+i^"
^ . ^ - - - - ■ - ^ L - _ _ _ _ _

--^-^
-
-■.■-■-■.■ _ _ J _

-«.
■_■■■■■■■ - - - - I I L L . ' . V . ' I I I r L L _ .■.-.-.■.

-x^»t***-

w e already have hinted at, is hope­ ^

t
>---^'OJt

< -^
-.
--
___j __ ■-■-■ an ■ ■ L - - - ■ . - . - . ■ . u i L L _ -

L_■■■-.-.-! i

" ^
i J- .-.■■■.

- *
IIJJ

. ^
-^- 1 -f - * r
■^■-■^i L ^ J A _

i ---^k-.n^ i u

- ' ■ -
■ L

* + -

fully equally obvious. The beginning

r
L^'--_^.-_-_ I I L__ *"■"■"■ r-- IT."

_H_^_ ^ - - j ^ . . -

^ s ;*> ^ - x -

of the tone burst excites t h e loud­ ^ t t ^ ^ ^ ^ ^ ^ ^ ^ F x t f - *

speaker w i t h a broad spectrum. Af­

ter a given response time to this
"shock" the loudspeaker will settle
to the steady-state value of the sine
1 0 0 ms
wave. Since w e measure only this
value, w e eliminate the influence of --^» _ _ _ _ _ 11 _ . . - . . - - - _ _ , , , _ _ _ ^ ^

the other frequency components

caused by the beginning and end of
the tone burst. Thus the frequency
resolution of the tone burst can
equal that of a continuous sine
wave.
0_0_-H----ri'
- r - - - - . - - . _ i - - --■-■-■-n_i ■ - i i i ■■■■■-.-.-■
-_ox^:^:J

^ ^WJH" ^ - - ^ O ^ ^ ^ ^ P ^ X ^ « O _ ^ ^ K ^ ^ . ^ ^ S _0_0_-H-- ccf -^O-K-P ■--. ^ --"4 frl

--xv^d> ( W ^ ^ ^ x ^ x ^ x i x ^ S ^ ^ ^
X^^>>-y_p^_¥_^_7
>x^S^^i¥|Wx^_o^xW^o^^ ^x^^^^>x^^^^^x^t^^+^^

i xrx' W f f l sx-- - o_o™ ^ o>W^: ^^—M-^o^xVM^^>Sx^t w '-^P>>™o^

Fig. 17 Frequency spectra of tone burts w i t h Repetition Rates o f 5 Hz and 15 Hz. Burst
w i d t h varies f r o m 1 ms t o 1 0 0 ms. Center frequency is 1 kHz, Displayed on linear scale
f r o m 0 t o 2 kHz w i t h a 5 0 dB vertical scale

7
 It is extremely important to em­ Briiel & Kjser Bruel & Kjaer Bruel & Kjser
DnanaaaaannDn anaaaaDDaDaannaaDaannnnnnaDDn □ a □ n □ a □[
phasize the criterion that the loud­ B r u e l & K j a e r P o t e n t i o m e t e r R a n g e d . — d B Rectifier: R W S
Lower Lim. Freq.: *20
" u , >«,.
Hz Wr. e 1. 160
Speed: m m / s e c Paper Speed: _?__ m m / s e c
Copenhagen ^
1CK75
speaker must have fully responded
and settled to the steady-state
value. If for some reason, the re­
sponse time of the loudspeaker is
longer than the duration of the tone
burst, the accuracy of the measure­
ment will be decreased. One cause
of the error will be that one does
not know which section of the signal
to select, since it has not settled.
Generally, of course, the section Rec. N o .
closest to the end of the burst Date

