BRUEL & KJAER 2131 User manual

I Instruction Manual
Digital Frequency Analyzer Type 2131
033-0178
t~ •!> . ~ •: '11 fCJ
"
44 * ·«~
.., ..+llo
~ - ~ :"
~-~
••
iliili
_.....,_ . .,.._,.,.,.,.._ ,
Ill I
2131
A real-time octave and 1/ 3 octave analyzer
featuring almost entirely digital construc-
tion, its operation being based on digit~! fil-
tering, detection .and averaging techniques .
The detected data
may be averaged either
exponentially or linearly with a constant av-
eraging time across all channels. Addition-
ally, exponential averaging with a constant
confidence level is available . Display of the
final spectrum formed is on an 11" cali-
brated display screen. "·
The 2131 is equipped with memories en-
abling the comparison of spectra, and has
full analog and digital read-in and read-out
facilities. T tie"~ e is also extensive 'remote pro-
gramming capability, allowing · ~ almost the
complete control of the 2131 from, e.g., a
computer.
+ BrUel & Kjcer

BROEL & KJA:R instruments cover the whole field of sound and vibration measurements .
The main groups are:
ACOUSTICAL MEASl,IREMENTS
Condenser Microphones
Piezoelectric Microphones
Microphone Preamplifiers
Hydrophones
Sound Level Meters
Precision Sound Level Meters
Impulse Sound Level Meters
Noise Dose Meters
Noise Level Analyzers
Standing Wave Apparatus
Calibration Equipment
Reverberation Processors
Sound Sources
ACOUSTICAL RESPONSE TESTING
Sine Generators
Random Noise Generators
Sine-Random Generators
Artificial Voices
Artificial Ears
Artificial Mastoids
Hearing Aid Test Boxes
Audiometer Calibrators
Telephone
Measuring Equipment
Audio Reproduction Test Equipment
Tapping Machines
Turntables
VIBRATrON MEASUREMENTS
Accelerometers
Force Transducers
Impedance Heads
Accelerometer Preamplifiers
Vibration Meters
Accelerometer Calibrators
Magnetic Transducers
Capacitive Transducers
Complex Modulus Apparatus
Bump Recorders
VIBRATION TESTING
Exciter Controls - Sine
Exciter Controls- Sine- Random
Exciter Equalizers, Random
or Shock
Exciters
Power Amplifiers
Programmer Units
Stroboscopes
STRAIN MEASUREMENTS
Strain Gauge Apparatus
Multipoint Selectors
MEASUREMENT AND ANALYSIS
Voltmeters
Phase Meters
Deviation Bridges
Measuring Amplifiers
Band-Pass Filter Sets
Frequency Analyzers
Real Time Analyzers
Heterodyne Filters and Analyzers
Distortion Measuring Equipment
Psophometers
Statistical Distribution Analyzers
Tracking Filters
RECORDING
Level Recorders
Frequency Response Tracers
Tape Recorders
Alphanumeric Printers
Digital Event Recorders
DIGITAl. EQUIPMENT
Computers
Tape Punchers
Tape Readers
Bruel & Kjeer
DK-2850 NJERUM, DENMARK · Telephone: + 45 2 80 05 00 · Telex: 37316 bruka dk
Instruction Manual
Digital Frequency Analyzer Type 2131
~
.
033-0178
-
...
llllli
~ ~ - ~ .
iillii
II
--
'• l ..
~ · .. .
II ·1·1 IIi fllliilli I
--- -
-·
2131
A real-time octave and 1/ 3 octave analyzer
featuring almost entirely digital construc-
tion, its operation being based on digit$11 fil-
tering, detection .and averaging techniques.
The detected data
may be averaged either
exponentially or linearly with a constant av-
eraging time across all channels . Addition-
ally, exponential averaging with a constant
confidence level is available. Display of the
final spectrum formed is on an 11" cali-
brated display screen. '·
The 21 31 is equipped with memories en-
abling the comparison of spectra, and has
full analog and digital read-in and read-out
facilities . THe re is also extensive remote pro-
gramming capability, allowing • almost the
complete control of the 21 31 from, e.g ., a
computer.
+ Bri.iel & Kjcer
I
I
-- -. ·---

DIGITAL FREQUENCY ANALYZER
TYPE 2131
Revision May 1 979

CONTENTS
1. INTRODUCTION AND SPECIFICATIONS (PRODUCT DATA) ................................................. 1
2.
CONTROLS .......................................................................................................................... 17
2.1. FRONT PANEL OF DIGITAL FREQUENCY ANALYZER TYPE 2131 . .. .. .............. 17
Controls used in General Operation .. ......... ........ ................ .. ...... .. ............... 17
Controls used in Averaging ........ .. ....... . .. ....... .. ............. .............. .. ............. . 20
Controls used in Spectral Storage and Comparison ...... . .. ....... ....... ..... .. .... .. ... 21
Input and Output Controls ....... . ........... .. .. ............ .. ...... ......... ............ .. ....... 23
2.2. DISPLAY SCREEN OF DIGITAL FREQUENCY ANALYZER 2131 . .. .... .. ..... ........ 24
2.3. REAR PANEL OF DIGITAL FREQUENCY ANALYZER 2131 ...... ....................... 25
3. OPERATION .......................................................................... . .............................. . ............... 28
3.1. PRELIMINARY ADJUSTMENTS ............... ........ . .. ..... .. ......... ................. .. ..... 28
3.2. CALIBRATION ............. ... ........ . .. .. .. ........ .. ...... .. .......... ............ . ............ .. .... 29
Calibration using the Internal Amplitude Reference of the 21 31 .. .. ................ 30
Calibration using an External Amplitude Reference .... .. . .. .. .. .. ....................... 31
3.3 . GENERAL OPERATION, ANALOG INPUT ..... ......... . ............. . .......... ..... ......... 31
3.4. THE AVERAGING FUNCTIONS .. .......................... .. ................... .. .......... .... .. 32
Operation of Linear Averaging .. ........ .. .................... .. .......... .. . .. ................ .. 36
Operation of Exponential Averaging ............... .. .. .. .. .. .. .. ....... . .......... ............. 37
3.5. THE STORAGE FUNCTIONS ........ .. ........................... ............ .. . .. .. .. .. . .. ...... .. 37
3.6. ANALOG OUTPUT FROM THE 2131 .. ...................... .. ........... .. . .. .......... ....... 38
Analog Output to a Level Recorder Type 2305 or 2307 . .. ...... .. .... .......... ..... .. 38
Analog Output to an X-Y Recorder .. ...... .. ................................................... 40
Analog Output to a Bruel & Kjcer X-Y Recorder Type 2308 .. .. ............. .. . .. .... 42
3. 7. DIGITAL INPUT TO THE 2131 ...... ........ .. .................. ......... .. .......... ............ 44
Digital Input from a Bruel & Kjcer Tape Reader Type 71 02 .......... .. ............... 44
3.8. DIGITAL OUTPUT FROM THE 2131 .. ........... .......... .. . .. ..................... .. ...... .. 45
Digital Output to a Bruel & Kjcer Tape Punch Type 6301 ....... ................ .. .... 45
Digital Output to a Bruel & Kjcer Alphanumeric Printer Type 231 2 .. . .. . .. ..... .. . 46
3.9. REMOTE PROGRAMMING OF THE 2131 .. ........ .. .. .. . .. ..... ........ .. .. .. .. ...... .. . .. 46
3.10. EXTENSION UNITS . .......... ....................... .. ................................ .. ............ 47
3.
11 . RACK MOUNTING OF THE 21 31 ... .. ...... .. .. .. .. .. .. ..... .. ........ .. .. . .. .. ....... .. ...... 48
4. DIGITAL DATA TRANSFER AND REMOTE CONTROL OF THE 2131
VIA THE IEC INTERFACE ..................................................................................................... 49
4.1 . IEC FUNCTIONS IMPLEMENTED ............. ..... .. . .. ... .. ...... ........ . .. . .. .. ........... .. .. 49
4.2. SELECTION OF ADDRESSES .............. ........... ....... .. . .. .......... ................. ..... 50
4.3. DATA FORMATS ......................... .................. .. ... .. ......... ........ ................... 51
4.4. PROGRAMMING CODES .............. ...... ........................ ..... .. ................ .. .... .. 53
4.5. PROGRAMMING EXAMPLES ........ ......... ............ .. . ....... .. ....... .. ...... .. ... .. .... .. 53
General Programming .. .. ................ .. .. .. .. .. ...... .. ....... .. .. ....... .... .. .. .... .. ......... 55

