Go with the coax.

Optical (or Toslink as it is sometimes know) tends to introduce a considerable amount of digital jitter. So what is jitter? Think of jitter as a miss-timing of the digital data. All the "1" and "0" of the digital data are the same but their time relationships have changed. Jitter messes up the word clock. The best example that I can think of right now is if you had the world's best marching band with a great bass drummer marking time. They would sound great. Now take that same band and substitute a bass drummer with no rhythm (say me for example) all of the information that the band is playing is the same, their sheet music and instruments haven

Hi

Id like to dispute this claim that optical produces the most jitter.

What exactly is jitter ? are we talking about distortion of the digital signal itself ? in which case error correction will fix this problem or are we really talking about quantization error which is the difference between the digitally sampled value and the actual analog value of the sampled wave.

Either way I cant see how it makes a difference between coax and fibre ?

Maybe you could explain how an inferior transmission medium such as coax can transmit a digital signal better than pure glass

Regards

Steve

Quote:
Id like to dispute this claim that optical produces the most jitter.

This is a fact with digital datastreams where the word and bit clocks are embedded in the audio data. Toslink's more limited bandwidth shifts the transition timings, which define the difference between "1"s and "0"s on an AES3 or S/PDIF datastream. See www.stereophile.com/features/396bits .

Quote:
What exactly is jitter ?

The correct data presented to the DAC at the wrong time. See
www.stereophile.com/reference/1290jitter and www.stereophile.com/reference/1093jitter .

Quote:
are we talking about distortion of the digital signal itself ? in which case error correction will fix this problem or are we really talking about quantization error which is the difference between the digitally sampled value and the actual analog value of the sampled wave.

With respect, you are taking a way too simplistic approach to the subject. Jitter doesn't involved data errors (unless it is gross) but the timing of the conversion of each data word to an analog equivalent. This produces spuriae in the analog signal.

Quote:
Either way I cant see how it makes a difference between coax and fibre ?

Read the articles referenced. The point to note, by the way, is that none of the transmission limitations matter until the point where the data are used to recreate the original analog signal. And that is where the jitter-induced spuriae affect the sound. D/A converters vary enormously in their ability to reject incoming word-clock jitter. See my measurements sidebars that accompany Stereophile's reviews of such products.

John Atkinson
Editor, Stereophile

John

Thanks for the detailed response.

From what your saying the AES standard is actually for synchronous transmission using an embedded clock and not as i understood it to be which is Asynchronous using just a known local clock to demultiplex the digital information.

This transmission method would allow for noise to be introduced due to subtle timing differences.

What I still dont get is how Coax can be better at transmitting a digital signal than fibre, I understand the principles of what your saying but from my telecomms network education the superiority of fibre as a transmission medium is immense, there is nothing short of a pure vacumn that is currently better. We can in the telecomms world transmit huge quantities of data at fantastic rates using current ATM and even later techniques.

What am I missing ? is it the TOSLINK standrad itself that is to blame ?

looking forward to your thoughts

Steve

Steve - I think the bandwidth required for telecoms is much lower than for music, since the primary concern has to do with the spoken voice; fibre optics would be adequate for this purpose, but, from what John has posted, perhaps, not broad enough for the full music spectrum.

Huh? The fiber optic system installed by both cable and locally Verizon handles BILLIONS of data bits....This is the 21st century. No such thing as just a phone line doing voice. But again, since this is audio nonsene, everything in reality is bent to suit shortcomings of crappy products, and mindless tweaks and useless junk. The cable system out there is fiber, it is only changed to cable at the pole to the house. Can you hear the difference? I would put money on it, if you didn't know if the DAC was hooked up with coax or optical, if you would know just by listening, and keeping volume levels exactly the same. Using the same DAC. Into the same pre amp.

Quote:
Go with the coax.

Joe,

What type of coax would you recommend? 50 or 75 ohm, RG6, RG59, quad shield, etc., etc. Thanks.

