This is only a preview of the July 1999 issue of Silicon Chip. You can view 36 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. Articles in this series:
Items relevant to "A 10µH to 19.99mH Inductance Meter":
Items relevant to "An Audio-Video Transmitter":
Articles in this series:
Items relevant to "An X-Y Table With Stepper Motor Control; Pt.3":
Purchase a printed copy of this issue for $10.00. |
Just how DO you test a loudspeaker?
CLIO: PC-driven
loudspeaker evaluation
Testing speakers, particularly hifi speakers, has always been something of a
problem. Either you had a fully set up anechoic chamber with a raft of
professional (read expensive!) test equipment . . . or you relied on your ears.
Review by Ross Tester
F
or the most part, the latter has been the “norm”. Not
that this has been a necessarily bad thing – after all,
it’s your ears that are going to be the final arbiters
anyway. But as test equipment, they suffer from a few
major drawbacks.
First, ears are subjective. Did I really hear that or am I
just imagining it?
Second, it’s very difficult to find two sets of ears calibrated exactly the same. In fact, calibration often varies
between otherwise matched pairs, especially as they age!
Third, and perhaps most important, ears cannot be calibrated anyway – so what’s the standard?
If only there was a low-cost way to objectively measure
and test speakers . . .
You’ve probably already guessed that all this is leading
to just that: a (relatively) low cost but accurate speaker test
and measurement system. It’s called CLIO and is manufactured in Italy by Audiomatica SRL.
The CLIO system has two basic components: a dedicated
8-bit PC Card which slots into a vacant ISA socket in an
IBM-compatible PC (386 or higher) and the software to
run it.
Audiomatica recommend a minimum 386-DX33 with
2MB RAM, VGA video card and a hard disk drive. A math
co-processor is not essential but is highly recommended.
Of course, running CLIO on a 486 or better automatically
gets a maths co-processor.
Also supplied with the CLIO system under review were
a 2.75m long RCA to RCA “noiseless signal cable” (oxygen-free copper), two 1m long RCA to alligator clip leads,
an Audiomatica MIC-01 calibrated condenser electret
microphone.
The 25cm long microphone is accurate to within ±1dB
from 20Hz to 10kHz and within ±2dB from 10-20kHz. It
comes with a mounting bracket intended to be attached to
a small microphone stand.
An optional Audiomatica amplifier, calibrated to the
electret microphone, is also available. This 10W, 0.004%
THD amplifier makes the system self-contained. With internal switching, the impedance and frequency response
80 Silicon Chip
of a loudspeaker can be checked without changing wiring.
For this review, we used our own audio amplifier so the
supplied amplifier was not required. Its main function is
for automatic or manual quality control setups.
What does CLIO do?
CLIO works as a precision A/D and D/A converter frontend for your PC. Using the power of the PC, it generates a
range of audio signals to drive an amplifier connected to
the speaker under test. Using a microphone calibrated to
the system, it listens to the speaker output and compares
this with the test signals.
CLIO uses several measurement methods, possibly the
most important being the well established maximum length
sequence (MLS) analysis technique.
car speaker installations – it also automates IASCA scoring), Fast Fourier
Transforms (FFT) with the ability to
switch back and forth between time
and frequency domains.
An inbuilt control panel also gives
you the ability to set and display a wide
range of input and output settings and
even display the output on a screen
based “oscilloscope”. It even has an
inbuilt L/C meter.
As you can see, CLIO is an extremely
versatile system. And we have only
talked about some of its testing capabilities. It does a lot more than this!
Getting it going
We must admit that we had some
difficulty in getting the system to work.
This is the “heart” of the CLIO system, an 8-bit card which plugs into a vacant
We’ll explain why so readers won’t
expansion slot in any PC from a 386 up. It should be fitted as far away from
experience similar problems.
the video card as possible to minimise interference. At right is the card end-on,
Fitting the card is simple: you simply
showing the input and output RCA connectors. The top connector is channel A
find a vacant slot (as far as possible
input, next down channel B input, next channel A output and the bottom
from your video graphics card) and
channel B output. During setup the channel A input and output are shorted.
plug CLIO in as you would any other
This has become an international standard for accurate
expansion card. The software, likewise, loads easily to
anechoic analysis and for room acoustics. In MLS the
your hard disk from the INSTALL command on the flopimpulse response is measured very quickly and with high py disk supplied. Theoretically, that’s all there is to basic
accuracy, with the computer analysing the data.
installation but in our case . . . not quite.
From the impulse response it is possible to obtain a
The first computer we tried to use was a 300MHz Pentium
variety of measurements, such as frequency response, II machine but it turned out to be too fast. (The software
phase response, minimum phase, phase with group delay has since been changed to allow even the fastest PCs to
removed, the energy time curve, cumulative spectral decay operate). When we had no joy there, we went to a 50MHz
(or “waterfall”) and reverberation time.
486 machine but it still wouldn’t behave. Different probCLIO analysis allows a wide range of control over these
lem, though: a “run time” error which we simply couldn’t
various tests to suit either the equipment under test, the eliminate.
environment, or both.
Perhaps there was a conflict in I/O addresses? The CLIO
Another analysis method is the tradition sinusoidal board has a jumper to adjust the address from 300H (factory
measurement, which can test frequency and phase re- default) to 310H. So we changed the jumper – alas, still
sponse, distortion, impedance and automatic evaluation
no joy. A lot of head scratching and to-and-froing between
of Thiele-Small parameters.
ourselves and the Australian distributors of CLIO (Audio
It will undertake third octave analysis (very popular in Consultants, of Stockport, SA) eventually solved the prob-
Two plots representing the same thing: the impedance (in ohms) vs frequency of a quality speaker system. On the left is
the plot produced by SILICON CHIP's Audio Precision Test Setup, while the plot on the right is that produced by CLIO. As
you can see, with only a minor difference between 10 and 20Hz, the plots are virtually carbon copies of each other.
