TOSLink Cables All the same? Any degradation over Length?

Status
Not open for further replies.
Jump to Latest
#21
hifiZen is right. TOSLINK is inferior to coax, the interface is highly unreliable, and it is not single-mode (it is multimode).

ANY transmission method can be plagued by poor interface and/or impedance irregularities. Fiber is especially susceptible in short lengths as too much energy is reflected back from the receiving end. This causes the light source (which in the case of TOSLINK is really cheap) to do all manner of wierd things. The end result is pulse distortion, which then translates into jitter.

S/PDIF is based on AES/EBU which is a real-time digital transmission protocol for use in studios, often over high quality balanced cabling.
Click to expand...

How to make a bad idea worse.........

The original version of this kludge allowed for, I believe, something stupid like 3 transmitters and up to 7 receivers on a single line. Obviously, no one is this group knows squat about transmission lines. If they did, they would not have chosen an XLR connector for a 110 ohm transmission line. Nothing high-quality about how this is done. The cable may be nice, but on the whole, it stinks.

And you wonder why recordings sound terrible these days.......

Jocko
 
#22
Re: Error correction? What error correction?

Daryl Low said:
The S/PDIF format has at best a parity bit. This is *NOT* an error correcting code! The big difference between S/PDIF and other "reliable" digital protocols is that S/PDIF is unidirectional with no flow-control.

S/PDIF is based on AES/EBU which is a real-time digital transmission protocol for use in studios, often over high quality balanced cabling. With that in mind, the protocol designers decided they could sacrifice error detection / recovery for better real-time response. After all, studios could better afford the expensive cabling and equipment.
--------------------------------------------------------------------------

Whatever the specs, AES/EBU cables also sound different; nearly as much as SPDIF. As does 75R BNC connections! I2S seems to be better although there are not many I2S (enhanced) cables.
Click to expand...
 
#23
I would hope I2S would sound better, as one does not have recover a clock from a serially encoded pulse train loaded with interface problems.

And speaking of 75R BNCs.....let me say it again.....for those who might be new....

There is no such thing as a 75 ohm RCA. Anyone want to prove me wrong?

SPDIF with proper 75 ohm cables, connectors, and terminations is the only way this format can sound good.

Jocko
 
#26
Jocko, some time ago in some other thread you mentioned that the receiver end on the Toslink can also be much improved compared to the standard way thay are done. Can you give some pointers of what you had in mind....

I know it's not the best way to transoprt a signal but I'm planning a new DAC in my head and Toslink would be used only for signal coming from PC. There I think it's not too bad as it decouples the grounds and PC's is a very noisy one.

Ergo
 
#27
Schmanthony said:
Hi,

I'm more than willing to abmit I don't know what I'm talking about...
Click to expand...

Ok, I know very little in this matter scienticly. I suspect that much of what has been said here are opions, nothing more. Does optic fibres really have problems with reflexions? Which factors are really degrading for the sound so you can actually hear it?

One thing is for sure: There are differencies in mechanical strength. A pro has gone mad with TOSLINK with it's tiny plastic also the short transmission length. There are also differencies in electrical characteristics but it's very seldom reported.

Discussing specific components and particular cables creates always hot debates. I wonder why? Also there are lot's of people out there who wants to make money of this which I don't like.
 
#29
peranders said:


You probely very right since RCA plugs originally not was intended for mathced impedance situations but, hey, how important is this fact really when we talk about rather low frequencies (low bit rate) and short mechanical length of the plug?
Click to expand...

What is the transmitter rise time? 4ns? 0.5/4ns=> frequency content to 125Mhz, regardless of bitrate. I don't know how bad the mismatch is to calculate the reflection...any figures from you Jocko?
 
#30
Jocko Homo said:
I would hope I2S would sound better, as one does not have recover a clock from a serially encoded pulse train loaded with interface problems.

There is no such thing as a 75 ohm RCA. Anyone want to prove me wrong?

SPDIF with proper 75 ohm cables, connectors, and terminations is the only way this format can sound good.

----------------------------------------------------

The issue is: what is good? In my experience, RCA terminated cables can sound good when compared with the same cable terminated in BNC. Some very expensive and highly touted cables consistently sound wrong over a range of systems and terminations. Time the industry sorts out the interfacial problem. There is enough money in it!
Click to expand...
 
#31
Okay, hope I can shed some light (before being shot down in flames). I am attaching below a forum article I put into Home Theatre Spot forums, but to add....

1) Multimode, single mode, etc. are going to make absolutely no difference in a 1 or 2 metre run. There is so little degradation of the light travel down this short of link that whether you had multimode plastic, single mode glass, multimode glass, you would not see any difference.

2) The fact that light travels down multiple paths in the cable is going to make no difference. We are talking 1-2 meter lengths typically and travel differences on the order of centimeters at 300,000,000 meters per second.. hmmmm.

Implementation is everything.....