Hi,

In case your system allows for using one or the other at the flip of a switch, try both and see what you think.

You can get either type of interconnect at affordable prices (Audio Advisor has some deeply discounted cables of both types) and see what your ears think.

I loved the old Audio Alchemy devices because it allowed for easy switching.

I preferred coax over optical, but liked AES/EBU best - go figure.

The problem with TosLink is that the bandwidth is insufficient to allow the timing of the data transmission to be retained. Bandwidth limiting causes timing errors in the zero voltage crossings used by the DAC to determine the timing of the signal.

I think the confusion some have is that bandwidth in this context has nothing to do with how much data is being transferred. In the context of an audio digital interface the frequency bandwidth is like voltage slew rate. An infinitely high frequency bandwidth would allow an interface to pass a perfect square wave of 1

Jeff

Sorry, your totally wrong here.

Your confusing bandwidth with traffic, while a single audio signal could indeed have a wide bandwidth its nothing compared to telecomm traffic.

Im not sure if you realise but technology exists for every phone call in the world to be simulateneously transmitted down a single fibre.

Steve

Hi

Again im going to have dispute these claims.

I know im beginning to sound like DUP but there definetly is a hifi bias coming out here. Bandwidth is purely the amount of information that can be carried down a single transmission link, the bandwidth of even a crappy plastic toslink cable is significantly higher than any coax except seriously high grade heliax style cables that cost hundreds of dollars per foot and are really microwave guide tubes.

I dont know what the answer is, all the hifi talk just doesnt match up to the long established priciples of data transmission and its mediums. As stated elsewhere data is date and its purpose is transparent to the transmission medium therefore all digitally encoded music or video is nothing more than a string of 1 & 0 's multiplexed in some way to be transmitted. The fact that fibre is a far superior way of transmitting in every other application in the world except hifi just smells of snake oil again.

Sorry to be so persistant but as a 25 year Telecomms network engineer I do feel very qualified to dispute this particular subject

Steve

You are certainly entitled to believe anything you choose.

I can accept that a quality glass fiber connected to a well enginered output of a transport coupled to another properly designed input of a DAC with excellent jitter rejection could do a great job. However in the typical application coax does a better job.

The induced jitter caused simply by a TosLink cable is readily measured. I haven't taken a look at the articles cited above, but JA typically points to well written articles which fully cite their sources (he is an engineer as well if I recall correctly). If you are interested, read the articles and check the cited references. I am sure they explain the science much better than I.

As with Dup, if you can't hear the difference between cables - digital or analog - don't worry about it. Buy the least expensive product that works for you and enjoy your system.

I was just suggesting a possibility based on what I apparently misinterpreted of what was posted. I certainly wasn't stating anything definitively -- it just seemed logical that the narrow band the spoken voice occupies would require less of the pipe in terms of frequency and amount of information than music ranging from 20Hz-20kHz, making for less likely timing errors and jitter.

Wow

As a great unwashed heathen I will step out into the fields in my ignorance. I will strive to increase my knowledge of all things hifi and worship at the temple of never doubt a hifi god with sound technical knowledge.

Im sure DUP will join me in not apologising for actually challanging some of this technical babble that is being preached in the interests of selling product and concepts

Steve
Steve

Performers -> dozens of mixers and effects -> clipped/hypercompressed mastering -> you think a few extra ps of jitter matters?

Even the Hi-Fi Gods will use a toslink cable to review a piece of equipment...go figure!

RG

Quote:
Bandwidth is purely the amount of information that can be carried down a single transmission link, the bandwidth of even a crappy plastic toslink cable is significantly higher than any coax...

I am sorry, Steve you are just plain wrong. A typical low-cost Toslink fiber and its transmitter/receiver have a bandwidth that is much lower than coax. This is shown in the Hawksford & Dunn article to which I referred you earlier. The lack of bandwidth leads to the introduction of timing uncertainty, which leads, in turn, to the corruption of the analog noise floor of the DAC. Read Barry Blesser's digital audio primer, published in the October 1978 issue of the Journal of the Audio Engineering Society, available from www.aes.org .

