Differences in sound: DAC vs. DAC + Pre-amplifier

[barrows]

4 minutes ago, hopkins said:

Using optical fiber and thinking it actually provides isolation is complete nonsense because you are still putting an ethernet chip inside the DAC. 

 

There are no measurements to back up the fact that this is a superior solution. 

not sure what an "Ethernet Chip" is?  There really is no such thing.

 

Your post misses the point I was making, perhaps I was not clear.

 

I was talking about an Ethernet Renderer connected to a DAC via USB, vs. a Server connected to a DAC via USB.  the point is that the Renderer presents a lower noise profile to the DAC than that of the Server.

[The Computer Audiophile]

3 minutes ago, Summit said:

 

Have you compared your preferred Endpoints to a SOTA server like Grimm Audio MU1, Innuos Statement or Extreme?

I've done some comparisons, but not with those exact models. That's why I made sure to note, "in my experience." I would love to compare more.

 

[Summit]

1 hour ago, barrows said:

Exactly, and by using Ethernet you keep noise from the server coming in.  Specifically, best practice is to use optical fiber Ethernet.  No matter what extremes one goes to trying to make a server as silent as possible, that server will never be as silent as well designed Renderer.  I suggest, that the better approach, is not bother about making the server silent.  Put it in another part of the home, and do not let its electrical noise to get to the audio system, by sending the music to the audio system over Ethernet (which is isolated by transformers), or even better, with optical fiber Ethernet (which does not pick up and carry electrical noise at all).

The the only noise which gets to the audio system is that generated locally, in the audio system.

 

There is another big advantage to Networked Audio as well:  You can do as much processing in the server as you may want, run room correction, for example, or much more sophisticated oversampling programs like HQPlayer (which can be a big sonic advantage, especially HQPlayer oversampling and a simple DAC which does no additional processing onboard, reducing noise in the DAC even further).  All this processing in the server makes a lot of noise, but by isolating the server away form the audio system, connected by only an optical fiber cable, the noise never gets to the audio system.  

 

I know and understand the theoretical advantage of Renderers. But theoretical advantage and practise is not always the same. I have no proof of it but I believe that if the digital signal has been polluted some noise will “ride” with the signal and it’s very hard (to some extent impossible) to get rid of.

 

My own experience has showed me that the sever matters with my uR and I have said it many times. I was one of the first to get a JCAT net board and to use LPS on my switch because of that.  

 

One more time, can you show me evidence (measurements or listing evaluations) to support that Ethernet Renderer is better than SOTA server like Innuos Zenith se, Innuos Statement or Extreme?

[davide256]

55 minutes ago, barrows said:

There is no possible technical explanation for why this would be so.  and it is entirely contrary to my, and thousands of others' experience.  No matter what extremes one goes to, a server can never have a s low a noise profile as well designed Renderer connected by optical fiber Ethernet.  The server, by definition, generates a larger noise profile.

I would suggest that the above expressed viewpoint is an outrageous one, not supported by any valid technical theory, and as such would be in opposition to any accepted understanding of how electrical systems, and audio systems actually work.  In order to support such an outrageous claim, I would suggest that a lot more than listening impressions of a few people would be required.

Can you provide any technical theory for this claim?

Can you provide any measurements showing how this approach could possibly make any DAC perform better?

 

You  know, try as I can, I've never been able to get server hardware that was not USB attached to the DAC to matter as much as the software it was running.

In all of my experimentation 85% of improvements came from tweaking/upgrading the USB endpoint. Perhaps it matters if you have a really resolving system

in the $100k price range but it certainly doesn't seem to at $10K worth of system investment.

[bodiebill]

21 minutes ago, hopkins said:

Using optical fiber and thinking it actually provides isolation is complete nonsense because you are still putting an ethernet chip inside the DAC. 

 

There are no measurements to back up the fact that this is a superior solution. 

 

I agree. And it is not absolute, but all depends on someone's requirements:

 

If one absolutely need network for whatever reason and a heavy lifting PC for HQPlayer upsampling, then the dual PC endpoint solution can bring benefits.

 

However there is a possible world in which one does not need network and does offline file upsampling, moving the hard work to outside the real time realm. There a low footprint PC (or even better: a minimalist SD transport) without network connection will bring even greater benefits.

 

Don't we all agree that network activity to some degree degrades SQ?

 

[Qhwoeprktiyns]

23 minutes ago, barrows said:

not sure what an "Ethernet Chip" is?  There really is no such thing.

