[amirm]

Thanks for the explanation.

However, aren't we really seeing the noise floor of the USB to S/PDiF interface?

There can be an argument there but the data doesn't support that. What we see instead is that USB by itself is cleaner than S/PDIF on the same DAC. And it is not just noise. Look at one of the other measurements:

What we see in yellow are jitter spikes at regular frequencies when using S/PDIF input. The USB input (red) generates none of that. So clearly that is not in the USB signal or the PC source.

What we are really seeing here is that manufacturers are doing a much more excellent job on USB than S/PDIF. They know it is a noisy signal and they have worked to clean it up.

In the interest of fair reporting, I do have a USB DAC that is horrible in this regard (Schiit Modi 2). The answer there is not to polish the turd as the saying goes but to simply get a better DAC which need not cost more money at all.

The other point here that is important is that we can measure all of this. No need to speculate blind. Noise measurements is one of the most common things in electronics. So before we promote any theory of problems, let's do that. Let's get data and only then say X and Y and Z. Not before.

 

[robertbudding]

I miss the good o;e days when audiophiles painted the edges of their CDs green and claimed that it improved sound quality. Or the magic wooden discs that were placed about the room that were also alleged to improve sound. Yes, some things do really matter, and some not so much. I don't even trust my own ears because humans perceive slight differences in volume as quality difference. Legitimate comparisons need to be matched to 0.1 DB, and test should be blind.

None of this means that I don't enjoy really good gear. It just means that I want to spend my money where it matters most.

 

[amirm]

Oops, I forgot that the above graph was actually with the S/PDIF stream created by my Audio Precision analyzer. So it is as pristine as one can get and not sourced from USB.

 

[bigshot]

-100 dB is inaudible for playback of music in the home.

robert budding, it's ironic that audiophiles spend so much time fussing over what they can't hear and neglecting what they can't.

 

[gregorio]

[1] I'm not confident it's fair to extrapolate from there that USB is inherently quieter than coaxial.
The noise floor measurement notwithstanding ([2] either number is beyond perception, especially with a signal present), I have found USB to, at best, equal coax in a particular system. [3] I've never heard it outperform coax in any system where a comparison was possible.

1. Agreed. Going on these measurements alone we do not know if USB is inherently quieter than coaxial, all we know for sure is that the output of this particular DAC is cleaner when fed a USB signal than when fed an S/PDIF signal but these measurements do not indicate which of these two possibilities is the cause: A. That the USB signal is actually cleaner/quieter than the S/PDIF or B. That the USB signal is noisier/dirtier than the S/PDIF signal but that this DAC is capable of cleaning-up that noisier USB signal to a higher degree.

2. I wouldn't agree that these numbers are beyond perception, beyond audibility certainly but not necessarily beyond perception. There are no numbers/levels which we cannot perceive, we can perceive differences of ANY magnitude, even when the magnitude of difference is zero (there are no differences whatsoever)!

3. If "either number" is beyond audibility then obviously you've never heard USB outperform S/PDIF, just as obviously, you've never heard USB under-perform S/PDIF either! Of course though, you haven't been comparing systems with this particular DAC, you may have used a much poorer quality DAC but for the noise to reach audible levels it would have to be a particularly crappy (or malfunctioning) DAC.

Further those results are only on HIS system. There are far too many variables, different DACS, different USB transceiver cards, etc. to come to any worthwhile conclusions with those results.

We can of course make some conclusions from those results but whether these conclusions are "worthwhile" to you personally is a different question. We can make at least two conclusions: 1. With an average consumer laptop and generic USB cable this $129 DAC can reduce whatever artefacts (interference/noise) is present at the output of the generic USB cable to levels well below audibility. 2. An audiophile grade USB cable COULD NOT make any audible improvement to this setup.

These two conclusions raise some obvious questions: 1. If a $129 DAC can reduce USB noise/interference from a consumer laptop with a generic USB cable to well below audibility, would you not expect any other similarly priced or more expensive DAC to perform equally as well or better? 2. Even if we assume that an audiophile USB cable performs significantly better than a generic one (despite the fact there is little/no evidence to support this assumption), if you have a DAC which performs audibly worse with a generic USB cable, surely the best solution is not an expensive audiophile cable but just to buy a competently designed DAC (for $129)?

