Review: Battle of Schiit Audio DACs

[Veri]

Some belated comments about the Schiit multibit DACs ( I have only discovered this forum /blog recently).
I do not think that the truncation of 24 bits audio data to 16 bits reflects technical incompetence, I think it is a result of Schiit's adherence to the "bit-perfect" philosophy.

What? How is truncating the original 24 bits to 16 in a lossy way, 'bit-perfect' in any way? And it was technical incompetence, they have admitted the 'bug' and implemented proper dithering for better linearity in all new versions via firmware update.
Plus, if you want your old multibit schiit DAC to dither you need to pay shipping as well as service fee for the firmware update...

 

[Chee-Hing]

Well, I asked Schiit about it ( the truncation) quite some time ago when their multi-bit DACs came out, so they could not have been ignorant of it and I had to assume it was a deliberate design decision.

 

[Veri]

Well, I asked Schiit about it ( the truncation) quite some time ago when their multi-bit DACs came out, so they could not have been ignorant of it and I had to assume it was a deliberate design decision.

They have told the public there is no 'audible difference' either way. Which is probably true when you look at the performance, the dithering is the least of your concerns... but still, it has nothing to do with proper bitperfect playback if that's what you thought. Truncation most definitely hurts your 24-bit data when fed to the DAC.

 

[Chee-Hing]

My simplistic view of dithering is that the untruncated part of the original data samples will get modified, so in that sense it violates the bit-perfect philosophy. Of course the truncation itself is a violation but the rationalization could be that true 24 bits recordings are unicorns and any significant musical info in the lower 8 bits would be swamped by noise and imperfections in the complete audio chain ( including our hearing ).

I agree that it is not a practical concern, the majority of my audio files are 16 bits anyway.

 

[Veri]

My simplistic view of dithering is that the untruncated part of the original data samples will get modified, so in that sense it violates the bit-perfect philosophy. Of course the truncation itself is a violation but the rationalization could be that true 24 bits recordings are unicorns and any significant musical info in the lower 8 bits would be swamped by noise and imperfections in the complete audio chain ( including our hearing ).

The problem with this thinking that unless you are feeding your files at 100% volume, 24-bit has a great advantage in terms of digital volume control on the client/software-side. If you are just throwing away those last 8 bits that's indeed a rather lossy approach. Iff you're using bit-perfect 16-bit files, those final bits should all be zero's in the first place so no problem there...

Anyhow, it could not be a problem. It's problematic however, that they could have foreseen the possibility of truncation being an issue yet they chose not to implement any dithering. The fact that they fixed it now, does seem to me that they accept it as a mistake made

 

[rwortman]

It's not so much about "bit perfect" as it is Mike's fetish for closed loop mathematics in the digital filter. He wants to be able to theoretically recover the original data from the end result. D/S DAC's make this impossible. There is no correlation between closed loop math and the accuracy of the analog output. It is just his preference.

 

[jules]

OK so I bought bagged a Mimby (v2) off ebay at a good price....

My original source was a Cyrus CD8x cd player (£1000 new 12 years ago) which had developed a fault but I really had no problems with it sonically - it had good detail across the board and excellent timing and dynamics. I wanted to try streaming so picked up a Chromecast Audio in December...

Well the CCA playing Tidal cd quality on it's analogue outputs sounded not bad at all for the price (£30!), compared to the CD8x:

slightly less HF details
similar or just less detail in the midrange
(amazingly) a bit better bass - more powerful and detailed but slightly less punch
reduced dynamics
the big one was timing... rhythmic stuff sounded slow and as soon as lots of instruments came it was a bit of a mess.

So I installed the Mimby on the CCA optical output. It takes the best bits of both the CD8x and the CCA... details across the board are as good as either but the timing is better than the CD8x and miles better than the CCA. Listening to drums and it's as if all the frequency components that make up the drum sound are all lining up at exactly the time time and sound pretty amazing. Bass is as fast as the CD8x but full and extended just like the CCA.

I played First it Giveth by Queens of the Stone Age and it's almost overwhelming to hear the guitars and drums tearing through the speakers.

Downsides? The mimby is perhaps less refined than my previous players and there is a touch of harshness in the upper mid-range that some could find tiring but I've got used to it after a few days.

So like everyone here I read the bad measurements but all I can say is the timing on the Mimby is better than anything I've used with my system and all the other hifi aspects are at least good and for me good enough to keep up with the rest of the components.

Items used:
Cyrus CD8x
CCA
Schiit Mimby v2
Cyrus PreVs2 (pre)
Cyrus X-power (amp)
Leema Xone (speakers https://www.whathifi.com/leema/xone/review )

 

[Veri]

I can say is the timing on the Mimby is better than anything I've used

Ahh, the much eluded "timing" aspect of digital audio. If only there was a way of measuring this :/ Oh wait there is. And it has been measured, it's and nothing special on the modi. If you're hearing anything different it's not " timing" I'm afraid.

I'm not saying that it's impossible to A/B the modi multibit. It's badly distorting enough that you might be able to discern it from another DAC. But timing is a bullshit term I can't agree with, nothing subjective about time.

 

[SIY]

@jules could you please tell us how you did the level-matching, switching, and double-blinding controls that are the basics for valid listening comparisons?

Thanks in advance.

