[Torq]

  I don't think I understand the USB noise bit very well. How does the USB not transfer noise into something like the Singxer SU-1 when USB is being used to send data to it? If it does, and the noise gets removed in the Singxer, why can't that same function happen inside the Yggy eliminating the need of a DDC in between?

 
It can.
 
Just a question of how much focus the designer wants to put on that particular issue - in this case, and particularly where Yggdrasil is so aggressive in terms of price/performance, it's easier to say "use a different input" rather than up the cost by doing heroic things on the USB input that simply won't be used by everyone.
 
Also, you're still largely stuck with the electrical noise generated by the USB receiver circuitry (which is a lot more complicated than an AES or S/PDIF receiver) inside the DAC anyway.  So unless you perform some actual electrical isolation on the USB input side of things after it's received by the DAC (which can be done with transformers or optical isolation), then there's only so much you can achieve anyway.
 
A properly isolated S/PDIF or AES connection (per spec, they're both supposed to be electrically isolated at the source-end but often aren't in the consumer space ... e.g. most "Sound Cards" that have COAX S/PDIF outputs) fixes those issues.
 
Of course, now you have to care about clocking (which you can largely forget about in USB Audio as there's no sample-clock in the USB Audio stream), but most good DACs, including Yggdrasil, make that largely moot.

 

[axle_69]

  The advantage of USB is its ubiquity and its convenience.  Whereas there are people here who like the sound of USB, the majority are with me, perhaps not to my extreme dislike.  It has been a major project of mine to elevate the sonic standards of USB, and I may be on the threshold of fixing it to my liking.  Until then, I find it functional but sounding less than wonderful.  As I have said elsewhere, it is like eating fast food - not delicious but it will form a stool.

That's good news. Especially considering Apple dropped all other ports in the Macbook and the only option now is Thunderbolt 3/USB, not even the Toslink output inside the headphone jack survived. Maybe other manufacturers will follow and for a lot of people having a desktop with a SPDIF or AES3 output is not very convenient, neither to have to have an expensive converter/audio interface in the middle.

 

[Defiant00]

  Interesting... there is a claim that Yggy truncates 4 bits to get 20 bits instead of rescaling data to 20 bits. Also there is a comment regarding what looks like a small energy glitch due to several switches changing state at the same time being due to the choice of 20 bit converters (fig. 7). Not being an engineer I'm getting lost... would someone like to comment on that?

 
Based off of recent other events I can understand why you got the response you did, but since no one offered an actual explanation, I thought I'd do so (with the caveat that this is all second hand, but I'm sure someone will correct me if I've missed something).
 
1) Directly from Jason, Yggy rounds to 20 bits, not truncates. Apologies if you already know this, but truncation means you just discard the last 4 bits regardless of their value. Also, if they rescaled everything to 20 bits then that would be mucking about with the original samples, which is something they are generally against.
 
The reason this doesn't matter (beyond a numbers game) is that the last 4 bits are very significantly into the noise floor of the mics/electronics/recording source, and would require a listening room well beyond what is typically possible, and assumes you have a file that takes full advantage of the available dynamic range, which due to recording and hardware limits, you likely do not.
 
2) Multibit has a (very small) zero crossover energy glitch (not sure if that's the technical term), when the signal goes across the zero point there is a small amount of extra signal/energy produced due to how multibit works. The point worth noting is the scale of these graphs; in Stereophile's measurements they played the various test tones at -90db, which is very quiet, because otherwise you wouldn't be able to see the 'glitch' at all. So sure, if you listen to test tones at -90db or so on the Yggdrasil vs some other DAC you might (maybe) be able to hear the difference, but for music at a normal level, the glitch energy is so inconsequential that it's...let's go with basically impossible, that you'd be able to hear it even if you specifically set up a test to emphasize it.

 

[watchnerd]

 
The reason this doesn't matter (beyond a numbers game) is that the last 4 bits are very significantly into the noise floor of the mics/electronics/recording source, and would require a listening room well beyond what is typically possible, and assumes you have a file that takes full advantage of the available dynamic range, which due to recording and hardware limits, you likely do not.
 

 
Even this is an understatement.  Most of the performance mics I own have SNR of about 100 dB, AKA about 17 bits.

 

[bwanaaa]

  PCI(e) professionnal card with AES output = problem solved 

 
Ali

Why is this better? A Lynx E22 (or RME or whatever) is plugged into the PCI bus. why would that data highway be quieter electrically?  And you still have a bunch of circuits on the card to convert to AES.
 
Wouldn't a purpose built USB card that isolates the power of USB to come from a quiet external linear power supply be better?

 

[Torq]

 
Why is this better? A Lynx E22 (or RME or whatever) is plugged into the PCI bus. why would that data highway be quieter electrically?  And you still have a bunch of circuits on the card to convert to AES.
 
Wouldn't a purpose built USB card that isolates the power of USB to come from a quiet external linear power supply be better?

 
Professional AES implementations generally provide the proper electrical isolation that's in the basic AES standard specification.
 