The cheap LEDs and photodiodes used in TOSLINK, and even arguably in the ST type connections are the biggest contributors to signal degredation. The jitter is not only very easy to measure (dead simple in fact! at these data rates and the amount of jitter), but highly audible depending on how well the DAC performs. Most receiver chips out there, Crystal, ADI, TI, do attentuate a lot of the jitter, but the best I have seen still have what would be considered measurable, and audible distortion. DACS as well have mechanisms that attempt to attentuate distortion, but getting to some of my other posts, they are very good at reducing average distortion, but not instantaneous transient irregularities.


Sorry, this was a response to another forum post, so we have "interesting thoughts" raised by another forum member interspersed in.

Originally posted by ciper:
[QB]The argument made about jitter and buffering does not agree with me. Conversion from electrical to light does not magically add noise to the signal. This type of conversion can be done with less than 5 dollars worth of parts, doesnt matter how many cool features you may add.

*** Actually, it does "magically" add noise. When you convert photons (light) to electrical, there is an inherent noise source call "shot noise". Shot noise is proportional to the square of the light signal level. Hence, it is a noisy process.

"Straight electrical transmission over co-ax, even using cheap components is prone to much less jitter, especially with the typical cable lengths in home theatre equipment. That is why it may often sound better."

These links arent low speed so much as they are low bandwith. The signaling speed may be different but sync errors should never come into play. If anything degradation of the signal will occur tens of thousand of times sooner when using coax compared to optical. Of course you would have to have 100 foot cable runs or more for this to be an issue. With this in mind you could still use relativly inexpensive fiber for a 300 foot run but a pretty costly coax would be needed for the same run in order to preserve the signal. Repeaters may even be needed.

*** I could have been clearer here. Bandwidth would have been a better term. And I did not want to imply that practically you could do a really long cable run better than optical. I was thinking more of the 6 foot lengths or less in home theater.

*** So I did some diging. Toslink is pretty cheap stuff. I picked some significantly better quality parts from Agilent, HFBR-14x4 designed for 20 megabit/sec links. The stated link jitter -- typical -- so the max would be worse is 7 nsec. These are much better quality parts than your standard TOSLINK connections in a stereo. I am working off memory, but I think TOSLINK jitter can be 3 times worse or more depending on who designed it. You are talking parts with rise times on the order of 50 nanoseconds. Depending on how the co-ax link is implemented (often a logic buffer as a driver and a comparator as a receiver or similar - buried in an IC somewhere probably from Crystal semi or Yamaha), you are typically looking at about 1 nsec of jitter. I am going off memory on the formula, but I believe 20 nsec of jitter (like toslink link) will result in only about 70db of signal to noise at 10KHz. I am probably off by a few DB, but the S/N loss is significant. So why does it not sound terrible. The receiver chips have PLLs in them that maintain as stable a clock as they can off the input signal. However, the less jitter on the input, the less jitter on the output.

Here, Ill use the logic found in some other posts to PROVE THAT OPTICAL IS BETTER.

[a$$ mode=on]The signal from the optical connection travels faster than the signal over copper. Since the optical signal cant be muddied or modified in any way during transmission this is like having the two devices directly connected. On the other hand not only does the digital electrical connection have a weakness against interfearance it can also act as an antenna and draw more interfearance into your components. With this added noise the curcuitry that changes the digital signal back into an analog signal has to work harder.

*** Over these short links, the issue is not so much the transmission as conversion from electrical to optical to electrical. The motherboard on your PC and the associated power supplies are phenominally noisy. Still, signals still move around with jitters on the orders of picoseconds. Fast edge rates generally results in low jitter rates. While we are talking cable, at 100MHz, over 6 feet, an electrical signal will only be attenuated by about 3% with good quality co-ax.

Another reason optical is better is voltage levels. The device transmitting this signal has to flip flop a voltage on the line in order to create this signal. This flip flopping is much higher than what an audio amplifier would EVER see. Most devices that generates this digital stream are fairly inexpensive (500 or less) so the quality of these voltage swtiches may be an issue. What if the voltage only reaches 80% of the peak? Or what if its 120%? With an optical signal the data is transmitted as a literal off and on. No matter what brand or type of the cable is the light will either be off or on. The component producing this light does not easily have a variable brightness. There is no way for it to produce less light. Its like trying to dim an led. For example, sophisticated curcuitry (compared to just using less voltage for a normal lamp) is need to dim a normal led.
[a$$ mode=off]

*** Had to give my head a shake on that one. You need to "flip" a signal around to drive a LED, which is exactly what is driving that TOSLINK cable. A LED's output is directly related to current level. That LED is probably driven by a voltage through a resistor. While you talk about that voltage being 80% low or 120% up, the likelihood is it may be a bit low, but not likely high. However, in either case, the voltage level makes no difference as long as it is "enough" what will be of more interest is how fast it changes, which will be much faster than the optical driver and receiver. And, if that source driver for the LED is changing in voltage, then so will the output of the LED. It is very easy to change a LEDs brightness in fact. All you need to DIM a led is to change the current which is done by changing the voltage or resistance if you have a simple circuit. You could not get any more easy than that. However, lets talk about the optical variables. Not all LEDS produce the same amount of light for the same current. Not even close, just look at those LED TV displays. The optical interface from the LED to the cable and the cable to the LED will be different from unit to unit, and the phototransistor or photodiode that converts the light to electrical will have huge differences from unit to unit. I can work it out if you like, but I think you will find the variation from system to system for optical pretty large, and certainly much worse than a 6 foot co-ax.


