Well, I don't really want to start an argument of optical vs. Coax, but here's what I have learned from looking at scope traces of both and actual listening sessions of optical, coax, AES/EBU. What was my path of progression? Optical to coax to AES/EBU and never looked back.
Let's first forget about the sound quality and impressions and interferences.
On a scope, if you measure the "receiving" end of the optical (TORX receiver head) and coax. You will find that the square wave coming from optical have rounded edge, giving it a "sawtooth" kinda look, while on the coax, it is much "squarer" (unless your digital source is really screwed...).
If you look at how digital signals are received, you will see that the rounded edge of the signal stretched the ambiguous state between 0 and 1 (See note below). This is one of the major contributor to jitter in your signal.
Why is optical having that problem? Take a look at the spec, Toslink transmitter and receiver only work up to tens of MHz. While a good coax goes way, way, way over 1GHz (IIRC BNC connector goes to about 700 something MHz)
On actual listening test, I have tried toslink (cheap free ones that comes free with almost everything) and coax (also uber cheap ones that is intended for composite video signal). I did this test about 5~8 years ago, when my friend and I are not at all into audio. We are just trying it out for the fun of it, and we expected toslink to win because we "think" that it won't get interfered at all. (My friend just got a new Kenwood AV integrated amp, we just had to try everything out...)
Well, to our surprise, coax sound better, WAY better... both cable costs us next to nothing, so I would classify them to be in the same league. Both are new from the packaging, so there is not much chance of pre-owned abuse.
Fast forward to a couple years ago, while talking to a stereo shop owner, and they also agreed coax sounds better. However, they have just gotten a source that outputs AES/EBU as well as optical and coax. I asked about AES/EBU's performance, and the owner didn't know. They've never tried that, and they don't have a XLR cable for digital anyways. I dug around their shop a bit and found a low end Audioquest analog balanced cable. I figured "what the heck, just try it out for the sake of trying it out."
The instant it was on, it felt as if the DAC we used (Micromega DAC1, nothing to do with benchmark) had just doubled in price. It is an even bigger difference than swapping toslink to coax. I don't have a clue how to measure the difference of that against coax or optical, but I doubt I can see much improvement on the scope with my bare eyes anyways (the coax was quite square already.)
Digital transmission is non-lossy with data, yes. However when you are considering the real time transmission of data that have a very important clock embedded into the signal. Any jitter you received will affect the embedded clock signal. Imagine it as if you are playing a tape, if the motor "jitters", the sound will just go to hell. Well, spdif data is like a tape with the clock acting like the motor of the tape player. Both are important, neither one should be ignored.
Note: if you check, a lot of digital device will have a defined low, high and "undefined" state, where voltage below certain point is considered 0, above certain range is considered 1, and anything in between is undetermined. Due to manufacturing issues, there are always chances that the chip will recognize something in the "undetermined" zone, but there are no guarantees on repeatability and reliability there. Which is why manufacturers cut that section out and tell users not to send signal in that state.
Let's first forget about the sound quality and impressions and interferences.
On a scope, if you measure the "receiving" end of the optical (TORX receiver head) and coax. You will find that the square wave coming from optical have rounded edge, giving it a "sawtooth" kinda look, while on the coax, it is much "squarer" (unless your digital source is really screwed...).
If you look at how digital signals are received, you will see that the rounded edge of the signal stretched the ambiguous state between 0 and 1 (See note below). This is one of the major contributor to jitter in your signal.
Why is optical having that problem? Take a look at the spec, Toslink transmitter and receiver only work up to tens of MHz. While a good coax goes way, way, way over 1GHz (IIRC BNC connector goes to about 700 something MHz)
On actual listening test, I have tried toslink (cheap free ones that comes free with almost everything) and coax (also uber cheap ones that is intended for composite video signal). I did this test about 5~8 years ago, when my friend and I are not at all into audio. We are just trying it out for the fun of it, and we expected toslink to win because we "think" that it won't get interfered at all. (My friend just got a new Kenwood AV integrated amp, we just had to try everything out...)
Well, to our surprise, coax sound better, WAY better... both cable costs us next to nothing, so I would classify them to be in the same league. Both are new from the packaging, so there is not much chance of pre-owned abuse.
Fast forward to a couple years ago, while talking to a stereo shop owner, and they also agreed coax sounds better. However, they have just gotten a source that outputs AES/EBU as well as optical and coax. I asked about AES/EBU's performance, and the owner didn't know. They've never tried that, and they don't have a XLR cable for digital anyways. I dug around their shop a bit and found a low end Audioquest analog balanced cable. I figured "what the heck, just try it out for the sake of trying it out."
The instant it was on, it felt as if the DAC we used (Micromega DAC1, nothing to do with benchmark) had just doubled in price. It is an even bigger difference than swapping toslink to coax. I don't have a clue how to measure the difference of that against coax or optical, but I doubt I can see much improvement on the scope with my bare eyes anyways (the coax was quite square already.)
Digital transmission is non-lossy with data, yes. However when you are considering the real time transmission of data that have a very important clock embedded into the signal. Any jitter you received will affect the embedded clock signal. Imagine it as if you are playing a tape, if the motor "jitters", the sound will just go to hell. Well, spdif data is like a tape with the clock acting like the motor of the tape player. Both are important, neither one should be ignored.
Note: if you check, a lot of digital device will have a defined low, high and "undefined" state, where voltage below certain point is considered 0, above certain range is considered 1, and anything in between is undetermined. Due to manufacturing issues, there are always chances that the chip will recognize something in the "undetermined" zone, but there are no guarantees on repeatability and reliability there. Which is why manufacturers cut that section out and tell users not to send signal in that state.