Even if you had a perfect external USB source, there's still a comparatively noisy (compared to the simple circuitry required for AES or S/PDIF reception) USB receiver in the DAC.  So if the DAC doesn't provide it's own internal isolation from it's USB receiver there's only so much you can do from the outside.
 
In other words you can't fix all of USB's issues from the outside.  You can improve matters, but you can't avoid them entirely.

 

[ColtMrFire]

And this is why I just use CDs.  I don't have to buy anything but a Coax cable.

 

[Torq]

  And this is why I just use CDs.  I don't have to buy anything but a Coax cable.

 
COAX is not a universal panacea (neither is AES).
 
First, to get around the electrical noise issues the COAX (or other) output needs to be implemented properly.  There are probably more implementations that don't implement electrical isolation on their COAX or AES outputs than do ... at least until you get into the realms of the more seriou  transports.
 
Second, unless it's a reasonably high-end CD-transport it probably has comparatively ropey jitter performance.  So, if your DAC isn't re-clocking S/PDIF data internally, and your CD player doesn't have a solid clock implementation, then CD will generally perform both measurably and audibly worse than all-digital transports from a clock/jitter perspective.
 
A lot depends on your DAC and your transport, of course, 

 

[ColtMrFire]

   
COAX is not a universal panacea (neither is AES).
 
First, to get around the electrical noise issues the COAX (or other) output needs to be implemented properly.  There are probably more implementations that don't implement electrical isolation on their COAX or AES outputs than do ... at least until you get into the realms of the more seriou  transports.
 
Second, unless it's a reasonably high-end CD-transport it probably has comparatively ropey jitter performance.  So, if your DAC isn't re-clocking S/PDIF data internally, and your CD player doesn't have a solid clock implementation, then CD will generally perform both measurably and audibly worse than all-digital transports from a clock/jitter perspective.
 
A lot depends on your DAC and your transport, of course, 

 
Yes.  A good transport is going to help.  But even a decent, low priced CD player sounds great via Coax into most Schiit DACs I've found.  I also find that the varying levels of performance from USB is too nebulous for me to bother with right now, the lengths I had to go to get it sounding decent was not only barely worth the effort but cost too much.  And Coax has sounded better than USB (from my sources) on all the Schiit DACs I've owned and auditioned.  CDs are a simple, cost effective and great performing solution for me.  YMMV.

 

[theveterans]

   
Yes.  A good transport is going to help.  But even a decent, low priced CD player sounds great via Coax into most Schiit DACs I've found.  I also find that the varying levels of performance from USB is too nebulous for me to bother with right now, the lengths I had to go to get it sounding decent was not only barely worth the effort but cost too much.  And Coax has sounded better than USB (from my sources) on all the Schiit DACs I've owned and auditioned.  CDs are a simple, cost effective and great performing solution for me.  YMMV.

 
To get the same performance as a dedicated CD player on a PC/MAC, you need to spend 5x the money and daisy chain USB cleaners and SPDIF converter / reclockers; or better yet completely eliminate the noisy PC/mac source through Audio over IP IMO.

 

[hifimckinney]

To get the same performance as a dedicated CD player on a PC/MAC, you need to spend 5x the money and daisy chain USB cleaners and SPDIF converter / reclockers; or better yet completely eliminate the noisy PC/mac source through Audio over IP IMO.

That is another standard, one more conversion layer another set equipment to make holes in our pockets. Was't players like Logitech SB already working over ethernet?

I am sure very soon some company will launch WHIFI.

 

[pdickerson]

Does not appear to be any interest in my impressions for Tidal/MQA through YGGY.

 

[watchnerd]

Does not appear to be any interest in my impressions for Tidal/MQA through YGGY.

 
If already heard it and have my own impressions.
 
Aside from that, the fact that MQA is incompatible with DSP is a deal-breaker for me.
 
Oh, and lastly, it doesn't solve a problem I have. 
 
So, no, sorry, not interested.  Nothing personal, I'm just not a fan of MQA.

 

[gefski]

Does not appear to be any interest in my impressions for Tidal/MQA through YGGY.

Sure, can you give comparisons for same tracks in 16/44 vs first unfolding to 24/96? Also what player you're using. In my rig Audirvana+ doesn't do any unfolding, so comparing Tidal's 16/44 file to same 24/48 file offers no audible improvement. IMO

 

[7ryder]

Sure, can you give comparisons for same tracks in 16/44 vs first unfolding to 24/96? Also what player you're using. In my rig Audirvana+ doesn't do any unfolding, so comparing Tidal's 16/44 file to same 24/48 file offers no audible improvement. IMO

The only meaningful comparison is if we're comparing tracks that are from the same master.
 
Without that guarantee, we don't know if MQA sounds better (or worse) because of what it claims to do or if the improvement is because the MQA file is from a better master. Same goes for redbook & hi-res file comparisons. 
 
Case in point, the remastered hi-res version of Nirvana's 'Nevermind' sounds worse than it's Redbook equivalent because they cranked up the volume when they remastered it. DR on the original 1991 CD release is 12 and on the 2013 remaster it is 7.
 
edited to fix "DNR" to "DR"