Schiit Byygy Dac Just Landed

Ehh, I would not put a Yggdrasil in the same category as it’s still relatively affordable to most “normal” people who are still happy to blow thousands on other stupid shit. Ferrari level lifestyles indicate an entirely different mindset where it becomes more about status symbols than actual enjoyment.

Nobody really needs a Ferrari and probably don’t even enjoy driving it as much showing others you can afford a Ferrari. You don’t need a Yggdrasil either but at least there are real gains to be had with regard to musical appreciation and enjoyment. I’m always really impressed when I still see Yggys in megabuck setups because it shows the listener is probably still relatively level headed.

 

would love a loaner on this one

 

Reminds me of Pretty Woman where the 'ho shows the stupid lawyer how he should be driving his Porsche.

I got no answer on my n-bit D/S question. Can someone point me at a paper describing why we need n-bit D/S when we already have PCM?

 

I asked a question about n-bit D/S, and @SoupRKnowva gave a brief explanation of the technology. Look back and you’ll see it.

That said, n-bit D/S is a method of digital-to-analog conversion, whereas PCM is a digital audio format. They are two different things.

 

If that's the case then how was d-to-a done before D/S was invented?
Was it all resistor arrays?

btw, I have searched the web pretty aggressively and cannot find anything on n-bit D/S as the search results are flooded with how 1 bit D/S is this and so and other not-helpful results.

 

I shared a slide deck earlier on the thread that discusses it, but in somewhat impenetrable technicalities that someone here objected to I own n-bit D/S DACs, Linn Organik, they are my favorites, the first to beat (to my ears) HQPlayer with the right filters and modulators > Holo May. I've signed up for Byggy in the hopes that it will have some of that for my headphone systems.

 

Yeah I think of Schiit's high performance, lower cost stuff like cars to some degree. Like, you're never gonna quite get to MSB ORFAS level or whatever because MSB is spending a gazillion dollars on boutique caps and billeted super thick shielded chassis and power supplies and such that Schiit is not going to bother with. It's cost no object, relatively, like Robin Leach stuff that is eeking out n% more performance or at least saying it is so that the ORFAS dudes have something to talk about when talking about their new toys. Schiit is "good enough," which there is a place for.

 

The thing is latest MSB are not R-2R. Not even sure they still use discrete resistor arrays, since the measurements are insane. DCS uses discrete resistor arrays - but that is also not R-2R. They use some form of random switching of the arrays to get rid of compounding noise.
I've not had a proper chance to test the Wadax stuff. From brief show conditions I could tell there is potential.

Or something made in US that has similar performance - Dodge Viper (sad that this one is out of production), Corvette.

The hi-end manufacturers have set up their business in a certain way. Not that they potentially could not produce for the middle class segment some great stuff, but it is actually harder to enter the market and be successful in this segment.

I ma a sucker for car analogies as well. Forget Ferrari, Lambo and Porsche. The uber stuff is Koenigsegg, Pagani, Bugatti.
See if I am right that most people of SBAF care about sound more than the 10 other attributes of hi-end.
In car world that would be purpose built track car. Hyper car will cost some millions. A great track car will cost 10x less and possibly lap the circuit in same time or quicker. A modest track car is almost as much fun and cost 100x less.

I have tested some hi-end dacs that cost 10..20x Yggdrasil price and would listen to Yggdrasil instead.
Now MSB and DCS are tiers higher in sound performance.
I believe in time people like Schiit will catch up and give people something 90% there.
Somewhere I put down the wish for a 5k Schiit totl dac. As that is where I see almost no compromise in component level is achievable and a lot of people could still afford one.

 

I went through that, somewhat carefully over 2 days. I didn't see anything in it saying why n-bit vs 1-bit D/S or anything about previous/alternative d-to-a mechanisms. Taking another look, I guess I'll have to go through it again, thanks!

It did seem to talk a lot about how their D/S on a chip was the best thing ever.

 

Given Byggy changes all other boards for a premium over the yggdrasil, I'm very surprised they didn't decide to swap out the motherboards with the newer style transformers, higher rails, and the like. I wonder what the decision was there (besides potentially cost since it's keeping the same motherboard as before)-- but as the upgrade is already so much more expensive you'd think people wouldn't mind paying a little more for an even beefier motherboard/power supply?

