UptoneAudio EtherREGEN

[Discopants]

That can get to be an expensive philosophy John, if I wanted to improve on my Lyngdorf TDAI 3400 without losing room pefect which I value highly I would have to drop £30,000 for a Steinway Lyngdorf model S. Thats not going to happen unless I get a truly massive win on the premium bonds.

Considering all my sources are streamed I’ve found tweaking my network to be a fun and SQ rewarding endeavour.

 

[mansr]

However, to me, a "decent system" is one that does not respond audibly to changes in the digital network, as long as the digital network remains functionally solid and well-enough dimensioned to transport the data. In my philosophy, the digital network has no valid role in changing the sound; and my own experience is that I can achieve that independence.

If someone else likes the idea of tweaking audio by swapping digital network components then it's their hobby and that's OK. But if changing a digital network component caused audible changes here I would be straight on the case to find out which part of my audio system was not well enough engineered.

Exactly. If something that shouldn't make a difference still does, it means there's a fault in the system.

 

[Coda II]

Exactly. If something that shouldn't make a difference still does, it means there's a fault in the system.

I am aware that the above reply was related to a point with the specific context of digital networks but just for the purposes of clarification would it be OK to stray just a little from that?

If you were to draw a Venn diagram of 'things that shouldn't make a difference' and 'things that might make a difference' would there be an area of intersection between the two; would there be an area for things that shouldn't make a difference but still might?

As an example (which may or may not be considered part of the digital network): the connection between streaming device and DAC could be optical or electrical, I don't think it is particularly contentious to say that there can be differences between the two in the same system and there are reasons why this is the case. However, as all it is doing is transferring digital data, it would be possible to claim (and in some implementations it is probably the case) that both are identical, that it shouldn't make a difference. But generally most on here would allow that it might (I think).

To bring it back to the question of the network before it gets to the streaming device, all I am trying to gauge is how absolute views are on 'things that shouldn't make a difference so we should not give any time at all to considering how they might' as opposed to 'if this might make a difference what could be done to alter it, for better or worse, short of breaking it completely?

 

[John Phillips]

That can get to be an expensive philosophy John, if I wanted to improve on my Lyngdorf TDAI 3400 without losing room pefect which I value highly I would have to drop £30,000 for a Steinway Lyngdorf model S. Thats not going to happen unless I get a truly massive win on the premium bonds.

Considering all my sources are streamed I’ve found tweaking my network to be a fun and SQ rewarding endeavour.

Absolutely fine that that's fun and rewarding. No challenge from me to the OP or anyone else who finds that a good way to pursue the hobby. I would certainly hope a Steinway Lyngdorf model S system could be network agnostic if that's what the designers intended. But however expensive it can get, in my experience an alternative philosophy does not have to be expensive.

It will only necessarily be expensive if immunity of the audio system to variations in digital transport is necessarily expensive. To cherry-pick from my own experience, after experimenting for a while I finally settled on a two Raspberry Pi network connected via WiFi &USB into an ATC CDA2 Mk2 Preamp/DAC/CD player. At under £3k the CDA2 was not (to me) expensive. Whatever I threw at the CDA2 from the network side when deciding what to settle on, made no difference to my ears on the audio side. And since for a short period I was professionally working in digital audio R&D (but not for HiFi) I had enough learning to challenge the CDA2 more extensively than most. This performance might not be pleasing to the OP or others but it pleased me.

Ultimately if anyone enjoys pursuing their objectives by changing digital network components that's fine. The OP puts forward experience that I don't dispute. Presumably for the purpose of initiating discussion since there's no other indication of purpose. My comment is that there are other philosophies. And as in my own example above I have found that a different philosophy can be pleasing to me and does not have to be expensive.

 

[Coda II]

I finally settled on a two Raspberry Pi network connected via WiFi &USB

I'm curious about your Pi set-up - if it is detailed elsewhere could you point me at it please? Alternatively: is one a server one an end point and what (if any) HATs are in use? Thank you.

 

[Heckyman]

I’m the OP, my initial enthusiastic observations were well over a year ago. I was simply sharing what had worked for me, in the hope that others might chime in with their own experiences, positive or negative, to add to the collective knowledge base.

That said, I sold my ER after reconfiguring my system to use WiFi or local files with the new Mac mini M1 direct to DAC. With this arrangement, I have the option to just download the files and disconnect the network completely. I find I very rarely bother. If there’s any difference at all (between network and no network) it’s a very subtle one.

 

[John Phillips]

I am aware that the above reply was related to a point with the specific context of digital networks but just for the purposes of clarification would it be OK to stray just a little from that?

