Older is better - D/A chip?

Audioengr--

The Nyquist Sampling Theorem requires the use of a reconstruction filter to remove ultrasonic images of the audio signal no matter what dac implementation one uses. This filtering can be implemented either digitally or by analog circuitry.

Digital filters were usually used in conjunction with ladder DACs and most nos dac chips. Back in 1982 the first Sony CD players (CPD 101) used no oversampling and analog brick wall filters at 22 khz. The Phillips players introduced oversampling and digital filtering (and used 14 bit dacs!). The controversy of the time was frequency dependent phase shift caused by the analog brick wall filters.

By 1985 or so everyone used oversampling and digital filters along with ladder dacs.

In the late 80's single bit dacs were introduced which required high oversampling rates (typically 64x or better) digital filtering and noise shaping just to function. They were really cheap to build since they didn't require the labor intensive calibration that multibit dacs needed on low order bits and became standard in consumer gear.

That launched the multibit vs single bit wars of the 90's.

A few high end manufacturers returned to the early days of no oversampling. Some even chose little or no filtering which dumped lots of ultrasonic energy into downstream components. These manufacturers also favor nos laddar dac chips. The choice to forgo (digital) filtering (yikes!) has nothing to do with the dac chip.

Along the way of realizing that good sound is timeless, I've also stopped worrying about keeping up with the latest technology. One CD player I own never really garnered much attention when it was offered, and as I have rediscovered it, find myself enjoying it more than some of the expensive machines I own.

Ghost - Thanks for the history, but I'm aware of it.

Many D/A chips do not give you the choice of avoiding the digital filtering, so the D/A chip does force this much of the time, and they auto-select the filter for each sample-rate. This simply sounds bad, not so much because of the digital filtering, because the digital filter implementations are so poor. If these were ideal, it would be great and the NOS DACs would not have such a following IMO.

But we live in a non-ideal world. I personally have had mostly poor audio experiences with typical digital filtering. This is why I made it selectable in my own DAC design. It is better IME to use more analog filtering instead.

However, I recently heard a DAC that had reasonable digital filter design, and that is the W4S DAC2 using the Sabre chip. This was using the I2S input on it driven from an Off-Ramp. World-Class SQ.

Steve N.
Empirical Audio

Elberoth - the Metrum Octave is a great piece. Many of my customers have one, but they also tell me that many of them cannot playback 176.4 or 192. Does yours?

Steve N.
Empirical Audio

Some of the newer DACs have addressed the pre/post-ringing in the impulse response with minimum phase and apodizing filters. Ayre, with the Wolfram chip, documents that well. SACD also got that much improved. Another star goes to the dithered digital volume control of Audiolab, Amarra and probably others, which often depend on upsampling.

Helped play around with DIY NOS DACs (stacking was the trend back then) with friends and while many people that listened thought it was better, I thought it was "romantic" sounding and less than honest although it was impossible to explain how. Haven't heard much of the commercial NOS products.