KeithEmo
Member of the Trade: Emotiva
- Joined
- Aug 13, 2014
- Posts
- 1,698
- Likes
- 868
First - thanks everyone for the kind words about my posts
I'd also like to take a moment to expand on my standpoint on "science" and "magic" (or "aesthetics"): I don't think you can safely separate the two in the context of our discussions here. It's fine to say that Van Gogh's painting of the sunflowers is great art, and that my picture of daisies is pretty bad, and to then go on and talk about his great use of color, which is a lot better than mine. That's all a combination of art AND science.(It's art because his is much more "pleasing" than mine; it's science because, underneath it all, there's something in the way we perceive color, or something else about the picture, that CAUSES his to be more appealing than mine to most people.)
HOWEVER, once we start seeing other painters claiming that their paintings are "as good as Van Gogh's" BECAUSE they use the colors he did, or that their competitors paintings are as bad as mine BECAUSE they use the same colors I did, a line has been crossed. We can't really evaluate their claims and make sense of them until we figure out whether the colors really ARE what makes the difference - because, if the colors are that important, then their claims may be true but, if it's not the colors after all, then their claims are bogus and can't be trusted.
For a given audio signal, the signal with the biggest rate of change will be at the highest frequency and the highest amplitude. So, for a Red Book CD, the signal with the highest rate of change possible will be a 20 kHz sine wave at 0 dB. This is a simple fact; it is NOT in dispute. So, if a Delta-Sigma DAC has "tracking speed problems", as some people seem to be suggesting, then it MUST produce high levels of distortion with a 0 dB 20 kHz test signal. And, to turn that around, if a DAC can deliver a 20 kHz signal at 0 dB with 0.05% THD, then it DOES NOT have this problem. Therefore, if a given Delta-Sigma DAC can turn in THD performance of < 0.05%, from 20 Hz to 20 kHz, at 0 dB, then we know that this IS NOT an issue for that DAC. (And so, whether some D-S DACs theoretically have this problem - we can tell easily enough which ones do and which ones don't.)
Now, as to "speed and pitch". All modern digital circuitry uses clocks that are pretty accurate and stable by human standards; a really crappy quarts clock is accurate to a few tens of parts per million (thousands of times more accurate than most turntables and most record cutters). No human is going to hear this error as a change in pitch, or as a difference in the tempo of the music. If you hear a difference where one DAC seems to have "better pace" or another seems to "slow down the music" - it's the acoustic equivalent of an optical illusion. If you actually measure the speed of the beats, or the tempo, you're going to find that they're spot-on ... or, at the very least, much closer than a human can possibly hear. Therefore, when you THINK you hear that happening, there's really something else going on. (And recalculating a sample here or there, when the samples are as close together as they are, isn't going to make a difference that you will hear as a time difference. A one foot difference in speaker distance produces a delay of about 1 millisecond, which is 44 full samples on a Red Book CD. If you dropped a sample altogether on a CD, it would alter the timing as much as moving your ear 1/4" closer to or further from the speaker. (Now, changing the relationship between the same signal in both channels may shift the image an inch or two - because our brains are very good at picking out such relative shifts, but it would still be pretty minor.)
When we talk about jitter, we're talking about time variations in the range of several hundred TRILLIONTHS of a second. You absolutely, positively, aren't going to hear that. What you hear with jitter is changes in the output that occur because the jitter causes the D/A conversion to work imperfectly.... which creates distortion.
Now, there IS a type of "timing error" that occurs with DACs, but it's related to oversampling. The digital filters used in oversampling produce ringing - which means that, when you feed a NON-CONTINUOUS signal through them (like a drum beat), the filter "smears" the start and stop of the signal in time. Different filters produce different amounts of ringing, and different filters put more or less of it before or after the original (intended) signal, all of which can make it sound slightly different. This doesn't happen with steady state signals, like sine waves, which is why it doesn't show up on distortion specs; however, it's pretty obvious with transients, so it does show up quite plainly on transient response scope photos. NOTE that this happens because of oversampling; so an oversampling D-S DAC will have the same amount of ringing as an R2R DAC that uses the same oversampling filter. (But there are scientifically valid reasons why a given amount of jitter MIGHT cause different amounts of different types of distortion depending on whether it happens to the signal being fed to a D-S DAC or an R2R DAC.)
