This is only a preview of the June 1995 issue of Silicon Chip. You can view 27 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Build A Satellite TV Receiver; Pt.2":
Articles in this series:
Items relevant to "A Train Detector For Model Railways":
Items relevant to "A 1-Watt Audio Amplifier Trainer":
Articles in this series:
|
Review by BOB FLYNN
Audio Precision
System One Analyser
In 1985, a new company, Audio Precision, Inc,
introduced an automated test set for the measurement
of audio equipment. Ten years later, the system has
been considerably upgraded & is still regarded as the
world’s best. Here are our impressions of the system
after several months’ use.
Since the earliest days of audio,
engineers have needed to know the
performance of the equipment they
have designed. As time has passed,
the range of tests has become more
extensive and detailed. For example,
the basic parameters to describe the
performance of a power amplifier
may be as follows: Gain, Frequency
Response, Power Output, Signal-to86 Silicon Chip
Noise ratio, Total Harmonic Distortion
and Intermodulation Distortion.
Parameters such as gain, bandwidth
and power output are relatively easy
measurements to make, requiring no
more than a sinewave oscillator with
a variable output and flat across the
audio band, an RMS-reading voltmeter (with a bandwidth greater than
that of the device being measured), a
dummy load and an oscilloscope so
you can spot the onset of distortion
(ie, clipping).
Signal-to-Noise ratio requires the
above gear and an AC millivoltmeter,
although for really quiet equipment,
you need a millivoltmeter which will
read down to microvolts. For example,
a noise measurement of -115dB below
20V equates to a reading of only 35µV.
For these tests, you also need bandwidth limiting and weighting filters
(eg, for A-weighted tests).
For harmonic and intermodulation
distortion, the test equipment list
grows longer. Originally, for harmonic
distortion tests, you needed an audio
generator to provide the test signal
and a harmonic analyser. Setting the
generator’s frequency was never a
bother but trying to match the notch
filter in the analyser to that frequency
was another story.
In our lab, we have an old Radford
“Distortion Measuring Set” and this is
quite a beast to drive. As well as a large
handspan dial labelled “Rejection
Frequency Tuning” and a “Rejection
Frequency Range” switch, it has two
knobs marked “Coarse Tuning”, two
marked “Medium Tuning” and two
more marked “Fine Tuning”; seven
tweaks to drive you up the wall! Besides the inevitable drift in the generator frequency, you were also fighting
the analyser’s own drift. Finding a
true null with this device was almost
a miracle.
Later test sets combined the audio
generator in the same case as the
distortion analyser. The tuning of the
notch filter was then coupled to the
setting of the test signal frequency.
Servo circuitry kept the notch filter
in track with the generator, to relieve
the operator of the fiddly task of nulling. Further circuitry automatically
adjusted the voltmeter so that the
analyser input level was automatically set at 100%. So it could be said
that these instruments were partly
automated.
This was a big improvement over
previous instruments but making a
“THD versus frequency” test, for example, still meant the operator setting
each plot frequency on the generator,
waiting till the analyser settled, then
plotting each reading on graph paper.
Producing a series of such plots under
different operating conditions could
take many hours or even days.
Automated testing
All this changed in 1985, when Audio Precision produced their System
One audio test set which was controlled by an IBM PC. The test set reviewed here is the System One, Model
SYS-22. The SYS-22 is a stereo analog
audio test set comprising a 2-channel
generator and 2-channel analyser.
Available options for the System
One include weighting filters, tone
burst generation, intermodulation distortion measurement, wow and flutter
measurement, and a digital signal processor (DSP). External options include
the DXC-127 DC (a DC voltage source,
a DC voltage and resistance meter and
a digital input-output device) and
SWR-122 switcher modules (these
Fig.1: this is the first screen used to set up the System One. The lefthand panel is
for the Generator while the middle panel sets up the Analyser.
