"isw" wrote in message
...
In article ,
"Ron Baker, Pluralitas!" wrote:


--snippety-snip--

You said you are a physicist/engineer.
What does "linear" mean?

Let's not get too far off the subject here. We were discussing whether
the "tuning beat" that you use to tune a musical instrument involved a
nonlinear process (ie. "modulation").

Then linearity is at the core of the matter.
What does "linear" (or "nonlinear") mean to you?

OK, if you insist -- *in this case* it means "linear enough to not
produce IM products of significant amplitude".

Good enough.
Then spectrum analyzers and the human auditory
system are not linear.
Stay with me here.


I said that it does not, and that
it could be detected by instrumentation which was proveably linear
(i.e.
not "perfectly" linear, because that's not required, but certainly
linear enough to discount the requirement for "modulation").

No nonlinearity is necessary in order to hear
a beat?
Where does the beat come from?

As the phase of the two nearly equal waves move past each other, there
is simple vector summation which varies the amplitude.

Consider two sine waves of precisely the same frequency, where one of
them is adjustable in phase -- use a goniometer, for instance. Use a set
of resistors to sum the two signals, and observe the summing point with
a 'scope or a loudspeaker. By altering the phase of one source, you can
get any amplitude you want from zero up to twice the amplitude of either
one.

Now just twiddle that phase knob around and around as fast as you can.

You've just slightly altered the instantaneous frequency of one of the
generators (but only while you twiddle), and accomplished pretty much
the same effect as listening to the beat between two guitar strings at
nearly zero frequency offset. With no nonlinear processes in sight.

Isaac

You put some effort into that. I give you
credit for that.

The socratic thing isn't working, so here
you go.

Is an envelope detector linear? The answer is no.
But how can that be? If you put in a sine wave of
amplitude A you get A volts out (assuming its gain is 1).
If you put in a sine wave of amplitude 2A and you
get 2A volts out. Linear, right?
Now you put in a sine wave of amplitude A at
455 kHz plus a sine wave of amplitude A at
456 kHz. (Consider the envelope detector
of a typical AM radio here.) What do you get out? A
sine wave of amplitude A/2 at 1 kHz. Intermodulation.
An envelope detector is not linear. No envelope/
amplitude detector is linear.

The typical envelope detector is a diode rectifier
followed by a lowpass filter.
The diode rectifier is obviously nonlinear and
gives you all sorts of intermoduation. With a
single sine wave input you get a DC term and
various harmonics of the sine wave. The lowpass
filter filters out all the harmonics and leaves
the DC.
If you put in two sine waves (assuming their
frequencies are above the cutoff of the subsequent
lowpass and their difference is within the
lowpass) again the diode nonlinearity results
in intermodulation. You get a DC component,
the difference frequency, the sum, and various
higher frequencies. The filter leaves only the
difference frequency and the DC. In an AM
receiver the DC is subsequently blocked too.

Do you see how this applies to spectrum analyzers
and the human auditory system?