We need a new standard in transferring digital signals between audio equipment.

[John Dyson]

9 hours ago, R1200CL said:

......maybe it’s even possible to mix Ethernet and USB protocols in same cable 😀 ?

(Maybe not my smartest idea)

I have been thinking the same idea.   Note that on ethernet transported netowrks:  ethernet packets encapsulate IP packets (usually) which encapsulate TCP/UDP and friends.  Very often, on local networks, there are packet forms that talk direct ethernet or use a generally non-routed IP packet.

 

As long as the ethernet BW isn't totally eaten up, then the scheme should work.  However, if the bandwidth is eaten up, then the latencies will start being very bad.  OTOH, just using 1Gb ethernet for audio alone, that should be perfectly okay if used on a consumer level app, perhaps even sharing the network at lower rates.

 

Talking 'ethernet' is trivial, but handling collisions/etc and using proocols/routable protcols on ethernet that starts being real work.

There might not be enough bandwidth on 1Gb for everything, but calculating an extreme example:

768000samples/second * 32bits/sample * 8 channels, that would give approx 200Mbits/second.

 

So, if you have two bidirectional communications, there just might be enough bandwidth for that high end application.

The calculation was based to minimize collisions -- going above 1/2 capacity starts accumulating collsion problems,

even though, with ethernet type protocols, the SW must be written to handle collisions.

 

A reliable UDP style thing might be best.  TCP and friends do have some ease-of-programming matters, but there are apparently

reasons to prefer UDP because there is more intimate control of retries..   Using a standard protocol might allow easier networking

compatibility, but of course realtime can be

a problem on a shared net unless the other devices on the network are cooperating.   Less than my extreme example would

probably be much more cooperative (e.g. 96k * 32bits/sample * 2->5 channels) -- that is MUCH MUCH easier to slip in to

an active network.

 

Of course, this is off-the-cuff, and I might have missed somthing.

Using dedicated ethernet protocols, telling people not to use the cabling/fiber for other applications -- that would probably

work pretty nice.   All of the hard, HW work is already done, and there is a lot of experience doing the higher level protocols.

Why not?  Proprietary control, maybe?

 

John

 

[jabbr]

22 hours ago, R1200CL said:

......maybe it’s even possible to mix Ethernet and USB protocols in same cable 😀 ?

(Maybe not my smartest idea)

That’s essentially Thunderbolt

[John Dyson]

5 hours ago, jabbr said:

That’s essentially Thunderbolt

Does Thunderbolt use a USB like transport?  I don't know or not...  Ethernet is so compatible already...  Of course, with the fiber version of Thunderbolt, it would be more practical to wire your house.   Ethernet and ethernet/fiber can already do that.

 

Frankly, back before USB was introduced, the Intel people came over and were giving me advanced information on it -- I didn't like it at all.  However, as USB became more useful and advanced, it started to make sense.   I am a terrible predictor of the future.

 

Except, what I was doing conceptual work at AT&T Bell Labs ended up being our current cell network (video, transports et&al.)   I didn't think that it would be practical, but I was thinking too much in terms of technology in 1983.   Should have patented it for the future -- I would have gotten my patent bonus, and whomever owns AT&T patents today would be more rich.

 

 

 

[jabbr]

1 hour ago, John Dyson said:

Does Thunderbolt use a USB like transport?  I don't know or not...  Ethernet is so compatible already...  Of course, with the fiber version of Thunderbolt, it would be more practical to wire your house.   Ethernet and ethernet/fiber can already do that.

 

Frankly, back before USB was introduced, the Intel people came over and were giving me advanced information on it -- I didn't like it at all.  However, as USB became more useful and advanced, it started to make sense.   I am a terrible predictor of the future.

 

Except, what I was doing conceptual work at AT&T Bell Labs ended up being our current cell network (video, transports et&al.)   I didn't think that it would be practical, but I was thinking too much in terms of technology in 1983.   Should have patented it for the future -- I would have gotten my patent bonus, and whomever owns AT&T patents today would be more rich.

 

 

 

https://en.m.wikipedia.org/wiki/USB4
 

Also NVMEof so the concept of “converged fabric” ... the new NICs at least can run whatever protocol is desired. The newest NICs have a processor on the NIC itself so this design pattern of something like a Zynq as a universal protocol interface is becoming ubiquitous. Unsure how this helps audio because most DACs use copper USB-2 input so really if you want fiberoptic SFP, you are limited to either Ethernet or a custom hardware which requires you to implement both ends of the link. Even Ethernet needs some embedded software eg NAA or DLNA 

[R1200CL]

4 hours ago, jabbr said:

Unsure how this helps audio because most DACs use copper USB-2 input so really if you want fiberoptic SFP, you are limited to either Ethernet or a custom hardware which requires you to implement both ends of the link.

Well, someone has to start the show 😃

 

And I would assume, like me that has an old DAC (that still sells for over 4K used), and I’m dependent on USB converter, that a new converter will be a good start. 
 

This is also why I asked if one SPF(+) interface can cover all standards if a company would like to develop something. There could even like you paid for what interface you wanted out to your DAC. 
(USB, i2s, AES/EBU, SPDIF)

 

One could even think of going back to the (old?) model of a PC direct to a DAC, and just forget the whole thing about endpoint and streaming boxes.

I don’t know if this requires a proprietary SW to implement, or if it could be done so nice that most companies could implement something like this if it ever will happen.

 

I’m also thinking clock interfaces should be fiber. Should make things much easier.

Well, if external clocks is needed in the future. Maybe if needed, the clock signal is just part of fiber transfer and a device that requires a better clock, just extract that part from the bus. 
 

Not sure I understand all possibilities, of what could be done or not, and what make some sense. 
Could probably extend to multi channel as well and of cause 8K video.

Why haven’t Sony or similar developed HMDI over fiber 😃

Maybe they rather go directly to so sort of local 5G and everything wireless.

 

Brings me back to if some sort of ioT chip would do it for audio transfer as well. 
 

[lmitche]

 

On 3/15/2021 at 5:08 AM, PeterSt said:

But of course ! This is already because there are many after market products for this. I mentioned Adnaco elsewhere recently (they were the first with this IIRC, more than 10 years ago). It will give you an idea.

But it won't sound better because of it. There wil be isolation, but what this does is "relative". All together it is more harmful than useful.

 

Etc. etc. etc.

(can of worms stuff)

Hi Peter,

 

My Adnaco sat on the shelf for two years until I figured out how to get the best out of it a year ago.

 

Once done, this is the best sounding "can of worms" you have ever heard.

 

But to each his own.

 

Larry

[PeterSt]

15 minutes ago, lmitche said:

My Adnaco sat on the shelf for two years until I figured out how to get the best out of it a year ago.

 

Haha, Larry, so now you are keeping that a secret ?

(or I missed (or forgot) it somewhere - apologies in that case)

[lmitche]

10 hours ago, PeterSt said:

 

Haha, Larry, so now you are keeping that a secret ?

(or I missed (or forgot) it somewhere - apologies in that case)

Pareto Audio Listening Experience - Music Servers - Audiophile Style

[kdubious]

Sincere Question: what problem are you trying to solve?

 

I just read the thread to catch up, and I can’t figure out the ultimate goal. You end up needing to clock I2S or DSD into a DAC chip, or a ladder, right?