• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Schiit Yggdrasil V2 DAC

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
Interesting, thank you. Does this look anything like sine-wave on the DAC output? The 20-bit dithered signal that is.
The filter is so strong and so "good" that it does what member lowvolume says. It can render beautiful sine waves out of total garbage produced by the DAC (and ADC in our analyzer). It is like demonstrating how dirty a plate in a restaurant is, after washing it a 100 times!
My understanding of washing a plate 100 times...
https://www.audiosciencereview.com/...easurements-of-benchmark-dac3.3545/post-85981
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
I think what atomicbob is doing is what is in AES17. The protocol for dynamic range is to measure with a -60 db 1 khz signal. Notch it out. Yes how the notch is done may make a little difference. Get a noise level reading for that bandwidth 22khz, and then take the difference between -60 db FS and the noise floor, add 60 db to it for dynamic range. So what he shows is what would be called something like 125 db dynamic range( well 121 or 122 db from what is shown). And that is why AES17 dynamic range and SNR are not the same thing. Some DACs modulate the noise floor with signal. But the dynamic range from max to low level signals would be this larger number because as signal level drops the noise floor may drop with it some few decibels. While a max signal vs the underlying noise floor might be a few decibels less due to the noise floor modulation.
Here is an explanation with infographic showing what the AES17 is doing.
http://ixbtlabs.com/articles2/creative-audigy2-zs/index-p2.html
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
I think I'm more aligned with Rob Watts on measuring linearity. You can have a DAC of excellent linearity poorly applied. You can test to see if the DAC is putting out the proper levels even if it is being obscured by noise from a poor implementation. Or you can see the DAC simply isn't linear regardless of implementation. For one thing measurements of the Yggy indicate even with filtering isn't capable of super excellent linearity.

I get the idea you don't get to listen to the filtered linearity, but we apply tests all the time to see how something is functioning for which that could be said. It is one reason I like SFDR (spurious free dynamic range). That is where you take your max level and see what the highest level of any aberration for any reason is. You know you have this much clean dynamic range. Whatever spurious signal is highest sets that. It could be jitter sidebands, 60 hz noise, other noise, idle tones anything sticking up from the noise floor. This is what you get to hear as the clean range of sound.

As briefly mentioned before I'm with you (and Watts) on this one. Fundamental Linearity and noise are separate issues. The noise should be excluded, hence Robs use of very high res Fft which is the same method I seperately concluded to be correct. If you just filter the filter characteristics will affect the ultimate reading, so it's somewhat an arbitrary result based on what arbitrary filter characteristics you choose. OK you might correctly say that an Fft is essentially a filter, but one bin can easily be small enough to exclude all but the fundamental. Long term averaging will remove most remaining noise.

Of course this is not what we hear, but we are not trying to establish what we hear with a linearity test. You should establish or measure the effect of noise in a different way.

Also agree with you regarding SFDR.
 
Last edited:

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,425
Likes
3,979
Location
SoCal
It would seem that the "Fundamental Linearity" test as it is defined has little value for the consumer then as it sounds like it is better suited for characterizing D/A chips or used in other specialized scenarios.
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,253
@amirm , you are getting haunted again by what one could call the theorization and commensurability problem.

To make science, one needs a set of definitions that everyone agrees upon (theorization). I believe audio is there. Everyone talks about SNR and such, and agree that S should be as highs as possible and N should be as low as possible.

Where trouble starts has to to do with commensurability. How do we measure things? What are the right measurements? What are established standards for measurements? This is where you have diffferent standards than the other guys.

Science is about uncovering the truth. So science is by nature consensus seeking. What strikes me in audio measurements is the absence of consensus on commensurability. The question for the scientist is then: Who makes REAL science?

This is not a critique of YOUR work. But I hope you see that you still haven’t convinced every researcher out there on the gold standard of audio measurements.

A couple of questions to all ASR readers: Is it possible to arrive at a gold standard of audio measurements? Why is there no consensus in audio on how to make measurements?
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Interesting point and good question.

Science is about uncovering the truth. So science is by nature consensus seeking.

Just want to clarify what you mean by this? The "so" makes it seem like you're conflating consensus with truth.
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,253
Interesting point and good question.



Just want to clarify what you mean by this? The "so" makes it seem like you're conflating consensus with truth.

No, you misinterpret me. Even though there is consensus on true matters (the earth is round), consensus isn’t always a reflection of truth.
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
No, you misinterpret me. Even though there is consensus on true matters (the earth is round), consensus isn’t always a reflection of truth.
For me Science is as often (should be) about disproving ideas and all agreeing something is wrong.

There’s knowing where to fish but it’s just as important to know where not to fish .

I will leave notions of ‘truth’ to others, I’d stick to our ‘best knowledge ‘ . ;)
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,253
For me Science is as often (should be) about disproving ideas and all agreeing something is wrong.

There’s knowing where to fish but it’s just as important to know where not to fish .

I will leave notions of ‘truth’ to others, I’d stick to our ‘best knowledge ‘ . ;)

Ok, replace «truth» with «truth seeking» which opens up for moving targets.