It's entirely possible that these conclusions and resulting questions are not "worthwhile" to you, in fact they might be quite the opposite of "worthwhile" as they contradict a belief/opinion which you obviously don't want contradicted. That doesn't mean they are not "worthwhile" to others or in general though!

You just need a DAC whose designer knows USB signal is not clean.

I would go a step further and suggest that all you need is a DAC designer who knows what USB is. The mistake made by many audiophiles (inspired by marketing BS) is the erroneous notion of some ideal/perfect USB signal which cannot be achieved in practise and therefore using an audiophile cable which improves this imperfect real world USB signal will result in better/quieter performance from the DAC. This notion is erroneous because the USB protocol does not assume an ideal/perfect signal, quite the contrary in fact, it specifically requires an imperfect signal! For example, the USB specs does NOT specify a square wave, it specifies a range of "eye patterns". A USB specified DAC should therefore operate optimally with any USB signal within this USB specified range. Even if an audiophile cable did "improve" this eye pattern it should not make ANY audible difference to the output of the DAC, as a generic USB cable is still transferring a signal within the USB specifications by virtue of the fact that it is a USB compliant cable. If there is an audible difference the DAC is not operating as a USB certified DAC, it is faulty. The USB specs also include a 5v power supply and isolation from EM/RF interference.

In other words, the USB specifications explicitly defines a non-perfect/noisy signal in the first place and this is therefore what a USB specified device MUST deal with. If it can't, if it needs some sort of additional USB signal cleaner (device, cable or whatever) then it is NOT a USB compliant device, it's a non-compliant device, a faulty, malfunctioning and/or falsely advertised device!

All the above is not generally the issue though. Few DACs are so incompetently designed/falsely advertised in the first place and few audiophile cables would "fix" those incompetent designs anyway. The majority (probably vast majority) of the time, the issue is audiophiles perceiving inaudible differences, IE. Issues/biases with their perception rather than anything related to sonic differences.

G

 

[SeeHear]

1. ...3. If "either number" is beyond audibility then obviously you've never heard USB outperform S/PDIF, just as obviously, you've never heard USB under-perform S/PDIF either! Of course though, you haven't been comparing systems with this particular DAC, you may have used a much poorer quality DAC but for the noise to reach audible levels it would have to be a particularly crappy (or malfunctioning) DAC...

G

You are right, I have never used a DAC of the quality of the SMSL Sanskrit (I'd never even heard of it until this thread). The DACs, and systems, I use are much more resolving than I would expect a 100 dollar DAC to be. The most relevant (to this conversation) comparison I can share would be using the Hegel HD 30 DAC. Using a Woo WA2 preamp, Sunfire amplification, ADAM Compact Mk3 speakers, and my own built computer with the previously mentioned Juli@ card, I found the USB inferior with the Hegel HD30, the Marantz SA-10, and my old Musical Fidelity M1 (no surprise given its age). With the Pioneer N-50, performance is too close to call. With the Mytek Brooklyn I'm currently listening to, I prefer coax from this computer, though the USB from my Macbook sounds pretty good, too.

My criteria for sounding good, or outperforming, has to do with the quality of what I can hear - the timbre, the soundstage and imaging, the micro and macro dynamics - in short, the musicality of what is reproduced. So, no, a noise floor 120 dB down is not appreciably different from one 130 dB down to me. In my experience, the measurements commonly available are useful as a baseline but are not sufficient to judge the musical satisfaction a particular piece of equipment can provide.

I've also compared interfaces using the Hegel HD12 fed by a bluesound node2 (which has USB and coax outputs). This system is even more resolving than my personal one. It comprises a Backert Labs Rhumba Extreme (on my next to get list), several DACs including the HD12, Luxman D-06u, Marantz SA-10 and the Technics SU-R1. Amplification is a pair of Carver Raven 350s in to either Technics R1 speakers (my favorite) or Carver Amazing Line Source (also very good, but a bit much for the room). With any of the DACs, I prefer the coax from the Bluesound, though the USB is pretty good.

