"Ray Andraka" wrote in message
news:deZqf.31975$Mi5.3388@dukeread07...
Hey Joel, what are you doing over here. Are you a pilot too?

Hi Ray!

Hmmm... no, I'm not a pilot, I just got sucked in by the cross-posting
(starting from sci.electronics.design) and the topic has drifted considerably
since it started.

I've used this for subsampling, although you have to be very careful of
clock jitter when you sub-sample. a couple picoseconds of jitter on the
sampling of a 100 MHz signal is going to add substantial noise to the
signal.

Yes it is... I suspect that's why that projects such as GNURadio (which
sub-samples using something like 80 or 100MSps ADCs) tend not to be as
sensitive as more traditional analog receivers. (Someone made the comment
that the FM decoder in GNURadio doesn't really even work as well as a $5
transistor radio, which is true enough albeit perhaps missing the point of how
cool/fun it is to be able to write any modulator/demodulator you like if
you're not looking for the ultimate sensitivty.)

(For example, if you have a signal centered at 100 Mhz that only has a 10
MHz bandwidth, you can sample it at something less than 100 MHz and still
recover all of the information. The more generally held belief is that you
would need to sample it at greater than 200 MHz in order to not lose
information).

I believe that folks who think you need to sample at 200MHz (the intuitively
reaosnable answer) are those who were never made to /had the opportunity to
open up an undergraduate signals & systems book. :-) Thinking about things
like modulation are so much cleaner in the frequency domain once one gets the
whole "multiplication in one domain is convolution in the other" bit down.

---Joel