I don’t quite see how it would help with jitter, though.
1 Like
According to those that claim to understand the fiber standards, jitter isn’t allowed.
Especially 10GB and use SFP+ ports.
There is no jitter in USB.
It’s impossible, by definition.
Wikipedia defines “jitter is the deviation from true periodicity of a presumably periodic signal”.
Since audio over USB (and Ethernet) is asynchronous, it is not presumed periodic and therefore cannot fail to be periodic.
If you don’t try, you can’t fail. Nobody can say my tennis serve is sloppy, because I don’t play tennis.
I know people say otherwise when the sell devices that claim to fix jitter, but they are wrong.
A more detailed description:
In both USB and Ethernet, the bits are transferred in a packet.
A packet is like a piece of paper: the sending device writes a bunch of numbers on the piece of paper and sends it. At the receiving device, the low level software receives the piece of paper, then taps the higher level software on the shoulder and says, I’ve got something for you, give me a buffer (a piece of memory) big enough the hold this many numbers. It copies the numbers from its piece of paper to the buffer, and the higher level software reads the numbers in the buffer and hands them to the digital-to-analog circuitry to be converted. The conversion circuitry is sensitive to timing, so in that reading-and-handover they have to be aligned to a precise clock, which is part of that equipment. But there was no timing in the transfer of the numbers from the piece of paper to the buffer. That’s what asynchronous means, the information is transferred at random time, and the recipient is responsible for creating precise timing.
Somebody may say that the USB signal itself has timing. Sure it does, almost everything in a digital environment has timing. But that timing has nothing to do with the audio timing, that’s just how the transfer of the piece of paper works. That USB timing is forgotten long before the numbers are copied from the piece of paper to the buffer.
6 Likes
These arguments really crack me up. Reminds of “street smart” vs “book smart” argument.
Those who do not hear a difference great for them. Same argument as speaker cables making a difference. Is it worth the money and do you like it is the biggest question.
I can speak from my experience that everything makes a difference. I would point to the electrical noise as being the biggest driver. If you have a very transparent system and you still can’t hear a difference then you should be excited to save money.
I recently changed my Ethernet switch from a low end Linksys to a higher end Cisco and it made a difference with the noise floor. The Cisco has an external power supply.
Off the shelf computers generally don’t have noise isolation. So USB out will carry noise through the cable. Innuos is not off the shelf, I think it has a custom motherboard and linear power supply. I would imagine their components may work better as a direct connection.
This whole thing is dependent on how transparent your system is. For a sub-$5K system then the difference will be probably less noticeable than a $30K+ system.
3 Likes
This is another carefully tended audio myth. Just because a marketer gets someone to pay too much for equipment does not make that equipment better. But many wealthy but technically ignorant people have been convinced of that myth, and it still circulates.
9 Likes
My ISP recently changed their access equipment… text is crisper, images bolder, and the odd $ cents on Amazon and ebay offers are now rounded down. Great update. /sarcasm
4 Likes
Very true, I have heard $100K systems and thought it sounded worse than a $1K system. I meant more on the transparency side, which doesn’t necessarily make it better. Though I am sure a sub-$5K system can have some level of transparency as well.
For those who love their systems that’s great, not saying spend more to get better. What I am saying is just because someone hasn’t heard a difference doesn’t mean it doesn’t exist. There are no absolutes in audio. It’s a personal preference.
3 Likes
If we are to believe the measurements of Amir and Archimago, you can get a pretty darn excellent level of transparency for less than $5K. The devil is in the speakers and how they interact with the listening environment. But you can get a very good pair of headphones these days for relatively little money. Overall, I’d think a really good headphone set-up shouldn’t cost more than $2K.
3 Likes
Definitely agree speakers and room are big drivers in transparency.
I wasn’t considering headphones. Sometime I wish I just went with a higher end headphone setup. It’s more paractical when you have kids and far cheaper.
2 Likes
Nope an off the shelves Supermicro in the pics I’ve seen of opened Innuos units.
No kidding, that’s pretty expensive for a standard motherboard with a custom linear power supply and software.
It seems to be the way of things. For example, Roon’s Nucleus uses a standard Intel NUC mobo.
I"ve often joked to friends and family I should open a side business building custom PCs for audiophiles.
4 Likes
Anders
This time you’re totally wrong. There is or potentially can exist jitter in Ethernet transfer. It’s actually an requirement in high speed internet specifications. (To have jitter below a threshold).
I’m also sure you are aware of John Swenson’s etherRegen and his white paper.
You can google “White Rabbit Switch” and you will find documents and procedures how to remove and deal with jitter. And phase noise.
There is of cause many types of jitter, so hopefully you will get back on track 
Below is a sample from a Finisar SFP module.
This is so true, hence why I suggested fiber interface on Nucleus.
Here is a cheap 10GB fiber Ethernet switch “ MikroTik CRS305-1G-4S+IN”
1 Like
Well, for what it’s worth, I’ve found a definite improvement in moving from ethernet to USB from my Rock. The Rock heats up a bit more though.
I also found that I could bypass my dac (an old ps audio perfect wave 2 with a bridge 2) and connect directly to the usb input on my McIntosh 6700 - there was no material difference, to my ears anyway. I had connected the dac via xlr to the amplifier using decent cables, incidentally.
I wonder if there is simply an old fashioned advantage in minimising the number of components in the chain from the source through to the speakers. An internal connection must be better than a cabled one; things should be better matched (so a dac built into an amplifier should match the amplifier); you might have one noisy power supply instead of two or three (eg with a router); the longer the chain, the greater potential for timing issues and so on.
In any event, I think its worth trying it out and seeing for yourself. I certainly found things which were contrary to accepted wisdom when I did it.
Cheers
Gareth
But we don’t claim superior sound quality. Fanless yes, but no audio voodoo.
Nucleus is the turnkey fanless Roon OS. It’s the best experience to run Roon.
3 Likes
But you aren’t sending audio over ethernet. You are sending an error corrected reliable protocol over it. That protocol then has audio encapsulated in it. This is the point of digital streams. They aren’t subject to the errors of the analog domain.
The beauty of digital streams is that you truly can and should ignore what they sit upon.
Once again:
2 Likes
Yes they are.
Or maybe we speak about two different issues.
I’m not saying there will be bit errors.
People do get better sound from fiber, 10GB Ethernet, or just change to a high quality switch.
Edit:
I read your last chapter in the link now. Yes, we’re not discussing the same issue.
The specifications for 10GB Ethernet require a stressed eye pattern test to pass.
This is a verification of almost zero jitter.
I think the idea of moving to high quality Ethernet and fiber would benefit audiophiles.
3 Likes
Were measurements ever provided to support their white paper?
That is down to your thought processes deciding what drugs are, alcohol is a drug.