Square Wave Response

[Samuel Groner]

That's simply the upsampling filter. As I said before, there's no need for digital filter to have symmetrical response (look at the PDFs I linked).

Gibbs Phenomenon basically relates to perfect (read: sinc-based) lowpass-filtering with different cutoff-frequencies. An upsampler may be approximating sinc filters but not necessarily so. And if the square wave has "odd" frequency with respect to sampling frequency another variability comes into play as the edge of the square wave lies "between samples", so to speak.

Samuel

 

[gnd]

Thanx, John, Samuel.

If I understand properly, the ringing I observe is part of EMU design. Limited bandwidth gives perfect Gibbs phenomena, and filter's phase shift causes assymetry.

Is there a way to improve this at output? Or is there any need at all to improve it? Could I say that such ringing is preety much beyond perception?

I tried changing samle rate to 192kHz. Almost perfect 100Hz square. Still some overshot, but much less ringing, and of course ringing moves to very high frequencies. But in the end 440Hz square at 192kHz looks very similar to 100Hz square at 44kHz.

...

 

[JohnRoberts]

All of the apparent artifacts are above the cut off frequency and while perhaps not obvious from inspection not extra signal but the consequence of the missing HF information.

JR

 

[gnd]

[quote author="JohnRoberts"]All of the apparent artifacts are above the cut off frequency and while perhaps not obvious from inspection not extra signal but the consequence of the missing HF information.

JR[/quote]

So it seems that all is fine, after all. Thnx.

 

[bcarso]

That wiki on Gibbs is a good one---thanks for posting the link gnd.