What actually is "phase noise"? Loudspeakers produce large phase shifts and the music playing in your room is bouncing off walls etc. Is "phase noise" from a DAC going to be significant in any way here? Turning your head a few degrees is going to cause phase changes.
They measure good in terms of SINAD and noise floor. That's not the same as audibly transparent.
ok but I really don't think intersample overs are what people are hearing when they say "x" DAC has a better sound than "y" DAC. Do you? Older CDs from the 80s dont produce any intersample overs.
Phase changes in a room is not phase noise. Totally different thing in a dac. There is convolution with phase noise (as viewed in the frequency domain) in a dac. Not the same as turning your head at all.
What's your source for this? Have you measured the phase accuracy of many DACs?
What do you mean by that question? How are you defining "phase accuracy?"
From the question, it sounds like you have no understanding of phase noise in a dac?
Anyway its like this: Every dac needs a voltage reference and a time reference. Both are analog signals. Noise in the time reference as viewed in the time domain is called jitter. Viewed in the frequency domain its called phase noise. Close-in phase noise is a special part of the phase noise spectrum that is important in fields such as audio and in radar.
Last edited:
But the cheaper DACs measure pretty much just as good distortion wise so I dont think the power supply is an issue in any of them. Do you?
Every dac needs a voltage reference and a time reference. Noise in the reference voltage power supply is a called amplitude noise. It is convolved with the audio signal as viewed in the frequency domain. Its a problem that doesn't appear as steady state distortion in an FFT.
Close-in phase noise is essentially the same as random wow and flutter. The clock phase or equivalently the clock frequency (which except for a constant factor is the derivative of the phase to time) has random fluctuations causing the frequency of the played back signal to vary proportionally.
I'm rather sceptical about the supposed audibility, because even with a very cheap crystal oscillator as time reference, the variations are a lot smaller than those of the best analogue studio tape recorder. Still, many people hold a different opinion on that.
It might. At least it's by far the grossest DAC imperfection I know. (Although you could also call it a mastering imperfection, of course.)
What if its a noise fold down effect?
Besides, its already noncontroversial that the published limit of audibility of jitter for a multibit dac is down around the level picoseconds or a few nanoseconds (which is at the level where it starts to sound like a distortion). Obviously that's still much smaller than the time deviation of tape wow and flutter.
I think it's very unlikely to be the reason people think one DAC sounds better than another. But I'm not going to dismiss intersample overs as a non-issue. From a technical standpoint it's something that should not be happening.
It almost certainly is a non-issue. Watch this video:
Also, if one channel is out from the other by a fraction of a millisecond is that not going to be dwarfed by the fact you didnt match the speaker distances to your ears down to 1 nanometer?
Last edited:
The video is wrong. It is already noncontroversial for some decades now that LF phase shift can be audible for some time domain waveforms, such as a low bass drum sound. Also, there is a known limit of audibility for phase shift. It is called the threshold of audibility for group delay. https://www.researchgate.net/figure/Group-delay-audibility-thresholds_fig2_352265058
I only watched a few fragments. As was to be expected based on the title, they have absolutely nothing to do with close-in phase noise, so it's clear that Miller-8 didn't understand post #89.
Did you read my comment below the video?
Do you also have evidence of phase noise being a problem with DACs and of the more subjectively highly rated DAC having less phase distortion?
The problem with the term "jitter" is that it lumps together the effects of close-in phase noise, far-off phase noise/the phase noise floor and non-harmonic spurs. Unlike Andrea Mori, I would be least worried about close-in phase noise (although it's always nice not to have it, of course), because noise skirts around the desired signal should be well-masked. If they weren't, analogue tape recorders would probably be unlistenable.
The phase noise floor and spurs can have quite an impact, especially in some types of sigma-delta DACs. Every picosecond matters there.
Why would this phase noise matter when your loudspeaker crossovers and room are causing all sorts of phase shifts that dwarf it?
In post #93, you mean? Yes, I did. That comment also has nothing to do with wow and flutter.
You are again mixing up phase noise and phase distortion. Maybe Wikipedia does a better job at explaining the difference than Mark and I do:
https://en.m.wikipedia.org/wiki/Phase_noise
https://en.m.wikipedia.org/wiki/Phase_distortion
I already presented a phase noise graph of SMSL DO300EX which BTW according to the manufacturer has a new self-developed CK-03 clock processing circuit that greatly reduces clock jitter. But according to measurements this circuit lowers noise floor (thus increasing SINAD) at the expense of close-in phase noise. Topping D90 had a similar phase noise graph but that was due to mediocre Vref regulator with high 1/f noise.
But don't get me wrong. I view these bad looking close-in phase noise graphs as only flawed implementations. I have not claimed that they have audibility implications. I'm not even aware of any studies or tests that have shown ultra-low close-in phase noise to have audible benefits.
- Status
- Not open for further replies.