Jitter, AIR and USB

[Antoine]

I sold everything and bought a Devialet to turn in a much simpler direction, and now... this.

LOL! The easiest way to be happy is to stay away from any forum or social media altogether. Don't read on and stay blissfully unaware.

It also saves the rest from having to read uninteresting and off topic posts so it's really a win-win!

[Jean-Marie]

This is a fascinating subject and that will lead to endless discussions but hopefully good reading

In full disclosure I am probably what some will describe as an objectivist so I trust in measurements, technical explanations etc...

When it comes to jitter, as long as we stay in the digital domain and do not exceed the capacity of buffers, jitter is a non issue and can be ignored up to the last re-clocking buffer before the DAC.

The only way I can imagine prior jitter to have an impact would have to be a very indirect one based on the fact that depending on how much this last buffer is filled up, the electronic gates working to manage it and that compose it will have variations in the patterns of the current they draw from the PSU, and if you did not design properly the powering of the DAC, you may have induced effects on the outcome. This being said, I high doubt it is the case because the variation patterns linked to the actual value of the words put in the buffers are as important as the ones mentioned before.

So we are left only the accuracy of the final clock of the DAC. Jitter in this clock will translate in tiny phase shifts and frequency shifts....

However I highly doubt they are what people are hearing because they are orders of magnitude lower that the phase shifts and frequency shifts induced by the doppler effect when your head is moving by just a few millimeters.

I know some people hate DBTs, and I can understand them because they are boring, painful and much less enjoyable than listening to real music, but they can be really enlightening.

One of the most challenging aspect is that our brain has been trained by millions of years of evolution to find patterns and changes. Spotting changes quickly was a matter of life or death, and therefore, this is deeply entrenched in our brain, to the point that if we are listening to the same thing, our brain will find differences and the only reliable way to determine that two things are the same or different is to rely on statistical analysis of answers in a DBT and only consider that there are differences when the answers deviate significantly from random guessing.

Jean-Marie

[BoyScout]

I sold everything and bought a Devialet to turn in a much simpler direction, and now... this.

LOL! The easiest way to be happy is to stay away from any forum or social media altogether. Don't read on and stay blissfully unaware.

It also saves the rest from having to read uninteresting and off topic posts so it's really a win-win!

... and saves me from people without no sence of humor.

[Confused]

The truth. Version 1.

I enjoyed reading Jean-Marie's earlier post, and I found myself agreeing with more or less every word. It also makes me think back to my recent experimentation with the Aqvox Ethernet switch. Irrespective of what I thought it might be doing, it was hard to escape the fact that when trying it both with AIR and the microRendu / Mutec combo. you could pull the 'output' Ethernet cable from the Aqvox and music continued to play for a second or so. It also claims to reduce 'noise', but in my set-up I can play one of those 'secret tracks' that you used to get at the end of CD's, with 5 minutes of silence before the 'secret track', I can play the silent bit, crank the Devialet uo to +10dB or something, and all I can hear is silence. What exactly is all this noise doing? Nothing it would appear. I have read much on here and other forums recently of people trying all sorts of noise reduction devices, upstream reclocking devices and declaring great 'discoveries'. If someone else writes that a 'veil has been lifted' (another one?) or their system 'is taken to another level' (again?) I think my head will explode! It is not inconceivable to me that that most, if not all, of what is being observed is in the range of confirmation bias, expectation bias, placebo effect and so on. Frankly, 99% of it is utter nonsense. It's good business for the people selling these devices I guess, but I think we would all benefit form being more logical, cynical and saving our cash for stuff that might actually work.

The truth. Version 2.

