[USAudio]

Can someone explain the differences between the 3 multibit DAC chips used by Schiit in their multibit DACs and the significance of those differences?

Here's the chip breakdown:

Model - DAC chip (quantity) - #bits

Yggdrasil - AD5791BRUZ (4) - 20bit
Gungnir - AD5781BRUZ (4) - 18bit
Bifrost - AD5547CRUZ (1) - 16bit

How does this relate to the resolution of the audio files played through it?
For example, how would a 24/96 file be processed by the Bifrost as it has a single 16bit DAC chip?

Thanks!

 

[Thenewguy007]

Never heard of those chips. All the review of DAC chips are from the older generation.
 
An Amazon link shows does show the AD5791 costs $150, so the Yggdrasil at the very least has $600 worth of DAC chips in it.

 

[Starchild1]

From Baldr:
 
2 weeks, 2 days ago

 

1. [img]http://cdn.head-fi.org/b/b7/100x100px-LS-b7d0f235_avatar-120-4.jpg[/img]

1. Baldr
2. Sponsor: Schiit Audio
3. 
4.  
5. offline

1. 119 Posts. Joined 5/2011

 
 
 
 
 
 
A few clarifications and amplifications re Bifrost Multibit DAC (Bimby).  The DAC is indeed a four year old 16 bit converter which for the glib, may not be exicting.  Those with the patience to study (yes, study) the datasheet will be exited when they realize the performance meets/exceeds that of many competing 18 bit dac chips.  The mega combo burrito time and frequency domain filter is also implemented in a SHARC chip identical to its bigger brothers Gumby and Yggy.  This permits the perception of the spatial cues uniquely proper to the Schiit filter.  The filter is identical to its bigger brothers' with just one exception:  the output rate on the Bimbo is limited to 192KHz rather than 384KHz. Since the filter is cascaded, the required taps halve to 9000.  The reason the unit must be returned to Schiit for upgrade is that the control microprocessor's flash memory must be reprogrammed.  I am proud of its price/performance ratio as a multibit D/A converter for $600.00.  I have been listening to one at my workstation for the last several months and find it quite satisfying.  Enjoy!

Edited by Baldr - 10/3/15 at 4:59am




 
From Jason:
 
2 weeks, 2 days ago
THREAD STARTER 

 

1. [img]http://cdn.head-fi.org/c/cd/100x100px-LS-cdcfd838_js+photo+1a.jpg[/img]

1. Jason Stoddard
2. Sponsor: Schiit Audio

 

1. 

 
 
 
 
 
 
 

1. offline

1. 1,185 Posts. Joined 6/2010
2. Location: Newhall, CA USA

Some fun facts about the AD5547CRUZ DAC used in Bifrost Multibit (for super techies only):
 
1. It features an integral nonlinearity (INL) plot that's better than +/-0.5LSB. This is a spec they never provide for audio DACs, because (a) it would be terrifying in the case of audio multibit DACs and (b) it is not possible to measure delta-sigma DACs in this way, since the output depends on the preceding and following samples.
 
2. The THD performance is actually scary good--far better than 16 bits. 16 bit level THD, from a theoretically perfect 16-bit DAC, is -96dB. Most 16 bit DACs from the Jurassic Age of Digital didn't hit this number. The AD5547 is -104dB, much better than 16 bit--and its noise level is down at the 22 bit level. And this is without the deglitching tricks applied to Gungnir Multibit and Yggdrasil.
 
3. It is a parallel input DAC. As in, the entire 16-bit word has to be written in at once, on 16 separate pins. This is very bizarre, since most DACs are serial in. Just one way that these are NOT easy to use for audio. 

 

[eugenius]

  Never heard of those chips. All the review of DAC chips are from the older generation.
 
An Amazon link shows does show the AD5791 costs $150, so the Yggdrasil at the very least has $600 worth of DAC chips in it.

 
 
The 1000 batch price is 38$. Still more expensive than other solutions but not MSB R2R module money.

 

[roddymcl]

Hello Jason,
 
and thanks for replying to my previous (ill informed, even confusing an amp product with a DAC product) post in another thread post linked here.
 
