dcollins wrote on Sun, 04 July 2010 13:07 |
bruno putzeys wrote on Sat, 03 July 2010 23:13 | So if we can now get people to understand that a dithered digital channel is no different from a noisy analogue channel, we'd be set to enter the 21th century...
|
So far, I've had zero luck trying to relate this concept to people.
They have read that "It's all stair-steps" or as I heard last night from a very reputable designer "In digital the lower the signal level, the more distorted it gets" so I wouldn't hold my breath.................
DC
|
Well, I dont think it is that simple...
Within perfect workingconditions they are the same, what gets in is what gets out.
Digital is digital and the faults introduced by digital (when the lo/hi boundaries are broken) is characteristic for digital, this is ofcourse in theory where everything is perfectly by the book. But digital technology today actually relies on analog components.
If you can manage to manipulate exactly one "analog" informationcarrier at a time with exacly the same conditions everytime, then it would be identical to digital theory, but thats more on the scale of nanotechnology rather than normal "big and heavy and less than perfect" electromechanical analog equipment, where each single informationcarrier is hidden way below the noise.A vinylneedle touches many molecules, and neither the vinyl or the needle has a perfectly shaped and perfectly smooth surface. Just as the tape has molecules that is less than perfectly organised on the surface and the head-molecule distance will also vary from molecule to molecule. So even if the tapemolecule itself is digital then still the threshold will change because the variations in the distance from the head.
It would be more correct to say that analog equals digital if the digital has enough selfnoise (with enough resolution below) to hide many bits.Or in other words: 96db of S/N Digital dont equal the same resolution as 96dB S/N analog. 96db Analog is more like maybe 120dB of digital resolution with 96db of noise.
And, Analog usually has a lot of faults that add a new dimension when the workingconditions arent ideal. Like nonlinearities close to the boundaries, some memoryeffects, like heat, from earlier signals and so forth.
And this could *maybe* with the ears integrationwindow and huge processingpower in mind also be regarded as a second level of dither.(Zero crossing varies slightly between each crossing because of random nonlinearities for instance, but put together they give a more precise picture of a single crossing)
And that analog "faulthandling" close to its boundarys with the sound from it that many describe in words like "natural" and "organic" or even "musical", also has to be taken into account when a new design is made regardless of on which technology "the medium itself" is made.
Well, just some fast thoughts from me, feel free to comment it and point out its faults