I often think about this issue as yin-yang (dark-bright), and a good product has this in balance - but what the correct balance is does depend somewhat on taste!
So yin - dark - is in technical terms, happens with zero noise floor modulation. Conventional DAC's have enormous levels of noise floor modulation. This means noise (bright hiss) pumps up and down with the music signal, and the brain can't separate a dark sounding instrument from the noise floor modulation - so smooth sounding instruments become bright. With Chord DAC's, including Mojo, there is no measurable noise floor modulation, so it innately sounds smooth and warm.
But its possible to artificially give the
appearance of more yin by contouring the sound. For example, add a lot of second harmonic distortion, and it sounds thicker and darker - but its an illusion, as everything sounds soft. You can also add LF errors too, to give the
impression of more weight to the sound - adding electrolytic caps, or letting the ref circuitry amplitude modulate the output from the signal envelope. Indeed, a lot of designers rely on this, as they do not have the abilities (stuck with using chip DAC's) to solve noise floor modulation, so have to use tricks to balance the sound.
On the yang side, natural brightness comes from two sides. First is transparency, and this resolves into detail resolution, and this is about how accurate the DAC/amp can resolve very small signals accurately. With my work on the reference DAC Dave, I discovered that there is no limit to how accurate the small signal needs to be - the smallest possible amplitude error is very audible, particularly in terms of sound-stage depth. Transparency is a complex issue, but comes down to two main issues - simplicity of the analogue section (each component degrades small signal linearity) and the performance of the noise shaper (before anybody says ladder DAC's these are awful for small signal linearity). Now Mojo has an extremely simple output stage - only one active stage and two resistors and two capacitors in the direct signal path, and this is done for transparency. On the noise shaper, it has 1000 times more resolution than conventional noise shapers, as the noise shaper runs at 104 MHz, not the usual 6 MHz of the best chip DAC's.
The second part of yang is timing. Now digital audio is sampled data, but the original signal in the ADC is a continuous signal, and the job of the DAC is to convert the sampled signal into a continuous analogue signal with the timing of the original signal in the ADC perfectly preserved. Now I talk a lot about reconstituting timing, and have had requests to show the problem. So here is a simple illustration of the problem:
[COLOR=FF4400]
[/COLOR]
Now this is a bit of a simplification - the burst signal is not bandwidth limited, but it serves to illustrate the problem of timing inaccuracies. Now how do these timing errors sound like? When the brain comes across timing errors, it can't deal with it - it can't make sense of the music. And when the brain can't process the signal, you then can't hear the transients. It is a bit like putting a picture out of focus, blurring the edges. What this does audibly is to make transients sound soft, and when one improves timing accuracies then the brain can perceive the starting and stopping of notes accurately - so things sound sharp and fast - more yang. Now what is curious about timing errors, is that there again is almost no limit to how small they need to be - before Dave, I used to think in terms of uS errors, now its definitely nS as being important - extremely small timing errors have a noticeable subjective musical impact.
Also it is very possible to use distortions to give impressions of good sound - use slew related noise floor modulation and you get the impression of good timing resolution - but its entirely false. The problem with using distortions like this, although it can sound superficially impressive - is that everything always sound the same. But the major problem with this approach is simply listening fatigue - I can listen to Mojo for 10 hours and still want more. It also illustrates the design nightmare of listening tests - is the sound quality "improvement" real or just more distortion or aberration? You have to be extremely careful on how one assesses sound quality.
So too conclude - Mojo can sound both rich and dark (immeasurable noise floor modulation) and very fast and dynamic (much lower timing errors) all at the same time. That's why we get so many different reactions to the sound of Mojo - some saying its rich and smooth, some saying its fast and dynamic - and the truth is both observations are correct.
Rob
