=?ISO-8859-1?Q?Re: S&H (was:Re: [sdiy] Vote for your favorite noise source)?=

[jhaible at debitel.net]

>Well, I suppose you´re right. But for equal distribution, you could just
>sample an ordinary triangle oscillator, that runs with much higher
>frequency than the sample rate (and out of sync of course).

I'm afraid this would lead to a very regular pattern.
The sampling will mix down the HF oscillator to the baseband, so you're
effectively sampling something that runs at f_osc / N, with N chosen such
that  -(f_sample / 2) < (f_osc / N) < (f_sample / 2).

JH.