Review and Measurements of Schiit Yggdrasil V2 DAC

[Rottmannash]

Tell us about it! Once a month someone refreshes this thread and a few like it and thinks based on lay understanding of audio they can school all of us on how the audio world turns. We could shut the thread down but maybe it is more entertaining this way....

Oh please don't shut it down-it's one of my few remaining entertainments.

 

[JohnYang1997]

Like period.

 

[bboris77]

I am not sure if this thread is the most appropriate for this quote, but Mike Moffat has come out with some fighting words on his blog at the other audio forum:

"Mosquito farts, lab coats, and incomplete "science"

In this missive I address the worship of one false audio god, namely the assumption of the valid “science” of audio measurement circa 2021. What’s wrong with measuring audio equipment? Nothing actually as long as you understand that these measurements have little to do with how well this equipment will cater to the type of sound you like. Utterly useless. The only exception is if the measurements are really, really awful.

A number of measurement parameters have been defined such as distortion (IM and harmonic), linearity, and noise. You can add wow, flutter, and speed accuracy for analog sources. Current measurement tech for DACs cover a dynamic range of greater than 120db even up to 140db. The machines that measure this kind of dynamic range have car type price stickers. Back in the analog days 70 db dynamic range was pretty damn good. I remember being at a AES convention in the early 80s (when notoriously bad sounding digital in the form of CDs); they sold t-shirts emblazoned with “Digital Finishes What the Transistor Began”. Early digital, for all of its 90-100 db dynamic really sucked, and most of the engineers admitted it. Back then, it was widely known that measurements were not completely defined to a level that could reproducibly be definitive with human hearing. Fast forward to today, there has been no change.

These guys that sell the pricey measurement gear will have you believe that low level (>80 db) is vital for good sound. Let me put that in perspective. If one attends a concert, -110db (or greater) levels are going to be at the level of mosquito farts. How about inaudible. Up above -80 or so db – that is where the measurements that we haven’t figured out yet are important.

Before you consider me to be hopelessly atavistic, I do feel that measurements can be important for example in production test to find faults in various products. Now if I want to be a poser, I could put on a lab-coat, strike a Napoleonic, hand in coat image, and brag about my newest, car priced Audio hand job analyzer, based on “science”. This analyzer is of course the ultimate arbiter of what you want for your system.

I tire of hearing about “science”. When I was a kid, big oil and petrochemicals owned the media. They ran public service ads on TV based on incomplete science or “science” – A slogan was “DDT is good for our kids” - More vegetables for them to eat. Yeah!

The moral is get what audio gear you like."

Personally, I try to always listen to both sides of the argument which has gotten me in trouble with everyone at some point I do agree with Mike's point that there has been gear that measured well in a limited set of measurements but did not actually sound great. However, this was 40 years ago. I wonder if it is possible to make an amplifier/DAC now that sounds bad but measures well in all the standard measurements used today. I somehow have a feeling that the answer would be "no". On the flipside, the eternal romantic in me still hopes that there may be other things that are not being measured that have an impact on sound reproduction which could lead to its improvements at some point in the future.

Where I start questioning whether this is more of a marketing statement is that very last line of the blog entry. It is clearly an appeal to people to not put much stock in measurements but believe in what they like (whatever that actually means). Worse than that, it is an Apple-like, ego-driven slogan where the word "you" is bolded giving the consumer a false sense of power of choice over their own destiny.

 

[SIY]

I am not sure if this thread is the most appropriate for this quote, but Mike Moffat has come out with some fighting words on his blog at the other audio forum:

"Mosquito farts, lab coats, and incomplete "science"

In this missive I address the worship of one false audio god, namely the assumption of the valid “science” of audio measurement circa 2021. What’s wrong with measuring audio equipment? Nothing actually as long as you understand that these measurements have little to do with how well this equipment will cater to the type of sound you like. Utterly useless. The only exception is if the measurements are really, really awful.

