This article was first published in 2007.
likes a powerful amplifier – it’ll develop better bass, be punchier and be much
less likely to be driven into distortion. But how do you know how much RMS power
an amp’s got? There’s usually no point in reading the numbers written on the
side – quite often they’re pure fiction.
amplifier power is actually pretty straightforward and can be done for near zero
cost. This story primarily looks at testing home sound system amplifiers but
exactly the same approach can be taken with car amps.
then when you have the measured power figure, it’s not hard to do some more
tests that also reveal things like frequency response and background noise
If you’re anything like me, you can’t resist
picking up a bargain. Especially when it’s a cheap piece of electronics that
someone in their, er, wisdom has decided no longer has the right fashion look.
Take amplifiers. Visit (in declining order of salubriousness) secondhand stores,
garage sales, roadside hard rubbish collections or the tip and you’ll find a
host of amplifiers that are available at ridiculously low cost.
Take the Rotel 712 integrated stereo amp shown
here. It cost me $8. Yes, that’s eight bucks. What was wrong with it, you ask?
Nothing.... Similar bargains can be had – recently I bought a Rotel (yep, I like
that brand) RX-203 stereo receiver for... wait for it, $3. So what was wrong with
that one? Well, I haven’t tested it yet but I’ll bet that again it works
But talking about testing.... Only a few days ago I
was like many of you – I thought that any meaningful testing of an amplifier
needed stuff like oscilloscopes, standalone frequency generators and specific
audio test gear that I don’t even know the name of. But then I found out that
all I’d need is a digital multimeter, my trusty PC, a few cheap resistors... and a
A jug element, hey? It sounds crazy enough that it
might just work...
But the first things that you need are you ears.
In the following discussion about technical testing techniques, it’s possible to
forget that the reason that you buy an amplifier is to listen to it. And hey, if
you like the sound that it makes, what the hell do any figures matter?
In the following testing regime your ears will be
used to assess the purity of a quiet test tone and also to listen to background
noise that might exist when there shouldn’t be any.
You’ll also need a decent digital multimeter.
Since these days ‘decent’ applies down to multimeters costing from about $60,
almost any near-current meter will do. It helps if it can measure frequency (but
that’s not vital, it’s just a good check) but it must be able to measure AC
volts and resistance.
A cable that connects the sound card output of
your PC to the amplifier inputs will also be needed. Typically, this is a cord
with a stereo 1/8th inch plug at one end and two RCA plugs at the
On the output side of the amplifier you’ll need a
monitoring speaker. This will only ever be used at low volumes (no test tones
will destroy it!) so it can be one of your normal speakers. To reduce the power
drawn by this speaker you’ll need a resistor – say a 150 ohm, 5 watt wirewound
jobbie. From an electronics store that will set you back about 30 cents. (The
resistor shown here was just one I had around – a physically smaller one is
And now we come to the jug element. Or elements.
Yes, you’ll need one or two electric jug elements (just the mains power sort that you
can buy at any hardware store or supermarket) and a Pyrex or high temp glass
bowl that you can fill with water before suspending the elements within it. If
you haven’t guessed, this is our amplifier load – and a bloody good load it is,
Finally, for a car amplifier, you’ll need a way of
supplying it with plenty of 12V power. Easiest is to use a fully charged car
battery – the full load tests don’t take long so the battery will have plenty of
juice still remaining when the testing is finished. (You can float charge it as
well if you like.)
Making the Load
Turing the jug elements into the load is easy. If
you use two (as I did), unwind enough turns of wire from each element that the
remainder left on each ceramic former poses an 8 ohm resistance when measured
with the multimeter. Then when you wire the modified elements in parallel,
you’ll have a very high power load with a 4 ohm total resistance.
Alternatively, you can make each element 2 ohms
and then wire them in series for an 4 ohm load. How you do it really doesn’t
matter – you just want as the end result lots of windings and a resistance that
matches the load the amplifier expects to see. (You can also make 2 ohm or 8 ohm
or 16 ohm loads just by adjusting the procedure.)
Place the elements in the Pyrex jug of water and
then use heavy gauge wire to connect the jug elements together and then to the
If you are testing a high powered amplifier, this water can become hot enough to
Generating the Test Tones
To generate pure sine wave test tones you’ll need
to download some frequency generator software off the web. This is available
either free of charge or with a 30-day free trial period. We used PAS Products
Frequency Generator version 2.6 (www.pas-products.com) but there are
plenty of programs around – a web search will soon find them.
Practice playing the generator through your normal
PC speakers until you are able to do two things – generate sine wave signals
over a 20 – 20,000Hz range, and vary the volume level. The latter is important –
some frequency generators make it difficult to vary the amplitude (volume) of
the test tones and so you should check that this function exists.
you want to, you can burn a CD with all the test tones on it and then play it
back through your car CD player. That way, you can easily power the amplifier
from the car, simply replacing the speakers with the jug element load and
running the standard speakers with a dropping resistor as the low volume
Measuring RMS Power
This one’s really exciting – everyone knows
about amplifier power, and being able to measure the output with your own eyes,
hands and ears is great fun. Here’s how you do it:
Fill the load with water and then connect it to
one channel of the amplifier.
Wire the monitoring speaker in parallel with the
load – that is, connect it to the same channel. Don’t forget that you’ll need to
put that 150 ohm, 5 watt resistor in series with the feed to this speaker.
Set the multimeter to VOLTS AC and then
connect it across the same channel.
Connect the input of the amp to the sound card
Open the frequency generator software and select a
frequency of 1000Hz.
This diagram shows the set-up.