John Atkinson
Editor, Stereophile

Quote:
Thanks for the detailed response.

You're welcome.

Quote:
From what your saying the AES standard is actually for synchronous transmission using an embedded clock and not as i understood it to be which is Asynchronous using just a known local clock to demultiplex the digital information.

That is correct. The local clock needs to be tied to that embedded in the data, which creates a causal link for the transmission of jitter.

Quote:
This transmission method would allow for noise to be introduced due to subtle timing differences.

Yes indeed.

Quote:
What I still dont get is how Coax can be better at transmitting a digital signal than fibre...

Because of its lower bandwidth, which in the worst case can be a few MHz. Again, see the Hawksford & Dunn article. I believe you are confusing what can be achieved with the best glass fiber and single-mode driver, which indeed is ultra-high bandwidth, with what is attainable from cheap LED-driven plastic.

Stereophile has published measurements showing how, with susceptible receivers/clock recovery circuits -- and those _are_ more rare these days -- S/PDIF over Toslink introduces greater spuriae in the reconstructed analog signal than the same digital data transmitted over coax.

John Atkinson
Editor, Stereophile

Steve,

What bandwidth in MHz do glass fiber telecom communication lines achieve? I imagine it is pretty astounding and puts plastic TosLink cables to shame.

As a separate issue, does anyone know the extent to which the additional conversion steps of electrical to optical, and then optical to electrical again affect the S/PDIF signal in a typical implementation of TosLink?

The bandwidth available on single mode glass fibre can easily exceed 2 GHZ with the correct optics.

Modern ATM transmissions are sending data at a rate of 565 M/bits per sec.

It seems as if this confusion stems from TOSLink cables actually being pretty crappy and not in fact the same as a multi or single mode fibre link, as a single mode patch cable costs approx $40 it would seem that a Toslink cable for about the same is very poor value

Sorry to push the issue, but it seems as if the truth came out in the end !!!

Steve

2 GHZ! Cool! This would meet any of my needs.

Also poor are the transmitters and receivers on TosLink connected equipment - they are typically pretty basic as I understand it.

Would the patch cable you refer to include lenses on the ends? (as an analog to what a TosLink cable has). I bet good lenses make a tremendous difference.

Quote:
Hi,

In case your system allows for using one or the other at the flip of a switch, try both and see what you think.

You can get either type of interconnect at affordable prices (Audio Advisor has some deeply discounted cables of both types) and see what your ears think.

I loved the old Audio Alchemy devices because it allowed for easy switching.

I preferred coax over optical, but liked AES/EBU best - go figure.

The A/B comparison is a good idea, Buddha. I have access to glass (not plastic) fiber, the needed connectors and the equipment to make my own cables as well as materials and tools to make coax cables. I'll try them both.

The numbers are there to prove the jitter argument. Besides, why waste time converting from electrical to optical and back if you don't have to? Open up any single box disk player and I highly doubt you'll find even one that uses optical connections internally. That being said I also doubt that the difference is audible. The TOSLINK optical really only makes sense on portable equipment where space is a concern. Otherwise just use the coax. The bandwidth argument is kind of silly as well when we're talking about digital signals in two different domains- optical and electrical. Maybe you can squeeze more bits through the optical over long distances but if it's a 1 or 2 ft. piece of cable carrying relatively low bandwidth digital audio signals whats the use of converting the data stream into the higher bandwidth optical just to send it a foot or two to another component and change it back again? This one just seems to be a no brainer. If the signal were somehow kept in the optical domain from the disk to the jacks it might make sense I guess. Speaking of which, another issue with optical is the typically very poor jacks that don't really make very nice connections and probably introduce "optical impedance errors" where say, a glass fiber has a slightly different optical characteristic than a polymer one. By the way, if TOSLINK optical is so great why haven't the computer companies jumped on the bandwagon? There now i've said it...