A number of measurement parameters have been defined such as distortion (IM and harmonic), linearity, and noise. You can add wow, flutter, and speed accuracy for analog sources. Current measurement tech for DACs cover a dynamic range of greater than 120db even up to 140db. The machines that measure this kind of dynamic range have car type price stickers. Back in the analog days 70 db dynamic range was pretty damn good. I remember being at a AES convention in the early 80s (when notoriously bad sounding digital in the form of CDs); they sold t-shirts emblazoned with “Digital Finishes What the Transistor Began”. Early digital, for all of its 90-100 db dynamic really sucked, and most of the engineers admitted it. Back then, it was widely known that measurements were not completely defined to a level that could reproducibly be definitive with human hearing. Fast forward to today, there has been no change.

These guys that sell the pricey measurement gear will have you believe that low level (>80 db) is vital for good sound. Let me put that in perspective. If one attends a concert, -110db (or greater) levels are going to be at the level of mosquito farts. How about inaudible. Up above -80 or so db – that is where the measurements that we haven’t figured out yet are important.

Before you consider me to be hopelessly atavistic, I do feel that measurements can be important for example in production test to find faults in various products. Now if I want to be a poser, I could put on a lab-coat, strike a Napoleonic, hand in coat image, and brag about my newest, car priced Audio hand job analyzer, based on “science”. This analyzer is of course the ultimate arbiter of what you want for your system.

I tire of hearing about “science”. When I was a kid, big oil and petrochemicals owned the media. They ran public service ads on TV based on incomplete science or “science” – A slogan was “DDT is good for our kids” - More vegetables for them to eat. Yeah!

The moral is get what audio gear you like."

Personally, I try to always listen to both sides of the argument which has gotten me in trouble with everyone at some point I do agree with Mike's point that there has been gear that measured well in a limited set of measurements but did not actually sound great. However, this was 40 years ago. I wonder if it is possible to make an amplifier/DAC now that sounds bad but measures well in all the standard measurements used today. I somehow have a feeling that the answer would be "no". On the flipside, the eternal romantic in me still hopes that there may be other things that are not being measured that have an impact on sound reproduction which could lead to its improvements at some point in the future.

Where I start questioning whether this is more of a marketing statement is that very last line of the blog entry. It is clearly an appeal to people to not put much stock in measurements but believe in what they like (whatever that actually means). Worse than that, it is an Apple-like, ego-driven slogan where the word "you" is bolded giving the consumer a false sense of power of choice over their own destiny.

It’s a pity when a smart guy has to be a mendacious huckster to move boxes.

 

[bboris77]

It’s a pity when a smart guy has to be a mendacious huckster to move boxes.

Well, the irony is that the backorder on Yggys is 10-12 weeks, so it's not like they have truckloads that are sitting there waiting to be shipped. It is really hard to move units when there is such a long wait period.

We are all spoiled, impatient brats these days. I want my Schiit next day or no deal LOL. I am being deliberately facetious, of course, but there is a bit of truth in there.

 

[majingotan]

Schiit ramping up their competition to Holo Audio May DAC with their More is Less Yggdrasil version. Almost 119 dB SINAD at 0dBFS FFT graph which is pretty much SOTA for an R2R DAC on a chip:

https://www.schiit.com/public/upload/PDF/APx Report--Yggdrasil More is Less (DAC11001).pdf

 

[MaxBuck]

Schiit ramping up their competition to Holo Audio May DAC with their More is Less Yggdrasil version. Almost 119 dB SINAD at 0dBFS FFT graph which is pretty much SOTA for an R2R DAC on a chip:

https://www.schiit.com/public/upload/PDF/APx Report--Yggdrasil More is Less (DAC11001).pdf

View attachment 152435

Pretty impressive Schiit.

 

[tmtomh]

I am not sure if this thread is the most appropriate for this quote, but Mike Moffat has come out with some fighting words on his blog at the other audio forum:

"Mosquito farts, lab coats, and incomplete "science"

In this missive I address the worship of one false audio god, namely the assumption of the valid “science” of audio measurement circa 2021. What’s wrong with measuring audio equipment? Nothing actually as long as you understand that these measurements have little to do with how well this equipment will cater to the type of sound you like. Utterly useless. The only exception is if the measurements are really, really awful.