Programming with the Tektronix 4051 .......................... ...... ......... .... ......... 56
Programming with the Hewlett-Packard 9825A .......................................... 59
4.6. SPEED OF SPECTRAL READ-OUT FROM THE 2131 .................................... 60
4.7. USE OF THE EXTRA STROBES AND SENSES ............................................. 61
4.8. CONTROL OF 1/12 OCTAVE ................................... .............. ................... 63
5. REMOTE CONTROL OF THE 2131 VIA THE REMOTE CONTROL SOCKET ............................ 65
5.1. REMOTE PROGRAMMING CODES ............................................................ 67
5.2. REMOTE PROGRAMMING AND TIMING DIAGRAMS ................................... 67

FEATURES:
• Real-time 1/3 octave
analysis in 42 channels
with center frequencies
from 1 ,6Hz to 20kHz
• Real-time octave analysis
in 14 channels with center
frequencies from 2 Hz to
16kHz
• Almost entirely digital
construction gives
exceptional operational
flexibility
• Digital true RMS detector
for wide crest factor
capability
• Linear and exponential
averaging with selectable
averaging time
• Exponential averaging with
selectable statistical
accuracy
• Both analog and digital
input and output
• Hold facility allows
instanta neous spectrum to
be held
• Independent memory for
storing a spectrum for
later recall and comparison
with other spectra
• Max. Hold facility allows
maximum level in each
channel to be held
18-020
type 2131
Digital Frequency Analyzer
•••
liili
iii I
.,
.. ~ · ..
Iii . •.•• .
Iii
I
-·
•
• Spectrum shown on a
calibrated, large screen
display with a 60 dB
display range
• Alphanumeric read-out of
channel level and center
frequency directly from the
display
• Pushbutton control for
ease of operation
• All major controls
externally programable
• Interface to IEC interface
bus
• 1I 12 octave analysis
USES:
•
Real-time analysis of
continuous .and impulsive
signals
• Provision of data for further
computation e.g., in aircraft
noise measurement and
certification
• On-line production testing
in quality control
• Phonetics and speech
therapy
• Acoustical research
• Analysis of community
noise
• On-line analysis of
complex signals

Introduction
The Digital Frequency Analyzer
Type 21 31 is designed to measure
and display octave and 1/ 3 octave
spectra in
real-time and may be
used in a
wide variety of analyses
of acoustic, vibration and other sig-
nals. It is almost entirely digital in
operation, in that it uses digital fil-
tering, RMS detection and averag-
ing techniques. The results ob-
tained are displayed on a calibrated,
large screen display, and may be
output to a variety of analog and dig-
ital peripherals.
The 21 31 has two modes of oper-
ation, pushbutton selectable from
the front panel. In the first, it gives
a
real-time measurement of the am-
plitude of the input signal in 42
1/ 3 octave channels having center
frequencies from 1, 6Hz to 20kHz .
In the second, it gives a real-time
measurement in 14 octave chan-
nels having center frequencies from
2 Hz to 16kHz, together _ with a lin-
ear channel. The input signal may
be A-weighted prior to analysis, if
required, using a selectable A-
weighting network in the input am-
plifier. An A-weighted spectrum is
then displayed, and the linear chan-
nel (in octave mode) gives the A-
weighted input signal level.
The digital principles of the 21 31
give
it a series of important opera-
tional advantages. Not least of
these is the extreme ease of use of
the instrument . Almost all of its
functions are pushbutton controlled
from an electronic control panel,
with LED indicators to show their
status . This electronic control in
turn allows an extremely wide
range of functions to be remotely
sensed and controlled over the
21 31 IEC interface, by an IEC inter-
face bus controller. Alternatively,
when the 21 31 is equipped with op-
tional interface ZD 0097, a B & K
Computer Type 7504 could be
used.
Advantages are also apparent in
the display of data, with text being
generated in Read Only Memories
to give a spectral display which is
calibrated in dB and channel num-
ber . The amplitude display range is
60 dB, while the frequency display
range is pushbutton selectable be-
tween 1,6 Hz 1,25 kHz and
2
25Hz - 20kHz, corresponding to
the vibrational and acoustical fre-
quency ranges respectively. Any
change in the input attenuation or
displayed frequency range is accom-
panied by an immediate automatic
adjustment of the calibration. A
channel selector can be used to
read-out the centre frequency and
amplitude of any of the channels
from alphanumeric displays on the
display screen .
It is often useful to be able to
store a spectrum, and then compare
it with another spectrum at a later
time. To allow this, the 2131 con-
tains a digital memory . A spectrum
may be read into this memory, and
then recalled to the display at a .la-
ter time to allow comparison with
new data. The new data may be
held in an additional store to facili-
tate the comparison. This store may
also be used to hold the maximum
level occurring in each channel . In
addition to facilitating the compar-
ison of spectra, the store may be
used
independantly of the memory
in the examination of analyzed data.
Spectra contained in both the me-
mory and the store can be output to
both analog and digital peripherals.
An analog output can be made to
an X/ Y Recorder and to a Level Re-
corder. Digital
input and output is
via
the IEC interface, which has
both a
manual and an addressable
mode. Ma11ual mode allows the in-
put and output of spectra over the
interface to an IEC compatible peri-
pheral, under the control of the
2131 front panel controls . When,
for example, this peripheral is an
IEC compatible digital cassette recor-
der,
this forms a convenient method
for the long term storage of refer-
ence spectra, which may be re-en-
tered into the 21 31 at will. Addres-
sable mode allows the input and
output of both spectra and remote
programming information over the
IEC interface bus. An IEC interface
bus controller is then required .
Many of the features mentioned
above arise as a direct result of the
digital principles of the 21 31. How-
ever, it should also be mentioned
that the techniques themselves add
further features and advantages.
For
instance, a digital filter has a
better controlled filter shape and a
greater freedom from drift than its
analog equivalent. It requires no
special trimming to maintain its pro-
perties as components age. Further,
it is inherently more flexible. How-
ever, probably the most important
advantages are that digital filtering
techniques greatly simplify the use
of a digital detector and a digital av-
erager .
By
using a digital detector, the
21 31 gives true RMS detection
without crest factor limitations,
apart from the natural limitations of
dynamic range and filter response
time. The digital averager, on the
other hand, gives a flexibility impos-
sible to achieve by analog means,
in that it can offer both linear and
exponential averaging . In linear av-
eraging, the detected data is aver-
aged over a
fixed time window,
while in exponential averaging a
continuous average is made. In
both, modes, 13 different averaging
times from 1/ 32 s to 128 s in a bi-
nary sequence, pushbutton selec-
table from the front panel, may be
used. These give
the same averag-
ing
time across all channels, to give
the same response time in each
channel . This is important, e. g., in
the analysis of impulses. Addition-
ally exponential averaging may be
made with a fixed 68% confidence
level, a . This enables averaging
with the same statistical accuracy in
each channel, which is important in
measurements on random signals.
A 68% confidence level of a< 2 dB,
a< 1 dB, or a < 0,5 dB can be push-
button selected from the front
panel.
The 21 31 is provided with an in-
ternal amplitude reference for easy
calibration . Its level is 100 dB RMS
referred to 1 Jl.V. A reference adjust-
ment is also provided enabling the
reference level to 'be adjusted in
steps
of 10 dB through a range of
±50 dB.

