Jump to content
  • 0
IGNORED

Is USB straight from a Mac computer to a DAC really that bad?


Question

Is USB straight from a computer to a DAC really a bad or should everyone always strive to isolate the computer's USB output from the audio stream???  I connect my Chord Quest directly to my Mac Mini; which is dedicated to only running ROON Core; with an AudioQuest Diamond USB and I think it sounds great.  But I according to manufactures of network streamers, eliminating the computer (or using an expensive audio optimized PC like an Innuos) will always sound significantly better.  I've also read that this is not necessarily the case and it really depends on how usb is implemented in the source and the DAC?  Specifically, I heard from Rob Watts of Chord explain that Chord DAC's are optimized for USB direct input.  So, I'm trying to decide if I need to try something like a SOTM SMS-200 Ultra or a Sonore UltraRendu but I'm hesitant to go to the expense and hassle of more boxes.   I recently read this update on this $150K system: https://www.soundstagehifi.com/index.php/opinion/1392-after-25-years-is-this-the-worlds-best-audio-system  Specifically:

"Some Facebook readers criticized me for not using an audiophile-grade music server or USB link. I responded that if anyone can show me a music server or USB link that actually sounds better that what I have in terms of resolution, tonality, soundstaging, imaging, whatever -- I’m all ears. But so far, I’ve heard nothing that has proven itself better-sounding or more versatile -- my computer plays any digital music format and file type from streaming services and my local music drive, and my USB link, with its lengthy length, transfers the bits just fine.

The reason I can get away with using a laptop has to do with the next component in the signal chain: the EMM Labs DA2 Reference DAC ($25,000). Designed by Ed Meitner, who’s been creating digital-audio products since the 1970s, the DA2 Reference seems immune to swaps of USB links, as well as differences in source components."

 

Am I missing something??

Link to post
Share on other sites

Recommended Posts

  • 0
2 hours ago, davide256 said:

in transition. selling Gungnir, mojo is interim solution enroute to Qutest.

 

I re-read and i guess the analog 2 is only for the yggi....how much do you want for your gungnir mb w/unison? And is it a newer gungnir that came with unison or an older one you had upgraded?

Link to post
Share on other sites
  • 0
12 minutes ago, The Computer Audiophile said:

B. 

i just edited answer above, and as you know I just echo others input... i am not a dac engineer, but from what I understand, the only purpose of upsampling is::

 

to quote PKANE::

The point of upsampling is to move the reconstruction/anti-aliasing filter well above audible frequencies, where it can be more gentle and not affect the audible spectrum. A high-quality filter at 22khz is hard to make in hardware without distorting phase and without cutting into frequencies below 20khz. A gentle filter at 88khz is much easier, and it can distort there all it wants without affecting the audible range. It's a simple engineering solution to a problem that could otherwise become audible... at least to some of us (not to me, not for a while )

to quote MONTY MONTGOMERY::

Monty Montgomery https://www.youtube.com/watch?v=cIQ9IXSUzuM&feature=youtu.be

If the ADC/DAC obeys the Nyquist rules (bandlimited) then your waveform will be perfectly reconstructed without additional interpolation. As mentioned its about making filtering easier, but that has become a bit of a moot point with current DACs.

----

these i would have to look up reliable sources, but i "believe" is universally accepted

Upsampling requires processing power and power which can introduce it's own noise if you have don't have quality power supplies.

All upsampling adds artifacts and changes the signal that has to be compensated for.

Is there any logical reasoning to upsample higher than 192K?

I will find reliable sources stating such if there is any problem you have with anything in this post?

I "believe" Miska would be best to answer what upsampling past 192K PCM actually improves on, and what are potential issues?  But without a logical explanation from him, that I am free to debate, I would have a hard time accepting.