should be most accurate. Since the 10 30 Hz 50 100 500 1000 2000 5000 10000 20000 40000D A B C NiT
750768
QP 1 1 2 4 M u l t i p l y Frequency Scale by Zero Level: 1612/2112 A B C Lin.
measurement w i not be on a set-
Fig. 18. Illustration of averaging due t o too short a tone burst. Dashed line shows free-field
tied signal, the value will be related
sine wave response, solid line shows the response measured w i t h a tone burst consist­
to the impulse response of the loud­ ing only of one period of its frequency
speaker to the frequency spectrum
of the center lobe of the tone burst.
Thus the frequency resolution of sured traditionally w i t h pure sines In the case of directional charac­
the measurement is reduced, and in an anechoic chamber. teristics, the broad frequency spec­
sharp peaks and dips normally seen trum of the tone burst will excite res­
in the sine wave response will be This does not mean that a meas­ onances and standing waves in
obscured. This is illustrated by urement is not possible on a single the loudspeaker and its cabinet. On
Fig. 18, which shows the frequency period tone burst. Provided that the axis, the level of these resonances
response curve obtained using only loudspeaker is good enough, it can is well below the direct signal and
a single period of the test fre­ have settled to the correct value in hence they are not seen. However,
quency. This then gives a frequency that single period. For this to be pos­ off axis where the level of the direct
resolution or bandwidth of the meas­ sible, the length of the tone burst signal is significantly lower, the in­
urement equal to the center fre­ should be equal to or greater than fluence of these resonances will be
quency of the tone burst. This aver­ the settling time of the loudspeaker. seen.
aging effect can be seen clearly by Otherwise the averaging pheno­
comparing Fig.7 w i t h Fig.8 w h i c h menon illustrated in Fig. 18 will oc­
shows the free-field response mea- cur.

Evaluation of Anechoic Rooms

The Gating System can be used frame in an anechoic room is seen removed the influence of the reflec-
as a powerful tool in examining an in the difference between the direc­ tion which was also seen very
anechoic room for reflective sur­ tional characteristics in Figs. 19 and clearly on the oscilloscope w h e n us­
faces. A loudspeaker can be aimed 2 0 . The use of the Gating System ing a tone burst.
at a suspected reflecting surface,
such as the construction around the •"^OQOOOOQGOQjJfJo
36" ___—■<*—-JIM i i 0 a

door, and using the Gating System,

reflections, if any, are readily ob­
served on the oscilloscope. The mea­
suring gate can then be adjusted to
only include the reflection, and the
amplitude of the reflection can be
plotted as a function of frequency
on the level recorder. This same
technique can also be applied to de­
termine the absorption coefficient,
as a function of angle, of various
materials. A significant advantage
of this technique is that the materi­
als can be measured on location in ^w^^:
the actual mounting. Fig. 19. Directional characteristics of loud­ Fig.20. Reflections from door frame
speaker measured using pure sine in (Fig. 19) are removed by use of the
anechoic room. Note the influence Gating System
An example of the influence of a of reflections f r o m 1 2 0 ° t o 1 7 0 ° .
relatively small and seemingly inno­ These reflections come f r o m the
cent portion of an exposed door door frame

8
Harmonic Distortion
Distortion measurements may be For distortion measurements, the
made using a tone burst provided filter will not be tuned to the funda­
that the response time of the mea­ mental as in Fig.21 but to some har­
suring filter is short enough to per­ monic. However, the same rules for
mit full response,
resDonse. The
i ne response filter response time still apply.
time (T) of a filter is related to its' Fig.22 shows the response of a fil­
bandwidth (B) by ter tuned to the second harmonic of
the frequency contained in the tone
1_ burst. The first peak corresponds to
T
B the filter's response to the begin­
ning of the tone burst, and the sec­
Thus for typical tone bursts of 3 ond peak to the end of the burst.
to 10 ms duration, bandwidths of at The steady-state portion before the Fig.22. Response of filter tuned to second
least 1 0 0 Hz to 3 0 0 Hz will be re­ end of the burst corresponds to the harmonic of the tone burst. Gate is
quired. This requirement is fulfilled second harmonic distortion of the adjusted to measure distortion com­
ponent
by the Heterodyne Analyzer Type loudspeaker to w h i c h the measur­
2 0 1 0 which has bandwidths up to ing gate is adjusted.
1 0 0 0 H z . Fig.21 illustrates this by
showing the response of a filter
w i t h a 3 1 6 Hz bandwidth to tone
bursts of various lengths.