Digital Filtering
For a complete explanation of digi-
tal filtering, reference to one of the
standard texts on the subject is rec-
ommended, such as "Digital Pro-
cessing of Signals", by Bernard
Gold and Charles M. Rader, pub-
lished by McGraw-Hill in 1969. A
generalised block diagram of a two-
pole digital filter is shown in Fig .1 .
This filter is of the type known as
recursive, meaning that feedback is
used
such that the output of the fil-
ter is always an explicit function of
the previous inputs and outputs. Its
properties, i. e., its shape, its rela-
tive bandwidth, and whether it is
high-pass, low-pass , bandpass or
bandstop, are a function of the mul-
tiplier coefficients Ao, A1, A2 , 81
and 82. The frequency range in
which it operates is a function of
the delay z- 1 . Assuming that the
addition and multiplication opera-
tions are instantaneous, the delay
z- 1 is equal to one sampling inter-
val. Hence, by altering the sampling
interval, it is possible to alter the
frequency range in which the filter
operates . For instance, doubling the
sampling interval, (i . e., halving the
sampling frequency), will mean that
the filter operates with the same rel-
ative bandwidth but in a frequency
range one octave lower .
The properties of a digital filter
with respect to response time,
phase shift, etc., are virtually the
same as for the equivalent analog
filter . The transfer function is writ-
ten using z-transform notation, and
the transfer function of the filter of
Fig.1 is :
A A
-1 A -2
o + 1z + 2z
H(z) = Ho B
-1 B -2
1 - 1z - 2z
The z-transform is a discrete
equivalent of the Laplace transform
with the z-
1
operator taking the
place of the Laplace operator s.
Conversion between the two can be
made
using the identity
z-1 = exp. (- st).
In the 21 31 filters, a special form
of the z-transform is used, known
as · the matched z-transform, and
the filter block diagram is modified
Signal
Input
Fig.1. Generalised block diagram of a 2-pole recursive digital filter
Signal
Input
Output
760111
Fig.2. Block diagram of 2-pole digital filter used in the 2131
to that of Fig . 2. The transfer func-
tion of this filter is modified with re-
spect
to the general case to be-
come :
H(z) = H 1
-1
-z
o 1 B -1 B -2
- 1z - 2z
In a practical digital filter, the mul-
tiplication processes take a period
of time which is significant when
compared to the sampling interval,
z-1 . The delay is hence modified
to make it plus the multiplication
time equal to the sampling interval.
The input to the filter is a series of
digitised samples representing the
time function. The filter modifies
them to give an output which is a
modified series of samples represen-
ting the filtered time function.
3

Description
General
Direct
Input
Preamp.
Input
Level Recorder
X/Y Recorder
Digital
Outputs
76011 2
Fig .3 shows the basic block dia-
gram of the 21 31 . As can be seen,
it can be divided
into three func-
tional sections . The first of these is
the Input Amplifier and Filter Sec-
tion, where the input signal is condi-
tioned to the correct levels and then
converted into digital form and anal-
ysed. The second is the RMS Detec-
tion and Averaging Section, where
the digital information coming from
the filter unit is squared, averaged
and then logarithmically converted
to the dB RMS level. Finally, there
is the Output Control and Display
Section, which controls the display
of the analysed and averaged data
on
the display screen and the trans-
fer of data to other peripherals, to-
gether with the digital input of data.
In the following , each section is de-
scribed in
turn .
Fig.3 . Basic block diagram of 2131
Input and Filter Section
The block diagram of the Input
and Fi Iter Section of the 2 1 31 is
shown in Fig .4 . It can be conveni-
ently further subdivided into the In-
put Section, the Analog to Digital
Conversion Section and the Filter-
ing Section . The Input Section con-
sists
of the Preamplifier and Input
Attenuator. The Preamplifier has
two inputs. The Preamplifier Input
is of the standard Bruel & Kjcer 7
pin type, allowing the connection of
most Bruel & Kjcer Microphone
Preamplifiers, to which it also sup-
plies power and microphone polari-
zation voltages. The input of signals
from other sources is made via the
Direct Input.
The Input Attenuator governs the
level setting for the rest of the Anal-
yzer. Its value is set from the front
panel controls . It can be used to re-
duce
the input signal level in steps
of 10 dB over a range
of 1 00 dB.
This 1
00 dB range, when coupled
with the 60 dB display range, gives
a
total operating range of 160 dB.
The
setting of the Input Attenuator
is transmitted to the Output Control
and Display Section such that the
dB scale of the display screen is au-
tomatically set. Any input overload
is indicated by a
light intensification
of the displayed data on the display
screen.
From
the Input Attenuator, the
signal is passed to the Analog to
Digital Conversion Section. The first
part of this is an Antialiasing Filter ,
where any high frequency compo-
nents in the input signal which
might cause aliasing distortion
when the signal is later sampled
are filtered out. It is a 12-pole But-
terworth low-pass filter ·, which
gives a maximally flat passband,
and has a
cut-off frequency of
27kHz and a roll-off of
72 dB/octave. The filtered signal is
entered into the Sample and Hold,
where it is sampled at a constant
frequency of 66,667 kHz. Each sam-
ple taken is then converted into a
1
2-bit two 's complement coded
word by the ADC, (Analog to Digital
Converter).
Effectively,
it is the sampling fre-
quency which sets the frequency
axis of the 21 31 . This means that it
is essentia1 that this frequency
should be accurately controlled.
Hence, 1t IS referenced to a 10 MHz
crystal controlled Master Oscillator.
The frequency of this Oscillator is
so
accurately controlled that any fre:..
quency axis calibration of the 21 31
is rendered
unnecessary . All timing
signals used in the 21 31 are refer-
enced to this same oscillator .
The remainder of the Input and
Filter Section of the 21 31 (exclud-
ing the Control and Timing). forms
the Filtering Section . In this, there
are five of the basic 2-pole digital fil-
ter units shown in Fig.2, three of
Input Section Analog
to Digital Conver si on
Section Filtering Section
Direct
Input
Preamp
Input
4
()iloo
.()iloo Preamp.
~
Signal line
Control line
rnpu .
information to Display L
Sample Bandpass 2 Pole
Input
~
Ant i al iasing
~
and
~
ADC
""""
~ Filter
~
Bandpass
~
Att . Fi
lter
Fi lter
Hold
r~ulti~exer
Unit
+
t Att setti ng
_[
I I I •
10 MHz
~
Lowpass 2 Pole
Master
I+
Control and Timing
,......
Filter
~
Lowpass
~
Filter
Oscillator rMultiplexer
Unit
I
Fig.4 . Block diagram of input and filter section of 2131
2 Pole
Bandpass
~
Filter
Unit
2 Pole
Lowpass
~
Fi lter
Unit
~
2 Pole U
Bandpass~
Filter
Unit
Buffer
~
Memory
-
Linear Channel
to RMS Detector
Analyzed Data
to R MS Detector
To Detector
Averager and
Output Control
760109

