DACs do indeed sound quite different - although, compared to, say, the difference between headphones,
you may or may not consider the differences important.
There are a whole bunch of issues which are being glossed over here.
The analog circuitry in the DAC certainly makes a difference, although the pervasive assumption that "discrete op amps"
are better than chip ones is not always true. There have also been studies that have shown that the better grade of op amp
chips don't sound especially different IF PROPERLY IMPLEMENTED. So, IN A GOOD DESIGN, changing the op amps won't
necessarily make a difference and, if it does, that difference may not be an improvement. Putting a "better" op amp into a circuit NOT designed for it may actually make things worse instead of better. Unfortunately, some op amps do sound noticeably bad
in certain types of designs, and, sadly, not all products are well designed. The point is that changing chips, or changing from
chips to discretes, may or may not be an improvement - depending on the overall design in which you are using them.
Simple specifications (like THD and S/N) usually don't tell much of a story for DACs because, for all except the poorest quality DACs, those numbers will be very good.... probably well below the limits of audibility. This is NOT, however, the same as saying that numbers don't matter, or claiming that there is some "intangible" involved. You just have to look at DIFFERENT "numbers". Things like "soundstage" and "depth" aren't intangible; we just don't have ways of measuring them directly. It's pretty obvious, for example, that soundstage has something to do with the phase relationships between various frequencies and the timing between them. (It's just not a simple and easily quantified measurement.)
One important "number" is jitter. Signal jitter definitely affects the way a DAC sounds, and DACs vary widely in their ability to eliminate jitter (or avoid being influenced by it). [It's also possible that extra jitter may artificially make the soundstage wider, which may be perceived as an improvement in some situations - with some music. If so, then eliminating that jitter, while more accurate, may NOT be perceived as an improvement.]
Another major differentiator between DACs is the performance of the digital and analog filters. Since those filters vary widely between DACs, and have a major effect on impulse response and phase accuracy, it's a fair bet that they have a lot to do with "soundstaging" and "transparency". Most DAC chips have digital filters built in, so it's reasonable to expect that units using a certain chip, and using the built-in filters that come with it, would have a similar characteristic sound - and may sound different than DACs using a different chip that has different filters. (Some chips have multiple choices, and some DACs use custom filters rather than the built-in ones. If you listen, even briefly, to a DAC that has switchable filters, you can cheerfully argue forever about which one is best, but it's pretty obvious that they are indeed subtly different.)
Headphones, being basically speakers, also have requirements in terms of being supplied with sufficient current and damping. Various studies have shown that many (but not all) headphones sound rather poor unless connected to an amplifier with a sufficiently low output impedance. Of course, again, this is something that can be measured and specified. Simple op amp based headphone amps, therefore, work very well with some headphones, and not well at all with others.
As has been noted, the differences between DACs are subtle enough that expectation bias is a major "risk" when comparing them. There are also enough variables that it can be difficult to analyze them all at once.
There ARE ways to isolate some of the variables, however.
For example, if you have a USB (and USB powered) DAC feeding your stereo, and you notice a lot of ground noise (the kind where you can "hear" what's on your computer screen chortling out of your speakers at low volume), try adding a simple "USB isolator" between the computer and the DAC. I'm NOT talking about an expensive USB-to-S/PDIF converter.... just a "USB isolator". Olimex and several other manufacturers sell them for about $50 (I bought one from these guys http://electronicsshop.dk). Most of them are based on the Analog Devices ADUM series isolator chips and should be equivalent. The one I tried totally eliminated the ground noise (and had no effect on anything else).
Originally Posted by bcschmerker4
I very much agree on this point (viz., the DAC being a source), and am therefore very much interested on who has the best discrete op amps in an offboard DAC, versus the National Semiconductor LM6172IN and LME91860NA and their integrated-circuit competition; some engineers have perfected the discrete-component route for parts of the chain from DAC to headset. Among the variables routinely discussed in the high end are what headsets perform best with which amplifiers, as every transducer has some weakness, in terms of turning input current into output sound pressure across a wide spectrum of frequencies, which can be neutralized at the amplifier level. My priority is the task of giving the amplifier/headset combination an accurate audio signal to work with.