4 ms·

The best part is the linked article's Conclusion section. It includes such gems as "zero and one values represent a lower and higher voltage in the cable and in

The best part is the linked article's Conclusion section. It includes such gems as "zero and one values represent a lower and higher voltage in the cable and in the sender or receiver. This is actually an analogue signal when we look at it on a scope. We audiophiles agree that cables sound different with analogue signals, so why not with Ethernet data?" and "Next, any interface for Ethernet has a correction circuit for errors. With many errors this interface has to work hard, that will result in a greater need for current from the power supply and the processor heating up."

wolfgang42 10y ago
Also, "So a decent switch will do a lot of work to correct errors, but it has no analogue sound circuit on board, that is why it doesn’t matter how much the switch suffers from retries and all kind of errors and interferences. In the end the switch nicely delivers the data to our music system and we have to make sure this last part is error free. We don’t want to use error correction in audio equipment after all."
- davidgerard 10y ago
  (OP here) A sysadmin in finance on my facebook suggested we could start selling audiophools the routers they use for high-frequency trading, which he found a similarly sensible field. (I found this somewhat disconcerting news.) Or perhaps the super-low-lag cards they sell to obsessive gamers.
  - wolfgang42 10y ago
    I've never heard the word 'audiophools' before--I like it :) It's never occurred to me that HF trading would have special routers, but I suppose that makes sense. This sounds like an excellent business model--or, better yet, sell custom routers with special 'audio quality algorithms' built in. I'm sure there's some manufacturer in China that would be willing to white-label a box with a few QOS setting tweaks, which can then be sold at an excellent markup. I will leave this as an exercise for someone with a better straight face than I.
smoyer 10y ago
Oh my - I'm not sure where to start. I designed 100BaseT and 100BaseFX equipment in the mid-90's and there is a very good reason Ethernet signals look analog on the wire ... We use the inductors and capacitors matching the drivers and receivers to the cable to round them off. Nice, perfect digital signals (square waves) are built from all odd harmonics of the original signal. It would be impossible to meet FCC emissions form the cables if you didn't eliminate these harmonics. The good news is that the receivers are really good at turning the differential signals in an Ethernet cable back into beautiful square waves on the circuit board. The Ethernet receiver also uses a PLL to lock on to the main frequency of the driver, so it even adapts as the driver frequency drifts. I can think of a couple reasons that one Ethernet cable might sound different than another but all of theses reasons involve increased error rates which would normally be caused by a bad cable, flakey driver or an overly sensitive receiver. Since the more complex and compressed the audio is, the more likely you'd break the decoding instead of hearing audio degradation. TLDR ... You might hear differences when sending audio over different types of cables but if you do'it's not because the Ethernet signals are analog.

wolfgang42 10y ago

Also, "So a decent switch will do a lot of work to correct errors, but it has no analogue sound circuit on board, that is why it doesn’t matter how much the switch suffers from retries and all kind of errors and interferences. In the end the switch nicely delivers the data to our music system and we have to make sure this last part is error free. We don’t want to use error correction in audio equipment after all."

davidgerard 10y ago
(OP here) A sysadmin in finance on my facebook suggested we could start selling audiophools the routers they use for high-frequency trading, which he found a similarly sensible field. (I found this somewhat disconcerting news.) Or perhaps the super-low-lag cards they sell to obsessive gamers.
- wolfgang42 10y ago
  I've never heard the word 'audiophools' before--I like it :) It's never occurred to me that HF trading would have special routers, but I suppose that makes sense. This sounds like an excellent business model--or, better yet, sell custom routers with special 'audio quality algorithms' built in. I'm sure there's some manufacturer in China that would be willing to white-label a box with a few QOS setting tweaks, which can then be sold at an excellent markup. I will leave this as an exercise for someone with a better straight face than I.

smoyer 10y ago

Oh my - I'm not sure where to start. I designed 100BaseT and 100BaseFX equipment in the mid-90's and there is a very good reason Ethernet signals look analog on the wire ... We use the inductors and capacitors matching the drivers and receivers to the cable to round them off. Nice, perfect digital signals (square waves) are built from all odd harmonics of the original signal. It would be impossible to meet FCC emissions form the cables if you didn't eliminate these harmonics. The good news is that the receivers are really good at turning the differential signals in an Ethernet cable back into beautiful square waves on the circuit board. The Ethernet receiver also uses a PLL to lock on to the main frequency of the driver, so it even adapts as the driver frequency drifts. I can think of a couple reasons that one Ethernet cable might sound different than another but all of theses reasons involve increased error rates which would normally be caused by a bad cable, flakey driver or an overly sensitive receiver. Since the more complex and compressed the audio is, the more likely you'd break the decoding instead of hearing audio degradation. TLDR ... You might hear differences when sending audio over different types of cables but if you do'it's not because the Ethernet signals are analog.