When you turn up the volume output of the
frequency generator software, you should hear a faint 1000Hz test tone coming
from the speaker and read an AC voltage level on the multimeter.
If everything is working as it should, turn up the
volume, listening intently to the test sound and watching the changing figures
on the multimeter. When the volume reaches a certain point – called ‘clipping’ –
the sound from the speaker will suddenly and clearly distort. Take note of the
multimeter reading just before this occurs and then quickly turn the volume back
On my $8 Rotel, the left-hand channel yielded a
result of 19.6V before clipping. So how do we turn this into a maximum power
figure? It’s easy – just square the number (ie multiply it by itself) and then
divide that by the resistance of your dummy load. So that was (19.6 x 19.6) over
8.5, which is 45 watts RMS.
I then repeated the test for the other channel and
got a figure of 48 watts.
Hey, not bad for an $8 amp, eh?
Measuring Frequency Response
Now you might be saying that it’s great that his
amp has 45-odd watts per channel, but that’s only at 1000Hz. What about over the
rest of the frequency range? This introduces the idea of frequency response –
how flat is the response of the amp? Testing this is again very easy – the
set-up is the same as shown above.
Leave everything in place as it was for the
previous test, but reduce the power output to 2 or 3 on the volume scale (ie
adjust it to what normally would be a quiet listening level) and decrease the
frequency of the test tone to 20Hz. (You will no longer be able to hear it from
your speaker – the frequency is too low). If using the car head unit, make sure
that all the tone controls are set to flat (ie zero adjustment) and switch off
the loudness button.
Measure the input voltage to the amplifier. For
example, say it’s 0.5V. Then measure the output voltage of the amplifier, and
note this value.
Next, change the input frequency to 100Hz and
adjust the input level until the input signal has the same voltage (0.5V in this
case) as for the previous measurement. Again measure the output voltage.
Keep doing this right through the frequency range,
up to 20,000Hz. Make sure that the input voltage doesn’t change – you may need
to tweak the volume control on the frequency generator (or car head unit) to
And of course, you don’t need to do the
measurements all in small increments – this table shows the results of this
testing on the Rotel.
Right Channel Output (V)
Left Channel Output (V)
As you can see, with a constant input voltage, the
highest output for the right channel was 2.2V and the lowest, 2.1V. So it
doesn’t vary much, does it? However, how do we express this variation in that
unit beloved of audio engineers – dB? Again, it’s easy. Simply divide the
highest figure by the lowest, log it, then multiply by 20.
So, from the table of data:
---- log x 20 = 0.4dB
(Incidentally, the ‘log’ function is as easy as
pressing log on a scientific calculator.)
So between 20 and 20,000Hz the biggest variation
away from ruler flat response for this channel is just 0.4dB. The other channel
measured a little worse at 0.5dB. Those are very good specs for an amplifier –
those eight dollars are looking better and better spent!
The signal/noise ratio is a measure of how quiet
an amplifier is: for example, during passages in a song when all the sound
stops, you shouldn’t be able to hear anything – no hiss, no hum. Well, not from
the amp, anyway.
Remember how when we tested max power output on
the Rotel we achieved a maximum output before clipping of 19.6V? That’s one of
the figures we need for this test. The other is gathered by again measuring the
voltage output with the volume control wound around a long way, but this time
with zero signal input. But if we just pull off the input lead it’s likely that
electrical noise could be picked up from the surroundings, so instead we wire a
¼ watt 1 kilo-ohm resistor across the input. (This also keeps the amplifier
happy as it’s seeing some input resistance.)
Again the dummy load, speaker and multimeter can
be left connected as we had them before. Wind up the volume control to the level
at which clipping previously occurred (eg ‘8’ on the scale) and read the
no-input-signal output voltage. It will be very low. In the case of the Rotel,
it was 2.6mV.
So at full power, 19.6V was outputted (that’s
19600mV) and at no input, a max of 2.6mV was outputted at the same volume
control position. To turn this into signal/noise ratio we do the same sort of
calculation as for frequency response:
-------- log x 20 = 77.5dB
Now a 77.5dB dynamic range isn’t wonderful – in
fact it’s pretty bad, and by far the worse spec of this amp.
And that brings us to another technique. Connect
the amp to its normal speakers (no resistor needed) and make sure that there is
no input signal (But keep the resistor across the input). Then turn the amp
right up and listen intently to the speakers. The more noise that you can then
hear, the poorer the signal/noise ratio of the amplifier.
It’s possible to gain a lot of information about
an amplifier with very little effort – and even some genuine excitement. Testing
an amp at full sine-wave power is like seeing a car giving its all on a chassis
dyno. In just the same way, expect things to get hot and bothered if anything in
the amplifier design is less than optimal (eg amp heatsinking, power supply
another cheapy buy but not in the same class as the Rotel covered in the main
text. I bought this Bose professional amplifier by tender (so no extensive
testing was allowed prior to purchase) for just AUD$480. That’s pretty good when
the new price is US$2000 and the thing can develop no less than a claimed 450
watts/channel into 8 ohm loads!
once I’d got it home I found that my new purchase wouldn’t develop any watts
into any channels – instead it just blew the circuit breaker on the external
power board. Inside the case, the transformer was simultaneously getting hot...
One VERY expensive transformer later (and there went nearly all of the
cost advantages of my ‘bargain’!) I was ready to do some testing.
how much power output then? Well, with one channel driven, try 560 watts into an
8.5 ohm load! And that’s with no apparent clipping – the amp incorporates
internal soft clipping circuitry. The signal to noise ratio wasn’t as impressive
though – a measured 88dB...