Fig.2: the second control screen for the System One. This screen sets up all the
settling parameters.
allow multi-point and multi-channel
testing of equipment).
Our unit has the options for inter
modulation distortion, tone burst tests
and the A-weighting filter. Perhaps the
most important other option is DSP
(Digital Signal Processor) and this
enables analysis of digital audio equipment such as CD players, Mini-Disc
and DAT recorders and also has Fast
Fourier Transform (FFT) analysis to
allow full audio spectrum analysis.
System One is large, heavy and
initially at least, quite inscrutable. It
has no switches or knobs, no dials or
displays or, in fact, anything that the
user can directly control. It is totally
controlled via an interface card which
is installed in an accompanying IBM
PC computer.
The System One measures 438mm
wide, 133mm high and 432mm deep
and it weighs about 15kg. The upper
left half of the front panel accommodates the generator output sockets,
comprising two three-pin male XLR
sockets and six banana jacks.
Below these sockets are three BNC
connectors, for a sync output, a trigger
input and a monitor output (eg, oscilloscope). The righthand side of the panel
is devoted to the analyser inputs, with
two three-pin female XLR sockets, five
banana sockets and a BNC connector.
Below these are five BNC sockets
arranged in a two groups. Three of
these are monitor outputs, while the
other two are for external filters.
Outputs of the generator may be
balanced or unbalanced, floating or
June 1995 87
Fig.3: this control screen is set up to measure the power bandwidth of an
amplifier at 1% rated harmonic distortion.
Fig.4: a self-test of the System One showing the overall frequency flatness,
using a measurement bandwidth of 500kHz. The specification is 20Hz to
20kHz ±0.05dB.
grounded. In floating mode, source
impedance can be 50Ω, 150Ω or 600Ω.
In unbalanced mode, it can be 25Ω
or 600Ω.
The analyser inputs are balanced
and the impedance can be set at 150Ω,
600Ω or 100kΩ. The low impedance
terminations are automati
cally removed if the input exceeds 30V.
Along with the test set and previously mentioned interface card,
the System One comes with a set of
program discs and a comprehensive
user’s manual. System One may be
88 Silicon Chip
run on any IBM or IBM compatible
machine through to 80486 machines.
The computer needs at least 640Kb of
memory, DOS 2.2 or later versions and
a Hercules, CGA, EGA or VGA monitor.
We installed the software and the interface card in a 386SX computer and
experienced no problems. Connection
between the computer and the System
One is made via a supplied cable fitted
with 25-pin D connectors.
The software runs under DOS (ie,
non-Windows) and as supplied with
our version, the program contains
about 80 test files and 13 procedure
files. Test files are compiled to carry
out a specific test; eg, THD+N versus
Frequency, THD+N versus Amplitude,
Crosstalk etc. The user can generate
custom test files or make modifications
to the supplied tests.
A procedure is a file that will execute a series of tests and sub-procedures in a fixed order. Procedures are
particularly useful for production tests
and quality control. Once a procedure
has been established by the production
manager, non-technical staff can run
tests on a product and every time the
procedure is run the test parameters
remain the same.
With everything ready to go we
were keen to put the system through
its paces. A good place to start is
SYS22CK.PRO, a procedure file with
eight tests to check key parameters
of the system. After this procedure is
finished, a report of the response of
the instrument to each test is shown.
If any of the parameters is outside the
specifications, then the instrument is
in need of recalibration or servicing.
The command S1 starts the program and the Audio Precision logo
appears with a command line below
it. This command line shows a menu
of fourteen single word commands; eg,
Run, Panel, Load, Save etc. One letter
commands are enough to produce
action and entering (L)OAD, brings up
the next command line with a further
10 commands such as Test, Limit,
Procedure, etc. Entering (P)ROCEDURE
displays a list of the Procedure files.
Picking SYS22CK loads that procedure, then entering (R)UN followed
by (P)ROCEDURE starts the series of
eight tests.