 

Link to post
Share on other sites
  • 0
8 minutes ago, beerandmusic said:

i just edited answer above, and as you know I just echo others input... i am not a dac engineer, but from what I understand, the only purpose of upsampling is::

 

to quote PKANE::

The point of upsampling is to move the reconstruction/anti-aliasing filter well above audible frequencies, where it can be more gentle and not affect the audible spectrum. A high-quality filter at 22khz is hard to make in hardware without distorting phase and without cutting into frequencies below 20khz. A gentle filter at 88khz is much easier, and it can distort there all it wants without affecting the audible range. It's a simple engineering solution to a problem that could otherwise become audible... at least to some of us (not to me, not for a while )

to quote MONTY MONTGOMERY::

Monty Montgomery https://www.youtube.com/watch?v=cIQ9IXSUzuM&feature=youtu.be

If the ADC/DAC obeys the Nyquist rules (bandlimited) then your waveform will be perfectly reconstructed without additional interpolation. As mentioned its about making filtering easier, but that has become a bit of a moot point with current DACs.

----

these i would have to look up reliable sources, but i "believe" is universally accepted

Upsampling requires processing power and power which can introduce it's own noise if you have don't have quality power supplies.

All upsampling adds artifacts and changes the signal that has to be compensated for.

Is there any logical reasoning to upsample higher than 192K?

I will find reliable sources stating such if there is any problem you have with anything in this post?

I "believe" Miska would be best to answer what upsampling past 192K PCM actually improves on, and what are potential issues?  But without a logical explanation from him, that I am free to debate, I would have a hard time accepting.

 

I just don’t see how any of that explains your comment here:

 

1 hour ago, beerandmusic said:

There will never be more audible music in high rate dsd compared to a clean 192K pcm signal, which is much easier to keep clean.

 

Perhaps you should ask @Miska, who actually designs filters, about this stuff. 

Founder of Audiophile Style

Announcing The Audiophile Style Podcast

Link to post
Share on other sites
  • 0
8 minutes ago, The Computer Audiophile said:

I just don’t see how any of that explains your comment here:

 

 

Perhaps you should ask @Miska, who actually designs filters, about this stuff. 

 

I edited post above and added MISKA would be the best to answer before you did (smile).  I agree, but even you answered it yourself...you have to design filters to clean up the mess....why create the mess in the first place, when it doesn't give you any more than 192K PCM...again, he could probably answer, and if he was honest, he would likely say that it is subjective (at best) whether it sounds better than 192K PCM without the additional upsampling and filtering.

Link to post
Share on other sites
  • 0
5 minutes ago, beerandmusic said:

 

I edited post above and added MISKA would be the best to answer before you did (smile).  I agree, but even you answered it yourself...you have to design filters to clean up the mess....why create the mess in the first place, when it doesn't give you any more than 192K PCM...again, he could probably answer, and if he was honest, he would likely say that it is subjective (at best) whether it sounds better than 192K PCM without the additional upsampling and filtering.

Oh boy. I kindly suggest you hold up on offering any more "information" about filters etc... It isn't in your wheelhouse. 

Founder of Audiophile Style

Announcing The Audiophile Style Podcast

Link to post
Share on other sites
  • 0
8 minutes ago, The Computer Audiophile said:

Oh boy. I kindly suggest you hold up on offering any more "information" about filters etc... It isn't in your wheelhouse. 

that's why i said to ask MISKA...i didn't say anything about filters...I know they are used for the reconstruction, that is all i know, and won't pretend i understand more.

I only suggest the higher rate DSD will require filters unneeded for lower rate PCM...both require filtering, but I "believe" even MISKA will suggest more advanced filtering is needed for higher rate DSD than lower rate PCM.

Link to post
Share on other sites
  • 0
37 minutes ago, The Computer Audiophile said:

 

TO quote mojo-audio::

 

DSD has significantly higher quantization noise than PCM, and the noise is much closer to audible frequencies, requiring significantly more sophisticated digital filters, as well as noise-shaping and upsampling algorithms. The algorithms native DSD DACs use often result in an overly smoothed over sound without the same immediacy, articulation, and harmonic coherency R-2R ladder DACs are known for.

 

Granted he does suggest using hqplayer to do quad dsd if you are going to do DSD at all, but he also suggest playing music in it's native format.