which are designed to operate as
bandpass
filters, the other two be-
ing
low-pass filters. The three band-
pass
filters are cascaded to form a
single 6-pole unit, henceforth re-
ferred to as the Bandpass Filter,
and the two low-pass filters are cas-
caded
to form a single 4-pole unit,
henceforth referred to as the Low-
pass Filter . (This Low-pass Filter is
used as an
antialiasing filter as the
sampling frequency is reduced). In
front of each of the two units, there
is a multiplexer used to control the
passage of information into the fil-
ters. There is also a Buffer Memory
for storage of partial results.
The ADC feeds a series of digi-
tised samples representing the time
signal into the filters, taken at a fre-
quency which is precisely controlled
to be 66,667 kHz.
Looking
first at the operation
within the highest octave, i.e. the
16kHz octave, all of these samples
pass through the Bandpass Filter
Multiplexer into the Bandpass Fil-
ter. Assuming that the Analyzer is
set
to measure in 1/3 octaves,
then the Bandpass Filter performs
three different calculations on each
sample to produce three new sam-
ples representing the 1/3 octave fil-
tered levels within the octave. The
change from one 1/3 octave to
another 1/ 3 octave is effected by
changing the multiplier coefficients
used within the filter.
Operation within the lower oc-
taves is
obtained by using the same
multiplier coefficients as in the high-
est octave, but with a successively
reduced sampling frequency. For in-
stance, halving the sampling fre-
quency with respect to the 16kHz
octave will produce the 8kHz oc-
tave
and halving it again will pro-
duce the 4kHz octave. Each time
the sampling frequency is reduced
it is
accompanied by a low-pass fil-
tering operation to correspondingly
reduce the high frequency cut-off of
the samples, and prevent the intro-
duction of aliasing distortion. This
operation is carried out by the digi-
tal Low-pass Filter.
Simultaneously with being en-
tered into the Bandpass Filter, each
digitised sample produced by the
ADC is passed through the Low-
pass Filter Multiplexer into the Low-
pass Filter. After it has passed
through the filter, it is fed back to
the Low-pass Filter Multiplexer for
another passage. This is repeated a
further time, such that each sample
passes through the 4-pole Low-
pass Filter a total of three times to
produce 1 2-pole filtering.
For this first low-pass filtering op-
eration, the cut-off frequency of the
filter is set to half that of the Antiali-
asing Filter . Hence, the maximum
frequency which each sample can
represent is reduced by half. This al-
lows every second sample to be dis-
carded, thus halving the sampling
frequency. The samples retained for
further filtering operations are en-
tered into the Buffer Memory.
When they are later retrieved, then
because the sampling frequency
has been halved, in the Bandpass
Filter, they will produce the next oc-
tave of information, and in the Low-
pass Filter, the Filter cut-off fre-
quency will be halved, permitting a
further halving of the sampling fre-
quency.
The 66,667 kHz sampling fre-
quency means that there is a new
sample every 15 J.IS. To maintain
real-time operation, in this time,
the filter unit must calculate the re-
sults in the highest octave and in
one other octave. This involves six
1/ 3 octave
filtering operations and
six low-pass filtering operations
(since the samples pass around the
low-pass filter three times). Hence,
each of the 2-pole filtering opera-
tions must be completed in less
than 2,5 J.IS. The filter unit then pro-
duces six filtered samples for each
sample that is entered into it, at a
rate of one every 2,5 J.IS . The sam-
ples from the filter output are
passed
to the Detection and Averag-
ing Section for RMS detection,
along with the unfiltered samples
to give the linear channel.
Octave operation of the Filtering
Section is similar to 1/ 3 octave op-
eration, except that after their first
passage through the Bandpass Fil-
ter, the samples are re-entered into
the Bandpass Filter Multiplexer for
a second passage. This double pas-
sage
through the 6-pole unit gives
12-pole filtering.
The 1I 3 octave filters of the
2131 conform to IEC 225-1 966,
DIN 45 652 and ANSI S1. 11, Class
Ill. In octave
operation, they con-
form to IEC 225 1 966, DIN 45 652
and ANSI S1.11, Class II. The
shape
of the octave filter is shown
in Fig.5, and a typical 1/ 3 octave fil-
ter, within an octave, in Fig.6. They
are plotted in terms of relative fre-
quency, since their shape is inde-
pendent of the frequency range in
which they operate .
c
.Q
"'
:::J
§
:;_
I I
ANSI
1
11
~ I \
Minimum
Limit
II
\
Class II
/I; 1/ I\ I\
1\
\\
0
dB
10
20
I 1\ \
I/B&K w ' I\
1/ "
30
40
I
I 1\
v '
50
60
0,5 2
frequency relative to centre frequency f/fm
760114
Fig.5. Octave filter characteristic of the
2131 analyzer
0
dB
10
20
30
40
50
60
I
II/
I I
//
I' if
I J
1/ 1/
/ J
v v
J
0,5
\ I II
ANSI
Minimum
Jl
Limit
1--
\ I; Class Ill
I\\
1 \,
I\,
[\
B&K -
~
~
\\
1\ \
1\
I\
2
frequency relative to centre frequency f/fm
760115
Fig.6. 1/ 3 octave filter characteristic of the
2131 analyzer
5

Lin ear:_ Channel
Analyzed Data
from Filter Unit
-Sign al Lin e
--C ontrol Line
Fr
om Ma ster
Oscillator
Detector
Averaging Time
Averaging Section
68 % C onfidence
Level
Lin /Log Conversion
And Square Root
Extraction
Log RMS
Output to
Display
Control
760 113
Fig.7 . Block diagram of detection and averaging sections of 2131
The Filtering Section of the 21 31
operates on a 15 bit word basis.
The
multiplier coefficients are en-
coded in 9 bits . Hence, each multi-
plication produces a 24 bit word ,
which is then rounded off to 15 bits
again .
RMS Detection and Averaging
Section
The block diagram of this section
is shown in Fig .7. The 15 bit serial
data coming from the Filtering Sec-
tion first enters a Serial to Parallel
Converter , which, in addition to con-
verting the data into parallel form,
rounds it off to 13 bits, (12 bits of
information plus sign) . This is then
entered into the Parallel Buffer to
await squaring . After squaring in
the Squaring Circuit, there are 26
bits of information . However, since
two of these bits are sign bits, and
the output of the Squaring Circuit
can only be positive, they are dis-
carded to produce a 24 bit word,
which is entered into the Averaging
Section.
Of the information entering the
Averaging Section, the highest oc-
tave represents one half of the fre-
quency range and one half of the
samples . (Remember that in one
sampling period of the ADC in the
Input and Filter Section, the Filter-
ing Section calculates the results in
the highest octave and one other oc-
tave.)
This makes it convenient to
use two Averagers , one which oper-
ates on the highest octave only, the
6
Single Octave Averager, and one
which operates on all lower oc-
taves, the Multi Octave Averager.
The information is switched be-
tween the two using the Single Oc-
tave/
Multi Octave Multiplexer, and
before each of them, there is a Buf-
fer for the temporary storage of in-
formation. Note that the linear chan-
nel, (scaled up from 12 to 15 bits
before its entry into the Serial to
Parallel Converter), is only detected
and averaged when the 21 31 is
measuring in octaves, when it takes
the place of the highest 1/ 3 octave
in the single Octave Averager.
KA,_1
Control
KA, _1 + T,
The block diagram of an Averager
is shown in Fig.8. Each Random Ac-
cess Memory, (RAM), is used to
store the current average within a
particular 1/ 3 octave. While the
Single Octave Averager needs only
one bank of three RAMs, one RAM
for each 1/ 3 octave on which it op-
erates, the Multi Octave Averager
must have 13 such banks, since it
operates on 13 octaves . In all other
respects, however , the two Averag-
ers are identical. The averaging it-
self is carried out using the Adder,
(I), the Subtractor, (-1:), the Tap
Register,
and the Tap Selector .
-Signal line
-- Control line
Switch
14------111111() --o---t---~1'-.,. Output
KA ,_
1 + T
,- A, _1
Tap Selector
Tap Register
760108
Fig .8 . Principle of averager