I wonder: does schiit plan to never upgrade the Yggdrasil motherboard, or expect Byggy owners to cash out more for a motherboard upgrade down the line?

 

I suspect it's because they've always promised the Yggdrasil to be future-proof. It may be a case where they're damned if they do, damned if they don't. Don't? Get potentially less performance. Do? Keyboard warriors complaining about "but you said my DAC was future-proof!"

Rogue Audio uses this model. Their stuff is fairly priced, but there's that temptation to upgrade the caps and various other innards. In some ways, not much different from (expensive) tube rolling.

 

Not described explicitly as n-bit on the deck, but if you look closely, some of the circuits shown operate that way. Ignore the stuff focused on single-chip implementation, what matters is the chosen "better" circuits in the deck (writing from memory, don't have time now to go hunt for the slide numbers that show the n-bit designs).

 

Yeah I’ve wondered about the motherboard stuff too, but it’s like, how much are we really getting from that change? If it’s a practical one like Marv mentioned (higher voltage rails for Nexus), that’s one thing. If it’s OCC silver wound transformer and film caps, it’s another. I’m not really saying that those latter things can’t make a difference, they can, but the delta is small and the price premium often high and the whole thing kind of goes against some of what I gather to be Schiit’s ethos.

Still, I imagine one day we’ll be getting a main board change too, for whatever reason. I also wouldn’t be surprised if they do a silent revision (or already have) from parts availability or whatever other issue.

 

Already +/-24V with choke input discrete shunt regulated power supply. It's fine.

 

That is sort of the opposite of how audio gear junkies respond to change, but I trust y'all until you prove me wrong! Schiit, I'm all wrapped up in this release and I can't even play at getting it. So little seems rational with this hobby! Just so you know Jason, Schiittalker constantly refuses to send me complimentary gear! I even asked nicely!

 

Schiit Spring AMA on May 12th. Maybe we get more answers on Byggy then.

 

Slide 5 compares the designs. The suggestion here is that we originally had "multi-level" DACs (this could mean resistor-ladder designs or N-Bit designs maybe?) and that we then moved to 1-bit oversampling DACs. There is no explanation accompanying because it's assumed you know what the labelled "interpolation filter" is. In theory, oversampling in conjunction with delta-sigma noise shaping is sufficient to make the "multi-level' DAC implementation unnecessary. The reason a 1-bit design is used is likely for simplicity. Not necessarily because it is superior. It is the overall design that matters (or so they reason).

The difference introduced with the "Modern" circuit is the "thermometer encoding" stage, so that must be the advantage. Given that the thermometer encoding takes in the same k bits as the 1-bit designs of yesteryear but passes 2^k bits out to the conversion stage, we can infer that the thermometer encoding stage is working on a log scale and we can also infer that it is adding information (or interpolating or cancelling information or something). Presumably in an attempt to mitigate errors from other sources. Where it gets this information from, I do not know.

Slide 9 is titled "Single-Bit and Multi-bit Trade-off". Many of the acronyms here are explained on previous slides, but I will explain the takeaway point here as simply as I can. Basically, according to slide 7, there are 3 main sources of error:
 DAC Element Mismatch
 DAC Asymmetrical Switching (ISI)
 Clock Jitter & Amplifier Nonlinearity
A 1-bit DAC suffers badly from the second problem, asymmetrical switching.
A multi-bit DAC suffers badly from the first problem, Element mismatch BUT "there are algorithms to address this". Recall the block labelled "thermometer encoding"? Yeah, that's what it is doing. Correcting errors algorithmically. By this argument, the slide is implying that the third design from slide 5 has the advantage because we can algorithmically correct for physical element mismatch errors, and this design does not suffer so badly from the errors which affect other designs.

Slide 12 would be relevant as it is contrasting a pure 1-bit design + lots of filtering with a multi-multi-part dac with no filtering. However, all that math simply says is "Assuming a set number 'L' of equivalent elements, we can shape the noise all we want, but there will still be the same total amount of noise". This is almost trivially true since they have defined noise as a function of L alone. All the math is just for show.

The rest of the slides move on to ideas which progress far beyond your question. I don't see anything relevant.

I am not an electrical engineer. I just have some interest in math, computing and physics, And I too don't have tons of time. It originally just took me an hour to peruse the slides while listening to music and the same amount of time again to write this. So please excuse me if there is something I have missed which is obvious to a real electrical engineer.