I'd just mention that I think the -104dB THD spec for the AD5547 is a reference multiplying spec.  That is, it's what you get when you fix the DAC code, and modulate the reference voltage, rather than what you get when you fix the reference voltage and modulate the DAC code, which I think is what matters for audio applications.
 
The excellent AD5547 is quite an old part, and the datasheet information reflects that.  More recent DAC developments have included more complicated DAC evaluations which result in datasheets that include both types of distortion measurements, so I don't exactly know how different the AD5547 spec would be for the latter type.
 
As an aside for uber-geeks, I believe that for these non-audio DACs, there is often a surprisingly good correlation between major-code glitch energy and code-change-dependent THD which would matter for audio.  I think that if you do a survey of a large number of non-audio parts, there is a surpringly consistent mathematical relationship between: frequency, voltage magnitude, glitch energy, and THD.  In ratio terms, think of the the integrated absolute magnitude of a sine wave, compared to the (similarly integrated) glitch energy.  That dimensionless ratio of units in Volts.seconds, in most DACs, is usually a great guide to code-dependent THD.  But of course, there are some exceptions, which can normally be explained by unusual DAC LSB-subranging architectures, and I think these might include the unusually good AD5547.
 
I may have said too much, but I'm still intrigued about how "non-audio" DACs are working in audio applications ...
 
Cheers,
Roddy

 

[USAudio]

Still wondering how a 24/96 file would be processed by the multibit Bifrost as it has a single 16bit DAC chip?
It supports up to 24/192 for all inputs.

 

[yage]

This is somewhat covered in the FAQ section of the Bifrost product page:
 

But what happens when I use 24 bit music?

We transform it to 16 bit, and it plays just fine. Just like the 2- to 5-bit delta-sigma DACs do. Except with a lot more bits.

 

[USAudio]

We transform it to 16 bit ..l

Not exactly sure what "transform" means, what happens to those other 8 bits?

 

[Oklahoma]

From the Schiit FAQ for Yggdrasil
 
Quote:

But the Arglebargle has like twelve 32-bit DACs in it! Yours only has 21 bits! Hell, that’s not a full 24 bits even! What about my 24-bit recordings? If your 24 bit recordings actually have 24 bits of resolution, we’ll eat a hat. And those "32-bit" DACs? Well, they have this measurement known as “equivalent number of bits.” This means, in English, how many bits of resolution they really have. And that number is 19.5. And 21 is better than 19.5, in all the math books we know.
 
Wait, what is this about 32-bit music?
How much 32 bit music do you have? (Not that it will ever exist—we can't get the noise floor that low. Period. Unless Dr. Who pays us a visit and drops some alien tech on us...) Also, how many giga-rate DSD recordings do you have? None, because they don't exist.

 
 
And From the Schiit FAQ for Bifrost
 

  I can’t get over the fact that Bifrost Multibit is only 16 bits!
You didn’t have any problem with delta-sigma being 2 to 5 bits, did you?
 
But, 16 bits!
Yeah, and most music is still 16 bits—99.9%+, in fact. 
 
But what happens when I use 24 bit music?
We transform it to 16 bit, and it plays just fine. Just like the 2- to 5-bit delta-sigma DACs do. Except with a lot more bits.

 
The number of chips is different because of how those chips are used.  For both Yggdrasil and Gungnir they use 4 of each chip because it is actually hardware balanced and one does the conversion for the + line and another does the conversion for the - line in a balanced system they are each referenced against each other instead of summed together and referenced against the ground like in an unbalanced system (2 per channel and 2 channels is 2x2=4).  For Bifrost it only uses one because the one chip can do both channels in parallel or at the same time.  
 
A big difference is the filter that Schiit uses.  Their filter does all the conversion and keeps the originals unlike basically every other filter.  The other difference is in how it all works to do the conversion.  The chips used in these dacs were never intended for audio use and are mainly used in medical devices where this level of precision is needed.  With Delta Sigma dacs they approximate the levels and just go from there and usually are only getting around 4-6 bits of resolution even with 24bit audio.  
 
