The Logic Off Topic list

> From: Spectro <spectro@...>
> 
> The problem lies in allowing >> nyquist frequencies to get to the
> the output, as for example would happen in an unfiltered saw or
> square oscillator at even modest frequencies with a less than ideal
> filter at the converter stage.

Actually, the problem is in how you generate synthetic waveforms.  In the
digital (discrete time) world, you could programmatically generate a
waveform in such a way that it's alreay aliased, so that no amount of
post-filtering could save it.  There's no rule that says that just because
you're in the digital realm you can't make aliased tones.  Once you realize
that, then you realize that it actually is not that easy to generate
non-aliased waveforms, and that you have to be very careful not to
accidentally generate aliasing with some process (such as clipping).