Question What’s best, Optical or Coaxial Inputs?

[Tony T]

I believe that to a certain extent it may depend upon what DAC you are using and what you are connecting to but as a general rule what input is technically better, Coaxial or Optical?
I labour under the impression that coaxial is better but have no real idea why. I may of read that Optical is often implemented cheaper, poorer component etc but can’t remember if it was a opinion or based on facts?
One thing I have noted is the apparent lack of dressed up or fancy inputs with Optical, the receptacles look like they are all made in the same factory biggest difference being colour of input flap.

While we’re at it, is AES/EBU or BNC a better option if available and why? i2s from my very limited understanding is not really appropriate for connections between boxes but is used in rather short paths between components on the same or adjacent boards in a single product.

Maybe a member with some actual technical knowledge can provide a bit of input to set me straight?

 

[rabski]

Implementation, implementation and implementation.

Optical theoretically should be better, as it inherently provides complete electrical isolation, so removes a major potential source of noise. It is also potentially immune to other problems.

The big 'but' is that both the send and receive ends involve additional circuitry to convert electric signals into optical signals and vice versa.

Wth regard to the different electrical connections, frankly at the sort of data transmission speeds and amount of data involved in domestic audio, a bit of wet string would do. It also depends on what is being transmitted and (again) on the send and receive methods. For coax, technically BNC is best, and BNC to BNC cables of about 1.5 metre. The characteristic impedance should be 75 ohms, and only BNC is. There is no such thing as a 75 ohm RCA connector, though some are better than others. The reasoning is that it is possible for internal reflections inherent in cable/connector to coincide with the actual signal and therefore attenuate it. Overall, AES/EBU is technically better, in exactly the same way as true balanced is better than single-ended topology for audio signals. In the same way, however, it needs to be correctly implemented at both ends. A rubbish transformer with badly specced impedance and bandwidth can undo all the good.

I2S is designed for short-distance interface internally (between integrated circuits). I don't care what bs some people come up with, it is unsuited for anything else. It has the potential to introduce errors and sync problems and the charactristic impedance means pretty much any cable is wrong. It usually works, simply because (as above) for 'our' purposes it manages to do the job.

In my world, I tend to look at what is used in pro applications for data transfer, as most of that deals with massively more data at massively higher speeds than domestic audio.

 

[Maverick]

my personal experience is that those optical sockets are usually very cheaply made and deteriorate very quickly so I hate them.
Maybe technical isolation arguments for optical vs. coax ?
Although if optical is better the cable sellers would be all over this surely?

 

[garn63]

I like Coaxial me

 

[bencat]

Well I will give this a go but be assured more technically gifted will be along in a minute to give the full details . As I can only give my opinion then here it is .
Philips and Sony devised the first agreed specification for home digital replay and this of course was for CD and is know as the Red Book . This sets out most if not all the agreed parameters and connection it specifies that it should be a true 75 ohm cable and sockets between two digital products at that time a CD transport and DAC .
Now to my knowledge the only true 75 Ohm system is BNC which has sockets and plugs which are both 75 Ohm and you then of course use 75 Ohm cable in between . For Red Book specified connections this is the real and best connection and in those very rare times when I have been able to use it I would agree .
CD was a mass media product and needed to gain traction and a place with enthusiasts and the general public as quickly as possible so the general connector used was the RCA phono socket and plug and so SPDIF was allowed on this connection despite the fact that most phono plugs , sockets and even cable is nothing like 75 Ohm . But they were easy and cheap to fit and already an very widely accepted connection so in they went .
Digital Audio in studios were most music was being recorded , mastered and mixed use AER/EBU balanced cables and connectors so for mass acceptance in that area a digital balanced connection on AES/EBU was needed and to conform to general professional standards this is specified at 110 Ohm . In my view this is the next best connection for general use as there are already excellent plugs and sockets that are consistent and relatively cheap that you can use . I think it not as good as BNC but it is very close and in my view better than RCA. Again probably a very controversial view but I also think it is cable standard agnostic in that a well made cheap cable offers the same performance and standard as a very expensive one .
Lastly they wanted to offer a new connection that would offer a more space age persona and also give the advantage of braking the earth chain and so theoretically being quieter . There were originally two optical systems TOSLINK which was the general one with a simple system and cable . However there was also the early option of AT&T which was a professional and more expensive one taken from the telecoms industry . AT&T was a first class system and in the only two systems I have ever heard using it the best non sounding cable ever it performed as per the amost perfect cable of being no cable . But due to cost it was not adopted by any major mfg and was consigned to history . TOSLINK became the defacto Optical connection and as happens very sadly with mass production everything involved in the chain was built down to a price . Original glass cables were ditched for cheap plastic ones . The system depends on a sender at one end and receiver at the other end these were reduced and made very cheaply and become the limiting factor in the quality of the light transmission . In theory Optical is the best system of all the different standards but only when it engineered and installed using the best components which just does not happen . Boutique and Audiophile makes will not implement decent TOSLINK systems either as it already has a bad reputation for being inferior to RCA SPDIF among Audiophiles so they struggle to get them to believe that if implemented properly it is a real contender .
Now lastly I2S this is not included in the above as it is not really a structured or specified link system for audio. Philips developed the I2S as aa way of chip sets to communicate with each other using the tracks of a circuit board . So it has not real accepted and agreed standards . It has no agreed plug and socket types and it is not robust over any longer distance than probably 0.5 mtr . It has no earth or RFI protection designed in to it because it was never meant to be outside the equipment case . It has the advantage that it has three separate signals Left and Right plus Clock . This means that you get the full clock information delivered along with the signal information and this can be used to accurately keep the clocks of all chips and units involved at the same time and speed similar to what reclocking can do but without the additional unit. However that clever bit requires some really gifted and careful digital engineering and as there are no standards one method used by an mfg will pobably only work with that mfg's products .

