What Kind of Noise Annoys...?
Several years ago I wrote a basic article on how to buy a
stereo system for a magazine that only occasionally runs material on the subject. The
editors, making a play on the name of a defunct British rock group, called the story
"The Art of Noise," apparently in the belief that noise and sound are the same
thing.
I had to cringe because to me noise has always been the
enemy. Far from being synonymous with sound, noise is specifically unwanted sound,
and if there has been art in audio design over the years, it has at least in part been
involved with getting rid of it.
What constitutes noise is to some extent a matter of
convention. It's a form of distortion, in that it is something present in a reproduced
waveform that wasn't there in the original, and by that definition, noise and distortion
could be used interchangeably. In practice, distortion usually refers to changes in the
waveform itself, and noise to things that are added to it.
Even there, some extraneous material is treated separately,
notably radio-frequency interference (RFI) from the CD operator down the street.
Mechanical buzzes in speaker frames or enclosure are usually not considered noise either.
In both cases, the sounds, although annoying and obtrusive, are not inherent in the
equipment; they are, in fact, types of malfunction.
Under that rubric we should probably also place surface
noise in vinyl recordings. LPs certainly contained their share of conventional noise, but
in theory, a perfectly stored and handled record would have no added physical noise.
Anything that wasn't in the original tape, or added during the mastering process, can be
attributed to such mishaps as dust, mistracking, scratches, warps and the like, none of
which is inherent in the recording.
The two phenomena that are usually considered noise in
audio are hum and hiss. The former is caused by 60Hz alternating house current leaking
into the audio signal. What we actually hear, mostly, is its second harmonic -- 120Hz --
which is why we can hear hum from components with no appreciable bass response.
A tiny bit of hum is inherent in a component's performance.
Most of our audio gear is powered by house current, and although today's power supplies
are very good at filtering it out, a very small amount remains. It might be measurable,
but it should not be audible. But, like other forms of noise, hum is cumulative, so if a
system consists of numerous devices connected in series, hum that's not audible in the
individual components might become so in the system overall.
Most of the hum we hear, however, is caused by our systems
misbehaving themselves in various ways. Our homes are awash in electromagnetic fields, all
throbbing away at 60Hz. Equipment chassis and cables are shielded to keep these out, but
sometimes shields fail; often connections between components become loose or otherwise
corrupt, allowing hum to affect the tiny signals.
Hum problems of this sort can be notoriously difficult to
track down, but they are a form of system failure and in theory can always be cured.
Hiss is random noise, and it can be caused by several
things. In some applications, it's also called "white noise" because like white
light, which contains all the frequencies of the visible spectrum, it contains all the
frequencies of the audio spectrum, in roughly equal amounts. It can be heard in a fairly
pure form as the interstation noise on the FM radio dial, and if you turn that up to a
reasonably high level (only for a moment, though), you can hear that it contains
substantial low-frequency material. Because we usually hear it at low levels, however, and
because our hearing is much less sensitive to bass at low levels, we tend to hear mostly
the high-frequency portion, which accounts for the term "hiss."
Some random noise is produced by all electronic gear, in
the form of "thermionic noise" or "resistance noise," an inevitable
byproduct of the resistors in the circuits. Like residual hum, this noise tends to be very
small, and in fact the two are often lumped together in a single
total-harmonic-distortion-and-noise figure in the specifications of those components for
which neither thing is a problem.
But there have been two traditional areas of audio in which
hiss is a big deal: analog tape recording and FM radio. In each case, the noise is
described in terms of "signal-to-noise ratio" (S/N), expressed in decibels (dB).
In each case, an upper reference level is chosen -- a fully modulated signal in the case
of FM, a specific distortion level in the case of tape -- and this is what the constant
noise level referred to. It's often quoted as a negative number, meaning that the noise is
so many decibels below the reference level.
A compact disc has an S/N of better than -90dB, and is thus
virtually noise-free. The first cassette decks boasted something like -40dB, and sounded
as though a shower was running in the background all the time.
Magnetic recording is inherently noisy, and the need to add
AC bias for linearity means the "window" between the maximum recordable level
and the noise floor is quite small. And the narrower the recorded track, the higher the
noise: it increases by 3dB -- doubles, in fact -- every time you cut the track width by
half. Thus the tiny tracks used in cassettes were unbearably hissy until improvements in
tape made higher maximum levels possible, and Dolby Noise Reduction lowered the noise
floor.
Mono FM is reasonably quiet, but stereo uses controlled
cancellation of the signal to achieve the stereo effect, and this increases noise
considerably. At one point, Dolby Laboratories introduced a noise-reduction scheme for FM,
but it never caught on.
One characteristic of FM is that, as signal level
increases, the audio volume stays the same but the noise level drops. Tuner sensitivity is
specified as the signal level needed to produce a stereo signal with S/N of -50dB. That's
not a whole lot better than the original cassettes, but in fact it's not typical for
listeners in the city. Tuner sensitivity has improved enough over the years that FM noise
is now only a problem in fringe areas.
In analog equipment, problems such as hum and hiss exist
because the equipment can't distinguish them from the wanted signal, and although various
technical tricks have been devised to keep the noise to manageable levels, it's always
there to some extent. In digital storage and transmission systems, there may well be lots
of the same sort of noise, but the reproduction system just ignores it. For those of us
who have spent decades trying to achieve a clean signal, that's nothing short of
miraculous.
...Ian G. Masters
ian@mastersonaudio.com
|
