You asked for hardware recommendations to give you least disturbance of the audio stream. I have given you those from Roon. There are many different views about what has most impact or indeed if any of this has any impact at all.
Personally several years ago I did find that when using USB output from the server various tweaks to optimize the server improved sound quality. Many seemed to think that the biggest improvement came from separating the server and player activities. This is also Roon’s recommendation and with Roon being a “heavier” server I followed that recommendation and sound quality was at least as good as the tweaked server though rigorous A B testing was not possible. With that I have left all the tweaking behind and focused on enjoying the music.
One other point: these evaluations are not just technical, they have a financial component too.
Amir at Audio Science Review made an interesting evaluation of another cleansing device, a SPDIF reclocker and jitter reducer.
He found that it worked as advertised. But when used with a high-end DAC, it didn’t I prove things because the DAC was already immune to jitter, and in fact had dramatically lower noise than what the reclocker introduced — the DAC did better without help.
But when used with a low-end DAC the reclocker did in fact help.
But the point is this: the reclocker cost $1,500, the high-end DAC cost $1,200, the low-end DAC cost $150. Why would you spend $1,500 to help a $150 DAC instead of buying the better DAC for $1,200?
Same thing with these network endpoints: the Rendu endpoints with advanced power supplies cost as much as a Nucleus.
Great story Anders. And you mention you have a Hugo2 and @Jan_R_Smit has a Qutest. So both are fans of Rob Watts’ work (me too).
Rob found that a fifty dollar (US$50) Audioquest Jitterbug made a difference with his ten thousand dollar (US$10000) state of the art Chord Dave…
Rob on Head-Fi forum:
"Now Dave is galvanically isolated from LF and RF noise, and this makes a huge difference in SQ. But the galvanic isolation is not perfect as there exists a 2pF coupling capacitance. Now you may think 2pF is nothing to worry about; but at 1 GHz it is nearly 80 ohms impedance, so RF noise in the GHz will couple through - and I know that GHz noise is significant in causing noise floor modulation and hence changing the sound quality.
I used my design lap-top, an MSI Intel i7 machine, as it will be noisier than my music lap-top (HP Pavilion) which is very power efficient. This is so it would be easier to hear any changes.
So adding the Audioquest Jitter Bug did indeed improve SQ - it was fairly easy to hear it, with it sounding smoother, warmer, with less sibilance and glare. Instrument separation and focus was better. These are exactly the kind of change I hear with lower RF noise, so it was not unexpected."
3 Likes
James_I
(The truth is out there but not necessarily here)
347
Agreed.
I was really shocked when I first read that first post over at AS/CA that limits posts to listening experiences without the ability to discuss cause and effect scientifically.
I wonder if your comment would have been banned over at AS/CA? I guess if we want to determine if there are any potentially valuable tweaks there, we have to lurk and not discuss…
This is interesting if I understand it correctly. Let me see if I have this correct; the designer of a Hi-end DAC knows that there is a noise pathway inside his unit but did not correct it? And the noise pathway is significant enough that it is audible at the device output? And that the mitigation of the noise path is as straightforward as additional filtering on the digital input? Is that the take away from the post?
You ask questions way too difficult for my small brain
But I think his words need to be read very carefully… sometimes a few times over.
For example: But the galvanic isolation is not perfect as there exists a 2pF coupling capacitance. Now you may think 2pF is nothing to worry about; but at 1 GHz it is nearly 80 ohms impedance, so RF noise in the GHz will couple through - and I know that GHz noise is significant in causing noise floor modulation and hence changing the sound quality.
From personal experience, having spoken with the designers of some of the best DACs on the planet, I can say Rob is easily the most transparent…
i.e. he will tell you things that others also know but won’t tell you or are hesitant to tell you because they want you to believe their product is perfect…
Don’t forget Dave is quite transparent too. And measures pretty well:
I think the question was that if Rob Watts himself could hear and measure the difference with something as trivial as the Jitterbug and find the logic for the noise/impedance at 1 GHz, then why wasn’t it simply addressed?
