Digital Drive Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it.

---

[ Follow Ups ] Thread: [ Display  All  Email ] [ Digital Drive ]

---

[ Alert Moderator ] [ Edit ] [ Delete ]

---

Drive by Viola recital for the Audio Alternativist

Posted by Thorsten on September 19, 2000 at 13:13:21

In Reply to: Re: Question for 24/96 DACs posted by DK on September 18, 2000 at 23:42:20:

Hi all,

Now I'm hardly an authority on Digital, nor am I overly fond of CD Standard Digital Audio. It get's on my good days a comment of "sounds not very good", if I'm peved off about something and don't feel like being PC and polite and say it sounds awfull.

There are few exceptions and often not the ones one expects.

Now for upsampling. Upsampling is basically sample rate conversion and in turn sample rate conversion is oversampling with a non integer oversampling factor. As a result "upsampling" has been used in ALL CD-Players apart from the first generation Sony Machines.

Ultimatly upsampling and/or oversampling can be used to make the job of the analogue reconstruction filter easier, assuming that one accepts in the first place that uch is needed. I cannot help wondering what a first generation Sony Machine would sound like with a decent new analogue stage. Has anyone a working one to donate?

Now I have most of the big name stuff and quite a bit of the less big name stuff. Of what I have heard to date the Audio Note Japan Hibari DAC driven from an Ensemble Dicrono Transport (with ANJ Electronics and B&W Nautilus 802's) was simply outstandingly the best. Yes, I am quite familiar with the dCS Purcell/Elgar Combo. What struck me most was not the "improvement" in sound (sure it sounds different, not really better IMHO, but different) but the extreme sensitivity to the Digital interconnect between upsampler and dac.

Use the "wrong" one and the upsampled sound is worse than the straight connection. As said, the sound upsampled when optimised properly is different, not better or worse, but different. As comparisonsn where made in a system that I'm familar with but that runs contrary to my own tastes in sound I cannot be conclusive, but certainly overall this system compared to the one from ANJ was not even in the same class, playing cuts I was very familiar with.

My own personal experiences at home with non oversampling DAC's (my own "Andante") against stuff the CS8420 Upsampler again convince me easily that there is NO merit at all in Upsampling.

In the end my view based on simple information theory is this:

If a given signal is processed the maximum amount of information that can be decoded from the signal is equal to that encoded in it. Ususally a minimal loss is present in the decoding chain.

Using the TDA1541 without oversampling will get to 3db within the 16-bit theoretical limits, that is nearly the practical limit for the technology.

Anyway, any application of oversampling and/or upsampling will add samples to the signal. It matters little what kind of processing is used to create the samples, in the end we have two problems now. In order to retrieve the original inormation we now must accuratly decode the new sample rate and indeed, we have ADDED something to our original Signal.

By definition, the added component is a DISTORTION of the original signal, as oversampling is easily shown to "work" with sinewaves, reducing severely the presented THD. So how is less distortion actually more? Well, the INPUT SIGNAL (eg. the signal on the disk) actually had only so much resolution to record the changes of the Sinewave, hence measured harmonic distortion.

By having less distortion after the Digital Signal processing than before we have added an artificial component to the signal. Depending upon your view this may be viewed as noise or distortion, but the one thing it is not is the original Signal. Indeed, this is irrevocably lost and replaced by something SYNTHSIZED in the digital processing.

Okay, what is the upshot?

Using no oversampling/upsampling will lead invariably to something that measures very unconventional (that is if you want to be polite, if you don't it measures so piss poor that if the measurements actually mattered for a penny it should not even work). The sound quality will be dependend upon and essentially ONLY upon the excution of the principle and how the disadvantages are managed.

Using oversampling/upsampling will invariably (assuming at least semi competent implementation of DAC and analogue circuitry) lead to practically perfect measurements (so not only should all such items sound "perfect" but also identical if these measurements actually mattered for a penny). The sound quality will be dependend upon and essentially ONLY upon the excution of the principle and how the disadvantages are managed.

So who wins? Whoever makes what sounds best. To me that is so far Audio Note Japan.

Later T

This post is made possible by the generous support of people like you and our sponsors:

Kimber Kable

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2025, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---