Quote:
Originally Posted by P_A_W /img/forum/go_quote.gif
I am not sure what you mean by frequency response.
The power response (as a function of frequency) will tell you about how uniform the amplitude response is, but it gives you no information about the phase. To what extent can you actually hear a phase shift (between frequencies)?
|
With loudspeakers, "power response" typically has a different meaning than you're thinking. It refers to the amount of total power emitted into the room (i.e. not just straight ahead), while "frequency response" refers to the amplitude response measured on-axis (straight ahead). Eventually most radiated sound reaches your ear through reflections, so the power response is a much more accurate gauge of how a speaker sounds. This was the subject of an enormous amount of research in the 70s and 80s.
Any multi-driver speaker has a non-uniform power response, the extent of the non-uniformity being determined mostly by the size of the drivers, the crossover point, and the steepness of the crossover. Different issues affect the bass region.
There is a debate about whether phase response is audible. Some believe it is audible, but only in the midrange, others believe it is only audible in the bass region.
Quote:
Assuming a linear component, the impulse response (or step function response) would give you both amplitude and phase at all frequencies. Of course, it is unclear exactly how to interpret this data when comparing two different components. |
Yes, precisely. It is always possible to measure, but the difficulty is in determining what is audible. Unfortunately, the measures which are the easiest to quantify (linear distortion and THD+N) are the least useful. Get used to reading nonlinear distortion spectra and correlating them with sound quality.
Quote:
(ie Does that $3k interconnect really significantly affect the signal or is the dominant source of degradation the amplifier or the DAC. Sure, you could demo it, but who has the time?) |
No, $3k interconnects and power cords typically don't make any significant measurable difference and aren't worth spending any time with. That said, the impact of a long run of unbalanced cable will be measurable.
Quote:
It appears from the length at which you wrote about non-linearities that you believe this to be the dominant source of audible distortion. Is this universally accepted? |
For the most part, yes. It is not difficult to get the linear behavior of a system right (the exception being headphones actually, where their linear behavior is all over the map and there is no universally agreed upon ideal linear response, because the head has been subtracted out). There are some cases where the linear behavior has an impact, such as the power response of speakers and the shape of a DAC's digital filter.
Quote:
For instance, for a DAC what is the dominant source of distortion? |
Linear distortion? The digital filter. Non-linear distortion? The analog output stage and I/V conversion.
Quote:
When these components were designed, I guarantee that the design strategy wasn't twenty golden-eared audiophiles ABXing every component choice. |
When you get into the true high end ($5,000+ gear), they often don't do any measurements at all, they just design by ear. In most cases this produces poor sounding gear, which is why you should avoid the high end.
If you're talking sources, you won't get any better than the Benchmark DAC1 or the Lavry DA10, no matter how much you spend. Those two are pushing the state of the art in digital to analog converters, and both are reasonably priced and neither uses boutique components. As you would expect from high quality, well-measuring gear, those two sound more similar than different, though they both have a slightly different flavor due to their use of different DAC chips and different output stages. Despite the advertising, recent revisions of the DA10 use exactly the same jitter recovery scheme as the DAC1, which I find a bit of a letdown, but they're both good.
Quote:
Sure, one could get into DIY, but these are commercial products and people have already made the measurements on better equipment than I am ever likely to buy. These companies won't tell us what the specs are because it would allow us to compare equipment without bringing it home. |
Yes and no. You'll never be able to compare equipment just by looking at measurements, because there are too many variables. For instance, in a speaker, whether you prefer a slightly more uneven power response over a peak in odd-harmonic distortion around 5kHz is going to be a matter of personal preference. There are many engineering tradeoffs in designing a quality speaker. It's the same with amplifiers, which is why a well-designed tube amp is competitive with a well-designed solid state amp, despite THD+N numbers suggesting otherwise. What measurements are good for is winnowing out the garbage from the quality gear, so you know what to audition. Once you become familiar with your own sonic preferences (e.g. can you hear a peak in a speaker's frequency response at 18 kHz?), you can do this a lot more effectively.