 

Your post misses the point I was making, perhaps I was not clear.

 

I was talking about an Ethernet Renderer connected to a DAC via USB, vs. a Server connected to a DAC via USB.  the point is that the Renderer presents a lower noise profile to the DAC than that of the Server.

 

Ok. But there are many ways of lowering the noise from the source. 

 

Via USB, you cannot completely isolate the DAC from the source. So how are you to correlate the noise level in the source to an actual performance increase in a DAC. Very difficult. 

 

I quoted you yesterday, and that really summarizes your logic, which is  just guesswork:

 

"such measures when done well have proven to improve DAC performance to levels where further efforts in this direction are probably not going to result in any actual audible change" 

[barrows]

5 minutes ago, Summit said:

My own experience has showed me that the sever matters with my uR and I have said it many times. I was one of the first to get a JCAT net board and to use LPS on my switch because of that.  

Indeed, and the working theory for why this is the case is that clock phase noise is the single issue here.  The Jcat NiC has a good clock, hence the improvement.  And of course you are referring to an electrical interface, where noise may propagate, so a low noise NiC is a good idea, and may improve performance.

With an optical fiber interface, general noise from the server does not get to the Renderer, that is just a physical reality.  But there is a theory (and only a theory, at this point unproven) that clock phase noise does travel with the signal.  I am still skeptical, as the implications seem to me to indicate that the Internet itself could not possibly work reliably if this were true...  But, perhaps the measurement system will be developed well enough to actually show that this can happen.  In the mean time, I do advise for using an upstream device with good clocking of the Ethernet data stream.

[barrows]

11 minutes ago, hopkins said:

I quoted you yesterday, and that really summarizes your logic, which is  just guesswork:

No they are not, they are verified through measurements, and correlated with listening tests.

 

BTW, I spent a little time looking for measurements of EC designs DACs.  Given the very high levels of distortion products in the outputs of these DACs, I would not trust that any relevant opinions could be formed by listening tests, of source noise, using these DACs.  Any low level effects caused by source noise would almost certainly be swamped by audible levels of distortion at the DAC's outputs.  I am not surprised that EC Designs does not give any specs for their DACs.

[Summit]

14 minutes ago, barrows said:

Indeed, and the working theory for why this is the case is that clock phase noise is the single issue here.  The Jcat NiC has a good clock, hence the improvement.  And of course you are referring to an electrical interface, where noise may propagate, so a low noise NiC is a good idea, and may improve performance.

With an optical fiber interface, general noise from the server does not get to the Renderer, that is just a physical reality.  But there is a theory (and only a theory, at this point unproven) that clock phase noise does travel with the signal.  I am still skeptical, as the implications seem to me to indicate that the Internet itself could not possibly work reliably if this were true...  But, perhaps the measurement system will be developed well enough to actually show that this can happen.  In the mean time, I do advise for using an upstream device with good clocking of the Ethernet data stream.

 

If an optical fiber interface would make the digital signal immune to all types of noise. A switch like the ones from UpTone would not make any difference, right? Have you tested an EtherREGEN? If not maybe Alex can lend you one .

[barrows]

20 minutes ago, bodiebill said:

Don't we all agree that network activity to some degree degrades SQ?

Do you mean other Network activity, separate from the activity of playing the music file?  If one is concerned about such, then I would advise an isolated dedicated LAN for the audio system, this is not hard to implement.  I would suggest testing for such effects in one's own system before bothering though: how about streaming some hi res video to couple of TVs while listening, and then shutting down the rest of the Network and listening again.

If you mean the network activity only associated with playing the music file, I would suggest that optical fiber cable isolates the Renderer and DAC from that.  If we are playing music files from a computer, now matter how we do it, there is always some processor activity associated with the action, this is unavoidable.  So the question becomes, how much processor activity is acceptable (or even audible), and how do we reduce that activity.  Using a Renderer which runs with less than 25 watts of power (one can equate power usage, roughly, to how much noise will be produced) is a good way to reduce computer activity which may effect the DAC to the lowest level possible.

Oversampling in the computer, and using a DAC with no onboard oversampling, is another way to reduce noise inside the DAC, and comes with the side \benefit of being able to apply more sophisticated and accurate oversampling algorithms.

[barrows]

7 minutes ago, Summit said:

f an optical fiber interface would make the digital signal immune to all types of noise

I did not say that, please go back and re-read the posts.

