SV: Re: [sdiy] Nonlinearities in IR3109 filters

[René Schmitz]

Hi Sean and all,

Sean Costello wrote:
> All digital filters use delay in the feedback part. Not by choice, but by
> the nature of discrete time systems:
> 
> onePole1Input = input+feedback;
> ...bunch of operations...
> feedback = OnePole4Output; // computed last, as it is the output of the
> filter
> 
> The feedback can't be added back into the system until it has been computed,
> which means that you have a delay in the feedback loop. Filters with
> zero-delay feedback loops are not realizable, which is why digital
> translations of analog filters are not trivial to do.

Well, this exists as well in an analog filter. But the signals travel 
faster there. So if you give your digital emulation the same bandwidth 
(->fs/2) as the originals circuits have, you end up with a closer 
emulation.
Actually its trivial to raise the sampling frequency. (Brute force :-))
Not trivial is to try to do if with minimal computational resources, 
where some corners have to be cut.

Cheers,
  René


-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159