Since I’m at a lose end for a moment, here is a bit of an overview of how I see a range of technical issues.
DACs. Over the years DACs have improved astoundingly. The early (i.e 80’s when CD players came on) DACs were quite primitive, and many could not actually resolve the full 16 bit range of a CD. Over time tha amount of digital logic that could be thrown at the digital part of the system has grown for essentially zero, to almost arbitrary amounts. A DAC is a funny device, it lives half in the digital world and half in the analogue. What is technically very interesting is that the boundary between the halves can be moved. In implementing any DAC the devil is in the details. As has been alluded to above, DACs may be sensitive to issues that are not obvious or easy to comntrol. The S/PDIF interface is a fundamentally flawed design because it emebeds the clock in the same stream as the data, and it is possible to recover signal correlated timing artefacts from the clock - a recipe for weird and objectionable problems in the sound. A lot of DAC design over the years has concentrated on resisting jitter - which is the term used for timing noise.
The basic figures of merit for DACs are essentially useless. Most numbers are the simple raw total harmonic distortion (THD) which does not convey most fot he problems a DAC can exhibit. Some DACs will have a jitter figure, but even this is actually useless without further qualification. It is easy to make a DAC resist high frequency tiiming noise, but much harder to make it resist lower frequency noise, and it is the lower frequency noise that causes the trouble.
DACs are extraordinarily sensitive to layout and power supply issues. Exactly the same circuit laid out by someone who really understands the issues versus a run of the mill designer only used to either analog or digital design, but not the issues of the two together, can exhibit significantly different final performance. This is IMHO a vastly understated problem. In the past many very expensive DACs were designed by people who came from a HiFi analog background, and there were total disasters in design that still went for silly money. It is possible to make a very good DAC with cheap parts if the design and layout is done carefully, and trivial to make a mediocre DAC with very expensive components if even one mistake is made in layout.
DACs in computer sound cards are a crapshoot. The test figures are essentially meaningless, and issues that come when you actually connect the output to your sound system - with ground loops, other noise mechanisms, noise from the other digital components - making it hard to get the intrinsic performance you might hope for. But of you are looking for reasonable sound and cheap, it is a good way to avoid spending money.
Amplifiers have similarly changed a lot over the years. The big advance has clearly been in usable class D amplifiers. Tripath led the way, and even although they went bankrupt, you see their chips still used in many cheap and good systems. Many other manufactures make class D chips now, TI for one. At the high end you get companies like B&O selling their Ice power designs to other manufactures, and at the very high end the likes of N-Core making really good amplifiers. However many conventional (class AB and class B) designs are still used. In the realm of home theatre you see Pioneer with Class D, but Onkyo, Marantz, Denon still use Class AB.
Improvements in understanding of design theory and newer components have led to much better amplifier designs and cheaper.
Specifications on amplifiers are mostly useless. Power is close to the only one of a manufacturers specifications that is meaningful. The rest do little more than make you feel good, although with some technical knowledge you can gleam some other facts about the design if you know what to look for. The only power worth looking at is RMS. Things like peak, or “music” power are just flights of fancy. RMS power gives you a solid grounding in the actual capability - covering thermal and power supply performance. Numbers like damping factor are essentially meaningless. Conventional moving coil loudspeakers expect to be driven by a voltage source, and a variation in output impedance of the amplifier from 0.4 ohm (dampling factor of 20) to 0.01 ohm (damping factor of 800) will make a just measurable difference in the very low bass response, and nothing more. In a conventional amplifier the output impedance is nothing more than the output device impedance divided by the feedback factor. It is pure marketing drivel. Total harmonic distortion figures are of marginal value. One thing you can look for is the point where the distortion goes very high. You sometimes see numbers like 0.05% THD, and 1%THD at 20 Watts. This tells you that the amplifier is at its absolute design limit at 20 watts. In reality such designs are more like 12 watt amplifiers.
Power ratings and needs for speakers are not well defined. Speakers have two main power limitations - how much power is needed to drive the bass driver to the point the voice coil slams into the pole-piece and is wrecked, and how much power will exceed the thermal limitations and burn the voice coils out. Both tweeters and bass driver can be burnt out depending upon the type of music. A maximum power rating for a speaker is hopefully going to be suggesting an amplifier power that is below the level you are likely to kill the speaker, and a minimum one where the manufacturer feels you may not get full benefit for the speaker. But it isn’t something to worry a great deal about.
Impedance ratings on amplifiers are an indication of the safe operating capability of the amplifier. A conventional amplifier, acting as a voltage source, will deliver more and more power into lower and lower load impedances, until either the power supply or output devices can’t handle any more. An amplifier is limited by the voltage of its power supply and the maximum current the supply can deliver, and the safe operating conditions of the output devices. This means you can, in principle, use a speaker with less than the recommended impedance, so long as you don’t ask the amplifier to deliver its full rated power. However there are also other issues. It is harder to make an amplifier deliver good distortion figures into lower impedances, and so pressing an amplifier into service in such a manner is probably to be avoided. The need to match amplifier impedance to speakers only matters with tube amplifiers, and in this case the load impedance is selected on the amplifier and is determined by the design transfer curve of the output devices and the impedance ratio of the output transformer. (Most tube amplifiers will tolerate a 2:1 mismatch without much issue.)
Something to beware of when using a computer as an audio source is that unless you are careful to operating system will mess about with the audio stream. It may re-sample the audio to a different sample rate, it may pass it through various mixing stages, and it may reduce the dynamic range. Turing off all this unwanted mess is worth investing time into.