Depending on the position · of the
switch at the -r input, the Aver-
ager can operate in a linear or ex-
ponential mode. The two modes are
compared in Fig.9, which shows
the same series of samples, (solid
line),
averaged both linearly, (dotted
and dashed line), and exponentially,
(dashed line).
In Linear Averaging, the average
is
carried out over a fixed period of
time, the Averaging Time. Each
sample entering the Averager is
then divided by a factor K, which is
chosen to correspond as closely as
possible to the number of samples
occurring during the Averaging
Time. 1/ Kth part of each sample is
summed until the Averaging Time is
complete, when the average is term-
inated. If the value of the rth sam-
ple taken during the Averaging
Time is Tr, and the sum of 1/ Kth
part of each of the previous sam-
ples is Ar-1 , then the process may
be represented by the following
equation :
Tr
Ar-1 +-
K
Due to purely technical reasons,
however, it is easier to produce the
circuitry for this process if the for-
mula is multiplied throughout by K
to give:
which is the formula used in the
2131 for this calculation.
In Fig. 9, K has been arbitrarily
set to 10, and the averaging period
is terminated when r is equal to 10.
The value of Ar, at r = 10, (i . e.,
the average vaIue of the first 10
samples) is then held by the Aver-
ager, all further samples being ig-
nored .
In Exponential Averaging, a con-
tinuous averaging process is ob-
tained using the equation:
T r- Ar-1
Ar-1 + -----
K
Here, new values of Ar are calcu-
lated for each sample using the
previous average and the newest
15
~ 14
-~
0. 13
E
<t: 12
11
10
9
8
6
End of Linear Averaging Period
· ~ . -<>-. -<>- ·-<>-·-o-·-<>-.-<>- ·~· -o
--Instantaneous Values
_ x- Exponentially Averaged
x Values
\ ·~·Linearly Averaged Values
\ X
\ \X
\ \ X 'x, >
\.
'
x-x
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time
160106
Fig.9. Comparison of linear and exponential averaging
sample, irrespective of the value of
r. In Fig . 9, K has been set to 5,
since it gives the same Averaging
Time as the Linear Average. At r =
1 0, the Exponential Average and
the Linear Average produce results
having the same statistical accu-
racy, (but not necessarily the same
numerical value) . However, the Ex-
ponential Average goes on to calcu-
late new values of A r, for r > 1 0, ex-
hibiting an exponential decay when
the values of Tr drop to zero.
In Linear Averaging, the switch
shown in the block diagram of
Fig.8, is open. The Averager accu-
mulates KAr, rather than Ar, and as
a
new value . Tr, enters the Adder,
KA r- 1 is retrieved from the RAM
corresponding to the channel being
averaged, and passes through the
Tap Register to the Adder, which
outputs KAr-1 + Tr. At the same
time, KAr - 1 passes through the
Tap Selector, which divides out K to
give Ar -1 . However, although
A r-1 can be transferred to the Av-
erager output , it cannot enter the
Subtractor, because the switch is
open . Hence, KAr-1 + Tr flows
through the Subtractor unmodified
to be entered into its respective
RAM as KAr.
In Exponential Averaging, the ac-
tion of the Averager is similar ex-
cept that the switch is closed .
Hence, when KAr - 1 + Tr flows
through the Subtractor Ar - 1 is sub-
tracted, from it to give KAr-1 + Tr
-
Ar-1 . This becomes KAr and is
entered into the respective RAM.
Both Exponential and Linear Aver-
aging can be carried out with a con-
stant averaging time across all chan-
nels . This can be varied from
1/ 32 s to 1 28 s, in a binary se-
quence, and can be selected from
the front panel controls .
A Linear Average stops automati-
cally when the chosen averaging
time has elapsed, after which the
Averager continuously outputs the
value of Ar at that time, until the be-
ginning of a new averaging cycle in-
itiated from the controls . Provided
that the same averaging time is
used, A r
may be used as the start-
ing point for the new averaging cy-
cle, the new results being added to
those already obtained . More nor-
mally, however, the Averager is
reset to zero before the start of a
new cycle. This . allows a different
averaging time to be selected, if re-
quired . If needed, pauses may be in-
troduced into the averaging cycle, al-
lowing, e. g., the change of signal
sources .
With Exponential Averaging, the
averaging time may be changed at
will while the average is in prog-
ress. The average will only stop
when this is requested from the
front panel controls of the 21 31 .
7

Gives the amplitude of the selected
channel
Gives the center frequency of the
selected channel
Display of spectrum calibrated in dB
and
channel number. Intensified co-
lumn shows which channel is se-
lected by the channel selector. The
spectrum shown is the reference
signal
dB line calibration automatically fol-
lowing attenuator setting
Channel numbers in accordance
with chosen frequency range
For
entry of a signal from co-axial
sources
For entry of a signal from most
B & K Microphone Preamplifiers
Allows adjustment of the reference
level in steps of 10 dB
Preamp . Input
r
Gives continuous adjustment of the
input gain
8
R...,_ Ad;lm . dB Input
0
•
Selects whether the signal analyzed
comes from the Preamplifier or the
Direct Input
Internal amplitude reference for cali-
bration
Allows selection of the Averaging..___
Time in Linear and Exponential Av-
eraging

aging
DigQIF~A~
Type2131
f""IU~tneY Renge
Kz
Input/OutpUt Control
r-- Input ---,r-- OutpUt ------,
..
• •
-----·
_J
.....---- OutpUt Spectt~~m Seleetor-.,
lf1Put A~ Stored
SfQw
,- Input Spectrum --, ,...-- l\lernory--,
Function Function
Cont. Hold Mex . Hold Store Protect
•
I I I I
--
Scale 8rightr14tu
Avereging Control
Reeet Proc;eed Stop
•
I I
_._
68% Confidence Level
0' < 0, 5d8 u < 1 dB o < 2dB
1 -J .
Intensity
3
•
System ReMI
Linear or Exponential Aver-
Attenuates the input signal in steps
of 10 dB
A-weights the input signal
Selects
whether the 21 31 mea-
sures in octaves or 1/3 octaves
Selects which frequency range is
displayed by the 21 31
Controls the output of data
Selects whether data entering Input
Spectrum Store is analyzed data
from the analog inputs or a digital
read-in from the digital input
Controls whether the display and
outputs are connected to the Input
Spectrum Store or the Memory
Controls the entry of data into the
Memory
Controls the entry of data into the
Input Spectrum Store
For
selecting a channel on the dis-
play
For control of the appearence of the
display
Resets the 21 31 always to the
same combination of function set-
tings
For control of Averaging
Selects 68% Confidence Level in Ex-
ponential Averaging
9