A number of measurement parameters have been defined such as distortion (IM and harmonic), linearity, and noise. You can add wow, flutter, and speed accuracy for analog sources. Current measurement tech for DACs cover a dynamic range of greater than 120db even up to 140db. The machines that measure this kind of dynamic range have car type price stickers. Back in the analog days 70 db dynamic range was pretty damn good. I remember being at a AES convention in the early 80s (when notoriously bad sounding digital in the form of CDs); they sold t-shirts emblazoned with “Digital Finishes What the Transistor Began”. Early digital, for all of its 90-100 db dynamic really sucked, and most of the engineers admitted it. Back then, it was widely known that measurements were not completely defined to a level that could reproducibly be definitive with human hearing. Fast forward to today, there has been no change.

These guys that sell the pricey measurement gear will have you believe that low level (>80 db) is vital for good sound. Let me put that in perspective. If one attends a concert, -110db (or greater) levels are going to be at the level of mosquito farts. How about inaudible. Up above -80 or so db – that is where the measurements that we haven’t figured out yet are important.

Before you consider me to be hopelessly atavistic, I do feel that measurements can be important for example in production test to find faults in various products. Now if I want to be a poser, I could put on a lab-coat, strike a Napoleonic, hand in coat image, and brag about my newest, car priced Audio hand job analyzer, based on “science”. This analyzer is of course the ultimate arbiter of what you want for your system.

I tire of hearing about “science”. When I was a kid, big oil and petrochemicals owned the media. They ran public service ads on TV based on incomplete science or “science” – A slogan was “DDT is good for our kids” - More vegetables for them to eat. Yeah!

The moral is get what audio gear you like."

Personally, I try to always listen to both sides of the argument which has gotten me in trouble with everyone at some point I do agree with Mike's point that there has been gear that measured well in a limited set of measurements but did not actually sound great. However, this was 40 years ago. I wonder if it is possible to make an amplifier/DAC now that sounds bad but measures well in all the standard measurements used today. I somehow have a feeling that the answer would be "no". On the flipside, the eternal romantic in me still hopes that there may be other things that are not being measured that have an impact on sound reproduction which could lead to its improvements at some point in the future.

Where I start questioning whether this is more of a marketing statement is that very last line of the blog entry. It is clearly an appeal to people to not put much stock in measurements but believe in what they like (whatever that actually means). Worse than that, it is an Apple-like, ego-driven slogan where the word "you" is bolded giving the consumer a false sense of power of choice over their own destiny.

This is just silly, not least because his company has revised and improved their products based on flaws found by @amirm 's measurements here. This missive is just macho d**k-waving. Nothing to see there.

"Before you consider me to be hopelessly atavistic," -- too late.

 

[gvl]

Schiit ramping up their competition to Holo Audio May DAC with their More is Less Yggdrasil version. Almost 119 dB SINAD at 0dBFS FFT graph which is pretty much SOTA for an R2R DAC on a chip:

https://www.schiit.com/public/upload/PDF/APx Report--Yggdrasil More is Less (DAC11001).pdf

View attachment 152435

Wowsers! Distortions 130dB under fundamental? How did they pull it off with a 105dB THD chip? I want to see THD(F), based on some preliminary data THD goes up on these chips with frequency.

 

[majingotan]

Wowsers! Distortions 130dB under fundamental? How did they pull it off with a 105dB THD chip? I want to see THD(F), based on some preliminary data THD goes up on these chips with frequency.

Guess @T.M.Noble can send a unit to Amir to find that out. It would be interesting to see nonetheless, and if IIRC the Schiit closed form digital filter should attenuate very steeply past 20KHz so it shouldn't interfere with the THD+N wideband sweeps

 

[gvl]

The filter will attenuate in the digital domain to suppress images, how would it help with THD rise with frequency that happens due to the DAC chip imperfections in the analog domain? Perhaps I’m missing something.

This was posted on DIYaudio, 192kHz sampling on a prototype developed by a TI app engineer. I hope Schiit found a way to drastically improve this.

 

[Kw6]

But I thought measurements didn't matter?