 

Again, why go from 44.1K PCM to quad rate DSD and all the expense and process power by modifying and filtering just to play something that was recorded at 44.1K....when that rate encompasses our hearing spectrum.

 

He also states::

The truth is that in recent blind studies they've proved that high-resolution PCM and DSD are statistically indistinguishable from one another.

Link to post
Share on other sites
  • 0
2 minutes ago, beerandmusic said:

TO quote mojo-audio::

 

DSD has significantly higher quantization noise than PCM, and the noise is much closer to audible frequencies, requiring significantly more sophisticated digital filters, as well as noise-shaping and upsampling algorithms. The algorithms native DSD DACs use often result in an overly smoothed over sound without the same immediacy, articulation, and harmonic coherency R-2R ladder DACs are known for.

 

Granted he does suggest using hqplayer to do quad dsd if you are going to do DSD at all, but he also suggest playing music in it's native format.

 

Again, why go from 44.1K PCM to quad rate DSD and all the expense and process power just to play something that was recorded at 44.1K....when that rate encompasses our hearing spectrum.

 

He also states::

The truth is that in recent blind studies they've proved that high-resolution PCM and DSD are statistically indistinguishable from one another.

i would never quote Mojo Audio on anything. 

Founder of Audiophile Style

Announcing The Audiophile Style Podcast

Link to post
Share on other sites
  • 0
23 minutes ago, The Computer Audiophile said:

i would never quote Mojo Audio on anything. 

haha....maybe MISKA will step in and correct him...then i would believe it...i trust everything MISKA says (smile)...imho, Miska is the most respected person on this site, with Jabbr a close second.

 

I believe i remember at one point that miska while debating dsd over pcm suggested that one of the main benefits is you can use a cheaper dac....correct me if I am wrong?

 

Miska, if you are there, is there any benefit of Quad rate dsd over 192K PCM if the original recording is 44.1K PCM?  (assume same hardware or if it makes more sense consider all same hardware except dac e.g. schiit gungnir mb w/unison vs RME)

Link to post
Share on other sites
  • 0
43 minutes ago, The Computer Audiophile said:

i would never quote Mojo Audio on anything. 

I mostly quoted him for the part where it says:: DSD has significantly higher quantization noise than PCM, and the noise is much closer to audible frequencies, requiring significantly more sophisticated digital filters, as well as noise-shaping and upsampling algorithms.

 

I rarely will give any credence to subjective opinions without my first hand experience....but above sounds "factual"?

Link to post
Share on other sites
  • 0
23 minutes ago, Miska said:

 

If the DAC perfectly obeys - which is certainly not the case. And it is certainly not a moot point.

 

 

Yes, certainly. But we should first split the topic into oversampling (upsampling) and delta-sigma modulation.

 

At 192k PCM, you are going to have very hard time to design analog filter that will suppress images below -144 dB, without having adverse effects such as unwanted phase shifts and distortion in the audio band. So far, what I've measured, reaching 1.5 MHz PCM upsampling rate allows fairly good reconstruction with a typically used analog filter. Still not -144 dB but at least close to -100 dB to allow reconstruction up to 16-bit RedBook accuracy.

 

Trying to use much higher PCM rates than 1.5 MHz through R2R type DACs poses challenges on settling time, because such multi-bit DAC would need to settle to +-0.5 LSB in way less than half of the sample period. This is then where we get to SDM DACs to allow high speeds and high resolutions.

 

With SDM (DSD) output we can run digital filters with for example 240 dB stop-band attenuation up to ~12 MHz sampling rates. This way we can reconstruct even hires material originally sampled at 192 kHz PCM with very good quality.

 

Then another topic would be performance if different delta-sigma modulator designs.

 

 

Not true, but it can remove distortion artifacts from the original data. It can both reduce artifacts and compensate for inaccuracies of both the source and the DAC. The whole point is to reduce artifacts in the DAC output!

 

Thanks for your response...much appreciated.  Most of it is gobbly-gook to a layman like myself, so please work with me.