My other "computer source" is an Aurender N100h. It only has USB out. I have to believe they put considerable expertise and effort into their USB implementation as it is (at least) the equal of coax from the Bluesound and the DLNA Ethernet interfaces. And, it is very revealing of the differences between various USB cables. I would not presume to explain why, but without mentioning the various (more expensive) others that sound worse, I will say the Oyaide with a jitterbug consistently yielded the preferred sound among those who bothered to sit through the comparison. Not that any of them were "bad", but at this level people have preferences that are important to them, and all preferred the Oyaide/Jitterbug combination.

Incidentally, do you know why what appears to be harmonics in the spdif trace is attributed to "jitter"? Or, is the scale mislabeled and not actually frequency?

 

[SeeHear]

There can be an argument there but the data doesn't support that. What we see instead is that USB by itself is cleaner than S/PDIF on the same DAC. And it is not just noise. Look at one of the other measurements:

What we see in yellow are jitter spikes at regular frequencies when using S/PDIF input. The USB input (red) generates none of that. So clearly that is not in the USB signal or the PC source.

What we are really seeing here is that manufacturers are doing a much more excellent job on USB than S/PDIF. They know it is a noisy signal and they have worked to clean it up.

In the interest of fair reporting, I do have a USB DAC that is horrible in this regard (Schiit Modi 2). The answer there is not to polish the turd as the saying goes but to simply get a better DAC which need not cost more money at all.

The other point here that is important is that we can measure all of this. No need to speculate blind. Noise measurements is one of the most common things in electronics. So before we promote any theory of problems, let's do that. Let's get data and only then say X and Y and Z. Not before.

I appreciate your patience, this is very informative, I have a question about the "jitter spikes." Why is that "jitter" and not harmonics? I thought this was a snapshot in time and those were the harmonics of the 12k fundamental? Wouldn't jitter manifest over time?

 

[Mediahound]

If a $129 DAC can reduce USB noise/interference from a consumer laptop with a generic USB cable to well below audibility, would you not expect any other similarly priced or more expensive DAC to perform equally as well or better

A $129 DAC will not resolve as well as something like a Chord DAC with a higher number of digital taps than traditional DACs, negating the need for a better USB cable- on a $129. DAC.

-100 dB is inaudible for playback of music in the home.

Speaking of noise floor audibility, check out Rob Watts (Chord's DAC designer) presentation here where he states humans can actually perceive a much lower noise floor than traditionally thought as well as noise floor modulation. I think he states humans can even perceive down to -200dB level. It's an interesting talk overall.

 

[robertbudding]

-100 dB is inaudible for playback of music in the home.

robert budding, it's ironic that audiophiles spend so much time fussing over what they can't hear and neglecting what they can't.

I love excellent sound, but I will have two sons in college next year, so I need results for my $$$. Next up: Focal Clear headphones, but with the 'stock' cable! That way I can tell my wife that I economized.

 

[robertbudding]

LOL! Do people believe that ones and zeros become different based on the cable? Either the cable passes the data stream unaltered, or it doesn’t. And if it doesn’t, then it can be measured.

I expect claims that seem absurd to be backed by solid experiments. Soluble blind tests with volume matched to 0.1 DB is needed. Until then, I have to believe that people are paying up to enjoy the placebo effect.

 

[Mediahound]

LOL! Do people believe that ones and zeros become different based on the cable? Either the cable passes the data stream unaltered, or it doesn’t. And if it doesn’t, then it can be measured.

That's a common misconception. The signal that moves along a USB cable isn’t digital – NOT little ones and zeroes – but an electrical-pulse representation of those ones and zeroes. This therefore analogue signal is prone to disturbance from EMI emanating from the host computer and electrical noise arriving over the air, otherwise known as RFI. Greater vulnerability to noise can degrade a cable’s ability to do its job.

Everyone here understands this and this is not what we're debating. What's being debated is to what level that degradation in the USB cable, (or lack of degradation) can be perceived.