The current theory is that clock phase noise may be able to travel embedded in the signal, although this theory is highly speculative at this point.  Because of this, a good clock handling the Ethernet data stream is recommended for absolute best performance.  Hopefully, at some point, there may be verification of this theory.

General, server borne, electrical noise is not transmitted over optical fiber interfaces.

[Qhwoeprktiyns]

22 minutes ago, barrows said:

No they are not, they are verified through measurements, and correlated with listening tests.

 

BTW, I spent a little time looking for measurements of EC designs DACs.  Given the very high levels of distortion products in the outputs of these DACs, I would not trust that any relevant opinions could be formed by listening tests, of source noise, using these DACs.  Any low level effects caused by source noise would almost certainly be swamped by audible levels of distortion at the DAC's outputs.  I am not surprised that EC Designs does not give any specs for their DACs.

 

As I said, I am not affiliated to them, and i suggested that you address  your questions and speculations directly with them. I send you a link yesterday to illustrate another point of view on the topic of isolation,  to which you basically replied: optical sucks because DSD is better! 

 

Well, believe it or not, USB ain't that great either.

 

We live in an imperfect "audio" world, with a lot of "unknowns", and i think the quote i gave above illustrates well that point. 

 

At the end of the day, it all comes down to listening (which is by the way what you admit to doing to evaluate the superiority of DSD). I suggest you do more of that... 

 

 

 

[Summit]

52 minutes ago, The Computer Audiophile said:

I've done some comparisons, but not with those exact models. That's why I made sure to note, "in my experience." I would love to compare more.

 

 

To me it’s clear after reading reviews and post by people that have first-hand experience with those servers, that they hold them as “sounding” better than all Ethernet Renderer.

 

I have no experience with any of them myself and find Ethernet Renderer together with a good server to be very good and much more affordable option.  

[bodiebill]

8 minutes ago, barrows said:

Do you mean other Network activity, separate from the activity of playing the music file?

...

If you mean the network activity only associated with playing the music file, I would suggest that optical fiber cable isolates the Renderer and DAC from that.

 

In my experience both have their effect, the second more than the first. But even when I played files locally, not via the network but with network attached, I got better sound when removing the network connection, alas losing my remote access of course.

[barrows]

14 minutes ago, hopkins said:

 

As I said, I am not affiliated to them, and i suggested that you address  your questions and speculations directly with them. I send you a link yesterday to illustrate another point of view on the topic of isolation,  to which you basically replied: optical sucks because DSD is better! 

 

Well, believe it or not, USB ain't that great either.

 

We live in an imperfect "audio" world, with a lot of "unknowns", and i think the quote i gave above illustrates well that point. 

 

 

 

Respectfully, the above illustrates that either you are not actually reading my posts, or you are misreading them in some way:

 

I am not speculating here, at all.  Again, I am referring to things which have been shown in measurements and correlated with listening tests-not speculation.

 

If you prefer to believe in "magical" solutions to audio "problems" which are demonstrated to be non-existent, i am fine with that.  But this is not my approach, and I do not share "magical" beliefs with you.

 

There are not as many "unknowns" as many people seem to like to believe.  Electronics, and audio, are very well understood.  Perhaps not perfectly understood, but very nearly so.  developing audio equipment is a process of engineering, it is science, and it actually adheres to the laws of physics.

[Qhwoeprktiyns]

Once again

 

"such measures when done well have proven to improve DAC performance to levels where further efforts in this direction are probably not going to result in any actual audible change"

 

Peace. Will no longer reply here. Have a great day! 

[barrows]

4 minutes ago, bodiebill said:

I got better sound when removing the network connection, alas losing my remote access of course.

Interesting, was this with an optical fiber Network connection?  And did you ever try and verify the test with someone else removing the Network connection while you listened, without your knowledge, and were able to accurately discern the exact moment the connection was removed?

 

[bodiebill]

3 minutes ago, barrows said:

Interesting, was this with an optical fiber Network connection?

 

Yes, with optical isolation also, using a fiber NIC in the endpoint. However in the end I changed back the last downstream part with copper ethernet as to my great surprise this clearly sounded better to my ears.

 

3 minutes ago, barrows said:

And did you ever try and verify the test with someone else removing the Network connection while you listened, without your knowledge, and were able to accurately discern the exact moment the connection was removed?

 

No, never did that so I cannot disprove placebo 🙂

However I am usually not comparing short term A/B but long term effects by living with the setups for days, being aware of the frequency of goosebumps or at least the amount of joy and involvement.

