[sdiy] Screwing with Square Waves

[Andrew Simper]

On 2 November 2013 00:29, Neil Johnson <neil.johnson71 at gmail.com> wrote:
>...
> The following video demonstrates these two points quite well:
>
> https://www.youtube.com/watch?v=cIQ9IXSUzuM&t=1040
>...
> Cheers,
> Neil

A little note on this video, the square wave showed isn't a
"bandlimited square wave" as stated (21:37) , it's a bandlimited
reconstruction of a set of points sampled from from a non-bandlimited
(ie trivial) square wave. He does use the word "apparently perfect"
earlier on which was a nice work-around, and I double checked with
Monty and he conformed this is one of his "glaring omissions" to make
the video flow better.

The upshot is that if generate a theoretically perfect step transition
and sample it using an infinitely long sinc function you never end up
with a set of points that only has two values, you always have wobbles
near the transition no matter at what fraction of a sample you start
from:

https://dl.dropboxusercontent.com/u/14219031/Chat/analytic-bandlimited-step.pdf

Andy