If you were to draw a Venn diagram of 'things that shouldn't make a difference' and 'things that might make a difference' would there be an area of intersection between the two; would there be an area for things that shouldn't make a difference but still might?

As an example (which may or may not be considered part of the digital network): the connection between streaming device and DAC could be optical or electrical, I don't think it is particularly contentious to say that there can be differences between the two in the same system and there are reasons why this is the case. However, as all it is doing is transferring digital data, it would be possible to claim (and in some implementations it is probably the case) that both are identical, that it shouldn't make a difference. But generally most on here would allow that it might (I think).

To bring it back to the question of the network before it gets to the streaming device, all I am trying to gauge is how absolute views are on 'things that shouldn't make a difference so we should not give any time at all to considering how they might' as opposed to 'if this might make a difference what could be done to alter it, for better or worse, short of breaking it completely?

There are certainly things I think should not make a difference that can and probably do make a difference. There is theory (and philosophy) and there is the engineering approximation to theory. They aren't exactly the same.

Draw a line through the middle of a DAC box. The digital bits on one side and the analogue bits on the other [1]. That's where I see the network ending and the audio system beginning. Ignoring details of the D/A conversion that happens across the line [2], in my philosophy that line should be a perfect barrier to unwanted imperfections from the network getting onto the audio signal.

The network and the digital side of the DAC are always noisy. Time noise (jitter) and amplitude noise. These imperfections should definitely not alter the digital data, but do always leak from the digital side to the analogue side. The question relevant to me is "is the leakage below my threshold for what matters?" S/PDIF, TOSLINK and USB interfaces are differently impacted. The levels of leakage today can be very low indeed for not too much money provided the network is not egregiously noisy. If that's achieved it's what I like but it does make network-side tweaks irrelevant which may not please others.

But even if they can, no-one can say that all DACs will do well at rejecting the bad stuff that happens in the digital domain. Also, I suspect it is quite possible for networks to still work but have egregious amounts of noise that no sane DAC will reject well enough. So network-side tweaks might make a difference and some will undoubtedly enjoy that line of experimentation.

[1] OK, if you know what happens inside a DAC that's usually not precisely possible but near enough.

[2] There are theory and philosophical issues here too, about what is "perfect". But I won't address them here.

 

[Coda II]

There are certainly things I think should not make a difference that can and probably do make a difference. There is theory (and philosophy) and there is the engineering approximation to theory. They aren't exactly the same.

Draw a line through the middle of a DAC box. The digital bits on one side and the analogue bits on the other [1]. That's where I see the network ending and the audio system beginning. Ignoring details of the D/A conversion that happens across the line [2], in my philosophy that line should be a perfect barrier to unwanted imperfections from the network getting onto the audio signal.

The network and the digital side of the DAC are always noisy. Time noise (jitter) and amplitude noise. These imperfections should definitely not alter the digital data, but do always leak from the digital side to the analogue side. The question relevant to me is "is the leakage below my threshold for what matters?" S/PDIF, TOSLINK and USB interfaces are differently impacted. The levels of leakage today can be very low indeed for not too much money provided the network is not egregiously noisy. If that's achieved it's what I like but it does make network-side tweaks irrelevant which may not please others.

But even if they can, no-one can say that all DACs will do well at rejecting the bad stuff that happens in the digital domain. Also, I suspect it is quite possible for networks to still work but have egregious amounts of noise that no sane DAC will reject well enough. So network-side tweaks might make a difference and some will undoubtedly enjoy that line of experimentation.

[1] OK, if you know what happens inside a DAC that's usually not precisely possible but near enough.

[2] There are theory and philosophical issues here too, about what is "perfect". But I won't address them here.

Just to make sure I have understood: in the context of your current set-up, it is the ATC that is doing what it is supposed to in keeping the leakage out rather than anything on the digital side that has been optimised to keep noise to an acceptable level?

 

[John Phillips]

I'm curious about your Pi set-up - if it is detailed elsewhere could you point me at it please? Alternatively: is one a server one an end point and what (if any) HATs are in use? Thank you.

Details are on the information tab of my profile page.

 

[John Phillips]

Just to make sure I have understood: in the context of your current set-up, it is the ATC that is doing what it is supposed to in keeping the leakage out rather than anything on the digital side that has been optimised to keep noise to an acceptable level?