The result of each test is displayed
on the monitor in either graphical or
tabular form. The total time to run the
eight tests and the tolerance report
was less than 90 seconds. As you
would expect, the instrument comfortably exceeded the specifications
for all tests.
Sample tests
The following is not intended to be
a blow by blow description of how to
use System One but rather to give a
rough idea of what needs to be done
to set up a test.
If a test is loaded and (P)ANEL is
entered then the screen shown in Fig.1
appears. The left hand panel titled
GENERATOR shows the generator
setup with the key functions such as
WAVEFORM in the leftmost column. If
the highlighting bar is moved, either by
the keyboard cursor keys or the mouse,
to cover a word to the right of any of
the key functions, then the command
line at the bottom of the screen shows
the options available for that field.
For example, if NORMAL, to the right
of WAVEFORM, is highlighted then
the COMMAND options available are
NORMAL, BURST, TRIG, GATE.
Highlighting the figure next to
FREQUENCY allows you to key in the
generator frequency, or by pressing
the + key, increasing the frequency by
the factor shown next to FREQSTEP
if “*” is selected to the right of FREQSTEP. Similarly, highlighting the figure
to the right of AMPLITUDE allows you
to key in some other signal voltage or
by pressing the plus key, increasing
the amplitude by the voltage shown
next to AMPSTEP if + is selected to
the right of AMPSTEP. Highlighting
the field to the right of OUTPUT gives
you the option of selecting OFF, (channel) A, B, A&B, A&-B. In other words,
you are setting up the generator with
the keyboard instead of switches and
potentiometers.
The centre screen panel is titled
ANALYSER and here again by highlighting the words to the right of the
key functions, the analyser can be set
up. Immediately to the right of MEASURE is a field allowing you to choose
channel A or channel B.
Further to the right again is a field
giving measurement options: AMPLITUDE, BANDPASS, BANDREJECT,
THD+N , SMPTE , CCIF , DIM W+F,
2-CHAN and CROSSTALK. Similarly,
to the right of the function READING
are the options %, dB, PPM, X/Y and
OFF. READING is just what it says;
the parameter being measured. LEVEL
can be set to V, dBm, dBu, dBv, dBr,
W, OFF. This is the signal level into
the analyser before any filtering or
tailoring.
To the right of BANDWIDTH the first
field allows you to choose the low
frequency cutoff and the next field the
high frequency cutoff.
The righthand panel titled SWEEP
(F9) DEFINITIONS allows the user to
set up the tests with sweeps of frequency or amplitude measurements
versus time. To the right of DATA-1
near the top of the panel can be selected the parameters to be plotted, ANLR
(analyser), GEN (generator), DCX (not
The Audio Precision test set can be fitted with a large range of options,
including DSP analysis for equipment such as DAT & MiniDisc recorders.
Fig.5: a self-test of the System One showing the residual harmonic distortion &
noise between 5kHz and 100kHz, with a measurement bandwidth of 500kHz.
fitted) or DSP (not fitted).
Further to the right can be selected
RDNG, LEVEL, FREQ, PHASE, NONE.
DATA-2 can be changed to SOURCE-2,
HOR-AXIZ, or STEREO. SOURCE-2
allows two types of parameters to be
swept in the one test.
Hor-axis permits two measured
values to be plotted against one another. Stereo generates consecutive
sweeps, the first through one chan-
nel after which the generator output
and analyser inputs are switched by
System One and the same sweep is
performed on the second channel.
SOURCE-1 is the swept independent
variable and can be set to GEN, ANLR,
SW1, DCX, DSP, EXTERN. With GEN
selected, for example, the next field to
the right gives the following options:
FREQ, AMPL, TB-ON, TB-INT, TB-LVL
or NONE. Frequency allows you to
June 1995 89
adequate and ENABLE (the top line)
to SWEEP. When the power output
versus frequency sweep is now made
the generator amplitude will be varied
to maintain the measured distortion
level at 1%.