Selection of the value of K for a
particular averaging time is made
using the Octave Address Modifier,
(see Fig.7). However, since the num-
ber of samples occurring over a set
period of time will differ from oc-
tave to octave, it is not possible to
use the same value of K in all chan-
nels . Hence, the value of the averag-
ing time and the address of the oc-
tave being averaged are entered
into the Octave Address Modifier,
which then generates the correct
value of K and enters it into the Av-
erage Control.
In addition to averaging with a
constant averaging time, Exponen-
tial Averaging also gives averaging
with a constant 68% Confidence Le-
vel. The 68% Confidence Level, a,
is defined by the - equation:
1
a =
2~
where B is the channel bandwidth
and TA is the averaging time. In a
measurement on random data,
there is then a 68% chance that the
measured results will be within ±a
of the true mean value, and a 96%
chance that it is within ± 2 a. For
this mode of averaging TA is varied
to give a < 2,0 dB, a < 1 ,0 dB, and
a < 0,5 dB. Selection of the value of
a is from the front panel controls .
The values of K needed to produce
this are now read into the Average
Control from the Fixed Confidence
Programmable Read Only Memory,
(PROM).
The output of each Averager is en-
tered into a Slave Memory . The
Slave Memories are necessary be-
cause each time a spectrum is trans-
ferred to the Output and Display
Control the output of the Detection
and Averaging ~ection must be held
constant for approximately 2 ms. To
take this output directly from the
Averager would require that all aver-
aging ceased for this period. Since
the Output and Display Control is
updated with a new spectrum every
22 ms, it would mean that only ap-
proximately 90% of the information
being detected would be averaged.
Use of the Slave Memories enables
100% averaging .
The Averagers operate using a 48
bit word. The information that they
10
produce is read into the Slave Mem-
ories in the order of the highest oc-
tave, least significant bit first. At
the same time, the number of bits
is reduced to 24 . When the informa-
tion is read out to the Lin/Log Con-
version and Square Root Unit, this
order must be reversed to be lowest
octave, most significant bit first.
This reversal is controlled by the
Slave Address Modifier. With each
transfer of a spectrum to the Output
Control and Display Section, the
Slave Memories are isolated from
their respective Averagers. The
Slave Address Modifier then con-
trols the flow of data into the Lin /
Log
Conversion and Square Root
Unit, which in turn generates from
this data the RMS spectrum .
Because the 21 31 uses an all dig-
ital Detection and Averaging Sec-
tion, the only crest factor limitations
experienced are the natural limita-
tions of filter response time and dy-
namic range. For the Detector
alone, then the crest factor perfor-
mance is purely a question of
where the signal being detected lies
in the dynamic range. The RMS De-
tector is built to have a 9 dB Crest
factor capability for signals which
are at full scale, i. e. a pure sirie
wave can be 6 dB over full scale.
The crest factor capability will in-
crease with decreasing signal level
to 69 dB at the bottom of the dy-
namic range.
The use of a digital Detection and
Averaging Section in the 21 31 is
greatly simplified by the use of digi-
tal filters. Use of such a Detector
and Averager with analog filters
would be extremely difficult, since
in order to have suffi'cient dynamic
range and in order to obtain sam-
pling frequencies as required by
sampling theory, it would require a
13 bit ADC operating with a sam-
pling frequency of some 400kHz.
Use of digital filters, however, al-
lows the use of a fairly low-cost
ADC, operating with a sampling fre-
quency of 66,667 kHz.
Output Control and Display
Section
The block diagram of the Output
Control and Display Section of the
21 31 is shown in Fig. 1 0. The Dis-
play is updated with a new spec-
trum from the Detection and Averag-
ing Section every 22 ms . This rate
of updating is determined by the
frame frequency of the display
screen, which is 45,5 Hz. Each
spectrum entering the Output Con-
trol and Display Section is repre-
sented by a series of counts, each
count being proportional to the log
RMS amplitude in a particular chan-
nel. These enter the BCD Counter,
which converts each one into a dB
level between 0 and 66,0, encoded
in BCD. The Adder converts these
levels for the Input Attenuator and
Reference Adjustment settings. As-
suming that the input gain of the
2131 is correctly adjusted, (this can
be controlled using an internal refer-
ence), and that the Reference Ad-
justment is set to 0 dB, the output
of the Adder will be a series of BCD
r.1umbers
representing the analyzed
spectrum in dB referenced to 1f.lV.
The Reference Adjustment allows
the reference level to be varied in
10dB steps by up to ±50dB. Note,
however, that levels below 1 11V
RMS cannot be displayed. Assum-
ing that analog input is selected,
the spectrum passes through the
Analog / Digital Input Selector into
RAM I.
RAM is the Input Spectrum
Store of the 21 31 . Its status is set
from the Input Spectrum Function
controls on the front panel. When
set to Continuous Update, it is up-
dated every 22 ms with the latest
spectrum to come from the Detec-
tion and Averaging Section. When
it is set to Hold, its input is isolated
such that whichever spectrum is in
it is stored . When it is set to Maxi-
mum Hold, the spectrum stored in it
is compared, channel by channel,
with each new incoming spectrum.
Any channel where a higher level
than that in the stored spectrum is
found is updated. Hence, a spec-
trum consisting of the maximum le-
vel to have occurred in each chan-
nel is formed .
When Analog / Digital Input Selec-
tor is set to Digital Input, a spec-
trum can be read into RAM I from a
digital input. It hence becomes pos-
sible to hold spectra such as refer-
ence spectra, in digital form exter-
nally to the 21 31, and then read
them back into RAM I as required.
Any spectrum in RAM I can be
transferred into RAM II. RAM II is
the Memory of the 21 31, and it is

