Gibbs phenomenon caused by digital signal processing?

[DrOnline]

Hi,

I'm an electrical engineer for a few years now, but it's been a while since I had to deal with this kind of stuff, I turned out to become mostly a programmer in the end, but i was thinking: is Gibbs phenomenon, which was demonstrated to me during my studies while working on Fourier series, something which is caused by recreating a digitized signal, or does it also exist in purely analogue electronics? I find the first scenario understandable, but not so easily the second.

This is not some urgent matter I need help with, I just wanted to see if I could get some explanation.

I guess I generally find, that a lot of the things I learned while studying, are harder to glue together into cohesive and lasting understanding, than it was to simply parrot back on the day of the exam!

 

[phinds]

As best I can recall, the Gibbs phenomenon occurs at discontinuities in a wave. Think of a sine wave that has an instantaneous phase shift of 180 degrees. Wouldn't the Gibbs phenomenon happen for that as well as for a square wave?

 

[Klystron]

[...]is Gibbs phenomenon ... something which is caused by recreating a digitized signal, or does it also exist in purely analogue electronics? [...]

IIRC ringing, over/under-shoot, and similar phenomenon are present while sampling analog circuits depending on filter designs and noise conditions. Not always unwanted as early musical synthesizers and 'fuzz tone' modulators took advantage of these phenomena to achieve natural sounding and special effects.

 

[sophiecentaur]

I would appreciate someone pointing out where the Gibbs Phenomenon needs to be looked upon as something more than an artefact due to under sampling. ( Apparently it was noticed way before the advent of modern processors and, I suspect, before Nyquist came up with his theorem (?). Digital signal processing has to follow the 'rules' and discontinuous functions are not suitable for application without some form of windowing to tame them first.

 

[Rive]

an artefact due to under sampling.

I'm new to this, but apparently this hard to kill thing is the reaction to 'discontinuous' input, and the main point is that the depth of sampling or the bandwidth of FFT used has little effect on its amplitude.
At least that's what I've found about this.

 

[sophiecentaur]

Discontinuous input implies infinite bandwidth. So all assumptions involving sampling at finite rates must be invalid. There’s the problem; the model’s flawed.