On Sat, 07 Oct 2006 16:39:21 +0200, rtstofer <rstofer@...> wrote:
As Stefan points out, the 50 Msps limits the frequencies that can be
sampled. The absolute maximum signal that can be accurately sampled,
according to the Shannon Sampling Theorem, is 1/2 of the sample
frequency or about 25 MHz. And that requires that the signal be
periodic. All bets are off for one shot signals.
I don't think the shannon theorem is very useful when looking at scopes.
It can be applied for real-time sampling (like for audio) but it doesn't
apply to sampling scopes (equivalent time) for repetitive signals.
If you use many periods to sample a signal there is no requirement to
sample at least twice each period. You can sample now, and again 5 periods
later plus t, and then ten periods later plus 2*t, and then 20 periods
plus 3*t, doesn't matter.
You can buy 100GHz sampling scopes with only 10Msps. Clearly not
fulfilling Shannon.
<>
Neither is it all that practical for real-time sampling scopes, unless you
correctly apply it, which you did, to the highest harmonic you are going
to need. But estimating which harmonic you might want isn't much more
accurate than guessing how many points you'd like to have in a period of
your signal.
ST
