On a recent Podcast an engineer making streamers (read biased) went on about the timing of those packets being the issue for an audio stream compared to a normal data stream where timing is irrelevant and it’s simply a matter of getting the packets through unharmed - and that it wasn’t as simple as just using a buffer to ensure proper timing. That fx the path from the stream having been received to the digital signal being converted to analogue can have multiple points of issue in this regard, praising some streamers/dacs for having multiple high Quality clocks, isolated from the noise of the rest of the circuit, particularly the processing power used to run a streamer.
That and the mention of differences in jitter.
These might have been mentioned elsewhere - but for clarity’s sake, is there any merit to either? (Didn’t trust the guy as far as I could throw him, but never know).
This month's Stereophile has a product to help soothe some of your fears...
Without going to the library to look it up (I'll be in there later, definitely first thing in the morning, so I can update this), the premium for peace of mind is around $6000 for an ethernet box and its optional external power supply, both of which look like they will fit in one hand while you clear space on the rack to install them.
Or, you can use your personal experience to help form and opinion of need.
Is there really a problem to solve, or only potential problems?
---
Disclaimer: As a Technician with an Honorary Engineer title on my business card (some customers were stuffy and required us to send Engineers to their job sites) I worked with many things digital, and the intra and inter continental transmission and necessary packet switching via point to point and SONET rings for NEC, WorldCom, and NorTel, for customers like Sprint, Baby Bells, C2C and Genuity, before two of the three of those vendors and one of the customers blew up, so, what would I know about audio? Nothing? Never heard a complaint like "My Napster is all jittery!"
Rarely would there be any real mystery about why something didn't work as designed. Bent pin, broken wire or fiber, dirty fiber, bad firmware/software...
"and that it wasn’t as simple as just using a buffer to ensure proper timing" - coulda fooled me. Lots of buffers in the beast below.
The packets of data come ZAP------------ZAP------------ZAP
It is then buffered ahead of the DAC, and the data stream clocked into the converter at the appropriate rate.
Unless it's a USB cable to the DAC, then it's ZAP------------ZAP------------ZAP again.
---
The Traffic Engineers did have to consider the transmission line and intermediate switches (say, from New York though Atlanta through Dallas Though Phoenix through LA to SanFrancisco) as part of the receive "buffer", because at 600+ mbit/s there could be a bit or so (per channel) in transit in every two feet of the fiber, not including the intermediate switch buffering, which could come to 9mbit of data (travelling at lightspeed) in one channel in the fiber between those two endpoints. The Traffic Engineers had a small nightmare on their hands.
When do we send a "Stop transmitting I'm full" to the previous machine in the path? Losing data, where the endpoint user would need to retransmit, would just make the congestion worse.
Of course we had much more expensive doodads on those jobs, for sure, like combining laser wavelengths (see DWDM) onto one fiber to (at the time) get about a half terabit transmission rate on a fiber, amplifying the light and correcting dispersion in-situ every 60 miles or so, spreading out those combined frequencies via a diffraction grating at the distant end, collecting the data and switching packets onto different routes as they exited the node.
NEC Asynchronous Transfer Mode Switch for Sprint, 1998:
OC-12 is a network line with transmission speeds of up to 622.08 Mbit/s (payload: 601.344 Mbit/s; overhead: 20.736 Mbit/s), which was kinda cutting edge at the time.
Send a bunch of those packet streams into the "switch" which mixes them all up (dang buffers again!) and routes them to the appropriate output streams.
That's not me in the picture.
I got to watch them work and then test their creation before taking it to the Customer in the US for more testing before deployment.
I haven't detected any issues with the cheap boxes and junk cables I have here at home.
Except the Router locks up every couple of weeks, and needs a power-cycle but that's all. It's a ten year old refurbished LinkSys rebranded with a Cisco logo.
---
Ok, I took so long above I went (so to speak):
Nordost QNet Unmanaged Ethernet Switch and QSource Linear Power Supply, $3199 and $2749.
It's round, it must be good. The cheap ones are all rectangular.
Oh wait, the QSource is rectangular.
No measurements provided, lots of opinion and listening impressions, I suppose, Didn't read it (yet, if I do).
Stole the picture from elsewhere, same stock photo in the 'Phile.
Solving a problem that doesn't really exist, at least, not here.