1 ms

■■■■■■■■■.-.■ ■.■■■■■■■.■.■ .-^-- .-H

V - - I"I i -r---- ■-■ --■ J----J i ih i I Hi i i u'---'^

>■'! +

.-. w v^ w ^J^ \" J

Fig.23 shows a typical instrument ond and third harmonic distortion

WBi^mm£$¥<°&<fawh£MM "^^ set-up where the Tracking Fre­ curves measured w i t h such a set­
quency Multiplier Type 1901 is up are shown in Figs.24 and 2 5 .
3 ms
used to lock on to the chosen har­ Both pure tone and gated measure­
monic and to tune the analyzer sec­ ment curves are shown. It can be
tion of the 2 0 1 0 to that harmonic. seen that due to the wide filter
(Distortion Measurement Control bandwidth (31 6 Hz) used, the lower
Unit Type 1 9 0 2 could be used frequency limit is in the 5 0 0 to
in place of the 1901.) Typical sec­ 1 0 0 0 Hz range.

Flexible Driveshaft
~l
Heterodyne Analyzer
2010

r - ■ ^

♦
10 ms r 1 2.05 9 3

BFO Gating System

s
4440 Power Amplifier
L * * o
L__ 4133 + 2619
4 Tracking Frequency Multiplier « ^
u 2706 ■ ^

Level Recorder
1901

2307
750419
Fig.21. Response of 3 1 6 Hz wide filter to
various tone burst durations Fig.23. Instrument set-up for distortion measurements using the Gating System
 X-«.

B r u e l & K J E e r Potentiometer Range

Bruel Sf Kjger

50 dB Rectifier
Bruel & Kjeer

RMS Lower Lim Freq 20 Hz Wr Speed

Bruel & Kjser

SO m m sec Paper Speed 3 m m . sec

Bruel & Kjffir
n a G □ n n n z i c c D ' Z i a a a L : u n n n n r : n n D 3 D D a c c n D n n n n n n z ; D G
B r u e l & K j e e r Potentiometer Range 50
Bruel & K]3jr

dB Rectifier. RMS Lower Lim Freq . 20 Hr Wr. Speed

Bruel & Kja?r

30
n D D n a c
m m / s e c Paper Speed. 3 m m / s e c
T\
10 7 5
Copenhagen 5 0 f * 5
6 0 5
Copenhagen ^ ^ ~ F ~

d B dB
.'--'

4 0 20
Measuring Ob| ;_ Measuring Obj '_
Swept ,_ Swept
Distortion of Distortion of
loudspeaker ^Q 15 oi,dspeaker 301-15 45
in anechoic n a nee ho LC
chamber clumber
2010 2010
1901 1901
20 1 0 20

-10^
r

Rec No R e c No
Date _1_?/iZ5 □ate 17/4 75
Sign . H.M 0 L D
Sign H.M. Q
10 2 0 ' Hz 50 100 500 1000 2000 5000 10000 2oooouoooao A B c im 10 20^H? 50 100 500 1000 2000 5000 10000 20000 >D A B C *tt.

QP 1 1 2 4 Multiply Frequency Scale by Zero Level 1612/2112

7504!3
A B C Lin QP 1 1 2 4 M u l t i p l y Frequency Scala by Zero L e v e l . - " " 1612/2112 A
75041
B C Lffl.
1
■ * ■ * * *
f.H_
Bruel & Kjser Bruel & Kjser Bruel & K|»r Bruel & Kja?r Bruel 8. Kjeer Bruet & Kja?r
n a n z : D U L ; ; j ^ : : a n n n 3 D D a u i u n c ! D a u i : G G D D 3 a n n z i n a n n a u D u 3 n n n n u t □ □ n u z i c j U L J o u n n c a n n u n n n : : ] u z : n z : D D D C G L : L j a a a a ~ J n :: n :: r. n c n c D G :
Briiel&KlEer Potentiometer Ranqe dB Rectifier Lower Lim Freq . — H z Wr Speed m m / s e c PaperSpeed mm/sflc B r i J e l Si K j a e r Potentiometer Range 50 dB Rectifier DC Lower Lim Freq 200 Hr Wr Speed 200 m m sec Paper Speed. 3 m m sec
Copenhagen 5 0 25
1

d B dB
f—r rf.ji -f.|i_j ■ ;i ■ ■ 1 W^_
m -9- ■ " ^
-4-

Bandwidth = 316 Hz ^ J l s e i
0msec. Distance 1 m ^
^10^75

dB d B
Copenhagen 50
rZ5'

dB d B
• —
■_ I. I 7~~w T T T ^

+ n
:—
j -■
_
■ - f\ . . _ r^ _^
10m sec Distance 1 m --f -
T f ■ l10l75

1 - U B a n d w i d t h 3 1 6 Hz Puts
1 _
tse-r^dB'dB
i

+ -
4O-20r B-60 40 20 I r - ---r ZIBH60
F r ~ "

