"daestrom" wrote in message
...
Aliasing happens on the analog to digital conversion, not the digital to
analog conversion. That's why low-pass filters are put 'in front' of analog
to digital converters.

You might not call it 'aliasing' -- it's arguably more appropriate to call it
'imaging' -- but a DAC that only outputs unit impulses scaled by the desired
output level creates infinitely many replicas of a band-limited sampled input
signal. Adding a first order (sample-and-) hold thereby gets you infinitely
many replicas scaled by a sinc function and -- as you mention -- typically
needs to be corrected or 'smoothed.' It isn't uncommon to purposely make use
of one the replicas, though, just as it isn't uncommon to sub-sample a
band-limited signal at well below its center frequency.

But this is *not* aliasing ala Nyquist.

It's all just linear system convolution with sample functions, hold functions,
etc. 'Nyquist aliasing' is one of those kinda vauge terms where it's usually
clear from the context what's meant, but it doesn't have any particularly
formal meaning. (I can't tell you how many times I've seen people stating
something like, 'The Nyquist theorem requires sampling at at least twice the
highest frequency present in the signal," when of course it says no such
thing.)

---Joel Kolstad