 

[jules]

I'm here to learn so I'm 'all ears' as we say in the UK

Here is the Jitter graph for the CCA on analogue outputs:

So the CCA would appear to have better timing than the Mimby?

 

[Veri]

I'm here to learn so I'm 'all ears' as we say in the UK

Here is the Jitter graph for the CCA on analogue outputs:

So the CCA would appear to have better timing than the Mimby?

Two things
-The general 'flatness' of the graph doesn't mean much, it means the 'spikes' are buried in its noise floor, with the CCA this is at -140dB (good). If noise floor were better we could start seeing nastier things. Sometimes the noise floor makes jitter appear cleaner than it really is.

-Either way it's not an issue. From the audibility thresholds thread:

The green line is "good enough". The orange line is basically perfection already. We 100% can't hear any issues below this line.

 

[jules]

@jules could you please tell us how you did the level-matching, switching, and double-blinding controls that are the basics for valid listening comparisons?

Thanks in advance.

Just 20+ years of working with hifi and settinng PAs for bands etc.

BTW just reading through other threads such as this post:

https://www.audiosciencereview.com/...ast-audio-analog-performance.4562/post-114113

That got 6 likes and no one asked that poster if they had done double blind tests etc.

 

[majingotan]

Just 20+ years of working with hifi and settinng PAs for bands etc.

BTW just reading through other threads such as this post:

https://www.audiosciencereview.com/...ast-audio-analog-performance.4562/post-114113

That got 6 likes and no one asked that poster if they had done double blind tests etc.

That's because it's not a Schiit DAC

 

[Veri]

That got 6 likes and no one asked that poster if they had done double blind tests etc.

You can't expect people to be chasing after more subjective posts in each and every thread. Generally claims like timing/stage/'blackground' and all that jazz are frowned upon though. Ehh~~

 

[jules]

Two things
-The general 'flatness' of the graph doesn't mean much, it means the 'spikes' are buried in its noise floor, with the CCA this is at -140dB (good). If noise floor were better we could start seeing nastier things. Sometimes the noise floor makes jitter appear cleaner than it really is.

-Either way it's not an issue. From the audibility thresholds thread:

The green line is "good enough". The orange line is basically perfection already. We 100% can't hear any issues below this line.

Thanks for the info.

Perhaps we aren't talking about the same thing. When I talk about 'timing' I mean that the when I'm hearing a drum hit, for example, the HF, midrange and LF all are lined up and are replayed at the same time. Is there a measure of this quality? thanks

 

[SIY]

Just 20+ years of working with hifi and settinng PAs for bands etc.

BTW just reading through other threads such as this post:

https://www.audiosciencereview.com/...ast-audio-analog-performance.4562/post-114113

That got 6 likes and no one asked that poster if they had done double blind tests etc.

Ok, so if I understand you correctly, there were no controls and no valid ears-only comparison? If so, I’d start there if you want to get results you can actually trust. Human ears are reliable, human brains aren’t, and the fun part about audio science is understanding that difference and getting down to really understanding what the sonic differences are.

 

[majingotan]

Thanks for the info.

Perhaps we aren't talking about the same thing. When I talk about 'timing' I mean that the when I'm hearing a drum hit, for example, the HF, midrange and LF all are lined up and are replayed at the same time. Is there a measure of this quality? thanks

Perhaps you need to know what J-test jitter means. It's much more stressful in "timing" test than what you're suggesting. https://www.audiosciencereview.com/...derstanding-digital-audio-measurements.10523/

Now, if there's unwanted harmonics popping up on this test at audible levels, that would result in audible timing differences between the DACs. Fortunately almost all modern and competent DACs never have this issue, hence the timing is indistinguishable. Note that Amir states:

Most audiophiles have heard of the term jitter as marketing material for DACs is peppered with it. Typical specs might be “10 picoseconds of jitter” or a statement that some “femtoclock” is used for the DAC oscillator. Please ignore all such talk! For one thing, we don’t care how accurate a DAC clock is. I can slow down or speed up music by 0.1 percent and you would have no idea. That picosecond number is worthless because jitter comes in all forms and it is the nature of it which determines audibility, not what its single value sum total is.

 

[SIY]

One other thing- it’s important to distinguish “timing” in the context of music (scale of dozens of milliseconds) from “timing” in the context of digital clocks (scale of picoseconds, with audibility starting in nanoseconds). Marketeers like to conflate these.

 

[jules]

Ok, so if I understand you correctly, there were no controls and no valid ears-only comparison? If so, I’d start there if you want to get results you can actually trust. Human ears are reliable, human brains aren’t, and the fun part about audio science is understanding that difference and getting down to really understanding what the sonic differences are.

I understand this approach and Amir and all the guys here are doing a great job looking at hifi from the scientific point of view. For me I'm hearing certain things and I'm trying to see the data that could explain that occurrence.

However if subjective thoughts and comments are not allowed on this forum I will fully respect that and keep quiet about such aspects if required.

 

[Veri]

However if subjective thoughts and comments are not allowed on this forum I will fully respect that and keep quiet about such aspects if required.

Good to see you are good spirited. Subjective thoughts can be voiced of course, but will often not be "accepted" as such. At times these subjective posts are also hard to take serious or clutter up topics. That's why people can/will react negatively.