Measuring Obj _ •_ __;Jr~ Measuring Obj - _

3. Harmonic + 40 d B : -. —t r
- i — i —
Swept |_ Swept _ ].
J. -r-
Distortion of I- Distortion of
i
louo^pedKur 3Q i i 51 ■6'45 loudspeaker 3 0 1 5 ■6-45
i- -'
l'l Ofdiridry in ordinary i — i -
-r T
ruum r l- room
4440 4440
2010 2010
1901 20,10 4-30 1901 2 0 10 1 --- 4-130
L.

I-
._J .
K -

10:5 2-15 10 T-Z-15

h l

Rec No. _ . Rec No.:

Date 24/4:75 . Date 2 4 / 4 75
S.gn H.M._ QCQ f 0 Sign : H M. 0 L-G 0J
10 20 Hz 50 100 200 500 1000 2000 5000 10000 20001 ID A B (ft* 10 20 Hz 50 100 200 500 1000 2000 5000 10000 20000 40000D A B C lin"
750411 750412
Multiply Frequency Scale by Zero Level: - -- - 1612/2112 A B C L*n OP 1 1 2 4 M u l t i p l y Frequency Scale by Zero Level 1612/2112 A B C Lin
QP 1 1 2 4

Fig.24. Results of distortion measurements. The top chart shows the Fig.25. Results of distortion measurements. The top chart shows the
fundamental and second harmonic of a loudspeaker measured fundamental and third harmonic of a loudspeaker measured
w i t h pure sines in a free-field. The lower chart shows the w i t h pure sines in a free-field. The lower chart shows the
same measurements made w i t h the Gating System in an ordi­ same measurements made w i t h the gating System in an ordi­
nary room nary room

Tone Burst Response

The Gating System may also be use of tone bursts for this tech­ tics". The duration of the burst
used as a tone burst generator to nique, it is important to remember should be chosen to give an approp­
permit subjective evaluation of loud­ the frequency distribution of the en­ riate bandwidth related to the cen­
speaker transient response as ergy of the burst as pointed out in ter frequency.
viewed on the oscilloscope. In the section "Tone Burst Characteris­

Early Reflections
The Gating System Type 4 4 4 0 (Ref.10) have studied the so-called
may also be applied to the measure­ "Box S o u n d " and have suggested
ment of so called "early reflections" that early reflections may be an ex­
w h i c h indicates the energy distribu­ planation of this phenomenon. Thus
tion as a function of frequency in the lower the level of the early ref­
the milliseconds immediately after lections, the better the sound qual­
the tone burst was supposed to ity.
stop. Early reflections are due to
overhang of the speaker, and reflec­ Since the early reflections are
tions, standing waves, and reso­ strongly frequency dependent, it
nances in the speaker itself and may be difficult to find the point to
especially in the cabinet. which the measuring gate should
be adjusted. But by changing the
A measurement of early reflec­ frequency up and d o w n , the point
tions indicates how fast the re­ at which the phenomenon is most
sponse of the system is and how pronounced can be observed on the s*

much sound of its own the loud­ oscilloscope and the gate can be ad­
speaker box adds to the signal due justed to measure this. Such an ad­
to internal reflections and reso­ justment is shown in Fig.26. Fig.26. Adjustment of the gate for measure­
nances. Barlow (Ref.9) and Stevens ment of early reflections

10
 The results of such measure­
Frequence Response (steady-state)
ments may sometimes show that a
relatively small, and seemingly insig­ Sound Pressure 0,5 ms after the burst
nificant peak in the continuous sine Sound Pressure 1,0 ms after the burst
wave response curve may be re­
Sound Pressure 1,5 ms after the burst
lated to a much larger peak in the
early reflection curve. Thus the col­ Sound Pressure 2,0 ms after the burst
oration of the sound at that fre­ Sound Pressure 2,5 ms after the burst
quency will subjectively be much
worse than indicated by the sine
~r~ ~T~ T-
wave resoonse. (Fig.27). 2k 3k 4k 5k ft Hz]
1k
750736

Fig.27. Frequency characteristics of early reflections compared to the steady state response