 

For my personal interest, Let's "start" with this.

assume source is 44.1Khz (standard wav file)

assume all hardware except DAC is the same and that you have an optimized "quiet" usb.

Do you believe there would any noticeable improvement whether you play that file upsampled to quad DSD using an RME ADI-2 DAC as compared to playing the same file in it's native format using a schiit gungnir mb2 (I only suggest different dacs because you state:: At 192k PCM, you are going to have very hard time to design analog filter that will suppress images below -144 dB, without having adverse effects such as unwanted phase shifts and distortion in the audio band...

you state have a hard time, but i believe the schiit upsamples to 192KPCM, which is why i picked that sampling rate and believe it fully encompasses our full audible spectrum.

 

 

 

Link to post
Share on other sites
  • 0
3 minutes ago, beerandmusic said:

For my personal interest, Let's "start" with this.

assume source is 44.1Khz (standard wav file)

assume all hardware except DAC is the same and that you have an optimized "quiet" usb.

Do you believe there would any noticeable improvement whether you play that file upsampled to quad DSD using an RME ADI-2 DAC as compared to playing the same file in it's native format using a schiit gungnir mb2 (I only suggest different dacs because you state:: At 192k PCM, you are going to have very hard time to design analog filter that will suppress images below -144 dB, without having adverse effects such as unwanted phase shifts and distortion in the audio band...

you state have a hard time, but i believe the schiit upsamples to 192KPCM, which is why i picked that sampling rate and believe it fully incorporates our full audible spectrum.

 

Noticeable to measurement equipment or someone's ears? For measurement equipment certainly. But I cannot comment on anybody else's ears.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
  • 0
17 minutes ago, Miska said:

 

Noticeable to measurement equipment or someone's ears? For measurement equipment certainly. But I cannot comment on anybody else's ears.

 

I do not wish to go deep into the weeds, nor do i wish to push one technology over another.  I am unbiased, I have your software and could care less which direction I ultimately go....I just want to know for my purposes, if one clearly exceeds the other or if it is similar results with marginal subjective differences.

 

When you say for measurement, certainly, can you expand on the differences that could make an audible difference on the output?  again, keeping in mind the original recording is 44.1K.

Link to post
Share on other sites
  • 0
7 minutes ago, beerandmusic said:

When you say for measurement, certainly, can you expand on the differences that could make an audible difference on the output?

 

R2R (or any other DAC) running at such low rate creates significant series of image distortions across the output spectrum. Usually at least to around 1 MHz, but the series can extend even close to 5 MHz. Likely first image is down only some 20 - 30 dB or so. These are fully correlated distortions with same and inverse frequency spectrum. And difference tones between positive and negative frequencies falling into audio band (intermodulation products).

 

While DSD running through typical DAC leaks only low level amount of uncorrelated noise, and any intermodulation products of such are noise as well.

 

Then another aspect is linearity and what to do about that.

 

Third aspect is that if your source content is 192k hires PCM, then if the DAC runs at 192k, there is no space at all for transition band between Nyquist of the source and first image band. Also if the analog filter is designed to cut just above 20 kHz, it will have quite a bit of phase shift at top of the audio band, and also cut into hires content. While if it is higher up, it has much less anti-imaging power. Let's say the filter is typical 12 dB / octave analog low pass. You would need 12 octaves before it reaches -144 dB attenuation or 10 octaves for -120 dB. Digital filter output rate should be high enough that the analog filter reaches -144 dB before first image.

 

How much these kind of things make audible difference is also system dependent.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
  • 0
8 hours ago, davide256 said:
8 hours ago, 6aardvark9 said:

 

Quite the opposite for me... I consider jitter to cause digital glare and top end harshness

tended to find this linked more to source hardware and power supply quality. Of course a PS Audio Dlink III could make any source sound like that

 

what I've experienced is that timing degradation in the source causes background instruments to lose definition

 I agree.

 

How a Digital Audio file sounds, or a Digital Video file looks, is governed to a large extent by the Power Supply area. All that Identical Checksums gives is the possibility of REGENERATING the file to close to that of the original file.