It’s a fair and valid question. I have two Jitterbugs as well and have heard similar things with both my DACs on the USB input. Some DAC manufacturers will say there is no benefit or it won’t make any difference, but you hear things differently. At least with Rob, he admits it and gives a reason, which is better than most manufacturers playing it safe and quiet, but at the same time…why not fix the problem? Unless it has been in later DACs?
Ideally, I should be able to plug in a Jitterbug into the USB input of my DACs and it should not improve or degrade the sound. It should make no differences at all.
Because source matters! Whatever the source introduces, cannot be magically fixed by the DAC. It’s not DAC’s job.
If a top SOTA DAC like this is not isolated at input from bad source, no DAC can:
Quoting:
…few comparisons of the MacBook Pro with the dCS Network Bridge revealed that nothing sent from the MacBook Pro via USB sounded anywhere near as full and rich in colors as signals that bypassed the computer entirely and relied solely on the Network Bridge’s coax outputs.
Realizing that, regardless of the claimed jitter-free superiority of EMM Labs’ USB input, my USB source was compromised, I wrote John Atkinson to ask how I might get a better-sounding music server to replace the MacBook Pro.
I’ve tried to hint the answer may be there when I mentioned Rob’s words may need to be read carefully…
“But the galvanic isolation is not perfect as there exists a 2pF coupling capacitance.”
When you ask “why wasn’t it simply addressed?” can you share which USB DACs have perfectly isolated USB inputs?
By the way this coupling capacitance doesn’t really show up in Stereophile’s state of the art Chord Dave measurements I linked above, so linking to other USB DAC’s measurements may not help either.
Whether there is an audible consequence of this coupling capacitance is something else completely but we’re talking technical stuff here (and potential technical mechanisms) since I’m quoting the DAC designer himself, not an armchair expert…
What Anders can and can’t hear with his Hugo2 (or what I can and can’t hear) is probably something for a different thread perhaps.
The USB isolation design of Dave is identical to Qutest. Confirmed by Rob Watts on Head-Fi Forum a couple months ago:
“All my galvanically isolated designs are the same”
Trying to link it back to the topic of the thread… here is what I posted a few days ago and it (maybe) relates to this coupling capacitance Rob speaks of when talking about different USB sources potentially sounding different, even with very good (not perfect…) USB isolation…
Again, it’s very worthwhile reading Rob’s quotes very slowly and carefully. If you don’t believe him, everyone should reach out to the designer of their own DAC and ask if they agree or not with the technical mechanisms he speaks of. And ask them what is easy to solve and what is not so easy to solve and ask for the technical reasons… Or just trust @AndersVinberg and his ears!
All said in good fun of course. I’m interested in learning about these potential technical mechanisms. But I appreciate that not everyone is. I have absolutely no issues with people saying they just trust their ears. While the potential technical mechanisms really interest me, like most people, in the end, especially with buying decisions, I ultimately trust what my ears like, like most of us here.
I don’t think you guys understand my position and intention. I’m not saying that sources don’t matter. It’s the opposite. That’s why I have a Jitterbug for when I need my iMac as a source for watching stuff or casual music. I have a BDP-1 for an optimized player and is preferable to the MacBook Pro and iMac’s USB. Hell, I used to have a separate uncompressed WAV library at one time for MPD use with the BDP-1!
My question is for only the top DAC manufacturers that DO understand and recognize these things and hear what we hear. My question is what’s stopping them for perfecting it, as in to make no difference at all? Is it money? Yes, they sound great and measure well but still have that 2pF, which does somehow still end up making a difference audibly. What can they do to totally eliminate it? Has Rob or any other DAC designer commented on that?