Quote:
I have access to glass (not plastic) fiber, the needed connectors and the equipment to make my own cables as well as materials and tools to make coax cables. I'll try them both.

Cool. Let us know what you conclude. Nothing better than real experience to decide these things.

Elk,

I was able to borrow the Toslink inter-connects (both multi and single-mode glass, not plastic) from a vendor I frequently deal with. (Much easier than making them.)

Anyway, I wasn't able to hear any difference between the multi-mode plastic Toslink inter-connect and the multi-mode glass inter-connect. (From what I've was able to determine, apparently multi-mode plastic Toslinks are the "standard".)
Maybe it was the short length of the cable that made any difference unnoticable. There was an improvement, though, when using the single-mode glass inter-connect. The sound was more clear, for a lack of a better word.

I then tried the coax inter-connects, that I had made (using RG-6 quad shield cable). Here there was a difference between the copper and the Toslinks. The copper sounded better than the supplied (MM plastic fiber) as well as the glass MM fiber. But between the SM glass fiber and the copper there was a slight tonal difference. One didn't sound better than the other...just different. User's preference.

Mike

Very interesting.

Now if we could just explain all of this in an easy to understand fashion.

Thanks for letting us know.

The short report: coax and single-mode glass interconnects sounded best.

Toslink interconnects, that come with whatever piece of equipment you buy, have plastic optical conductors. (Plastic because it is cheaper than glass optical conductors.) Also, these Toslinks are made of multi-mode optical conductors. Glass refracts light better than plastic and single-mode glass refracts light in a different manner than multi-mode. So, in theory, a single-mode glass interconnect will transmit the light better than multi-mode plastic.

Try researching optical fiber on wikipedia...I'm sure they explain it better than I have.

If you want to be even more confused ask me another question.

Hey Elk, check this out:
http://www.audioholics.com/techtips/audioprinciples/interconnects/toslinks.php

Here they explain what toslink is all about. (Haven't read it myself, though.)

Quote:
Hey Elk, check this out:
http://www.audioholics.com/techtips/audioprinciples/interconnects/toslinks.php

Here they explain what toslink is all about. (Haven't read it myself, though.)

Broken link...

What can I say? I came across that info while doing a google search.

Been doing a lot of reading on this subject lately, trying to drag vague recollections of E&EE classes skipped over twenty years ago into the twenty-first century. Happily(?) not much has changed while I was away. While the level of understanding of real-world problems in digital circuits has advanced considerably, very little else has.

One major take-out for me has been the extent to which a desire to be backwards-compatible with the then-massive installed base of analogue studio equipment (inadvertantly) hamstrung the digital interface standards from the get-go and the degree to which the subsequent explosion in the installed base of digital equipment (both professional and consumer) has militated against efforts to do anything about it.

Anyway, here are some suggestions for further reading I've come across on my travels.

For the subjectivists: a lively blow-by-blow account of a Toslink/coax tag-team smackdown.

For the engineers:
this latest page-turner from the AES which covers some of the learnings of the last decade and how they are proposing to wrap them up into a set of new guidelines for the use of AES3 interfaces going forward. It is extremely comprehensive but also somewhat 'chewy'. For those who like to catch up on their reading backlog in the 'small room' I would caution that this will require a good long movement, and I don't mean Mahler. Don't forget to recycle.

Finally, for those who prefer the ride at a slightly higher altitude I would recommend this piece. I have no idea about the author's credentials but it seems pragmatic and well-written to me. There is a really good chain(?) of links at the end for the real insomniacs to get stuck into.

