Optical Cable

[tbewick]

I haven't heard about jitter relating to Toslink. Most respectable sources (like Dolby, Audioholics) say that either S/PDIF or Toslink will work fine.

 

[MDS]

Jitter is inherent in any digital audio system but the levels are extremely small - on the order of picoseconds - that is trillionths of a second.

S/PDIF is the protocol used for both coax and optical connections. Any jitter or lack thereof will be the same for either type of connection. The reason people confuse s/pdif and toslink as being two different things is that the s/pdif spec was written first for electrical connections (coax) and includes the electrical parameters for the physical interface. Toslink is the physical interface for optical connections (as well as the name of the connector).

 

[ZoFo]

lifatec.com

Lifatec was recomended by BlueJeans Cable who made all of my analog cables. They will build a Toslink cable to any length that you want and are very resonable, $12 for a 3ft Toslink if I remember correctly.

There fiber cables are very thin, no additional jacket is used. The connectors all have metal ferals and come in different colors, you can also get two Toslink cables molded together. They also make a "Armored" version where they cover the fiber with Romex and use aluminum connectors.

 

[xboxweasel]

I read that digital audio can transmit up to "x" channels of encoded audio (DTS, DD, DTS-ES, DD-EX) and 2 channels of non encoded audio (like from a CD player).

Can't remember what value x was. At least 7 though.

cheers.

 

[Davy9]

Optical vs Coax

I go with optical just for one reason:

Coax consits of a conductor while optical doesn't. The conductor in the coax is susceptible to interference from other sources or cables. That being said, I don't know how much this really makes a difference, I just like to do what ever I can not to comprimise the end result.

Optical fibre is just a glass tube that carries a light signal. The quality of it is determined by the amount of inconsistencies in the glass itself, or the ability to produce the highest percentage of "perfect" reflections. The higher the quality, the less inconsistencies.

 

[Hi Ho]

The way I see it is this. An optical cable transmits 1's and 0's. That is, the light is either on or off. There is no in-between. Now, if you are getting a signal at all, that would mean you are getting a perfect signal. How can some light signals get through and some not? It there is anything getting through then it's all getting through. Is this not correct?

 

[tbewick]

This was the point of what I said in an earlier post - the Toslink/electrical digital signal IS analogue, and is therefore susceptible to distortion and noise as a signal that is analogue not digitally coded. The whole point of using digital coding (with error correction (redundancy), self-clocking, phase-locked loops etc.) is to make the signal resistant to this distortion and noise in a way that a non-digitally coded signal isn't. Resistance to noise and distortion is what digital audio is all about.

 

[markw]

Y'all seem to forget that that light signal originated an electrical signal...

...which was created by interperting and error checking/correcting the original ligjht signal that was reflected off of the DVD's surface.

IOW, it goes through two additional conversions when you use optical.

 

[MDS]

This was the point of what I said in an earlier post - the Toslink/electrical digital signal IS analogue, and is therefore susceptible to distortion and noise as a signal that is analogue not digitally coded.

You are a bit confused tbewick. It is 'digitally coded'. A stream of ones and zeros can be transmitted over a wire using any number of modulation schemes. It's the same principle as sending data over the wire between computers. It's the same principle used to send mpeg-2 data over the coax cable from the cable company for 'digital' cable. Just to make it even more interesting, your IR remote sends digital data via an analog infrared signal. RF remotes send digital data using analog radio frequency waves.

The signal is analog but has either fixed frequency or fixed amplitude (depending on the modulation used) but the DATA is digital.

 

[tbewick]

I'm just reciting from the book 'An Introduction To Digital Audio', by John Watkinson, Focal Press, which I read a while back. This is the technically correct way of describing how any digital channel works, like the Toslink interface. The transmission medium itself is analogue, and is susceptible to distortion and noise. Hi Ho said that 'if you receive a signal, it is perfect' - this is not true, it will have noise, distortion, and jitter. The original digital information can be obtained from this by interpreting all voltages/current/light intensity above a certain threshold to be 'on', and all below a certain value to be 'off'. Jitter is rejected and the signal correctly clocked by using a phase lock loop. The use of a coding scheme is essential in transmitting the digital data. How the digital information is reproduced correctly so that the cable becomes transparent is a subject of some detail.