So after all that there answer to the OP's question is not balck and white but more it depends . If you want a real world answer that works straight out the box then my view is 75 Ohm BNC . I2S has the option and chance to be better but it depends on the connection being short as possible and the implementation being done by someone who really understands what they are doing . RCA SPDIF would for me come next because current companies know and can make consistent and good quality plugs , sockets and cables if you can find and get them all as close to 75 Ohm in value then they will be as close to BNC as you can get . TOSLINK is a an excellent system let down by being built as cheaply as possible and so losing all of the great strengths it theoretically had. It can still sound good and does make a very noise free connection still but the quality is compromised by the poor quality sender and receivers that are commonly used on even expensive equipment.

 

[Beobloke]

I’ve always preferred the coaxial connection in the past when I’ve compared, but this was on budget equipment, I haven’t tried it for years to be honest.
It’s 1’s and 0’s so I guess done properly they should be pretty similar in theory.

Ah but if your fibre optic connector is really good, you get more 1s, which makes it sound better.

 

[MF 1000]

The majority of high end digital kit is fitted with BNC connections

 

[Beobloke]

In my industry, we send data, video signals and control protocols down cables that up are up to 4000m long, and the best way to do it is using fibre optics. High data rates, immunity to the 3000V power lines that are running beside them and industry standard connections that are well proven mean that it’s the obvious way to do it.

I’ve also compared the coaxial and optical connections between my Marantz CD94 and CDA-94 to find that the optical connection sounded noticeably better.

in summary, when it’s done properly, I’d definitely go with fibre optic being the better option.

 

[bencat]

when it’s done properly

This .

 

[rabski]

when it’s done properly

As Andrew says, this bit.

Unfortunately, if often isn't. There's a reason a lot of designers add an extra 's' to Toslink.

 

[bencat]

Or even insert an h before the l so that it better describes the experience .

 

[freefallrob]

I’ve always preferred the coaxial connection in the past when I’ve compared, but this was on budget equipment, I haven’t tried it for years to be honest.
It’s 1’s and 0’s so I guess done properly they should be pretty similar in theory.

 

[It Cost How Much!?!]

Best ST Optical, then AES/EBU, then BNC Co-ax, then co-ax, then Toslink, then USB. Not tried HDMI.

 

[rabski]

Ah but if your fibre optic connector is really good, you get more 1s, which makes it sound better.

Woah there. You also get more 0s. Anyone knows that zero multiplied by anything equals zero, so this is obviously a major problem. Thus, the better the digital connection, the worse the sound. Probably.

 

[StingRay]

I tried both between my streamer and DAC, the coaxial was louder, even levelling the volume, thought it still sounded better. I may try with different DAC and see if its the same.

 

[Paul55]

Optical makes no electrical connection between each end. Might solve a problem.
Coax, which to meet the standard should be phono, but is much better on BNC, is the best, if implemented correctly with terminations and isolation transformers. If not then optical.
AES/EBU probably serves no purpose unless you want to repurpose existing sound tie lines or it gives you an extra input/output that you need.

 

[tuga]

Ultimately it's down to implementation.
@Fourlegs (apparently no longer a member) said that Rob Watts Innuos said the optical receiver will produce (measured) noise and thus defeat its noise-rejection advantage, and prefers S/PDIF for that reason.
Rob Watts says that USB can potentially be as good as optical.
And USB has the advantage of being able to output DSD1024 / PCM1.536MHz.

I would keep away from i2s, not convinced it can be implemente properly between boxes.
I tried it back in the mid-'90s with my Audio Alchemy 3-box CD playback system and preferred S/PDIF but a lot has happened since.



Edit: my memory failed me, it was Nuno Vitorino, not Rob Watts, the designer Nick was referring to

 

[Tony T]

I recall having a DPA setup in the 90’s that used a optical system called Deltran that required 4 or 6 optical cables to work, mine was a PDM 1 Series 3 which was a 2 or 3 box DAC. I think Rob Watts had a large input in their design so I wonder if his Chord products have well implemented optical inputs?
Apart from the DPA kit I owned I’ve never used optical for hifi, my interest was piqued as I’ve been trying 2 Allo streamers and wanted to compare them on as level a footing as possible with one being USB output and the other BNC or Coaxial outputs.
I use the Coaxial input on my DAC for my CDP so have had to put together a BNC cable>Impedance transformer>AES/EBU cable as i have no BNC input on my DAC but have AES. A bit of a faff all told.

 

[tuga]

I recall having a DPA setup in the 90’s that used a optical system called Deltran that required 4 or 6 optical cables to work, mine was a PDM 1 Series 3 which was a 2 or 3 box DAC. I think Rob Watts had a large input in their design so I wonder if his Chord products have well implemented optical inputs?
Apart from the DPA kit I owned I’ve never used optical for hifi, my interest was piqued as I’ve been trying 2 Allo streamers and wanted to compare them on as level a footing as possible with one being USB output and the other BNC or Coaxial outputs.
I use the Coaxial input on my DAC for my CDP so have had to put together a BNC cable>Impedance transformer>AES/EBU cable as i have no BNC input on my DAC but have AES. A bit of a faff all told.

Measured jitter is very low on the Dave so I would say that yes Chord products have well implemented optical inputs.
But optical is limited to 24/96.

 

[Tony T]

But optical is limited to 24/96.

Oh
All other digital inputs are at least 24/192 though are they not?