Review and Measurements of Schiit Yggdrasil V2 DAC

[andreasmaaan]

Of what practical value are those distortion metrics if new generation measuring devices, like the one @amirm uses, are not able to support them?

I'll take this as a rhetorical question...

I'd hope that future generations of these devices come round to incorporating these better metrics.

Alternatively, we can plod on with what's presently available, looking primarily at measurements other than THD when judging a device's performance.

 

[Krunok]

I'll take this as a rhetorical question...

I'd hope that future generations of these devices come round to incorporating these better metrics.

Alternatively, we can plod on with what's presently available, looking primarily at measurements other than THD when judging a device's performance.

My question was not rhetorical, I simply see no practical use of metrics that are not implemented in the measurement devices available on the market.

 

[andreasmaaan]

My question was not rhetorical, I simply see no practical use of metrics that are not implemented in the measurement devices available on the market.

This is why I suggested looking primarily at measurements other than THD that these devices are capable of taking.

It is also why I suggested that in future these devices should make efforts to include these superior metrics.

EDIT: I wouldn't bother with Gm though. It's experimental correlation to subjective sound quality seems to be better than THD but not nearly as close as the R(nonlin) and DS metrics.

 

[DonH56]

Quick comment on musical "distortion": The spurs in the signals from musical instruments I do not consider distortion. Overtones are part of the sound. If the second harmonic from a guitar is 25% that is the sound we record. It is distortion if our system adds 25% to what was recorded.

 

[Krunok]

This is why I suggested looking primarily at measurements other than THD that these devices are capable of taking.

Is this about something concrete? Is there something you would suggest @amirm should add to his standard set of measurements which is currently not included and his equipment is capable of measuring it?

It is also why I suggested that in future these devices should make efforts to include these superior metrics.

Have those "superior metrics" been backed up by listening tests? Can you post som links?

 

[Krunok]

Quick comment on musical "distortion": The spurs in the signals from musical instruments I do not consider distortion. Overtones are part of the sound. If the second harmonic from a guitar is 25% that is the sound we record. It is distortion if our system adds 25% to what was recorded.

That goes without saying. Different musical instruments sound different when playing the same tones because of their specific harmnic mixture and we should measure the deviation from their specific tone as a distortion.

 

[andreasmaaan]

Is this about something concrete? Is there something you would suggest @amirm should add to his standard set of measurements which is currently not included and his evie is capable of measuring it?

Well @amirm already measures a lot more than just THD, and most of the other measurements he posts are very useful.

Focusing on just harmonic distortion, I think it would be useful to measure the distortion components of a single frequency at frequencies other than just 1KHz, especially where the THD vs Frequency graph indicates that harmonic distortion is not largely uniform regardless of frequency. A measurement at 100Hz and say 6.67KHz would be good. But generally I think a lot can be discerned/inferred by looking at the 1000Hz distortion spectrum in conjunction with the THD vs Frequency graph.

An IM spectrum graph would be useful too, either CCIF or multitone.

But I should be clear that I think taken as a whole @amirm's measurements give us enough information to make a well-informed judgement of the DUT.

Have those "superior metrics" been backed up by listening tests? Can you post som links?

The most rigorous study I know of on this topic is this one.

Geddes and Lee's paper on their own metric is available here.

 

[Krunok]

Well @amirm already measures a lot more than just THD, and most of the other measurements he posts are very useful.

Focusing on just harmonic distortion, I think it would be useful to measure the distortion components of a single frequency at frequencies other than just 1KHz, especially where the THD vs Frequency graph indicates that harmonic distortion is not largely uniform regardless of frequency. A measurement at 100Hz and say 6.67KHz would be good. But generally I think a lot can be discerned/inferred by looking at the 1000Hz distortion spectrum in conjunction with the THD vs Frequency graph.

I agree. Also, beside THD vs frequency I think it would also be nice to have THD vs amplitude on DACs that have volume control as they can be used as preamps.

 

[DonH56]

That goes without saying. Different musical instruments sound different when playing the same tones because of their specific harmnic mixture and we should measure the deviation from their specific tone as a distortion.

Sorry, I was too lazy to quote. This is what I was responding to:

Music is distortion.

A little more (maybe, measureably, a lot more) isn't easily noticed (particularly without a clean reference with which to immediately compare).

Here's a single bass guitar note. 22.7% THD on the fundamental.

His point that we can tolerate quite a bit of distortion when the music (or movie) is playing is valid, but I disagree with his terminology.

