Like the sampled signal being a sinus wave :-) .. and not counting the samples to end up fx. at 0 ( 0. 180 degrees ).
The nyquist theorem only states the MINIMUM speed required to sample the signal :-) ... not what you actually need ... / regards, Lars Segerlund. On Wed, 24 Nov 2004 10:20:00 -0500 "Peter J. Desnoyers" <[EMAIL PROTECTED]> wrote: > Charles Lepple wrote: > > > > The Nyquist theorem (20 MS/sec would give you a theoretical upper > > bandwidth of just under 10 MHz) assumes ideal sampling, and probably a > > bunch of other caveats that I can't remember now. > > For a periodic signal that repeats exactly with a period t, and a > maximum frequency f, you need 2ft samples to represent it. That's not a > sampling rate, but a total number of samples. And you don't need to get > all the samples during a single period of the signal. > > This is the basis of sub-nyquist sampling - if the analog bandwidth into > the A/D, sampling jitter, and a few other things are good enough, then > you can sample at 2f/N for N (+some delta) repetitions of the signal. > > Basically you're doing what an analog scope does - you're capturing > multiple iterations of the same signal, and averaging them together. And > just like an analog scope, if there is any variation in the repetitions, > it will show up as noise. (although sometimes the noise can be harder to > interpret - I know from experience that certain digital scopes can show > timing jitter as an impossibly high frequency noise on top of the > signal, instead of the multiple traces you would see on an analog scope) > > -- > ..................................................................... > Peter Desnoyers [EMAIL PROTECTED] > UMass Computer Science (617) 669-4728 >