Sometimes this makes me conclude the opposite of the initial A/B assessment.

For instance if A has a higher level of detail I initially think it is better, but when I notice that I can listen to B much longer and stay involved, I choose B.

It is all very intuitive (sorry) so perhaps hard to replicate for others.

 

For years I had a complete optical network between server and endpoint, and was happy as in my mind optical was almost synonymous to noise isolation. Now I think that was an example of placebo, as when I went back to copper for the last stretch, just to try, I saw had been wrong, at least in terms of perceived SQ.

 

[matthias]

3 hours ago, Summit said:

To me it’s clear after reading reviews and post by people that have first-hand experience with those servers, that they hold them as “sounding” better than all Ethernet Renderer.

 

A good example is this one with the Taiko Extreme:

 

https://www.whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-150#post-631854

 

Maybe interesting for @barrows as well. 🙂

 

Matt

[barrows]

3 hours ago, bodiebill said:

 

Yes, with optical isolation also, using a fiber NIC in the endpoint. However in the end I changed back the last downstream part with copper ethernet as to my great surprise this clearly sounded better to my ears.

 

 

No, never did that so I cannot disprove placebo 🙂

However I am usually not comparing short term A/B but long term effects by living with the setups for days, being aware of the frequency of goosebumps or at least the amount of joy and involvement.

Sometimes this makes me conclude the opposite of the initial A/B assessment.

For instance if A has a higher level of detail I initially think it is better, but when I notice that I can listen to B much longer and stay involved, I choose B.

It is all very intuitive (sorry) so perhaps hard to replicate for others.

 

For years I had a complete optical network between server and endpoint, and was happy as in my mind optical was almost synonymous to noise isolation. Now I think that was an example of placebo, as when I went back to copper for the last stretch, just to try, I saw had been wrong, at least in terms of perceived SQ.

 

I really do not want to be promoting specific products here, and have tried to avoid doing so.  But, I would suggest that actual Renderers purpose built for audio, do have a technological (not imaginary) advantage in terms of noise versus general commercial computer gear.  Built for audio products generally have much more room in the BOM to spec more expensive components (like many ultra low noise linear regulators) and more sophisticated layouts, with more isolated sections.  For just a single example of what I mean: how many commercial main boards used as a renderer have a dedicated LT 3045 regulator, just a few mms from the USB output for the USB output power leg?

i agree that long term listening is often better for really evaluating a change, although it can be very difficult.  For me, short term comparisons often lead to "ah ha!" moments, where there "might" be a "difference", but the difference in often a sideways one, and not necessarily an improvement.  We humans seem especially subject to interpreting any change as "exciting" and therefore concluding it is better; long term listening seems to even out these first impressions, especially when a wide variety of music is tssted.  I always try and verify any conclusions based on short term A/B style comparisons with long term listening.

[barrows]

51 minutes ago, matthias said:

 

A good example is this one with the Taiko Extreme:

 

https://www.whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-150#post-631854

 

Maybe interesting for @barrows as well. 🙂

 

Matt

What a waste of time that was.  Same old nonsense, with no basis in technical reality and no actual facts, just a listening preference.  I asked for a technical explanation of these claims, and support of measurements, i see none here.

[matthias]

5 minutes ago, barrows said:

What a waste of time that was.  Same old nonsense, with no basis in technical reality and no actual facts, just a listening preference.  I asked for a technical explanation of these claims, and support of measurements, i see none here.

 

It is easy, if it sounds better it is better.

The technical explanation why it is better sometimes comes much later.

 

Matt

 

[matthias]

@barrows

 

Maybe a more technical explanation here:

 

https://www.whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-99#post-616560

 

Matt

[barrows]

12 minutes ago, matthias said:

@barrows

 

Maybe a more technical explanation here:

 

https://www.whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-99#post-616560

 

Matt

That is not a technical description.  What I mean by a technical description is one which gives an actual technical explanation for why a given approach results in a change in performance.  

[The Computer Audiophile]

It's all a matter of where to place the computer and how clean the output is. 

 

Some people think that a higher power PC closer to the audio device is best while others think a lower power device connected to the audio components is best. This isn't rocket science. We are connecting computers to our DACs. So far, I'm a big fan of low powered custom designed fiber connected endpoints such as the Signature Rendu SE optical. I'm totally cool if others like the opposite. I have no desire to dwell on it and make calls to "authorities" supporting me or refuting me. 

 

I don't think we get very far by posting definitive statements based on a couple guy's experiences.