No worries - Don

 

[Krunok]

His point that we can tolerate quite a bit of distortion when the music (or movie) is playing is valid, but I disagree with his terminology.

No worries - Don

I also don't agree with terminology, but he probably didn't mean it the way "music is distortion" sounds , at least on a first thought.

True. I believe our perception of distortion is very complex so we tolerate more on low frequencies and much less on highs, probably also tolerate more when several instruments are playing in the same time vs when only one is playing etc.

 

[Frank Dernie]

Music is distortion.

A little more (maybe, measureably, a lot more) isn't easily noticed (particularly without a clean reference with which to immediately compare).

Here's a single bass guitar note. 22.7% THD on the fundamental.

That is timbre though. A flute produces a fairly clear fundamental with very low values of harmonics, almost a pure sine wave. An oboe sounds different to a flute because it adds lots of harmonics to give its characteristic timbre, notably 5th iirc.
The combination of added harmonics, hence timbre, of string instruments is quite strongly influenced by where and how it is bowed or plucked since the extent to which the extra modes of the string's vibration are excited is effected by both, a bit like the excitation of room modes depends where in the room the speaker (excitation) is.
I find this endlessly fascinating...
It means that harmonic distortion changes instrumental timbre, maybe to an audible extent, and it doesn't surprise me that some people may prefer this change, even though it is not an accurate reproduction of the actual instrument being listened to.
Non-harmonic distortion is always noise though.

 

[zalive]

The way I see things:

1. If someone hears something as better than something else...is he biased (if it's not a blind test/listening)?
2. If someone measures a device...how does particular set of measurements describe what we hear as a quality?

In other words...I'm skeptic in both, and I can't take any as a definite reference.
Still I have to have my own personal reference point. And it's...my own ears not somebody else's ears, but mine

Related to the thread topic...I'd like to see results of a double blind listening test of Yggdrasil vs Topping DX7s (or some other Topping which measures well). After all this being done, I think it would do justice to all the effort Amir has done.

What I'm curious: would blind test result approve or disprove Amir's objective measurements results as a measure of what we actually hear as a sound quality?

 

[Grave]

Measurements are evil booooooooooooo.

 

[andreasmaaan]

The way I see things:

1. If someone hears something as better than something else...is he biased (if it's not a blind test/listening)?
2. If someone measures a device...how does particular set of measurements describe what we hear as a quality?

In other words...I'm skeptic in both, and I can't take any as a definite reference.
Still I have to have my own personal reference point. And it's...my own ears not somebody else's ears, but mine

Related to the thread topic...I'd like to see results of a double blind listening test of Yggdrasil vs Topping DX7s (or some other Topping which measures well). After all this being done, I think it would do justice to all the effort Amir has done.

What I'm curious: would blind test result approve or disprove Amir's objective measurements results as a measure of what we actually hear as a sound quality?

There have been a large number of studies conducted over the decades to try to determine at what threshold particular kinds of distortion become audible.

The whole body of research is of course complex, and there are no simple answers. But over the years, a reasonably good picture of the audibility thresholds for various kinds of distortion has been developed.

For example, we know that the human ear has approximately 120dB of dynamic range, but only under optimal (and quite artificial when it comes to music) conditions. We know that certain amounts of group delay and nonlinear distortion appear to be outright inaudible, say around 0.5 to 1ms and 0.001- 0.01% respectively, depending on various factors. We know that the audibility thresholds for lower order harmonic distortion are higher than they are for higher order harmonic distortion. Etc, etc...

I'm not sure if you're familiar with these studies, but they do allow us to look at measurements and say this or that type/degree of distortion/noise is likely/unlikely to be audible. If a particular thing is likely to be audible, studies also tell us a bit about whether it is likely to be acceptable/objectionable/preferred.

I'd also be a bit curious to know whether a difference could be discerned under blind conditions between a Yggdrasil and a DX7s, as I think the Schiit probably straddles the thresholds of a few of these metrics, while the DX7s appears to pass all the tests, so to speak.

 

[DonH56]

I've never seen a study stating 0.001% to 0.01% nonlinear distortion was audible; I have always read the threshold is more like 0.1% for test tones. Not saying someone with much better ears than I could not hear it, but that is an order or two lower in magnitude than what I expected. Maybe I am out of date, or just going deaf...

May be worth the reminder that most of us find IMD much easier to hear than HD since the distortion tones do not line up nicely on the harmonics for intermodulation distortion. But still 0.01% (-80 dB) and below is getting down there...

