Better: Coax, Optical or BNC?

[ROBSCIX]

I have been doing alot of reading lately on connections for S/PDif and electrical charachteristics of each. There seems to be alot of disagreement as to which is the best connection method for a S/pdif output to an external DAC.

Many suggest it is optical, no possibility for any interference through EMI/RFI and no electrical connection to the source.
Others suggest it is Coax, you take the optical receiver and transmitters out of the equation which can be of low grade.

Also, some suggest that BNC is the way to go.
I came to the conclusion that each connection methods has both pros and cons but there has to be more to it.


So, I was wondering what experienced people think is the best method for connection and what are their reasons behind it?

 

[jh4db536]

People say the coaxial sounds "more analog" then they say optical has jitter, but opto isolation is great for computer sources. And so on.

Then there's I2S connections, also USB.

The best method for connection is the one that sounds the best (if there are any differences) to your ears. After all digital is just 1's and 0's. The DAC interprets it the same under the same clock assumptions whether it's onboard or offboard/discrete or IC and whether it's a coaxial cable or a toslink.

 

[cobaltmute]

Galvanic isolated BNCs with 75 Ohm co-axial cable. I trust the RF engineers opinions.

 

[ROBSCIX]

Quote:

Originally Posted by jh4db536 /img/forum/go_quote.gif People say the coaxial sounds "more analog" then they say optical has jitter, but opto isolation is great for computer sources. And so on. Then there's I2S connections, also USB. The best method for connection is the one that sounds the best (if there are any differences) to your ears. After all digital is just 1's and 0's. The DAC interprets it the same under the same clock assumptions whether it's onboard or offboard/discrete or IC and whether it's a coaxial cable or a toslink.

Yes, I know there are other conenction methods, I am just talking about "consumer" S/pdif. I always go with what sounds good to me, I was just curious on the different opinions among the audio community.

 

[FallenAngel]

I stick to 75R coax terminated in 75R BNC with input/output pulse transformers for my gear.

 

[Wilf]

Quote:

Originally Posted by cobaltmute /img/forum/go_quote.gif Galvanic isolated BNCs with 75 Ohm co-axial cable. I trust the RF engineers opinions.

As an RF engineer in a previous life, I completely agree.

That and good, long interconnects for the coax.

 

[linuxworks]

long interconnects?

do you disagree with one of our experts, dan lavry? he's quoted as saying that shorter (up to realistic sizes) is better. longer only adds more reflections that make picking the waves out 'harder'.

his argument makes sense to me.

as for the better interconnect, any properly done interconnect will sound the same. the issue is: on YOUR dac and transport, is there a better *implementation*?

one is not better by design. the only thing that varies is the implementation. don't accept 'opto is better' or 'coax is better'.

"it depends" is the only valid answer. you can find good and bad in all interconnect types.

you can even find very high bandwidth opto blocks, so that's not even a determining factor.

 

[ROBSCIX]

Well that is what I am asking people is their opinions.
Like any other debate in AUDIO some have very strongs ideas about this is better then that or that is better then this.

I think the long interconnects comment may have just been a joke.

 

[linuxworks]

not sure it was a joke but I've heard people argue FOR it with full seriousness, before.

ideally, though, your cable should be as long as it takes for the reflection to travel the length of the cable during a typical song duration so that reflections from one song don't blur into the next one.

...ok, THAT was definitely not serious

 

[Pars]

The long spdif cable comment was not a joke. The RF engineers that I know of (previous poster, and Jocko on diyaudio/diyhifi.org) seem to agree on this. Dan Lavry does not. Since an SPDIF interface with a poorly conceived clock running at ~2.8MHz (for 16/44.1 CD) is an RF interface, I'll let you draw your own conclusions.

 

[cobaltmute]

I'm still digesting all the info in this thread, but it may be appropriate:

DIYHiFi.org • View topic - Homebrew TDR

 

[ROBSCIX]

I figured this topic could generate some good conversations.

 

[linuxworks]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif The long spdif cable comment was not a joke. The RF engineers that I know of (previous poster, and Jocko on diyaudio/diyhifi.org) seem to agree on this. Dan Lavry does not. Since an SPDIF interface with a poorly conceived clock running at ~2.8MHz (for 16/44.1 CD) is an RF interface, I'll let you draw your own conclusions.

I've read jocko's posts before and I'm sure he has a pretty good background. I'm not knocking him at all. but I'm going to side with lavry on this, just a gut feeling.

besides, if the cable is properly terminated, the reflection argument becomes less and less significant. add to that the ability of modern dacs to reclock and reject noise on the waveforms and the reflection issue, I think, become theory-only and not a real world issue.

personally, I've mangled spdif in almost every possible way, using the most absurd cabling imaginable and it still passes bits 'just fine' to any modern decent dac. I almost can't create issues with cabling; and so I have to trust my 'lying ears' on this one

 

[linuxworks]

Quote:

Originally Posted by cobaltmute /img/forum/go_quote.gif I'm still digesting all the info in this thread, but it may be appropriate: DIYHiFi.org • View topic - Homebrew TDR

a tdr can be useful; but one also has to ask, how much 'wave damage' can there be before YOUR dac shows 'signs of issues'.

that's the more important question. measuring SWR on the cable is fine but it is only a measurement and needs *interpreting*.

otoh, it can be useful to 'null out' any standing waves on the cable; but again, will that result in better audio? its not clear at all to me that reducing swr below X amount will be an audible improvement.

 

[cobaltmute]

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif besides, if the cable is properly terminated, the reflection argument becomes less and less significant.

But we all know that most are not terminated properly.

RCA connectors are not 75 Ohm. And there is other impedance change along the line causing reflections.

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif a tdr can be useful; but one also has to ask, how much 'wave damage' can there be before YOUR dac shows 'signs of issues'

Like many other things in audio, there are measurements and what the ear can hear.

Why do designers put termination resistors on I2S lines or clock lines? The DAC says that it can handle it, but it can make a difference.

As you've said, you can abuse the daylights out of the signal. It makes sound. But have you ever tried to get it right? True 75 Ohm end to end? And then compared the difference in sound?

I'll admit I haven't. But I've read enough of the theory that technically it seems worth the effort. And this is Head-Fi. We build 18W amps to power headphones. Why not spend the effort? I know I will on the next one I design.