 

[ShiZo]

The filter will attenuate in the digital domain to suppress images, how would it help with THD rise with frequency that happens due to the DAC chip imperfections in the analog domain? Perhaps I’m missing something.

This was posted on DIYaudio, 192kHz sampling on a prototype developed by a TI app engineer. I hope Schiit found a way to drastically improve this.

That is possible though right?

 

[shoto]

The filter will attenuate in the digital domain to suppress images, how would it help with THD rise with frequency that happens due to the DAC chip imperfections in the analog domain? Perhaps I’m missing something.

This was posted on DIYaudio, 192kHz sampling on a prototype developed by a TI app engineer. I hope Schiit found a way to drastically improve this.

you need to read the original post for the image

''There the main reason for this is the sample-and-hold circuit. The short of it is that during the hold state the output will begin to move in a non-linear fashion if the R2R side of the DAC is greater than ~0.6V of the output. This due to some limitations of the design. Basically, if a sample to sample step is greater than 0.6V you will see this, which results in some greater harmonics and some other spikes. This hurts THD+N

.
As the sample rate is constant, the sample-to-sample voltage step increases at higher frequencies. Reducing the output amplitude reduces the sample-to-sample voltage step, which then reduces the THD. A higher sample rate will improve this further until you reach the part where the device has trouble settling. I wish the system was a bit more capable so I could setup 384kSPS and 768kSPS. ''

 

[ElNino]

you need to read the original post for the image

''There the main reason for this is the sample-and-hold circuit. The short of it is that during the hold state the output will begin to move in a non-linear fashion if the R2R side of the DAC is greater than ~0.6V of the output. This due to some limitations of the design. Basically, if a sample to sample step is greater than 0.6V you will see this, which results in some greater harmonics and some other spikes. This hurts THD+N

.
As the sample rate is constant, the sample-to-sample voltage step increases at higher frequencies. Reducing the output amplitude reduces the sample-to-sample voltage step, which then reduces the THD. A higher sample rate will improve this further until you reach the part where the device has trouble settling. I wish the system was a bit more capable so I could setup 384kSPS and 768kSPS. ''

That's the 20 bit version of the IC. If you look at both sets of measurements Schiit just posted, it looks like the 16 bit version of that IC has less of this behavior.

 

[T.M.Noble]

This is just silly, not least because his company has revised and improved their products based on flaws found by @amirm 's measurements here. This missive is just macho d**k-waving. Nothing to see there.

"Before you consider me to be hopelessly atavistic," -- too late.

Now, I would in no way be interested in speaking for Mike Moffat. He is quite capable of doing that himself. However he did not ignore the contribution of measurements when creating our products. He clearly states, " I do feel that measurements can be important for example in production test to find faults in various products." Indeed, we make better products now than we did in years past. However, neither Mike nor Jason are convinced that maximization of measurements over every other metric is what makes a product superior.

 

[T.M.Noble]

Guess @T.M.Noble can send a unit to Amir to find that out. It would be interesting to see nonetheless, and if IIRC the Schiit closed form digital filter should attenuate very steeply past 20KHz so it shouldn't interfere with the THD+N wideband sweeps

Since Mike's interview, which was quoted on this thread, I am unsure if Amir is interested in testing our gear.

 

[T.M.Noble]

But I thought measurements didn't matter?

They don't. To Mike Moffat. We can chew bubble gum and walk at the same time. If Jason and Mike only ever built what they liked, we wouldn't be much of a company. We pay attention to what customers want and try to supply them with the best products we can at an affordable price. We stand by everything we produce, that isn't to say we we are void of preferences.

 

[Jimbob54]

Since Mike's interview, which was quoted on this thread, I am unsure if Amir is interested in testing our gear.

I bet he would if you sent him the stuff at the other end of the measurement spectrum than that he likes to see ;-)

 

[T.M.Noble]

I bet he would if you sent him the stuff at the other end of the measurement spectrum than that he likes to see ;-)

I will leave that up to the patrons here. I imagine if enough people request a review, he will reach out. We are always willing to send out a unit for review.