 

[andreasmaaan]

I've never seen a study stating 0.001% to 0.01% nonlinear distortion was audible; I have always read the threshold is more like 0.1% for test tones. Not saying someone with much better ears than I could not hear it, but that is an order or two lower in magnitude than what I expected. Maybe I am out of date, or just going deaf...

May be worth the reminder that most of us find IMD much easier to hear than HD since the distortion tones do not line up nicely on the harmonics for intermodulation distortion. But still 0.01% (-80 dB) and below is getting down there...

Ha, yes. I've seen one study that found one particularly egregious example of THD of only 0.005% to be audible. So I thought I'd quote the worst case scenario and then add a little buffer. Can link you the study later if you're interested? IIRC there was an unusually high amount of IMD given the very low level of THD in that stimulus, FWIW...

 

[Blumlein 88]

Ha, yes. I've seen one study that found one particularly egregious example of THD of only 0.005% to be audible. So I thought I'd quote the worst case scenario and then add a little buffer. Can link you the study later if you're interested? IIRC there was an unusually high amount of IMD given the very low level of THD in that stimulus, FWIW...

I'd like a link to that. You would be talking -86 db on the distortion. Not impossible, but seems unlikely to me. I could imagine a couple of tones at 19 and 20 khz with that level of IMD at 1 khz that would be heard by most adults as the two tones are not heard. I don't think that level would be heard otherwise like mixed in with music.

 

[DonH56]

Ha, yes. I've seen one study that found one particularly egregious example of THD of only 0.005% to be audible. So I thought I'd quote the worst case scenario and then add a little buffer. Can link you the study later if you're interested? IIRC there was an unusually high amount of IMD given the very low level of THD in that stimulus, FWIW...

I don't have a link, it comes from some old AES papers and my IHF study guides, long and deeply buried now... At one point I made some very short files with varying levels of distortion from 0.01% to 10% added to a 1 kHz (I think) sine wave. Nobody heard 0.01%, a few claimed 0.1%, more at 1%, and everybody at 10%. I am not sure if the files were ported over here when Amir copied all my old articles.

If you just do the math, 3IMD is ~9 dB above 3HD for equivalent maximum signal levels and the same distortion factor IIRC.

 

[zalive]

There have been a large number of studies conducted over the decades to try to determine at what threshold particular kinds of distortion become audible.

The whole body of research is of course complex, and there are no simple answers. But over the years, a reasonably good picture of the audibility thresholds for various kinds of distortion has been developed.

For example, we know that the human ear has approximately 120dB of dynamic range, but only under optimal (and quite artificial when it comes to music) conditions. We know that certain amounts of group delay and nonlinear distortion appear to be outright inaudible, say around 0.5 to 1ms and 0.001- 0.01% respectively, depending on various factors. We know that the audibility thresholds for lower order harmonic distortion are higher than they are for higher order harmonic distortion. Etc, etc...

I'm not sure if you're familiar with these studies, but they do allow us to look at measurements and say this or that type/degree of distortion/noise is likely/unlikely to be audible. If a particular thing is likely to be audible, studies also tell us a bit about whether it is likely to be acceptable/objectionable/preferred.

I'd also be a bit curious to know whether a difference could be discerned under blind conditions between a Yggdrasil and a DX7s, as I think the Schiit probably straddles the thresholds of a few of these metrics, while the DX7s appears to pass all the tests, so to speak.

I suspect there's something uncovered by objective measurements as being done today, which would strictly relate objective to blind test listening. And I don't just mean audible or not. I suspect testers might prefer some equipment that measure poorer at least in some setups.

Yggdrasil vs Topping DX7s might be a perfect experiment. Objective side makes it pretty clear, and expected possible results from the objective perspective is either Topping wins (distortions audible on the 'Yggy' side) or without conslusion (distortions on both proved to be inaudible in practice). But what if Yggdrasil won the blind test?

Only one way to find out: do the blind test comparison.

By the way, I never listened to any Schiit or Topping DAC in any system (I have Schiit Eitr USB/SPDIF converter). But what intrigues me is that regardless of really good measurements, some listeners are not satisfied with Topping's sound. Is this a bias, or those listeners actually do hear something being wrong?

 

[Grave]

I doubt anything below 0.01% (-80 dB) is audible. IIRC with my cheap DAC/amp the THD+N is 10x lower at 0.001% (-100 dB).