Yes. There are two sides to unwanted leakage from digital to analogue: the amplitude of the digital-side noise & the rejection of the CDA2. I have tested the question "can I hear rubbish on the CDA2 output?" The answer is "no". But I do not know if that's because the Pi 3A+ is not too noisy or the CDA2 rejects noise extremely well or some combination. I have done nothing to mitigate any rubbish from the Pi 3A+ to the CDA2 on USB. EDIT: And I do not know for sure whether there is some subtle impact from the network on the audio that isn't manifested as noise (although I detected nothing to spoil my enjoyment of the music when trying things out).

A further thought is that IMHO the entire information content of digital audio ought to be in the bits. No mysterious wanted side-channel. So if the bits are perfectly delivered to the DAC then IMHO any audio impact from the network can only come from unwanted side-channel noise. But I would be interested to hear if others think network impacts on audio come from other sources than the side-channels and indeed hear if network impacts on audio are considered a good thing (as will be seen I think they are always bad).

 

[Coda II]

Yes. There are two sides to unwanted leakage from digital to analogue: the amplitude of the digital-side noise & the rejection of the CDA2. I have tested the question "can I hear rubbish on the CDA2 output?" The answer is "no". But I do not know if that's because the Pi 3A+ is not too noisy or the CDA2 rejects noise extremely well or some combination. I have done nothing to mitigate any rubbish from the Pi 3A+ to the CDA2 on USB. EDIT: And I do not know for sure whether there is some subtle impact from the network on the audio that isn't manifested as noise (although I detected nothing to spoil my enjoyment of the music when trying things out).

A further thought is that IMHO the entire information content of digital audio ought to be in the bits. No mysterious wanted side-channel. So if the bits are perfectly delivered to the DAC then IMHO any audio impact from the network can only come from unwanted side-channel noise. But I would be interested to hear if others think network impacts on audio come from other sources than the side-channels and indeed hear if network impacts on audio are considered a good thing (as will be seen I think they are always bad).

Is there a purely practical reason in your set-up in having a separate server-Pi and renderer-Pi? I run a single Pi with USB SSD but get the impression that this is not what the majority of Pi users do. For example, the developer at Allo states that the USB input is noisy (not necessarily that it corrupts the signal in anyway but that it introduces noise to the unit) and therefore it is better to have the music files served via ethernet.

 

[John Phillips]

Is there a purely practical reason in your set-up in having a separate server-Pi and renderer-Pi? I run a single Pi with USB SSD but get the impression that this is not what the majority of Pi users do. For example, the developer at Allo states that the USB input is noisy (not necessarily that it corrupts the signal in anyway but that it introduces noise to the unit) and therefore it is better to have the music files served via ethernet.

I did start with a Pi 3B+ connected to the CDA2 USB doing both server and renderer duties. I tried this with Ethernet and with 5 GHz WiFi networking to investigate some of the common hypotheses about network impacts on audio. I then did the pretty inexpensive test of splitting the server away from the renderer and having a separate 3B+ server and a lower power (less noisy?) Pi 3A+ as renderer networked by WiFi.

This arrangement meant I could power the 3A+ from the system's IR-switched mains block. It's OK to hard power-off PiCorePlayer's Squeezelite renderer but risky to do this with LMS. And that's how the system stuck when I decided I was hearing nothing in any arrangement that spoiled my enjoyment of the music, and it was time to stop testing. (Well I did later get a Pi 4 for the server to give me a 3B+ spare for other experiments and in case of failure.)

Others report hearing differences so maybe I have cloth ears. But if so they're my cloth ears.

 

[duckworp]

Oh dear , the trolls have descended , they just know it's not going to sound any better because they work in IT or are networking engineers.

Not one of them have experimented with trying different switches to hear if it makes a difference because that would be a total waste of thier time..... however they need to save a few deluded audiophiles from spunking £40 on foo or worse still.

Cue 20 pages of pisstake , vitriol and demands for users to do ABX testing.....

I don’t understand the motivation for some folk (usually the same ones) to repeatedly pop up in threads which are discussing subjects such as network optimising or mains cables, simply to dispel a post as nonsense. People who hear differences in SQ are not going to change their mind. Why do it? It seems to be particularly bad on Pink Fish, which is a shame as it does discourage people from posting. The American sites don’t seem to suffer in the same way. Some of these bits are bits folk, such as John Philips in this thread, write in a more considered way which is good to read. Others don’t unfortunately.

 

[Octavian]

I don’t understand the motivation for some folk (usually the same ones) to repeatedly pop up in threads which are discussing subjects such as network optimising or mains cables, simply to dispel a post as nonsense. People who hear differences in SQ are not going to change their mind. Why do it? It seems to be particularly bad on Pink Fish, which is a shame as it does discourage people from posting. The American sites don’t seem to suffer in the same way. Some of these bits are bits folk, such as John Philips in this thread, write in a more considered way which is good to read. Others don’t unfortunately.