Recently I have been taking great interest in some of the threads over on 'Computer Audiophile'. There is a group of guys that have been experimenting with 'direct connection' of devices such as the microRendu and sMS-200, that is connection via Ethernet without router or Ethernet switches. Pretty much universally those that have tried this have claimed good results. It does not stop there though. More recently there has been some experimenting with using multiple clocks upstream of the DAC. What appears to be emerging is much evidence that using multiple clocking devices in serial, with the very best clock being the one before the DAC, is obtaining significant improvements in sound quality. An example of what these guys are trying is stuff like using an 'Ultra' modified sMS-200, together with the tX-USBultra, and using spare clock 'taps' to provide superior clocking to the upstream Ethernet switch. Fascinating stuff, and this genuinely appears to be yielding some solid results. This experimentation is ongoing and is yet to reach it's conclusion. For example, now it is established that adding extra clocks upstream works, then how far can this go? Presumably there will come a point where no further benefit is reached, or indeed thing get worse due to noise and losses. This is fascinating stuff and it will be interesting to see how it develops. Of course all this may appear too extreme for some, something for the fanatics and hobbyists only, but I am not so sure. Indeed, it would be easy enough to run the new say the sMS-200Ultra with a tXUSBultra, yielding end results that the more conventional hifi manufactures would charge you five time times the price to achieve. Of course it would be easy to dismiss all this 'Computer Audiophile' stuff using the tired old 'bits are bits' arguments and dismissing these 'new findings' as being delusions in the minds of fanatic hobbyists, with the reality being that they are just victims of confirmation bias, expectation bias, placebo effect. I believe this would be quite wrong, and indeed insulting to those concerned. Some of these guys are highly experienced in this game and very well respected. Some of them spend a considerable time on forums, so will be well aware of the psychological pitfalls of testing hifi kit, these guys are not 'newbies', they are very intelligent and highly experienced individuals. I for one am following their good work with interest. If you look at the new kit emerging from the likes of SOtM, UptoneAudio, Sonore, and so on, together with the very clever and experienced 'experimentalists' on Computer Audiophile, I cannot help thinking that we are entering a very exciting phase with 'digital front ends'. I think we could be on the verge of some real breakthroughs in low cost, high performance kit, and indeed some real breakthroughs in understanding why all this 'bits are bits' stuff is just so wrong.

[thumb5]

To my way of thinking, using the phrase "bits are bits" to characterise a point of view about this is misleading.  Surely no one is arguing that the bits are somehow changed depending on subtle things like single or even multiple re-clocking, are they?  If they were, that would be admitting that the engineering principles we rely on to run the world are somehow faulty and need extra help (especially for digitised audio) which seems to be nonsensical.  By virtue of the same engineering this would be easy to detect anyway - think CRCs used by Ethernet and USB, for example.

The more interesting part of the discussion to me, at least, is about how the reliably-delivered stream of bits gets converted to analog, and whether there is any way for that process to be upset by the "upstream" digital equipment, isn't it?  I think one can hold the view that the digital stuff "just works" - so the bits you put in are the bits you get out, no argument - but still agree that there are plenty of other factors that can cause noise, distortion, etc. in the conversion process, any of which could depend on what is connected upstream.

Another way of putting this is that although it's straightforward these days to deliver a stream of bits reliably from A to B at audio data rates, it's still non-trivial to design a low-distortion, high-resolution digital-to-analog converter.  It seems that the focus of attention for many is to keep the bad stuff out of the DAC.  That's got to be a good idea in principle, but how much effort one wants to put into it in practice is obviously a very personal decision: for some people that's the hobby in itself, for others it's a means to an end (listening to music), and for others it's just "utter nonsense" (to borrow your phrase).  Each to his own, as always!

[Confused]

So what does jitter actually sound like? Do any of us really know for sure? I stumbled across this recently and have only just had time to try. Per the link, have a quick read and then try the various samples provided. Absolutely fascinating I have to say! For me the later music clips were far more interesting and enlightening than the 1kHz sine waves at the beginning. Well worth a few minutes of your time I would say.

http://www.sereneaudio.com/blog/what-doe...sound-like

I also found this, not as interesting as the stuff per the link above, but it does feature a chap with a spectacular grey beard and an oscilloscope, which is very hard for me to resist to be honest.

https://www.youtube.com/watch?v=ZT_1UATci3c

[Axel]