Channel
Selector
---+
Selected
Ch annel r-
j_
10 MHz__.,
Display -
Clock Contr ol
Counter
{J
Inten sify Column
Ra ster and
Com-
Sync. signals
para tor
Frequency
Range Di
sp lay
II
Raster
Column
Li ght
Column
Display
Front
Panel
---+
Count er
s''""d ~
Scaling
Control
lig~
Screen
Control
Input from
D etector
Input Att .
and Ref.
Adju st Sett
-----..
_____.,
ing
Channel
Address
BCD
~Multiple xer - rv' ~ ext
Counter
00,0 -66,0 ~
om
D Data
Addr ess
Adder
Bu s
000,0 - -
150,0
D L '
Analog/
RAM I
P
ara ll el to
Serial Digital
~
----
Input
Converter
____..
RAM II
Selector +
Input Spectrum_____.
Function
RAM
Mem ory Fu nction ----.
Control
~ Isolation Buffer
Character
Address Bu s
~
Character
Rom
Input /
Output
Update
Counter
Level
Recorder
Control
i
Button _
Addr ess
Te xt
T ext
Counter
~
I I
ASCII
Serial
to
Bit Serial "
Converter
Bit Serial
to ASCII
Serial
I . Converter,
Form atter
Log
to Lin X!Y
Converter
DAC
Recorder
DAC Cont rol
I
H
~~Button
Addr
Group
ess
...
Remote
Control
~
I"
>
X/Y Ramp
Y Level
Remote
Control
ASCII or
Byte
Serial Input
ASCII Serial
Output
Level Recorder
Output
L__ __________________ ___..Level Recorder Control
' 760107
Fig.10 . Block diagram of output control and display section of the 2131
intended for the long term storage
of spectra. Transfer of •a spectrum
into the Memory leaves the Input
Spectrum Store free for further updat-
ing. Additionally, the Memory Con-
tents can be protected to prevent ac-
cidental erasure . The protection is
controlled from the Memory Func-
tion pushbuttons on the 21 31 front
panel, and when selected, it is
maintained for as long as the power
is switched on.
The output to the Display screen
and to the electrical outputs of the
21 31 can be taken from either of
the RAMs . It is also possible to alter-
nate the output between the two, al-
lowing , for instance, the compar-
ison of a reference spectrum stored
in the Memory with new spectra
held in the Input Spectrum Store.
Two alternating frequencies are pro-
vided, a fast one with an 88 ms dis-
play cycle (shown in Fig. 11). such
that the contents of the two RAMs
appear almost superimposed, and a
slow one with a display cycle of ap-
proximately 3,5 s. The display cycle
is
split between the two RAMs in
the ratio 3:1, so that it is always
possible to distinguish between the
two . The slow alternating frequency
also permits the simultaneous out-
put of the contents of both RAMs to
an X/ Y- or Level Recorder . The out-
put is then in the form of a continu-
ous line showing the envelopes of
both spectra .
Generation of the display on the
display screen takes place in a ser-
ies of Counters and Read Only Mem-
ories , (ROMs). under control of the
Display Control. The spectrum en-
ters via the Column Light Control
and the Raster Scaling, which set
the height of the columns represen-
ting each channel on the display
screen and the vertical scale respec-
tively. The horizontal scale, (i.e .
1,6Hz 1,25kHz or 25Hz-
20kHz). is set according to the Fre-
quency Range selector. The ampli-
tude display range is 60 dB, and cal-
ibrated grid lines are generated to
allow a quick read-out of the level
in all channels . More accurate read-
outs can be obtained using the
Channel Selector . The selected
channel is indicated with an intensi-
fied column, and its level and cen-
tre frequency can be read from al-
phanumeric displays on the display
screen . Fig. 12 shows an example
of an octave spectrum display.
Analog outputs of the spectrum
can be made to a Level Recorder
and to an X/ Y Recorder.. The output
to the Level Recorder is written on
frequency calibrated paper. The
speed of the output is controlled by
the Level Recorder . For an X/ Y Re-
corder, a staircase signal is pro-
vided to give the X-deflection . The
period of the staircase is 45 s for an
output from 25Hz to 20.kHz and
60s for an output from 1, 6Hz to
20kHz . Frequency and amplitude
calibration signals are provided.
Note that when an output is made
with an alternating display, the
speed of the output is slowed by a
factor of 8, or a factor 4 if a small
internal modification is made . Note
also that during all outputs (both an-
alog and digital) , an internal hold
function is generated for the period
of the output.
11

Displayed spectra may also be out-
put digitally to a range of digital
equipment via the Digital In/ Out
socket. The output is encoded in
ISO 7-bit code, (i . e., ASCII, but
without the parity bit). On request
of a digital output, the internal hold
function is immediately activated.
However, it is then necessary to
wait for up to 22 ms for the begin-
ning of the display update cycle.
Once the cycle has begun, it be-
comes possible
to read-out a spec-
trum at a maximum speed of 60 f..iS
per channel. Where read-out is to
slower peripherals, the normal sig-
nals are
available to control the
read-out at a slower rate. Note that
it is necessary to wait a minimum
of 44 ms between two digital read-
outs
to ensure that the display
screen has been updated .
When the 21 31 is connected to a
computer, a very wide range of rem-
ote sense and control functions be-
Fig.11. Alternating display of two spectra using "Fast" mode
Fig.12. Typical octave spectrum as displayed on the screen of the analyzer
12
come available . These are entered
via the Digital In/ Out socket in the
form of a code, and are decoded
into their actual functions in the
remote control and front panel con-
trol . Alternatively, after a small in-
ternal modification, these signals
may be entered via a separate Re-
mote Control socket.
In the standard version of the
21 31, an IEC Standard Interface is
included, enabling the connection
of other IEC compatible equipment .
However, it is also possible to sup-
ply interface card ZD 0097 as an op-
tion, enabling the connection of a
Bruel & Kjrer Computer Type 7504 ,
a Tape Punch Type 6301 , or a Tape
Reader Type 71
02 .
e u
~ I
e t u
l<>®r!d
·-
C;«:KU ..
••
··~
IA>S0l'Z02.t.:JV
lt>"-100 11! •nv
u.t
Fig.13. Split view of rear panel showing
plugs and sockets