If you'll allow me to pull out some of the, ahem, pithy bits from all of this, I would say (in reference various comments throughout this thread):

• Both Toslink and RCA (Coax) interfaces can produce good sound
• They have different theoretical strengths and weaknesses
• As far as the cables go there are differences, oh yes sirree, but those differences may be quite slight and price may not always be a good guide to quality.
• A number of factors are quite critical in cable design and many of these are not covered by the interface definitions as they weren't understood at the time the latter were written.
• The implementation of an interface is generally more important than the type. Real-world components are made down to a price and the specific compromises chosen in each implementation probably affect the sound quality more than the theoretical considerations.
• Your ears are the final judge, but do remember that upstream differences or flaws will only be discernable to the degree that the downstream system is transparent to them. Trying to hear the difference between two good digital cables in a mid-fi/poorly set-up system might be like trying to taste the difference between two vintages of the same wine poured into a red-hot chilli.

Purely out of idle curiosity I have added a mini cable survey to my to-do list. Current combatants are:
In the red corner (Toslink): Supra X-ZAC, Wireworld Supernova 5, Van den Hul OptoCoupler, Generic 'licorice' cable
and in the blue corner (Coax): Belden 1694A, Kimber Illuminati D-60, Transparent "High Resolution 75 ohm link Triple Shielded", an old Audioquest 'Joe Doe' found at the bottom of the cable box (postmortem ongoing).

Naturally I'll report back on any findings if I can sneak away from nappy-changing duty long enough to even embark on this folly. Breath-holding definitely not advised.

Struts,

I look forward to your findings and seeing how they compare to mine.

Another reference.

I just found this article which seems to be a first class summary of the issues and extremely approachable. Warmly recommended for those interested in the technical details but without the engineering background or the stamina for the AES papers.

Happy reading!

Another reference. But this one is really good. Honest!

Check out this great article over at Positive Feedback which IMHO sweeps the board as far as lucid, clear explanations of jitter go and is a great 101 on computer audio into the bargain.

Perhaps I should start a Todd-style thread comparing articles on jitter? Wait a second, perhaps I already did??

Stop!!! Jamaican me crazy!!

Mike, thanks for your listening notes. I've just done similar tests between Toslink, Coax and USB inputs on my RME ADI-2 DAC, before I read your evaluation.

My own notes are similar. USB and Coax inputs sounded identical to me at 96kHz/24bit and 192kHz/24bit. Toslink was a little worse. Bass seemed to have a little less depth, drums a little less clear snap. It's not a huge/unbearable difference. With Coax, the sound is just a bit sharper which makes a listening session is more pleasurable.

Toslink seems to compete better at lower sampling rates (48kHz) which suits me fine as I have a CD player and a Roku and an Airport Express (secondary sources) which all output at 44.1kHz and 48kHz and can share an optical switch. My iPhone (primary source) will reside on the Coax connection and my Mac (primary source) will use the USB connection.

The Toslink which I'm using here is all off-the-shelf standard plastic/MM Toslink. I haven't noticed any sound quality difference whether the Toslink is very thin or thicker or has more or less expensive looking connects. Of course, the better built an ordinary Toslink cable is, the better it will stand up to movement/abuse. If the cable just lies in one place though, unless you have the SM glass which Mike recommends Toslink connects all sound the same (in my experience), especially if the devices are going through a plastic MM switch (as in my case). I haven't tested extensively direct Toslink vs Toslink via switch. I did those tests when I first introduced the switch and they came back identical (given that I'm already using plastic MM Toslink connections this result is unsurprising).

Main testing method in my case: I have two USB to S/PDIF converters which offer both optical and coaxial output simultaneously. I used both of them (separately) to A/B the identical audio conversion on two inputs. Results were the same with both USB to S/PDIF converters though one uses the Cmedia C6631 and the other uses the XMOS U208 converter. with both conversion chips/converters, coaxial was slightly better.

Bonus test: hum. In my system, I did not find that there was any more hum at max volume on my speakers when using coaxial. If someone were to suffer from hum via coaxial, the easy solution would be to use an optical connection. In this case, the trade-off of an electronically silent connection against a slight loss of detail would definitely be worth it.