Hi markw,

The distinction between when the data is analogue or not was not made by me. When digital data is sent over distances it is sent over a 'channel'. The design of channels is a separate subject from when digital data is kept in an electronic circuit.

 

[jonnythan]

I fully maintain that with digital, the signal is either there or it isn't.

If these interfaces were not capable of transmitting the exact same data, bit for bit, regardless of interference or whatever else, then the $4 18" SATA cable connecting my hard drive to my motherboard would be incapable of copying a 400MB file without numerous disk and transmission errors, and the $20 optical cable connecting my Cisco PIX router with the T3 line at my office would be incapable of coherently transmitting our internet traffic.

A $100 optical cable is not going to have ANY discernable difference from a $10 one. They are transmitting the exact same data, bit for bit. The entire idea of frequency response changing is ludicrous and belies a complete misunderstanding of digital data. With a CD sound source, you're going to have the exact same number of bits per second whether the CD contains total silence, a singer's voice, or a 100-piece symphony orchestra, and those bits are all 1's and 0's.

 

[tbewick]

'If these interfaces were not capable of transmitting the exact same data, bit for bit, regardless of interference or whatever else, then the $4 18" SATA cable connecting my hard drive to my motherboard would be incapable of copying a 400MB file without numerous disk and transmission errors, and the $20 optical cable connecting my Cisco PIX router with the T3 line at my office would be incapable of coherently transmitting our internet traffic.'

The reason why this is possible is because digital systems can reject distortion and noise, and this is not only because 'the signal is either there or it isn't'.

'The entire idea of frequency response changing is ludicrous and belies a complete misunderstanding of digital data.'

Again, this is a fundamental misunderstanding. Jitter could reduce the bandwidth of the audio as it has the effect of increasing the noise level. Personally I doubt whether jitter is very audible (at all), but as I say this is my personal opinion and is not based on anything I have read. Chris (WmAx) posted some time back a reference to a scientific journal which had a study into the audibility of jitter.

As for understanding digital audio, I am certainly no expert, and approach the subject as a layman, but I have read a book on the subject.

 

[MDS]

Jitter does not reduce the bandwidth of a signal. Jitter is a TIMING error. It's effect, were it significant, which it is not, would be to change the data due to incorrectly interpreting a 1 as a 0 or vice versa.

The simplest layman explanation I can give is to use something like PWM (Pulse Width Modulation) where the 'width' of the pulse determines a 1 or zero. Say a 20 usec pulse was defined as zero and a 50 usec pulse was defined as a 1. Now what is the receiver of the stream to do if it *thinks* it saw a 40 usec pulse? Was that supposed to be a zero but the pulse was too long or was it supposed to be a 1 but the pulse was too short? That is jitter, but of course we are talking about timing errors on the order of picoseconds and all the PLL and re-clocking circuitry mentioned takes care of it.

Optical cables can easily affect the data because the principle is different. The light pulses travel down the cable due to 'total internal reflection'. [Anyone remember 'The angle of reflection is equal to the angle of incidence'?] A kink or too tight a bend radius or worse a broken cable will totally disrupt the travel. It CAN happen, but unless the cable is broken, it doesn't happen in practice. Same with electrical connections - they CAN be susceptible to interference but in practice reject it just fine.

All this talk about jitter/rfi/emi is usually much ado about nothing in the short runs of typical ht cabling. Saying 'it either gets there or it doesn't' is actually close enough to the truth.

 

[tbewick]

I'm talking about the audible effect of jitter. Most of the fundamental concepts, like jitter, bit-depth, sampling etc. are quite easy to understand, but the effect of jitter in a real world system is very complex physically. How much jitter is tolerable has to be resolved using listening tests, objective and subjective. This is just like deciding how many bits you need or how many samples to take for the added distortion and noise (in sampling and quantisation) to be inaudible.