So what does jitter actually sound like?  Do any of us really know for sure?  I stumbled across this recently and have only just had time to try.  Per the link, have a quick read and then try the various samples provided.  Absolutely fascinating I have to say!  For me the later music clips were far more interesting and enlightening than the 1kHz sine waves at the beginning.  Well worth a few minutes of your time I would say.

http://www.sereneaudio.com/blog/what-doe...sound-like

I also found this, not as interesting as the stuff per the link above, but it does feature a chap with a spectacular grey beard and an oscilloscope, which is very hard for me to resist to be honest.

https://www.youtube.com/watch?v=ZT_1UATci3c

I found that very revealing as well. I spotted some of the jitter, which on some tracks sounded like timing issues, and on others, sounded like distortion in the higher frequencies. I was only listening to an iPad Air through a set of Musical Fidelity IEMs, so not ideal. Perhaps it would have helped to listen on a more revealing source, but I still got the gist of what it was all about.
Loved the guy with the beard. Had a touch of the Denzil Dexters about him!?

[Hifi_swlon]

To my way of thinking, using the phrase "bits are bits" to characterise a point of view about this is misleading.  Surely no one is arguing that the bits are somehow changed depending on subtle things like single or even multiple re-clocking, are they?  If they were, that would be admitting that the engineering principles we rely on to run the world are somehow faulty and need extra help (especially for digitised audio) which seems to be nonsensical.  By virtue of the same engineering this would be easy to detect anyway - think CRCs used by Ethernet and USB, for example.

The more interesting part of the discussion to me, at least, is about how the reliably-delivered stream of bits gets converted to analog, and whether there is any way for that process to be upset by the "upstream" digital equipment, isn't it?  I think one can hold the view that the digital stuff "just works" - so the bits you put in are the bits you get out, no argument - but still agree that there are plenty of other factors that can cause noise, distortion, etc. in the conversion process, any of which could depend on what is connected upstream.

Another way of putting this is that although it's straightforward these days to deliver a stream of bits reliably from A to B at audio data rates, it's still non-trivial to design a low-distortion, high-resolution digital-to-analog converter.  It seems that the focus of attention for many is to keep the bad stuff out of the DAC.  That's got to be a good idea in principle, but how much effort one wants to put into it in practice is obviously a very personal decision: for some people that's the hobby in itself, for others it's a means to an end (listening to music), and for others it's just "utter nonsense" (to borrow your phrase).  Each to his own, as always!

I'm sure you'll all have seen this, but it's interesting that he echoes thumb5's thoughts - the bits are all there and always have been - it's the engineering behind the clocks, power supplies, etc etc that decide the final outcome on a device-level.

https://www.computeraudiophile.com/ca/ca...gies-r656/

Quoting a bit of the following discussion from Dominique Brulhart of Merging where he touches on this in abswer to a question:

2) Stricto-senso, the "quality" of the bits transfered by Ethernet, USB, Firewire, Thunderbolt or even AES or SPDIF is the exact same. Every bits reach the destination intact and in the proper sequence. If it weren't the case we wouldn't discuss about sound quality but about glitches, pops, scratches or bangs. So, on the transport aspect, all those technologies are equal (if we put aside latency, number of channels, resolution, which are quantity considerations and not quality considerations). However, the sending and receiving devices can be more or less accomodating to the technology involved by these transmissions, some are synchronous, some are asynchronous, some have more or less jitter, some are more of less precise, etc... and it is then down to the sending and receiving devices to be properly designed to accomodate with these constraints. In theory all those transmission mediums can sound equivalent if the hardware implementing them is properly designed. So in that sense Ethernet can have some advantages, like a better galvanic isolation, specially if using optical cables. Ravenna is a very precise, low jitter and low latency protocol, so meaning that it can have, in theory, advantages in the design of hardware communicating with it. But it's main advantage is definitely its flexibility. Quality is really down to the engineers not to the technology.

[maxijazz]

https://devialetchat.com/showthread.php?...5#pid59465

[Confused]

Something to try if you have any interest and a little spare time......

http://www.cranesong.com/jitter_1.html