It is very simple: you post extraordinary claims asking us to believe in your claims without any evidence. Not very surprisingly quite a few of us are not willing to believe you. That's the nature of public debate.

 

[Heckyman]

It is very simple: you post extraordinary claims asking us to believe in your claims without any evidence. Not very surprisingly quite a few of us are not willing to believe you. That's the nature of public debate.

Don’t wish to enter into a circular debate (again!), so I’ll just make one point. The idea of this thread was to post actual experiences in relation to a specific product (or at a pinch, similar products). No-one here is involved in selling said product or asking anyone to “believe” in anything.

The only people posting “claims” are those doing so in the negative, i.e. people like yourself who haven’t any actual experiences to report. Frankly, that’s off topic.

 

[watts]

I don’t understand the motivation for some folk (usually the same ones) to repeatedly pop up in threads which are discussing subjects such as network optimising or mains cables, simply to dispel a post as nonsense. People who hear differences in SQ are not going to change their mind. Why do it? It seems to be particularly bad on Pink Fish, which is a shame as it does discourage people from posting. The American sites don’t seem to suffer in the same way. Some of these bits are bits folk, such as John Philips in this thread, write in a more considered way which is good to read. Others don’t unfortunately.

Exactly. When I go searching for a thread about a product and its reviews, it would be nice if 2/3's of it wasn't clogged up with useless input from turd head cultists who are brainwashed by insufficient measurements.

Have we all seen the review of the so-called objectivists?

https://www.superbestaudiofriends.org/index.php?threads/audio-science-review-review.9827/

 

[kleinbje]

I have continually been amazed when power, vibrational(except spinning discs), digital, and NETWORK changes alter and sometimes improve my sound. I recently switched my Roon server from an Imac to a Nucleus+, did not hear the sound difference I was hoping for, though the upgrade was not initiated for sound but functionality, convenience, placement. Some network and digital changes that were revelatory were adding new Ethernet cables. Amazon had the Audioquests, so on a whim I grabbed enough cinnamon cables to connect modem/router/computer and one Vodka(later upgraded to a Shunyata Sigma) from network satellite to my Cary streamer. I just had to drive them 5 minutes down the road to return them, not even box them, too easy to resist. They didn't go back, wow, they had very meaningful improvement in my system(to the poster who commented about American forums being kinder to network discussion, not so much, see SHF). I was pretty much all analog, but the time on Tidal/Q has increased, I explored more now. I actually played some of the tracks the audio reviewers gawk about. I was really surprised given the asynchronous nature and re-clocking of ethernet, but hell if I wasn't the fool for waiting so long to have better music. An analog square-wave is subject to physics whether in a digital or analog circuit apparently. I then tried to improve the power supplies, I previously found used Bolder and Teddy supplies for my squeezbox touches, but could never be sure of their influence. Now that my system was revealing enough to hear a damn ethernet cable I purchased a HDPlex300 for the Nucleus+ and my router. The improvement was clear, spatially, cymbals sounded like cymbals, bass became very defined, and more importantly I listened more. Next I added an Ifi Elite to power the Network satellite connected to my streamer, thought I heard a difference for the better but unsure. I then switched the generic power chord for the Ifi for a Cardas Clear I had picked up second hand a many years ago for the bedroom system. Yes, now we were getting somewhere, best digital sound I had ever herd by far. It even eclipsed some(few) records, a complete shocker. So the question is do I need a switch between the satellite and the streamer. I would favor EE as I heard the Etherregen gets hot, and I already have some tubes in here. Throwing another box in the signal chain goes against what I believe in, but so did fancy digital cables, expensive network cables, and a overbuilt power supply for a damn router!!

Jeff

 

[Octavian]

Exactly. When I go searching for a thread about a product and its reviews, it would be nice if 2/3's of it wasn't clogged up with useless input from turd head cultists who are brainwashed by insufficient measurements.

Have we all seen the review of the so-called objectivists?

https://www.superbestaudiofriends.org/index.php?threads/audio-science-review-review.9827/

"Brainwashed", "turd head cultists"? Now, this is what I call a contribution. Truly the best a subjectivist has to offer.

 

[mansr]

Now that my system was revealing enough to hear a damn ethernet cable

I'd describe a system with that property as faulty.

 

[kleinbje]

I'd describe a system with that property as faulty.

You would be dead wrong, very well executed. Thanks for playing. Oh remind me, which ones did you double blind test??