JULY 1999 81
lem: some type of conflict between the CLIO card and a
network card fitted to the PC.
This should have been evident from the very first – a
conflict should show in the Windows control panel. But
for some reason it did not. Anyway, to cut a long (actually
very long!) story short, when we removed the network card
from the machine CLIO burst into life.
We also possibly made a mistake in trying to operate
the system under DOS. Our reading of the instructions
suggested it had to run that way.
CLIO requires 575K of free memory and we were having
some trouble achieving this with what was loaded in the
machine. In the end, a chance remark from Audio Consultants about operating under Windows 95 led us in that
direction – and success.
So what is the wash-up of all this? When installing
CLIO, put it into a “bare bones” computer (ie, no extra
cards) running only Windows 95/98 and you shouldn’t
have any problems. You were wondering what to do with
that pensioned-off 486, weren’t you?
Running the calibration procedure takes a few minutes
but is fully detailed in the manual, so we won’t repeat it
here. Suffice to say it basically runs itself.
Testing a speaker
Once everything is working satisfactorily, setting up CLIO
is quite simple. You need to verify system performance
and operation, then calibrate the system to your PC. The
instruction manual covers this more than adequately.
During the calibration process, a loopback cable is required – that is, the input and output of the “A” channel
need to be shorted with a suitable RCA-RCA lead (one was
supplied in the package).
One point to note, though, the four RCA sockets on
the card backplane are not labelled – in a normal PC the
“A” input will be the top socket, the “A” output will be
the third socket down. They are clearly labelled in the
instruction manual.
After all our (mis)adventures installing CLIO, this part
was a bit of an anticlimax. It ran like clockwork!
We put CLIO through a range of tests measuring a high
quality 2-way speaker system. The first test we ran was
(at least to us) one of the most interesting: we wanted to
compare the results obtained by CLIO against the results
of our laboratory Audio Precision test equipment.
The speaker impedance was first measured and plotted
by the Audio Precision (incidentally, about $20,000 worth!)
and then repeated using CLIO (at less than a tenth the
price!). The results speak for themselves – above about
20Hz, the plots are virtually identical.
We then ran a variety of tests using CLIO, some of which
are reproduced on these pages. (We actually ran many tests
over several days but space precludes us from showing
the results. The ones shown are typical tests but of course
CLIO is capable of much more than those shown here).
We were mainly interested in looking at the basic parameters of the speaker: its frequency response, for example,
is one of the fundamental tests and most-quoted figures
when a salesman is extolling a speaker's virtues!
(Like most quoted figures, though, frequency response
can be fudged, especially if no amplitude reference is
given.)
We also looked at the interaction between the listening
environment and the speaker. Unfortunately we were limited in the size of room in which we could conduct our
tests and this became very evident as we progressed. We
expected severe room interaction – and CLIO proved that
we got it! One of the beauties of CLIO, though, is that these
The CLIO instruction manual is basically well-written,
although there is some evidence of Italian/English
translation going just a little awry. Compared to some
Asian manual translations, though, it's good.
Accessories supplied with the CLIO system: two RCA to
alligator clip leads and a long (2.7m) high-quality RCA to
RCA lead. The lower pic shows the 25cm-long high quality
calibrated microphone mounted on its stand adaptor.
Setting it up
82 Silicon Chip
The MLS (maximum length sequence) test is a de-facto
standard for analysing room acoustics. The microphone
picks up a combination of sound from the speaker and
sound reflected in the room. As the system knows what the
speaker should have been reproducing itself, it can analyse
the effects of the listening environment.
effects can be cancelled out if required.
We checked the phase output,
showing just how good (or bad) the
speaker components (particularly the
cross-over) were. Speaker manufacturers go to extreme lengths to get the
cross-overs “just right”. Sometimes
they win, sometimes they lose.
CLIO in quality control
Having computer power to make
all the calculations gives CLIO a huge
advantage over other forms of testing.
Tests that used to take hours of measuring and calculating are performed
In this test, the speaker is “swept” with a 2.82V sine
wave from 200Hz to 20kHz and the microphone is placed
500mm from the speaker. The sound pressure level is
then plotted. Below 200Hz, room reflections (especially
in a small, non anechoic room) tend to make the readings
meaningless.
in seconds (actually in milliseconds!).
This makes CLIO an ideal candidate
for use in quality control applications.
Indeed, there is an option for CLIO
which is intended for just that. (The
QC option wasn’t supplied for evaluation but its operation is covered in the
manual. We have no doubt it would acquit itself with the same performance
as the rest of the CLIO package).
How much?
CLIO is not cheap – but it’s a bargain. The accompanying panel shows
the price and availability. We’re im-
Cumulative spectral decay, otherwise known as a “waterfall”
plot, looks at the way the speaker behaves immediately after
it is hit with a pulse. In a perfect world, the decay would be
linear with time but speakers are not perfect devices. The
results would have been much better in a larger room.
pressed with its seemingly endless
features, its ease of use and the way
it works. And, with only a minor reservation after our difficulties getting it
going, we give it the thumbs up! SC
Recommended retail price of the CLIO
system, not including amplifier, is
$1840.00 ($1551 if tax exempt). The
amplifier sells for $605.00
($457 tax exempt).
Enquiries to the Australian distributors,
Australian Audio Consultants,
PO Box 11, Stockport SA 5410.
Phone/Fax (08) 8528 2201
A perfect loudspeaker would be phase-linear; that is,
the sound output would be a perfect reproduction of the
input signal. However crossovers and even the speakers
themselves introduce phase distortion. This test shows the
difference between the input signal and the output.
JULY 1999 83
|