Measurements and Review of Schiit Yggdrasil DAC

[Dismayed]

It will empty your bank account at a much higher rate.

You will learn the secret handshake, and you can call yourself a Schiithead.

 

[amirm]

I believe the 20bits in the AD spec sheet is specified a minimum, not maximum.

That would be a first for any chip company, advertising minimum rather that typical. Here is the headline for that chip:

It is 20 bits but can have an error that is 1 LSB over or under. So worst case is actually 2 bits of error, making it an 18 bit part.

 

[Blumlein 88]

Hi Amir,

I remember you did a review of the Behringer UMC204HD last year which you classified as the king of low cost DACs. Not sure if it's still king or not, but how would the Behringer UMC204 compare to the Schitt Yggdrasil in terms of its technical performance?

Put another way, does the Yggdrasil have any advantages over the Behringer?

Amir placed the Topping above the UMC204HD.

Now I am not Amir, and he may disagree. I see only that the Yggdrasil may have a few decibels lower noise floor, and has higher output voltages. Otherwise it doesn't quite equal the UMC204HD in other areas of performance.

So the Topping is better performance, while the UMC204 is not bad vs the Yiggy for a fraction of the cost.

 

[svart-hvitt]

Put another way, does the Yggdrasil have any advantages over the Behringer?

Keeping this question to measurements and nothing else, the Yggdrasil will increase GDP more than the Behringer. It will also inrease the trade deficit.

Some have suggested to replace GDP with Gross Happiness Product. It’s ambiguous how the Yggdrasil would fare in a GHP context.

 

[amirm]

I remember you did a review of the Behringer UMC204HD last year which you classified as the king of low cost DACs. Not sure if it's still king or not, but how would the Behringer UMC204 compare to the Schitt Yggdrasil in terms of its technical performance?

Unfortunately I can't run most of my measurements on the Behringer because it doesn't have S/PDIF input. So can't make a definitive statement there.

 

[Blumlein 88]

Keeping this question to measurements and nothing else, the Yggdrasil will increase GDP more than the Behringer. It will also inrease the trade deficit.

Some have suggested to replace GDP with Gross Happiness Product. It’s ambiguous how the Yggdrasil would fare in a GHP context.

Happy as a pig in Schiit.

Or

The Schiit has hit the fan.


Oh we also left out the increased functionality of the UMC204HD vs the Yiggy.

 

[Thomas Savage]

Happy as a pig in Schiit.

Or

The Schiit has hit the fan.


Oh we also left out the increased functionality of the UMC204HD vs the Yiggy.

 

[Thomas Savage]

@amirm Mike Moffat has a message for you answering your inquiry based on the questionable measured performance of some of their products ..,

 

[rebbiputzmaker]

That would be a first for any chip company, advertising minimum rather that typical. Here is the headline for that chip:

View attachment 10808

It is 20 bits but can have an error that is 1 LSB over or under. So worst case is actually 2 bits of error, making it an 18 bit part.

Not sure as that was the first I had saw that data sheet. I was just commenting on what you had circled in red, "min" thought that would denote the minimum conditions.

 

[Rene]

That would be a first for any chip company, advertising minimum rather that typical. Here is the headline for that chip:

View attachment 10808

It is 20 bits but can have an error that is 1 LSB over or under. So worst case is actually 2 bits of error, making it an 18 bit part.

That would make it a 19 bit part, Amir.

The fact is, the Yggi could have achieved 20 bit performance if the chip had been used as the data sheet specifies.

When I looked at the Yggi pcb assembly there was no evidence that the chip was being properly supplied its reference voltages. The dac reaches full potential (per the data sheet) only when supplied with both positive and negative 10 volt references via low noise, low output impedance op-amps in feedback with internal nodes. I saw only a single TL431 voltage reference per channel and no op-amps. Running on a single 5 volt reference would be responsible for the observed 16 bit performance as per the data sheet's +/-4 LSBs max error.

Since Schiit is using two AD7591s per channel, we aren't sure what topology they are using: running in parallel only buys 3dB in SNR; push-pull the same. Sign-magnitude would have minimized the low-level glitch, but there is no evidence from measurements that they did that.

As for Yggi SNR, it would depend on what signal level it's measured at. It may measure only 90dB at full scale but 90dB at -20dB also, a good place to test in my opinion, since that is closer to the levels we listen at.

My own experience with listening to two units was that the Yggi is very resolving of small details, so much so that you may not enjoy listening to poorly recorded music.
However, with well recorded high-res sources from MA or Reference Recordings, the result is often stunning. I suggest that this where the subtle differences between dacs, and R2R vs sigma-delta dacs in particular, will be apparent.

 

[mindbomb]

Hi Amir,

I remember you did a review of the Behringer UMC204HD last year which you classified as the king of low cost DACs. Not sure if it's still king or not, but how would the Behringer UMC204 compare to the Schitt Yggdrasil in terms of its technical performance?