I would also say that commenting on the typical level of jitter (picoseconds) is meaningless unless you place that in the context of how much information you are sending. I would agree that such matters are trivial in regards to audible distortion but this is only based upon my own subjective observations.

Personally I like to understand the physical basis of most explanations, rather than just knowing 'digital is good'. This is probably the reason why people post threads asking whether optical is better than electrical digital etc. I would say it is more satisfactory to offer some technical explanation as well. Without having some physical understanding of the topic, it is very difficult to dispute the idea of needing to spend thousands on an expensive CD player, or of spending an excessive amount on cables. You can argue against these things using statistical explanations, like double-blind tests, but I believe some kind of physical explanation is also desirable.

 

[AudioSeer]

If these interfaces were not capable of transmitting the exact same data, bit for bit, regardless of interference or whatever else, then the $4 18" SATA cable connecting my hard drive to my motherboard would be incapable of copying a 400MB file without numerous disk and transmission errors, and the $20 optical cable connecting my Cisco PIX router with the T3 line at my office would be incapable of coherently transmitting our internet traffic.

Almost all digital communications like the ones you specify incorporate error correction to compensate for bit errors. The poorly designed S/PDIF interface does not to the best of my knowledge.

 

[tbewick]

The S/PDIF interface is quite similar to the professional AES/EBU interface, used in recording studios. This is a quote from Wikipedia:

'S/P-DIF is a consumer version of the standard known as AES/EBU; it provides small differences in the protocol and requires less expensive hardware.'

I've also heard criticism of Toslink on the Stereophile web site and on the Sound On Sound magazine web site. Sound On Sound said that bandwidth was reduced using Toslink. Stereophile said something about there being fundamental design problems in the interface. I would be interested to know what the audible effects of such poor design are in an objective (double-blind) test, which neither of these sources cited.

 

[tbewick]

Here's a quick diagram that I drew which hopefully shows more clearly what I was trying to explain earlier. The point I was trying to make was the importance of the digital system being able to reject distortion and noise, rather than being insusceptible to it.

 

[MDS]

I've also heard criticism of Toslink on the Stereophile web site and on the Sound On Sound magazine web site. Sound On Sound said that bandwidth was reduced using Toslink. Stereophile said something about there being fundamental design problems in the interface.

'Bandwidth' has two different meanings depending on whether you are talking about analog or digital. For analog, it is the highest frequency possible; for digital it is the amount of data it can carry over a given time period (the 'bitrate'). Sure the highest supported bitrate of s/pdif is lower than others but taking that quote out of context, one would assume that s/pdif somehow loses some of the information being transmitted. If you need to transmit 24/192 kHz uncompressed PCM, then it won't work, but that doesn't mean it 'reduces the bandwidth' - it's not like it will try and chop up the original data into smaller packets. It just doesn't work that way - it's not a packet switched network like tcp/ip.

That picture is nonsense. First, an optical connection has no analog component. The 'bits' are encoded using the same modulation schemes as electrical (coax) but they are pulses of light - not varying voltages or frequencies. I can transmit morse code 'digitally' by using a flashlight - short pulses for dots and longer pulses for dashes. That would be the equivalent of pulse width modulation. s/pdif uses bi-phase mark, which is slightly more complicated.

Everything can be 'susceptible' to problems, but they just don't occur over the short distances of HT cabling. You can build a null modem cable to transfer data from one computer to another using only 3 wires - receive, transmit, and ground. That wire would be susceptible to interference too and yet over the 3 feet it has to travel the number of errors encountered will be zero.

The Myths of Digital Audio are alive and well.

 

[tbewick]

All I can say is to refer you on to the book I read about digital audio, which was written by a respected author, and has received good reviews. If you were to engineer a system for sending digital audio over distances, then you would need to understand the properties of the cable and allow for distortion of the digital signal.

 

[MDS]

All I can say is that the book to which you refer as well as Principles of Digital Audio by Pohlmann are not a casual read (I've read them) and you have taken some simplistic, generalized statements and blown them out of proportion. Both are excellent books but require far more background in computer science, networking, and data communications to fully grasp.