Printouts
Fig.6: a self-test of the System One showing the residual harmonic distortion
and noise of both channels between 20Hz and 20kHz. Note that below 5kHz, the
distortion is less than .0005%!
set the sweep START and STOP to the
range required. Amplitude allows you
to set the START and STOP amplitudes
of the generator and the output will be
swept with a fixed frequency signal but
varying in amplitude. NONE gives a
single point measurement with tabular
display.
More options
Pressing “page down” displays a
second screen, as shown in Fig.2.
When you operate a manual test set, if
you alter the generator frequency and
the analyser is an automatic frequency
tracking instrument, you have to wait
for the instrument reading to “settle”
(ie, to stabilise) before you take the
reading. With System One, SWEEP
SETTLING allows the selection of
parameters that effect the settling as
it performs a sweep.
If, while a sweep is being run, the
trace stops at some point in its travel,
then a “T” will appear at the bottom of
the graph before the sweep continues.
The T indicates a “time out” meaning
that, at that point in the trace, the
required settling parameter was not
achieved. This can be due to noise in
the analyser signal. Some adjustment
of the parameters in the SWEEP SETTLING panel will be required.
Pressing “page down” a second
time displays a third screen, entitled
REGULATION, as shown in Fig.3. This
allows testing of a device while varying either the test signal amplitude or
frequency, while making a sweep. For
example, say you need to measure the
power bandwidth of an amplifier at a
distortion level of 1% across the audio
band. To achieve this, set “REGULATE
ANLR RDNG TO” 1% and “BY VARYING GEN AMPL” to the HI BOUND
and LO BOUND levels you think are
Key Specifications
Total System THD+N ���������� <.0015% from 20Hz to 20kHz, with 80kHz
measurement bandwidth; <.001% from 20Hz
to 20kHz, with 22kHz measurement bandwidth
Total System Flatness �������� ±0.05dB, 20Hz to 20kHz
Total System IMD ��������������� <.0018% SMPTE; <.002% DIM; & <.0005%
CCIF
Analyser Residual Noise ���� <1.5µV (-114dBu) with 22kHz measurement
bandwidth.
90 Silicon Chip
Having made a test and deciding
that the displayed graph is the one to
keep, what methods are available to
keep a record? If you have a dot matrix
or HP LaserJet printer connected to
your computer, then a screen dump
can be made by typing a <*> (asterisk).
However, the printout will only be as
good as your monitor’s resolution. If
though, you start the program with
<s1/g> and then after running your
test you press the escape key, the
command line appears at the bottom
of the screen. Typing (S)AVE brings
up another menu. Pressing (G)RAPH
ICS then allows you to save the graph
as a .GDL (Graphics Display List) file.
There are also two utilities in the program: Post and Plot. Post allows you to
convert the .GDL file to a Postscript or
EPS file and Plot will let you convert
it to an HPGL plotter file.
Well, having had the use of System
One for some months now, what are
my impressions? At first, while very
pleased with the instrument’s performance, I could not help feeling that I
had been removed a couple of steps
from the testing procedure. Maybe it
was the fact that I was now setting up
the generator and the analyser on a
keyboard, with no more twiddling of
switches and pots and no more waiting
for the instruments to settle. I cannot
say I missed plotting the results on
graph paper with a pencil though.
This feeling of being remote from the
testing soon passed.
The more you use System One, the
more things you find it can do. Now
I would hate to have to return to the
old manual way of doing the job. This
is truly automated audio testing. The
only thing that the operator needs to
be aware of is interference from the
monitor’s radiated timebase. While
the System One is very well shielded,
there is a need to take care to keep it
out of the device being tested.
The System One and other distortion test sets by Audio Precision,
Inc are distributed in Australia by
I.R.T. Electronics Pty Ltd, 26 Hotham
Parade, Artarmon, NSW 2064. Phone
SC
(02) 439 3744.
|