Frequency Response: A
Fundamental Audio Attribute
This is part of an occasional series
in which I'll provide basic information on the building blocks of audio and video: the
terminology, specs, and principles. There are very few places you can go to clear up
misconceptions about the language of home-entertainment electronics, so from time to time
I'll try to address this deficiency.
The most widely specified aspect of audio gear is frequency
response, so it's perhaps appropriate to start there.
What is it?
Sound is made up of a series of air-pressure differences. A
vibrating body causes the air molecules next to it to be alternately compressed and
decompressed (or rarefied), and these changes radiate outward from the body that created
them. Ultimately they hit our eardrums, moving them in and out. We interpret these
movements as sound.
The pressure changes can be fast, in which case we hear
high or treble sounds, or slow, in which case we hear low or bass sounds. The rate of
change is known as a sound's "frequency" -- how often it changes.
Both sound-emitting objects and our hearing apparatus react
very differently to different frequencies. Voices and musical instruments are defined by
these variations, and each of us has a distinctive physical reaction to sounds we hear. A
violin doesn't sound like an oboe, even if they're playing the same note, and that oboe
might sound quite different to you than it does to me. That's the way it works in the real
world.
A system designed to store or transmit these sounds mustn't
discriminate. The balance of frequencies created by that oboe must be maintained
throughout the system and delivered intact at the other end or it will no longer sound
like an oboe, however your ears or mine might interpret that.
An audio device, therefore, must respond equally to
incoming signals of a given strength no matter what their specific frequencies. Its
ability to do that is its frequency response.
How is it specified?
The basic unit of frequency is the number of complete
compressions and rarefactions in a second. That used to be called "cycles per
second" or cps, but is now universally known by the name of the man who defined the
phenomenon, Heinrich Hertz. One hertz (abbreviated Hz) is one complete cycle each second.
We are sensitive to frequencies over a fairly wide range.
Young people can often hear as high as 20,000Hz, although as we age this drops somewhat --
most middle-aged men cut off at 16,000Hz or lower. At the low end, some people can hear as
low as 16Hz, although again this changes with age. The usual range quoted for hi-fi
equipment is 20Hz to 20,000Hz, which is approximately correct and easy to remember.
Those numbers alone define a frequency range -- the
extremes at which we can expect a device to produce some output. To determine how fairly
it deals with different frequencies, however, we must add a tolerance. Since the middle
frequencies are usually the best behaved, a frequency of 1000Hz (1kHz) is usually taken as
the reference point, and the maximum deviation from that level stated, in decibels.
Decibels describe the ratio of one sound level to another
in a way that relates fairly closely to how we hear: 3 decibels (or 3dB) is a just
noticeable level difference. A response that denotes a frequency range "±3dB"
means that the level never varies by more than 3 decibels above or below that at 1kHz.
That's a fairly generous margin these days, allowing a total variation of 6dB from highest
to lowest. Most good equipment is now specified to within 1 decibel.
How does it affect equipment?
With most components, even though frequency response is
almost always specified, it doesn't mean very much today because equipment design has
reached a point where it's a simple matter to make the response "flat." That
means, when response is plotted on a graph with frequencies in the horizontal axis and
relative level in the vertical, the line is very close to straight.
Phono cartridges and cassette decks, especially if not
perfectly adjusted, might exhibit response less than ideal, but electronic devices such as
amplifiers and CD players almost always exhibit virtually perfect response.
The one big exception is the speaker. As an
electromechanical device, it rarely shows the nice clean curves of its electronic
brethren, and if detailed measurements are taken, different curves are recorded from
different directions.
In truth, although things have improved over the years,
there is still little agreement as to how to measure speakers or evaluate those
measurements, so many manufacturers don't specify response for their speakers. Where
curves (or even single numbers) exist they take considerable experience to interpret.
There's no small irony, therefore, in the fact that while
good frequency response is a basic necessity in any audio component, the numbers published
don't tell us much. Either they're close to perfect, as with electronics, or largely
useless, as with speakers.
...Ian G. Masters
ian@mastersonaudio.com
|