Peripheral Equipment For 21 31
Connection of Equipment over
IEC Interface
The 21 31 is equipped with a digi -
tal interface designed according to
the IEC proposal , "Interface for Pro-
grammable Measuring Apparatus
Byte-Serial Bit - Parallel" , TC 66
(Central Office) 22 . This allows it to
be connected to other instrumenta -
tion having similar interface . Fur -
ther , since the only significant differ -
ence between the IEC interface and
one de signed according to IEEE
standard 488 - 1975 / ANSI MC 1.1-
1975 is th e type of connector used,
connection to instrumentation so
equipped is also a possibility
Of particular interest is the con-
nection of a desk-top calculator to
the 21 31 over the IEC interfa ce.
The calculator can then operate on
data supplied by the 21 31 in auto-
matic test sequences . the syst em
forming a form i dable acoustic mea-
suring package .
Bri..iel & Kj<Er have surveyed the
range of IEC and IEEE compatible
desk - top calculators available, and
have decided to recommend two for
use with the 21 31 . These are the
Hewlett-Packard 9825A, and the
Tektronix 4051 . Connection to the
9825A requires Hewlett-Packard op-
tions 98034A , 9821 OA and
9821 3A . The minimum recom-
mended memory size is 1 6K . Con-
nection to the 4051 is direct. with
no Tektronix options being neces-
sary . Here . a 32K memory is recom-
mended .
Two general purpose programs
have been produced by Bri..iel &
Kj<Er to enable basic acoustical cal-
culations to be made by the above
calculators on data supplied by the
2131 . They are BZ 0011, for the
4051, and BZ 0012 , for the
9825A . BZ 0011 contains the fol-
lowing programs :
1. Input of spectra from 21 31 to
the 4051 into either of two
buffers at preset time inter-
vals, (176 ms minimum) .
2.
Average of spectra in either
buffer .
3. Di
fference between the aver-
age spectrum in each buffer .
4 .
Calculation of the perceived
noise level with tone correc -
tion after ISO/DIS 3891 .
5 ,. Calculation of Stevens loud -
ness .
6. 3
dimensional display of up to
65 1I 3 octave spectra.
7 . Calculation of reverberation
time .
8.
Calculation of sound power ac-
cording to ISO 3741 .
9 . Calculation of Leq , Lnp , TNI ,
L10. l5o . Lgo and Lxx (xx
user definable , e. g., L 17) .
10 . Control of 1I 12 octave .
BZ
0012 contains similar programs,
with the exception of the 3 dimen-
sional program , which is absent .
Note that the minimum read -in in-
terval between spec tra is now
44 ms .
Connection between 21 31 and a
4051 or a 9825A is made using
cable AO 0157 .
Alphanumeric Printer Type 231 2
The Alphanumeric Printer Type
2312 is shown in Fig. 14 . It may be
connected to the 21 31 over the IEC
interface , and used to obtain a hard
copy of a spectrum in the form of a
Fig . 14 . B & K Alphanumeric Printer Type
2312
print-out giving the level in dB in
each channel, such print-outs being
initiated over the front-panel con -
trols of the 21 31 .
The 21 31 can output data in
either of two formats . On delivery ,
the simpler format is selected, in or-
der to give the faster read - out. In
Fig.15. B & K/ Varian Computer Type 7504
outputs to the 231 2 . however. it
can be an advantage to select the
second data format , which will give
a more " readable" print-out with no
significant increase in the time
taken to generate it .
Computer Type 7504
The optional interface card ZD
0097 allows direct connection of a
21 31
to a Computer Type 7504 .
The Computer Type 7504 is shown
in Fig . 15 . In its standard version , it
has a 4K memory . The memory can
be expanded in modules of 4K up to
32K. A two pass Assembler is sup-
plied with both the 4K and 8K ver-
sions , while with the 8K version , in
addition , there are compilers for
FORTRAN and BASIC , expanding
the programming methods avail-
able .
Bri..iel & Kj<Er
can deliver two gen -
eral purpose programs for use in a
7504 when it is connected to a
21 31 .
The first of these is BZ
0008 , which allows the front panel
pushbuttons of the 21 31 to be rem-
otely controlled from the 7504 us-
ing FORTRAN. The second , BZ
0009, gives several general pur-
pose routines, namely :
1 . Input and output of spectra.
with autoranging of the 21 31
input .
2. Averaging of spectra .
3. Difference
spectra.
between
two
4. Calculation of Stevens loud-
ness .
5. Calculation of perceived noise
level with tone correction, for
a single spectrum .
Note that if it required to use the
Computer Type 7504 with a rem-
otely controlled 21 31 and a Tape
Punch Type 6301 operating in paral-
13

lei, it is necessary to equip the
7504 with an extra interface card
ZD
0057 having modified address
codes
for connection of the 6301.
However, assuming that the
punched outputs are relatively short
in nature, (e.g., reference spectra),
it should be remembered that they
can always be punched using a Te-
letypewriter Type 6401 , with no
extra
interface being necessary.
Optional interface card ZD 0097
also allows direct connection of the
Tape Reader Type 7102, and the
Tape Punch Type 6301, for the digi-
tal input and output of spectra, re-
spectively.
Examples of Use
Measurement of Sound Power
The · measurement of Sound
Power is the subject of ISO 3741
through 3745. Of these docu-
ments, ISO 3741 and 3742 have al-
ready been approved as standards,
while ISO 3743 through 3745 re-
main in the draft form. They specify
the measurement of Sound Power
under various conditions and to var-
ying degrees of accuracy. Although
they do not deprecate the use of ex-
ponential or RC averaging, these
documents are unanimous in recom-
mending the use of linear averaging
in the measurement of Sound
Power.
A 21 31 connected to the Rota-
ting Microphone Boom Type 3923
makes an ideal system for measur-
ing Sound Power according to these
standards and draft standards. In
them, the sound field emitted by
the object under investigation can
either be measured using an array
of microphones set up at predeter-
mined positions around the object,
or using a single microphone which
is traversed around the object on a
fixed path. A system comprising the
3923 and the 21 31 uses the latter
method. The greatest advantage of
this method is that the greater part
of the spatial averaging of the
sound field is carried out automati-
cally by the linear averaging pro-
cess of the 2131. At the end of
the measurement procedure, the
engineer is no longer left with a
complex and lengthy calculation pro-
cess requiring the use of a minicom-
14
Digital Frequency
Analyzer 2131
•-
~
..
• I 1_ _. • =I
--·
760908
Fig.16. Use of the Rotating Microphone Boom Type 3923 with the 2131 in Sound Power
Measurement
BrUel & Kjer BrUel & Kjaar BrUel & Kjatr
D DDDDODDDDDDDDDDDDDDDDDOOODDDDDDDDDDDDODODOOD~DDO
Bri.iel & KjCBr
100
Potentiometer Range :_ dB Rectifier -- Lower lim Freq .:_ _ Hz Wr Speed :- mm / sec PaperSpeed : -mm / sec
Copenhagen
Measuring Object : 90
- -- - 80
A - weighted level:-=
50 Hz 91 ,5 dBE
-- - - 70 60 Hz 95 2
dB~
Rec . No .: __
Date : ___ 60
Sign .:_ _ _
3347 :
Time Const .:
Weight . Netw .: _
Ref . Level:_ _
50
Th i rd Octave Band 16 20 31 . 540 63 80 12~60 250315 50~30 1 k~/5 2 2 .5 4
P
115~enter Frequency12 ,5 25 50 100 200 400 800 1.6 3, 15
10
20 40 Lon .
8
16 31 ,5 Weight . 740 74 2
6,3 12 .5 25 1e15 1 21n 0 A 8 C Lin .
Fig.17. Typical sound power spectrum of Reference Sound Source Type 4204, used in Sound
Power measurements with the 2131 in the comparison method
puter , but with simple data manipu-
lations which can be easily handled
using a relatively unsophisticated
pocket calculator, or which can
even be
carried out by hand.
The major requirement for the
use of a moving microphone for
Sound Power Measurements is that
the microphone should make one or
more complete traverses around the
sound source during one linear aver-
aging period.
The path of the tra-
verse, which is usually circular
around a sphere or hemisphere hav-
ing the sound source at its centre,
is specified by the relevant docu-
ment . The 3923 can be used to
give circular traverses having vary-
ing diameters and periods of 16 s,
32 s, or 64 s. Hence, it is only a
matter of selecting the appropriate
linear averaging time on the 21 31,
and synchronizing the start of the
3923 traverse with the start of the
averaging period on the 21 31 be-
fore these requirements are ful-
filled. At the end of the averaging
period, the 21 31 will display a spec-
trum of the average sound field
along the path of the microphone.
At the end of the measurement
procedure, a number of spectra will
result, each one corresponding to a
different traverse path. These
spectra may be averaged externally
to the 2131, or internally, by ad-
ding successive averages and then
correcting the spectral levels ob-
tained for the number of spectra
added,
(note that this process is not
recommended for large numbers of
spectra, since noise can begin to be-
come significant). The Sound Power
of the object under investigation
may then be calculated from this re-
sulting spectrum, various correc-
tions being taken into account ac-
cording to the method in use.
Hence, by
using the 21 31 with the
3923, the complexities of Sound
Power measurement are considera-
bly reduced.
Under reverberant conditions, a
further stage of simplification is pos-
sible by using the Reference Sound
Source Type 4204. This gives a
known Sound Power, and allows •
the comparison method of measure-
ment, specified in ISO 3741, to be
used.
Calculator Controlled Operation
The wide range of remote sense
and
control facilities available of the
21 31, and its IEC interface make