Still, I wish there was more consensus on measurements (which and how).
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,499
Likes
5,417
Location
UK
As briefly mentioned before I'm with you (and Watts) on this one. Fundamental Linearity and noise are separate issues.
I'm enjoying the discussion, we should do more of this. I'm not really qualified to comment but this is the internet so here goes.

I agree with the discard noise as much as possible at times during the development process, to find out what is really going on, but I'm not sure how much relevance it has to the listener, so when testing for the consumer it seems more relevant to include the noise that would mask the signal, and suppress the out of band noise.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,499
Likes
5,417
Location
UK
2. Use a filter that is just enough but no more. Here is the response of my filter relative to (one of two) Jude's filter settings centered around measuring linearity at 200 Hz:

index.php


As you can see, my filter in red not only doesn't filter as much (50 dB versus 60 to 70 dB for Jude's), but it has a much more well behaved frequency response. See the various troughs in Jude's especially one around mains frequency of 50 to 60 Hz which helps the DAC but not showing as much of its power supply in DAC output.

I have also worked to make sure that with or without filter, the output of the DAC at the main frequency of 200 Hz doesn't change. Digital filter can ring and levels can change. While this is not a big error in Jude's case, it is there nevertheless which I corrected in my custom filter.

Bottom line: would you like the measured accuracy of a DAC be through that blue curve or red? I hope we both agree that less is more and attention needs to be paid to the underlying signal processing to create a correct measurement.
Have you measured the sample you have with the blue line and red line settings on the same graph?
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
I'm enjoying the discussion, we should do more of this. I'm not really qualified to comment but this is the internet so here goes.

I agree with the discard noise as much as possible at times during the development process, to find out what is really going on, but I'm not sure how much relevance it has to the listener, so when testing for the consumer it seems more relevant to include the noise that would mask the signal, and suppress the out of band noise.
The intresting thing is the two measurements ( amirs less filtering vs Jude’s method applied to amirs data) did not produce wildly diffrent result I believe.

The real story is just wtf is Jude measuring as it’s nothing like the various yaggi’s amir has had his hands on.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Part of the problem here with standardization is the linearity has become an obscure measurement because DACs have gotten so good. I brought it back from the dead because it seems that people have taken their eye off the ball and producing products that flunk it again. And in the case of Schiit multi-bit DACs, hugely so.

By increasing its visibility, I think we are going to see people verify their designs with it and the problem should lessen in the future.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
I still maintain that one can easily do the linearity measurement with noise or excluding noise. You indeed get different bits (ouch) of information.

Amir was kind enough to send me his filter file, and I'll try to get some comparative examples posted this weekend.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Have you measured the sample you have with the blue line and red line settings on the same graph?
Please, please, don't send me back to measure this DAC again. So no, I don't have that measurement. But I do have another method that Jude used in another one of his graphs which I have already posted:

Schiit Yggdrasil Jude Script Linearity Measurement.png


So compare that red, to mine from the original review in red again:



index.php


As you see, they are identical in the error they show. The difference is that I use proper scales of +-5 dB and he uses +- 10 dB which visually reduces the error. But otherwise, we see the same jumping up and down due to improper quantization.

This is why I said the difference in measurement technique is quite slight compared to the errors in the DAC itself.

And that the difference between the outcome of his and my measurements is NOT due to measurement technique but the fact that he is testing a different sample than what customers have bought.

Before I got my third sample, I assumed that he would be testing a commercial unit as I had so I thought the massive differences had to be due to measurement technique. Turns out that is not a factor at all and the unit being tested is so different.

Likewise Jude thought I had a good unit like his and therefore I must have cheated in my measurements of the DAC. Both of us were wrong this way.

The fault is squarely on the shoulders of Schiit by not being upfront about what units they gave to Jude and AtomicBob versus what customers have already purchased. Those two individuals could also use a lesson in transparency in clearly stipulating where they got their units, and going back and questioning Schiit why they are seeing such different results than I am in linearity.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
The mystery is Amir's AES17 DR is still different from AtomicBob, and the disappearing of truncation artifacts in his measurements. So it is not a merely a problem of "gold standard" or not.
There is no question in my mind at this point that both AtomicBob and Jude have received a different unit than the three customer ones I have tested. No other explanation exists for the large difference in linearity. The difference in Dynamic Range could be explained by that, or the different implementation of this "scheme" as cooked up by AES. Or simply run to run variation between DACs.
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
There is no question in my mind at this point that both AtomicBob and Jude have received a different unit than the three customer ones I have tested. No other explanation exists for the large difference in linearity. The difference in Dynamic Range could be explained by that, or the different implementation of this "scheme" as cooked up by AES. Or simply run to run variation between DACs.
Given all three of you are such great friends why not get together and measure each other’s units .., maybe open a jar of the mrs pickles and sing some songs around the camp fire .
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Let's also not forget other differences I have found which again, they don't show. To wit, both units I tested with analog board "V2" had this frequency response error:

index.php


Both Jude and atomicbob showed flat response. No way to screw up such a simple test.

So once again, we are facing the simple fact that Schiit took notice of our work here and has gone back and either fixed these problems for customers at large or has some prototype units with the problems resolved that they then sent to Jude and Bob Smith.

Pride and PR are getting in the way of arriving at the truth.
 
Top Bottom