Headphone enthusiast thread! - Page 149

If you don't have a lot of background in electronics (even I don't, and I have a degree and a half in EE), the key phrase to pick up is "no feedback." You lose a whole lot of performance in many kinds of systems by going without feedback. Note that relatively simple discrete transistor-based designs are among the most primitive design of amplifier. Transistors are the building blocks of op amps and other electronics--if you were to instead use some op amps, dedicated headphone output chips, buffers, etc. (which contain transistors inside), there may be many models these days that would be significantly more advanced in relevant ways. You can build an amp with very high accuracy with discrete transistors, but it's a lot more difficult and would require a more complicated design...and you'd want to be using feedback. This is why I bring up the "simple" part and also why I think comprehensive testing would be important--it doesn't inspire a lot of confidence in the design prowess of the team to have the turn on/off transient issue. If you're making things harder on yourself, you need to know what you're doing.

The key advantage with a discrete transistor output stage is being able to handle a whole lot of power, at least compared to most op amps. Most headphones don't need all that much power though, and some op amps (not to mention some more expensive dedicated headphone amp chips or buffers) would be suitable for most.

This analogy sucks so don't read much into it, but it's like you want to make a building that has good structural integrity: you're complaining about a modern well-engineered concrete/steel building in an area that gets weak hurricanes, as opposed to a straw house with the self-imposed constraint that you can't use any columns.

Anyhow, how does a sound card get blasted by current? I agree that sound cards face challenges inside a computer, what with stray EMI and power dirtied by a lot of fast-switching digital circuits, and therefore I usually say that going external is better. It's just that the external design we're comparing to, is hamstrung by design (for effect maybe, but I think mostly for marketing bullet points).

The extent of these issues is going to really depend on the computer. Many don't even have dedicated GPUs, though I guess many who consider a Xonar Essence would have one. Arguably you could say that when most people are listening to music carefully, their GPU and CPUs aren't under heavy load--then again, with all the low-power states these days, maybe switching back and forth between medium/heavy and light loads would create even more problems.

There are some test results for Essence STX on Stereophile, and at least those are far beyond what you could do with a very primitive circuit design using a few MOSFETs with no feedback, despite the limitations of being inside the computer chassis. RMAA sucks, but also see some RMAA loopback tests of at least the Essence A/D and D/A chain, around the web. Some of those reviews may be with dedicated graphics cards installed. It's been shown the TPA6120 headphone output chip on the Essence (and FiiO E9) is very respectable when done right (not terribly done right but okay on the E9), so driving headphones should not be an issue other than the output impedance that circuit uses.