I'd also like to take a moment to expand on my standpoint on "science" and "magic" (or "aesthetics"): I don't think you can safely separate the two in the context of our discussions here. It's fine to say that Van Gogh's painting of the sunflowers is great art, and that my picture of daisies is pretty bad, and to then go on and talk about his great use of color, which is a lot better than mine. That's all a combination of art AND science.(It's art because his is much more "pleasing" than mine; it's science because, underneath it all, there's something in the way we perceive color, or something else about the picture, that CAUSES his to be more appealing than mine to most people.)
HOWEVER, once we start seeing other painters claiming that their paintings are "as good as Van Gogh's" BECAUSE they use the colors he did, or that their competitors paintings are as bad as mine BECAUSE they use the same colors I did, a line has been crossed. We can't really evaluate their claims and make sense of them until we figure out whether the colors really ARE what makes the difference - because, if the colors are that important, then their claims may be true but, if it's not the colors after all, then their claims are bogus and can't be trusted.
For a given audio signal, the signal with the biggest rate of change will be at the highest frequency and the highest amplitude. So, for a Red Book CD, the signal with the highest rate of change possible will be a 20 kHz sine wave at 0 dB. This is a simple fact; it is NOT in dispute. So, if a Delta-Sigma DAC has "tracking speed problems", as some people seem to be suggesting, then it MUST produce high levels of distortion with a 0 dB 20 kHz test signal. And, to turn that around, if a DAC can deliver a 20 kHz signal at 0 dB with 0.05% THD, then it DOES NOT have this problem. Therefore, if a given Delta-Sigma DAC can turn in THD performance of < 0.05%, from 20 Hz to 20 kHz, at 0 dB, then we know that this IS NOT an issue for that DAC. (And so, whether some D-S DACs theoretically have this problem - we can tell easily enough which ones do and which ones don't.)
Now, as to "speed and pitch". All modern digital circuitry uses clocks that are pretty accurate and stable by human standards; a really crappy quarts clock is accurate to a few tens of parts per million (thousands of times more accurate than most turntables and most record cutters). No human is going to hear this error as a change in pitch, or as a difference in the tempo of the music. If you hear a difference where one DAC seems to have "better pace" or another seems to "slow down the music" - it's the acoustic equivalent of an optical illusion. If you actually measure the speed of the beats, or the tempo, you're going to find that they're spot-on ... or, at the very least, much closer than a human can possibly hear. Therefore, when you THINK you hear that happening, there's really something else going on. (And recalculating a sample here or there, when the samples are as close together as they are, isn't going to make a difference that you will hear as a time difference. A one foot difference in speaker distance produces a delay of about 1 millisecond, which is 44 full samples on a Red Book CD. If you dropped a sample altogether on a CD, it would alter the timing as much as moving your ear 1/4" closer to or further from the speaker. (Now, changing the relationship between the same signal in both channels may shift the image an inch or two - because our brains are very good at picking out such relative shifts, but it would still be pretty minor.)
When we talk about jitter, we're talking about time variations in the range of several hundred TRILLIONTHS of a second. You absolutely, positively, aren't going to hear that. What you hear with jitter is changes in the output that occur because the jitter causes the D/A conversion to work imperfectly.... which creates distortion.
Now, there IS a type of "timing error" that occurs with DACs, but it's related to oversampling. The digital filters used in oversampling produce ringing - which means that, when you feed a NON-CONTINUOUS signal through them (like a drum beat), the filter "smears" the start and stop of the signal in time. Different filters produce different amounts of ringing, and different filters put more or less of it before or after the original (intended) signal, all of which can make it sound slightly different. This doesn't happen with steady state signals, like sine waves, which is why it doesn't show up on distortion specs; however, it's pretty obvious with transients, so it does show up quite plainly on transient response scope photos. NOTE that this happens because of oversampling; so an oversampling D-S DAC will have the same amount of ringing as an R2R DAC that uses the same oversampling filter. (But there are scientifically valid reasons why a given amount of jitter MIGHT cause different amounts of different types of distortion depending on whether it happens to the signal being fed to a D-S DAC or an R2R DAC.)