I don’t have the latest DACs or the newest streamers/players, but I still hear the same thing in the past. While both the newest DACs and streamers have gotten better, they still remain sensitive as to the source (hardware and inputs/outputs) and what is happening inside the source player (RAAT/DLNA/MPD). Both the optimized streamer/player AND the DAC should be rejecting these changes from going downstream, yet it’s still not perfect. Do they have any plans in mind to completely fix that last bit?
I get what you’re asking but in my reply I mentioned:
“everyone should reach out to the designer of their own DAC and ask if they agree or not with the technical mechanisms he speaks of. And ask them what is easy to solve and what is not so easy to solve and ask for the technical reasons…”
Side story (just one example)…
Ted Smith’s next DAC for PS Audio is the Ted Smith Signature DAC and is a 2-box solution.
A ‘digital’ box with all inputs and powerful FPGA for all DSP stuff. And a separate box for only the D-to-A conversion.
The 2 boxes will be connected via fibre optic cable (i.e. no electrical connection), which can be 100m in length if you want…
And funnily enough, Rob says TOSlink is the best sounding input with this DACs - TOSlink (optical) is free of incoming RF and incoming jitter is not a concern (for him). There are many Chord DAC owners who agree with him but also many who disagree with him and prefer very complicated USB chains… This SQ preference maybe relates to the same quote of Rob’s that I shared in this post:
But I’ll share the quote again here for those too lazy to click the link:
"As you know, RF noise creates noise floor modulation, as the intermodulation distortion from random RF noise is a white noise modulated by the wanted signal. This then results in noise floor modulation, and is very very audible. It accounts for the things sounding brighter and less smooth; additionally, when you reduce RF noise, things sound considerably warmer and darker, and one consequence of this is perception of tempo - more midrange gives the impression of a slower tempo, as individual instruments have much more body.
Now if somebody prefers the brighter sound from more noise floor modulation, then fine - that’s their taste and preference. But it’s not accurate."
This observation of improvement, it should be easily measurable. Since it is related to a measured fact of 2pf and 80 ohms at 1ghz. And noisy Floor modulaties. So where are the measurements to support this observation? Does audioquest themselve come with any proof? Juist wondering
The 2pF coupling capacitance refers to the USB digital isolator used in Rob’s DACs.
That’s the designer himself telling you that about his own DACs so I’m not sure what measurements you would need from Audioquest (or measurements from anyone else)?
The only surprising thing (to me) is you have a designer that is this transparent… I find it quite refreshing to be honest.
What Rob Watts experienced is what most if not all high end DAC manufacturers have found. That there were devices that improved USB, and that it is relatively trivial to introduce similar technology into their DACs today. Look at the evolution of USB on Schitt DACs for an example, obviously helped by having a separate USB module. All it demonstrates to me is the best people are always open to learning!
The issues of source interference cannot be solved completely, they can somewhat be improved upon, but not by "memory playback’ on a PC or Raspberrypi…and not by simply tuning buffers in squeezelite!
You are a brave man Stephane. Aren’t you leaving Roon soon? Going out with a bang !
All I’ve done in my posts is to ask the question, how does anyone know that this ‘LMS+squeezeplayer with memory playback’ request is actually not technically worse than Roon as it works right now? i.e. using Rob Watts explanation, is it possible it increases RF, giving the perception of better ‘detail’ when it is actually just increased IM distortion caused by increased RF?
And as I asked earlier, who’s job is it to investigate this? Are Roon supposed to do RF/noise measurements? Or is it the people making the claim that should be doing some measurements and bringing the results to Roon?
I’m all for any change that improves SQ but I’m not really for something that is technically worse, even if 100 people say it sounds better because it sounds more detailed than Roon as it is now… I’m not saying this request is the latter, only asking the question.
I wasn’t always thinking like this. I may have done a 180 (or maybe 270) the last 3 years That’s the problem when you talk to a lot of real experts (numerous designers of state of the art DACs for example) on the subjects and learn from them.