Someone else can chime in with more details as I am hitting about the limit of what I know and don't want to start giving out tons of incorrect info.

 

[landroni]

Yggdrasil - AD5791BRUZ (4) - 20bit
Gungnir - AD5781BRUZ (4) - 18bit
Bifrost - AD5547CRUZ (1) - 16bit

How does this relate to the resolution of the audio files played through it?
For example, how would a 24/96 file be processed by the Bifrost as it has a single 16bit DAC chip?
 

From what I understand Bifrost will take those extra 8 bits and toss them out the window. Then take the 16/96 file and stretch it through the R2R ladder and into the amp.
 
This is no different from all Delta-Sigma DACs out there, which will take a 24/96 file and convert it all the way down to 6 bits (throwing 18 bits out the window!), then take the 6/96 file and oversample it wildly and proceed with the digital trickery to digitally increase the bitrate and deal with the resulting noise levels.
 
A 16 bit file is NOT that bad, actually. Monty Python argues that given human physical constraints, that's pretty much all that we need until we evolve:
http://people.xiph.org/~xiphmont/demo/neil-young.html#toc_tdro1b
"16 bits is enough to store all we can hear, and will be enough forever."
 
And technically, if simplified, the number of bits simply serves to define the noise threshold, and for all practical intents 16 bit is all we need from audio files; 24 bit is, generally, overkill. So I wouldn't lose my sleep over 24 bit files, as not R2R nor DS techs handle them fully anyways.
 
For a longer version of the above see this post:
http://www.head-fi.org/t/779572/r2r-multibit-vs-delta-sigma#post_12038925

 

[USAudio]

This conversation reminds me of the DSD1793 converter used in the iFi Micro iDSD, which is a hybrid multi-bit AND delta-sigma DAC:
 
"It uses a 6-bit true Multi-bit DAC for the upper 6-bits of PCM ... Any bits below this are converted with a low order 256 speed Delta Sigma modulator (in effect DSD256)"
 
http://www.audiostream.com/content/qa-thorsten-loesch-amrifi#MZctDYEO4Y09rdL8.99
 
 

 

[Ian Billen]

Schiit are making quite a name for themselves. I have heard (and read) several great comments about their pieces such as 'unlike any DAC I've heard before'. To be frank .. (<< and yes.. you can still be Schiit ) .. I've not heard one thing that categorized any detail of the sound of their DAC's as being even so much as even 'decent' (every review and comment I've came across show nothing but significantly high marks). For their look, build and reputation thus far I truly commend Schiit. However I never find a direct answer as to why DSD is left out by the Schiit brand ... I am torn between venturing into a Schiit product or seeking something with DSD capability. Can you elaborate as to why you are not going the DSD route with your products and is this a possibility as an option in one of your products or a possible upgrade in the in the future? Perhaps Schiit is simply waiting to see what route the Industry takes as a whole .. or do you have something against DSD technology? 


Ian Billen
http://ianbillen.com

 

[landroni]

  However I never find a direct answer as to why DSD is left out by the Schiit brand ... I am torn between venturing into a Schiit product or seeking something with DSD capability. Can you elaborate as to why you are not going the DSD route with your products and is this a possibility as an option in one of your products or a possible upgrade in the in the future? Perhaps Schiit is simply waiting to see what route the Industry takes as a whole .. or do you have something against DSD technology? 
 

Schiit feels DSD is an irrelevant, stillborn format, e.g.:
http://www.head-fi.org/t/647942/dsd-to-support-or-not#post_9093131
https://vimeo.com/160027074
 
Search for @Baldr or @Jason Stoddard  mentioning 'DSD' for a more complete overview, or even their recent press release on MQA.

 

[Ian Billen]

I will do that and Thank you for the references (will research it more tonight). However while it never took off entirely .. It is still there... with the rest in the pack. 


How do 'you' feel about DSD sound versus PCM or when compared to other high end DACS that do not have the option? As well .. how do you feel about the possibility of DSD sticking around? 

 

[jorgeu]

This article seems to cover everything in detail 
 
http://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/