 

[castleofargh]

A $129 DAC will not resolve as well as something like a Chord DAC with a higher number of digital taps than traditional DACs, which would negate the need for a better USB cable on a $129.

Speaking of noise floor audibility, check out Rob Watts (Chord's DAC designer) presentation here where he states humans can actually perceive a much lower noise floor than traditionally thought as well as noise floor modulation. It's an interesting talk overall.

while very few people doubt his technical expertise as a designer, his approach to listening tests is often... well, his own. his attempts at correlating how he feels with objective data can often feel like a bad joke. if you trust what he calls listening tests and his guesses about the causes, he can notice stuff at silly levels like -150dB or even below, which of course isn't the case. you could put any signal you like at -150dB and wait for people to perceive it in a proper blind test. the wait might be real long though.and he himself admits that it's weird, but instead of looking at why he might reach the wrong conclusions, he goes on with his gut feelings on correlations like they're discoveries.

LOL! Do people believe that ones and zeros become different based on the cable? Either the cable passes the data stream unaltered, or it doesn’t. And if it doesn’t, then it can be measured.

I expect claims that seem absurd to be backed by solid experiments. Soluble blind tests with volume matched to 0.1 DB is needed. Until then, I have to believe that people are paying up to enjoy the placebo effect.

things are more complicated than that. a 1 still usually ends up being a 1, but delays, noises and various possible impacts of that noise on the chips or circuit, do end up having some impact. which should under normal use be inaudible.
but in the end I guess the arguments come down to people who think their ears in sighted tests are more reliable than measurements(they're wrong). and also a dispute over how to treat an anecdotal result. if a certain cable on a certain gear ends up having a massive impact, to some that's evidence of the importance of good USB cables. and to guys like me, it's evidence that something seriously wrong is going on. and that some gear(including sometime the original USB cable), or the environment are bad. bad as in, needs to be changed, bad.

it's the good old "one time is enough to prove I was right", against "dude, do you even statistics?". all cable topics are like that. horrible testing methods, too much self confidence and ego, then that.

 

[dmance]

@castle
Not quite. The bits are perfect...all the way from source into the usb receiver chip and to the innards of the D/A.
Think of an invisible fog of RF traveling along the cable ...so hard to get rid of. It conducts across metal and even air gaps.
All the concern is the /A part of the D/A. It's there that the circuit modulates the reference line voltage to the output signal. Any perturbation in this is audible.

 

[Strangelove424]

That's a common misconception. The signal that moves along a USB cable isn’t digital – NOT little ones and zeroes – but an electrical-pulse representation of those ones and zeroes. This therefore analogue signal is prone to disturbance from EMI emanating from the host computer and electrical noise arriving over the air, otherwise known as RFI. Greater vulnerability to noise can degrade a cable’s ability to do its job.

Everyone here understands this and this is not what we're debating. What's being debated is to what level that degradation in the USB cable, (or lack of degradation) can be perceived.

Nobody can perceive binary data in a USB cable. What you perceive is the analog wave converted out of the binary data. The accuracy of that conversion is not on a slope or a spectrum, it is binary too. Black and white. The data is either whole or not whole, on or off, working or not working.

To use an analogy to drive my point home, changing the lighbulb in a flashlight won't change the meaning of the Morse code message I create with it. Sure, one light might appear stronger than the other - but in regards to to the message being delivered, either you got it or you didn't. You may think of the binary data the same way. They may take the form of electronic waves, but they are no more of an analog signal than flashlights delivering messages of Morse code. The medium is analog, but the data is binary. The accuracy of transmission is a function of the binary data, not the analog delivery method. That's one of the greatest advancements digital brought us. The fact that our data, and the physical medium that holds it, are not inextricably linked. That is what allows you to back up a computer reliably, copy a file perfectly from one drive to another, or send a song to a DAC. Same reason there is no generational loss in digital.

 

[GChief]

Just an observation from a technician that has worked on digital electronics for over 30yrs. Some people in this thread really have no clue what a digital signal is.