Put another way, does the Yggdrasil have any advantages over the Behringer?

The topping d30 is curently the best measuring of the budget dacs. The yggdrasil is better in linearity (16 bits compared to the topping's 15 bits) and dynamic range (20 bits compared to the topping's 19 bits). And according to amir, the schiit yggdrasil did better in the smpte imd test also iirc. But one of the yggdrasils had a ridiculously bad thd+n vs frequency curve, and even the "good" one was still significantly worse than the topping in that area.

So it's not so bad that it will get swept by budget dacs. But at the same time, performance is uneven compared to other high end dacs.

 

[garbulky]

Well @Dimitrov are you asking will it sound better? Because that's not really addressed here. AFAIK most double blind tests will tell you that most of the poorer measuring DACs that are supposed to be orders of magnitudes worse than another will are still below the audible threshhold for regular listening though perhaps maybe just the tiny bit audible for listening to a series of tones.

So by following that logic, in terms of money spent, you should be using your iphone as your primary dac, or a $30 Behringer UCA 202 due to its performance and not a $1000 Behringer unless it has features that interest you.

This forum appears more interested in comparing measurements between DACs thanexpounding in detail just how different they will sound in real world music listening.

Not a bad thing, but in case you were thinking a Schiit Ygdrassil is going to produce terrible sound, well no it's not. In As far as DBT testing shows, most of the things I've read points most DACs including the yggy are "supposed" to sound equal to a $30 Behringer UCA 202. If it does sound different not by some huge amount but by a small teensy amount that you may have a lot of trouble telling apart at all.

Now is that something you are going to be basing your purchases on or if you hear something different will you defer to double blind testing results?

 

[Jerry Sobel]

View attachment 10809

I think it should read "you are bananas if you buy this Schiit"

 

[amirm]

So by following that logic, in terms of money spent, you should be using your iphone as your primary dac, or a $30 Behringer UCA 202 due to its performance and not a $1000 Behringer unless it has features that interest you.

This forum appears more interested in comparing measurements between DACs thanexpounding in detail just how different they will sound in real world music listening.

This is not the position of the "forum." Just because we add measurement data to the conversation, it doesn't mean anything other than that: i.e. we now have some objective data to chew on. And to compare to manufacturer's objective claims such as "21 bits of resolution," this and that Signal to noise ratio, and them being gods of design. Measurements shine a strong light on the device to show restrictions in design so that people don't run off and assume perfection when they perform their listening tests when data doesn't support it.

We have also discussed at length the requirements for a transparent delivery channel. That requirement calls for 120 dB of accuracy/20 bits in mid-frequencies. This is based on psychoacoustics which has a huge emphasis on listening tests.

There are literally thousands of DACs out there. Measurements and tear downs should let you filter out the ones that have no hope of achieving the above transparency requirements and competence in engineering.

Inversely, you would be hopelessly lost and at mercy of the manufacturer marketing machine without firm, objective measurements.

 

[garbulky]

This is not the position of the "forum." Just because we add measurement data to the conversation, it doesn't mean anything other than that: i.e. we now have some objective data to chew on. And to compare to manufacturer's objective claims such as "21 bits of resolution," this and that Signal to noise ratio, and them being gods of design. Measurements shine a strong light on the device to show restrictions in design so that people don't run off and assume perfection when they perform their listening tests when data doesn't support it.

We have also discussed at length the requirements for a transparent delivery channel. That requirement calls for 120 dB of accuracy/20 bits in mid-frequencies. This is based on psychoacoustics which has a huge emphasis on listening tests.

There are literally thousands of DACs out there. Measurements and tear downs should let you filter out the ones that have no hope of achieving the above transparency requirements and competence in engineering.

Inversely, you would be hopelessly lost and at mercy of the manufacturer marketing machine without firm, objective measurements.

Agreed on most of them. Are there double blind tests that show the threshholds of audibility for actual music not just frequencies?

 

[Rene]

Agreed on most of them. Are there double blind tests that show the threshholds of audibility for actual music not just frequencies?

I'm sure double blind testing was used to determine the perfection of 44.1/14 bits (the original Philips proposal for compact disc).
Tests have been done to show the audibility of noise in the presence of pure tones. Those tests informed the development of MP3.

Executing a double blind test is difficult at best. We always have to ask: who's doing the test, what is being listened to, what is the equipment being used for the listening and in what environment, and what is the expectation bias (yes, there is always a bias with humans).

As they say in another industry: YMMV.

 

[Jimster480]

That would make it a 19 bit part, Amir.

The fact is, the Yggi could have achieved 20 bit performance if the chip had been used as the data sheet specifies.