PROFILE UPDATED 13-11-2020

Link to post
Share on other sites
  • 0
25 minutes ago, Miska said:

 

R2R (or any other DAC) running at such low rate creates significant series of image distortions across the output spectrum. Usually at least to around 1 MHz, but the series can extend even close to 5 MHz. Likely first image is down only some 20 - 30 dB or so. These are fully correlated distortions with same and inverse frequency spectrum. And difference tones between positive and negative frequencies falling into audio band (intermodulation products).

 

While DSD running through typical DAC leaks only low level amount of uncorrelated noise, and any intermodulation products of such are noise as well.

 

Then another aspect is linearity and what to do about that.

 

Third aspect is that if your source content is 192k hires PCM, then if the DAC runs at 192k, there is no space at all for transition band between Nyquist of the source and first image band. Also if the analog filter is designed to cut just above 20 kHz, it will have quite a bit of phase shift at top of the audio band, and also cut into hires content. While if it is higher up, it has much less anti-imaging power. Let's say the filter is typical 12 dB / octave analog low pass. You would need 12 octaves before it reaches -144 dB attenuation or 10 octaves for -120 dB. Digital filter output rate should be high enough that the analog filter reaches -144 dB before first image.

 

How much these kind of things make audible difference is also system dependent.

 

 

THanks..again, assuming 44.1K hz source and assuming hardware is same, so throwing out last 2 paragraphs (for my purposes...others may be interested in that part though)....so concentrating on 1st paragraph only.....

 

creates significant series of image distortions across the output spectrum. Usually at least to around 1 MHz, but the series can extend even close to 5 MHz. Likely first image is down only some 20 - 30 dB or so. These are fully correlated distortions with same and inverse frequency spectrum. And difference tones between positive and negative frequencies falling into audio band (intermodulation products).

 

----

So to clarify, the schiit would create these "image distortions across the output spectrum"?

And you state usually around 1mhz? 

I thought the audible range is typically up to 50KHZ max?

 

I am not going to pretend i understand any of this, because I don't, and you can easily convince me.

 

I think I mainly want to get a schiit, because other dacs have "seemed" less lively for some reason...but maybe it is just my imagination.  I also want to think that with the unison USB, that it will make a difference with something i can nicely pair with my regular daily use windows PC....

 

I mostly play 44.1K wav files., but if you think that upsampling to quad dsd will give me a better output, i will likely head in that direction, even though i have done that before...but maybe with the gustard a18 with it's linear power supply it may make a better difference to my ears.

 

 

Link to post
Share on other sites
  • 0
8 minutes ago, beerandmusic said:

So to clarify, the schiit would create these "image distortions across the output spectrum"?

And you state usually around 1mhz?

 

Every second image band has inverse spectrum. This creates image pairs of positive and negative frequencies around multiples of the sampling rate DAC operates at.

 

I said up to. First image of DAC operating at 192 kHz from 44.1k source is around 192 kHz and spans form 169.95 kHz to 214.05, same repeats around 384 kHz, etc.

 

Just to simplify, for perfect reconstruction according to the sampling theory, all these images must be completely removed by the reconstruction filter. If there is something left, the reconstruction is incomplete and not accurate, and contains systematic  / correlated error. Point of upsampling is to move these images up in frequency so that the analog filter can do better job. In ideal case, the combined result of the digital and analog filter is perfect reconstruction.

 

19 minutes ago, beerandmusic said:

I mostly play 44.1K wav files., but if you think that upsampling to quad dsd will give me a better output, i will likely head in that direction.

 

You still need to pay attention to the DAC. Not all DACs perform the same.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
  • 0
52 minutes ago, Miska said:

Third aspect is that if your source content is 192k hires PCM, then if the DAC runs at 192k, there is no space at all for transition band between Nyquist of the source and first image band.

 

 However, in the real world there is very little genuine musical material beyond around 55kHz ,with that already at quite a low level,  and images of any musical material above that will be at extremely low levels. The microphones used in the attached by Barry Diament were only 1dB down at 40kHz which is a wider bandwidth than most .

