[Scott Dorsey]

Roy W. Rising wrote:

Intellectually, I know that using a 44.1 KHz sampling rate there should be
some kind of anomally when a sine wave of 11.025 KHz (44.1 / 4) is
recorded. After all, there are so few samples per cycle from which to
reconstruct the waveform!

There are PLENTY. It takes two samples to reconstruct a sine wave, and of
course everything is really the sum of sines. Those two statements are
the base on which the Sampling Theorem is based.

There is a good intuitive discussion of this in the FAQ.

But ... I've done the test ... A/D~D/A @ 44.1
KHz. As my sine wave generator sweeps upward from 10 KHz, the
ocilloscope's waveform display stays smoothly sinusoidal. At 11.025 KHz,
nothing changes! As Carson would say "How dey do dat?" Somewhere in the
darkness Ron Extes whispers "algorithms!"

No, not at all. It takes two samples, and you know everything about a sine
wave. It takes two samples per cycle of the highest frequency in a bandlimited
waveform, and you know everything about it.

It seems confusing at first, but JJ's explanation is good... and when you
get an intuitive idea of how it works, it's just plain neat.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."