When I looked at the Yggi pcb assembly there was no evidence that the chip was being properly supplied its reference voltages. The dac reaches full potential (per the data sheet) only when supplied with both positive and negative 10 volt references via low noise, low output impedance op-amps in feedback with internal nodes. I saw only a single TL431 voltage reference per channel and no op-amps. Running on a single 5 volt reference would be responsible for the observed 16 bit performance as per the data sheet's +/-4 LSBs max error.

Since Schiit is using two AD7591s per channel, we aren't sure what topology they are using: running in parallel only buys 3dB in SNR; push-pull the same. Sign-magnitude would have minimized the low-level glitch, but there is no evidence from measurements that they did that.

As for Yggi SNR, it would depend on what signal level it's measured at. It may measure only 90dB at full scale but 90dB at -20dB also, a good place to test in my opinion, since that is closer to the levels we listen at.

My own experience with listening to two units was that the Yggi is very resolving of small details, so much so that you may not enjoy listening to poorly recorded music.
However, with well recorded high-res sources from MA or Reference Recordings, the result is often stunning. I suggest that this where the subtle differences between dacs, and R2R vs sigma-delta dacs in particular, will be apparent.

I can say the same for my DX7 which is indeed a DS DAC. Poorly recorded music easily jumps out and artifacts in any recording will be clearly apparent.
Micro details are definitely present on any tracks which share them.
Accuracy vs live is extremely close with my Aeons and O2 Amp provided there is a proper live recording to feed it.
I can provide some if you wish to test, even available via Spotify (probably only useful if you have premium).

I'm sure double blind testing was used to determine the perfection of 44.1/14 bits (the original Philips proposal for compact disc).
Tests have been done to show the audibility of noise in the presence of pure tones. Those tests informed the development of MP3.

Executing a double blind test is difficult at best. We always have to ask: who's doing the test, what is being listened to, what is the equipment being used for the listening and in what environment, and what is the expectation bias (yes, there is always a bias with humans).

As they say in another industry: YMMV.

Yes exactly, in the medical field there have been tainted double blind tests especially if they are trying to paint a specific picture for support of their product.

 

[tili]

Yes and it’s our responsibility to present the best face of rational ( objective) thought in our approach to Audio ( that’s one of the founding principles of this forum by the way) . I will deal with the disingenuous trolls but on the whole the recent influx of new members posting are here out of their own curiosity. They still might struggle with the way we do things but I don’t mind that, I will deal with any circular argument dynamic that might occur.

You can’t moan so many are irrational easily lead etc in Audio then chase away the few that take a intrest in what we are doing. You want audiophiles to embrace a more rational objective view , dial down the hostility and give people a chance.

the community doesn’t deserve you or this forum just like humans don’t deserve dogs

 

[Thomas Savage]

the community doesn’t deserve you or this forum just like humans don’t deserve dogs

No one strokes me or picks up my schiit though

Anyway I’m a horse but don’t no one go thinking about taking me for a ride, it won’t go well for you

 

[Dimitrov]

Well @Dimitrov are you asking will it sound better? Because that's not really addressed here. AFAIK most double blind tests will tell you that most of the poorer measuring DACs that are supposed to be orders of magnitudes worse than another will are still below the audible threshhold for regular listening though perhaps maybe just the tiny bit audible for listening to a series of tones.

So by following that logic, in terms of money spent, you should be using your iphone as your primary dac, or a $30 Behringer UCA 202 due to its performance and not a $1000 Behringer unless it has features that interest you.

This forum appears more interested in comparing measurements between DACs thanexpounding in detail just how different they will sound in real world music listening.

Not a bad thing, but in case you were thinking a Schiit Ygdrassil is going to produce terrible sound, well no it's not. In As far as DBT testing shows, most of the things I've read points most DACs including the yggy are "supposed" to sound equal to a $30 Behringer UCA 202. If it does sound different not by some huge amount but by a small teensy amount that you may have a lot of trouble telling apart at all.

Now is that something you are going to be basing your purchases on or if you hear something different will you defer to double blind testing results?

Hi there. I was asking more about the technical performance than outright sound quality because the latter is far more controversial and subject to personal opinion, but if it measures better in almost every department it would be reasonable to suspect it would either :

1) Sound better
or ...
2) Sound the same, because the measured differences are still far below audible thresholds

I have no opinions on the Yggdrasil as far as sound quality goes, but I'm curious about the poor (relative) measured performance compared to far less expensive products. However the measurements done by Atomicbob over at Superbestaudiofriends (yes, I know, not great friends with audiosciencereview it seems ) paints a different picture on the measured performance.

No, I don't post over there.

Now I'm not sure whose measurements are more accurate/reliable as I don't have any frame of reference and I don't have any experience in that field. It would be good to know either way!

If there are inexpensive DACs that can rival or outperform far more expensive options then I'm all for it and IMO, the public should be made aware of it.

BTW, when I say outperform, I'm referring to the technical measured performance. If someone like me can spend $200-300 and achieve performance on par/"better" than a product costing $2000 then why not?