Personally like I have said before, I use both and both are great. Its really the PHYSICAL properties of your setup that will help you decide what to use.

What I was thinking of doing is going down to frys and purchase ALL THE OPTICAL TO COAX AND VICE VERSA. I would convert the signal from the dvd player back and forth as many times as I could and see if there was a sound difference. Even with these el cheapo products im willing to bet the difference will be non existant.

*** I am not stating that co-ax is a better transmission media than optical, though for short distances it often is. You could easily develop an expensive link that was very very good, and you would notice absolutely no degredation. However, the standard used on audio equipment is low speed and prone to jitter. Good audio equipment goes a long way to reducing, or possibly eliminating it from being noticable.
Click to expand...
 
#32
alvaius said:

*** So I did some diging. Toslink is pretty cheap stuff. I picked some significantly better quality parts from Agilent, HFBR-14x4 designed for 20 megabit/sec links. The stated link jitter -- typical -- so the max would be worse is 7 nsec. These are much better quality parts than your standard TOSLINK connections in a stereo. I am working off memory, but I think TOSLINK jitter can be 3 times worse or more depending on who designed it.

Transmitter jitter in the Crystal parts is about 1ns nominal (pretty much what you said), possibly less for a buffered driver. Over a good coax link with all impedances matched and a good reciever, I think the jitter added by the link is on the order of hundreds of picoseconds at worst. So optical is easily 3-10 times worse in terms of jitter. With transformer-coupled links ground loops are a non-issue, and we can certainly provide impedances of 1k or more between reciever and DAC circuitry to abate the noise issue.

Another reason optical is better is voltage levels. The device transmitting this signal has to flip flop a voltage on the line in order to create this signal. This flip flopping is much higher than what an audio amplifier would EVER see. Most devices that generates this digital stream are fairly inexpensive (500 or less) so the quality of these voltage swtiches may be an issue. What if the voltage only reaches 80% of the peak? Or what if its 120%?

I know these are not necessarily your comments, but I'd say this is exactly why we use a comparator at the input. Very tolerant of different levels.

Another thing to consider is that light is not necessarily all that much faster. I do not remember how fast transmission is in 75 ohm coax, but I believe it is 0.67c, which is not really significantly slower over these distances.
Click to expand...
 
#33
Well, have we beat the Coax vs TOSlink issue to death yet?

Schmanthony, In reply to your post, I'd say that you probably won't hear a difference from 3' to 50', and no TOSlink isn't lossless, and I'd bet that shielding is a marketing gimmick. Not to say that they aren't shielded, but to what advantage?

When in Rome, do as the Buddists do, and follow the middle path to enlightenment (or to the Coliseum). Get a TOSlink cable that has good fit and finish - not too cheap but not too expensive - and you'll be on your way to bliss.

On another point:

I will never argue subjective points about what somebody heard or didn't hear when they switched to expensive cables, but I can say that a friend of mine and I did an ABX test between Coax and TOSlink. We both failed miserably at being able to determine - by hearing - one cable from another, and I would say that we both have experienced ears with many different systems over the years. A lot of what people think they hear is psychological.

Unless you're an equipment reviewer, and change your "reference" equipment more often than you change your underwear - or if you were captain of your high-school debate team and love to argue - leave the bloody cables alone, stop bickering and just enjoy the music!

Well, at least once you've got some decent cables... 🙂
 
#36
Co-ax is inherently better than optical in almost all systems. Yes, I could create an audio component that makes co-ax worse, but it would be tough. On the other hand, I could also make a system that would make the two of them sound exactly the same. If you have a real good DAC on the end of either, I would not expect to be able to tell the difference. On an older Denon AV receiver with a JVC DVD as the source, the difference between the co-ax and toslink is astounding.

Alvaius
 
#37
Jocko, some time ago in some other thread you mentioned that the receiver end on the Toslink can also be much improved compared to the standard way thay are done. Can you give some pointers of what you had in mind....
Click to expand...

OK.......today's lesson.......

Suppose you were building a jet fighter that you wanted to make invisible to radar. What is one possible way to do so??

Absorb the radar pulse, and therefore reduce the reflection.

Works the same way on transmission lines.

Jocko
 
#38
What are we talking about really (now)?

Most SPDIF signals are generated with some kind of transmitter. I'm sure we all can agree about that. This generator is probely like CS8404/04/05 which has jitter in the ns area. How much more jitter does an optical device produce? How much jitter can we stand feed into a DAC?

How much does poor cables (optical/coax) interfere with the transmission?

I think (no harm in that) that we have lots of opinions here. I myself have tested a 1 m nylon fibre and a 25 meters long. Can't here any difference but the light is much weaker in the long fibre, still I'll get my ones and zeros.

To answer to thread starter:

Yes, you do get degradation over length. You come to a point where the "ones" gets too weak. The receiver will not recognize them as "ones". The degradation is rather "digital". Perfect transmission until you get to a "grey" area (can be rather long) and beyond that no signal at all.
 
Status
Not open for further replies.
Top Bottom