• •
Gating System
4440 (No. 1}
4133/2619
Noise Generator Power Amplifier Measuring Amplifier
1405 2706 2606

Gating System Hewlett-Packard

4440 {No. 2) HP 9825A

"I
"1

Digital Frequency Analyzer

2131

HP 9872A
L
l
L JJ
750738/1

Fig.28. Instrumentit set-up for real-time analysis of early reflections using gated white noise

A considerably more advanced This same instrumentation can tance from a wall to use its reflec­
technique for studying early reflec­ also be used to analyze the charac­ tions. By adjusting the gate to in­
tions is suggested in Fig.28. Here, teristics of loudspeakers which are clude the last part of the tone burst
the white noise output of Noise Gen­ designed to use reflections from and the first several reflections, the
erator Type 1405 is gated and fed walls or corners to give the desired response of this speaker type and
to the speaker. (It should be noted sound. Such types include omni-di- the room's early reflections can be
that gated white noise retains its rectional loudspeakers and speakers seen.
white noise spectrum). A second designed to be placed a given dis­
Gating System is used to process
the received signal. The signal is
fed through the transmitting section
of the second 4 4 4 0 which is trig­
gered from the Gate Output of the
measuring section of the first 4 4 4 0
(Fig.29). The output of the second
4 4 4 0 (Pulse Output) is then fed to
the Real-Time Analyzer Type 3 3 4 8 II^JV^ Time Signal
where the frequency spectrum of
the desired section of the early re­
A Trigger Gate (Receiving section of Gating System 1)

flection is desplayed. The delay of AC Measuring Gate (Transmitting section of Gating System 2)
the gate can then be varied and the
changes in the frequency spectrum W* Measured Signal {Output of transmitting section of Gating System 2)
750737
can be observed as a function of
time after the end of the burst. Fig.29. Setting of gates to measure early reflections with the instruments in Fig.28
Phase Response
A tone burst test signal not only tion Note No. 1 5 - 0 9 0 "Loudspeaker couple the receiving section of the
contains the amplitude information, phase measurements, transient re­ Gating System. For information on
but also the phase information. This sponse and audible q u a l i t y " can this, contact Bruel & Kjaer directly.
can be measured using a phase me­ also be used for phase measure­
ter w i t h a sufficiently fast response ments using tone bursts. However, The system has the same basic
time. to make this possible, t w o minor low frequency limits as for ampli­
modifications are required, one to tude response measurements . . . be­
The instrumentation and tech­ increase the phase meter's re­ ing dependent, of course, on room
niques described in B & K Applica­ sponse time, and the second to DC size.

Conclusion
Gating techniques provide a characteristics may be measured u n ­ w i n d o w , the single tone response is
powerful tool extending electro- der simulated free-field conditions. measured, w i t h o u t the influence of
acoustic measurements to non-ane- The transmitted tone burst con­ the broad band spectrum. By adjust­
choic rooms. By the use of the tains a broad spectrum of frequen­ ing the gate to other sections of the
Bruel & Kjaer Gating System Type cies. However, by selecting only the burst, the transient response and
4 4 4 0 , amplitude and phase re­ steady-state portion of the burst us­ early reflection characteristics of
sponse, distortion and directional ing a suitably delayed measuring the loudspeaker can be measured.

Appendix
Calculation of M i c r o p h o n e /
Loudspeaker distance and
pulse length
Assume a room (Fig.30) w i t h the
transducers equally spaced between
floor and ceiling (h, the height of
the room is assumed the smallest
of the room's dimensions). First, w e
w i l l only consider reflections from
side w a l l s , ceiling and floor. The
pulse length (t) must then be shor­
ter than the difference between the
time it takes to travel the reflected
( 2 l / c ) and the direct path (d/c).
Hence
L - * H
750415/1
2 2
21 — d Vh + d d
t ^ (D Fig.30. Travel distances f o r the first reflections w h e n the loudspeaker and m i c r o p h o n e are cen­
tered along all three axes of the r o o m

Solving for d w e obtain 2 2 2 The reciprocal of w h i c h gives the

Ct = h — c t
lower frequency limit f m j n
2 2 2 2ct
d^ h — c t
(2)
2ct f mm. = 595 (6)
h
The criterion that the microphone 2 2 2
3 c t = h
should be at least one wavelength
from the loudspeaker gives

at a distance between transducers

d s* ct (3)
of
t = h v3 (4)
3
w h e r e t is the period at the lowest
d = ct = c
h. V3 = 0,577h (7)
frequency w h i c h also corresponds c 3
to the pulse length w h i c h contains
one period at the lowest frequency. h
Setting Equations 2 and 3 equal w e t = (5)
595
obtain the optimum pulse length w h i c h is the optimum spacing be­
and corresponding transducer spac­ tween transducers for a given mini­
ing: m u m room dimension h.