02.Bye- Ya (Bolero) .jpg

 

How a Digital Audio file sounds, or a Digital Video file looks, is governed to a large extent by the Power Supply area. All that Identical Checksums gives is the possibility of REGENERATING the file to close to that of the original file.

PROFILE UPDATED 13-11-2020

Link to post
Share on other sites
  • 0
18 minutes ago, Miska said:

I said up to. First image of DAC operating at 192 kHz from 44.1k source is around 192 kHz and spans form 169.95 kHz to 214.05, same repeats around 384 kHz, etc.

 

 

but all of those are still above 50KHZ?

For my ears, i don't think i would even care about anything above 20khz?

 

Am i missing something?

 

i know you started off saying you can't comment on "others ears", so let me rephrase...

Is there any measurable differences on the output  that "could" be audible between the 20hz-40khz range

 

Link to post
Share on other sites
  • 0
35 minutes ago, sandyk said:

 However, in the real world there is very little genuine musical material beyond around 55kHz ,with that already at quite a low level,  and images of any musical material above that will be at extremely low levels. The microphones used in the attached by Barry Diament were only 1dB down at 40kHz which is a wider bandwidth than most .

02.Bye- Ya (Bolero) .jpg

 

Try with Sanken CO-100K ;)

 

Also with Sennheiser MKH-8020 you get pretty wide bandwidth for recording cymbals and such.

 

And not necessarily so low levels even at 100 kHz:

http://www.cco.caltech.edu/~boyk/spectra/11.htm#b

 

And of course with electronic music, or music using synths, you are not limited by microphones, you just get what the synthesizer algorithms create in digital domain.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
  • 0
30 minutes ago, beerandmusic said:

but all of those are still above 50KHZ?

For my ears, i don't think i would even care about anything above 20khz?

 

Am i missing something?

 

Because the difference frequencies (intermodulation) between those fall into audio band.

 

31 minutes ago, beerandmusic said:

Is there any measurable differences on the output  that "could" be audible between the 20hz-40khz range

 

Not sure I understand the question. But if you want to focus on 20 - 40 kHz range, the measurable differences are both digital filter differences and analog differences.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites
  • 0
7 minutes ago, Miska said:

 

Try with Sanken CO-100K ;)

 

Also with Sennheiser MKH-8020 you get pretty wide bandwidth for recording cymbals and such.

 

And not necessarily so low levels even at 100 kHz:

http://www.cco.caltech.edu/~boyk/spectra/11.htm#b

 

And of course with electronic music, or music using synths, you are not limited by microphones, you just get what the synthesizer algorithms create in digital domain.

 

Miska

 I have previously posted the same spectra above, however, as I stated , very few recordings have much genuine musical content above 55kHz due to the actual microphones used.

 Do you have you any recordings like you mentioned that have such content that you can post a copy of their frequency spectrum ?

 Have you heard anything about the Laser microphones that were mentioned some years back, with their extremely wide bandwidth  ?

 

Alex

 

How a Digital Audio file sounds, or a Digital Video file looks, is governed to a large extent by the Power Supply area. All that Identical Checksums gives is the possibility of REGENERATING the file to close to that of the original file.

PROFILE UPDATED 13-11-2020

Link to post
Share on other sites
  • 0
6 minutes ago, sandyk said:

 I have previously posted the same spectra above, however, as I stated , very few recordings have much genuine musical content above 55kHz due to the actual microphones used.

 

That is sort of irrelevant for the topic here though. For 192k hires you'd want to have analog low pass filter fc at least not lower than 100 kHz. Which already means that you have quite severe phase shift at 50 kHz, but decent phase shift at 20 kHz. But then if your conversion section runs at 192 kHz sampling rate having fc of 100 kHz means that you leak severe amount of images and thus have very poor reconstruction accuracy. OTOH, if you put fc at let's say 25 kHz, you get a little more attenuation for the images, but severely modified phase response in audio band.

 

Now compare that to having 100 kHz analog fc and conversion section running for example at 1.5 MHz...

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×
×
  • Create New...