It’s a common misnomer that digital has "steps." To be fair, it seems intuitive that it does. But it doesn’t, as proven here. And analog has limited resolution, too - just as with digital.
It is certainly not proven with this youtube. Digital is sampled, not continuous. The reproduction accuracy with digital is a function of the sample-rate and filtering to "smooth" those steps. Any NOS D/A chip will output these stair-steps and requires filtering to eliminate them. If you disable or raise the frequency of the digital filtering from any Delta-Sigma D/A chip, it will have these stair-steps. I have seen it on my scope.
There most certainly is a Nyquist theorem. A fairly good explanation of it is here.
The Nyquist theorem is true and often cited, however, it makes some assumptions such as the waveform is continuous and not transient. Transient waveforms cause the Nyquist theorem to break-down.
Not that it doesn't apply, but the sample rate required to get an accurate transient reproduced is much higher that Nyquist would predict.
Steve N.
Empirical Audio