12
 For reflections from the end walls Hence the length of the room differentiating w i t h respect to time
of the room along its longest dimen­ must be at least 15% longer than
sion (L), the length of the pulse the smallest dimension in order for x = 5 0 log e f»
must be shorter than the difference Equations 5-7 to be valid.
between the time it takes for the
first reflection to return to the micro­ However, w i t h reflections from
phone (L/c) and the time it takes for the end walls setting the limits, the
*=*LZf
f
the direct sound to reach the micro­ pulse length must be (from Eqn. (9))
phone (d/c). where x is the Paper Speed (P) in
t=-L = L (11) m m / s and f is the sweep rate in
Hence: 2c 688
H z / s . Hence
L — d w i t h an optimum distance between Pf
t < f = (12)
transducers of (combining Eqn. (3) 21,7
and (9))
or With the Gating System, the fre­
d = quency resolution (B) in Hz equals
2
d< L ct (8) the sweep rate divided by the num­
ber of measurements per second
Now reflections from the far wall (the Repetition Rate, R).
only become a limitation w h e n the Sweep Speed Derivation
minimum distance of Equation (8) is For the logarithmic Level Recor­ B =
equal to, or less than that of Equa­ R
der paper commonly used w i t h
tion (3). Setting the t w o equa Bruel & Kjaer generators, one de­
cade in frequency corresponds to BR = f (13)
L= 2ct (9) 5 0 mm paper displacement.
Combining Equations 12 and 13
and substituting t from Equation (4) Hence the paper displacement (x) gives
w e get is related to frequency (f) by Pf
BR =
21,7
|_=J2 V 3 h = 1 i5h (10) x = 5 0 log f
3 or

P = 2V7 BR (14)
f

References
Ref.1. Peter Levin Ref.5 Henning Moller Ref.9. O. A. Barlow
Calibration of Hydrophones Loudspeaker phase meas­ The sound output from
B & K Technical Review No. urements, transient re­ loudspeaker cabinet walls
1, 1973 sponse and audible quality. Paper C5 of 50th AES Con­
B&K Application Note vention
Ref.2. Bruel & Kjaer, Product Data 1 5—090
4440 Ref.10. W . R. Stevens
Ref.6 Henning Moller Sound radiation from loud
Ref.3. Franklin F. Kuo Relevant loudspeaker tests speaker cabinets
Network analysis and syn- in studios, in Hi-Fi dealer's Paper C4 of 50th AES Con
thesis demo-rooms, in the home vention
John Wiley & Sons, Inc. etc., — using 1/3 octave,
U.S.A. pinkweighted, random Ref.1 1 . Henning Moller
noise B&K Application 3-Dimensional acoustic
ef.4. R. B. Randall Note 1 5 — 0 6 7 measurements
Frequency Analysis of Sta­ —using gating technique
tionary, Ref.8. J . M. Berman B & K Application Note
Non-stationary and Tran­ Loudspeaker evaluation us 17—163
sient Signals ing digital techniques
B&K Application Note KEF Research Fellow, Uni Ref.1 2. Henning Moller
14—165 versity of Bradford Electro Acoustic Measure-
Paper of the 5 0 t h AES Con ments
vention B & K 16—035

13
 Bruel & Kjaer ^ = -
Bruel&Kjaer Instruments, Inc.
185 Forest Street, Marlborough, Massachusetts 01752 • (617) 481-7000 • TWX 710/347-1187
Regional Offices: MA (617) 481-7737; NJ (201) 227-6100; MD (301) 948-0494; GA (404) 422-5200; Ml (313) 522-8600; IL (312) 358-7582;
TX (713) 645-0167; CO (303) 469-7914; CA (714) 978-8066, (415) 574-8100; WA (206) 324-5905.