[-=Germania=-]

I have discussed digital cables, especially optical cables with people who actually know and research optics.

One of which deals with optical communication, another who does optical/quantum computing, and another still who does computational electrodynamics. Please keep in mind that I have taken optics-specific courses that deal with relation to electronic devices and integration.

Their answers were all that in a standard spdif signal over relatively short lengths, there would be no mathmatical difference as long as the cable was "good enough" to get the signal there. It was a matter of bandwidth.

The best example of digital signals in an experiment/project done in the first electrical engineering course a student takes. They construct a system with a cheap CD player, a red laser, and an optical receiver hooked up to a computer. The students put the music disc in the computer and record the waveform of the song. Then they move it to the CD player and the system which puts the signal over the air, and record the waveform. Then they find the difference in the basic experiment which is usually not even computable by the Matlab program. After that, the students have to blur the signal with things like wax paper, plastic pieces, tissues, etc. Well, what the experimentation shows is that the difference in the reproduced waveform is that the signal does not noticibley degrade until it is close to shutting out completey.

Onto the audio.

Well, my better judgement and sanity said that the difference between a $20 Optical and $100 Optical was nothing in terms of an audio system (especially when conversion has to happen anyway). I was at the audio shop last weekend and saw a boxed high end digital optical cable in a box marked $10 each. For $10 what could I loose?

The truth is that the digital cable which retails for $100 sounded no different than the $25 thinner glass cable that I was previously using (dist. under the Onkyo Japan). Both of those were better than freebies I had gotten made out of cheap plastic which makes sense.

Would I bother to return it for $10? No. In fact, I have been too lazy to remove it from my current setup. So while it doesn't sound any different, I am keeping it in there only because it has a thicker jacket and a more sturdy connector just for piece of mind when re-routing or re-wiring my setup in the future. Personally, I would switch back to the premium monoprice optical cables, If I hadn't managed to get a kink in them.

My experience is that premium digital makes no difference from a decent cable and people who do it for a living say the same.

 

[obobskivich]

thoughtfully written and well explained

on an unrelated note: your avatar's eyes follow me, its kind of scary

 

[cswann1]

Good stuff. Being a Computer Science major and audio buff I've known for a while that anytime a digital signal is used, as long as the 0's and 1's get to their destination intact, that's all that matters. And a cheap optical cable does the job as well as a pricey one.

Hopefully this blog entry will help educate the public and keep them from spending money needlessly.

 

[Lazarus Short]

Germania's experience reinforces what else I have read on this subject - that as long as you avoid Toslink cable, you're good to go. I do remember reading about some optical coupling goo to put on the ends of the glass cable, but I have no experience with it. My experience has been with coax digital cable, and I sold my best ones long ago. I now use a Tara Quantum II solid-core interconnect for a digital cable, and I have no qualms about it. I recall reading about some test results which indicated that a wire coathanger made a fine digital cable.

Laz

 

[linuxworks]

good show

at first read, I thought you were going to say it MATTERED. even though I know it doesn't matter, I read on, anyway

$10 for mech improvements is worthwhile. no issue there. it would have been neat to borrow a $100+ cable and see if there were bit errors or jitter getting 'better'.

btw, any ideas on how to quantitatively measure bit errors and jitter? I'm building an spdif mux circuit and other than looking at how the squares 'go thru' this circuit, I'd love to know how 'bad' my jitter is. have any way to measure this?

/bryan

 

[linuxworks]

laz, you missed the point - you don't have to 'avoid' opto cable!

in fact, I prefer it. more flexible, it never not-works (coax can sometimes have not-so-great shield connections on some rca's) and its 100% interference (sending and receiving) free. run those right next to power lines - for miles - and no one cares!

opto is great! only downside is that consumers can't cut and terminate to length like they can for coax.

 

[Rhynri]

Well, LW, you can cut it....

Whether it is still usable or not, well, that is your problem!

^.^

Great post!

 

[-=Germania=-]

Well, with coax, you still have to deal with the usual problems with transmission lines.

Plus, you have to deal with coaxial cable.

For instance, the characteristic load impedence can vary by 3-5 ohms or more depending in length. Plus, that introduces problems with reflections in the cable itself.

Last week, I was in the lab doing testing on a decent quality resonably high speed 30ft coaxial transmission cable. The characteristic impedence was supposed to be 55ohm. In actuality, it was testing at 58 ohms with the connectors and even the other 30ft cables were ranging in terms of their calculations based on the cable reflections (tested with infinite load, 25ohm, 55ohm, and 100). Then, you need to account for the speed difference of electrons versus light and the skin effect on a coaxial line. Then, the higher bandwidth/thicker the cable, the more capacitance it causes on the line and more difficulty to drive. You need to make sure that the chracteristic impedance is consistant throughout and you have equipment capable to send a recieve well.

The advantage of coaxial over optical in stereo circuits is that there is virtually no coupling like in optical. That, and most DAC's and equipment doesn't run high speed enough for any of it to really matter anyway.

An optical connection is just plain easy for a high speed transmission line and has less that can degrade it since it is well, light.

In 5.1 and 7.1 though, an optical connection is a clear winner IMO since you are doing much more multimoding.

 

[sgrossklass]

A slow change in characteristic impedance shouldn't do too much actually. It's sudden jumps that are critical (if occurring in pairs, you get some nice resonator with the resulting combing in the frequency response). Still, a few ohms are not likely to do anyone much harm. Terminating a 50 ohm line with 55 ohms gives a reflection coefficient of <5%, which should be harmless.

For a careless mix of 50 and 75 ohm cabling, it's 20%, where it's more easily imagined that trouble may arise for our digital transmission if signals aren't too great otherwise.

Hams consider an antenna to be half-decently matched if at an SWR of 2:1 or less, which if I'm not mistaken equates to a reflection coefficient of 50% (or a 150 ohm load in a 50 ohm system). In receiving situations, one can frequently get away with fairly extreme mismatch. Then again, radio stuff tends to be narrowband, so the limit for tx matching really is the power reflected back from the antenna (the final amps usually aren't too fond of that and may fry).

skin effect --> file under dispersion (effectively).

Thicker coax usually means less specific capacitance, not more. This is easily seen when looking up data for some types. The thin stuff usually is at 100 pF/m or higher (thin stock RCA cable might hit 200), while thick coax of the same impedance may have less than 50 pF/m. It's not like that would be terribly surprising - specific L/C ratio defines characteristic impedance, and thin cables typically have higher inductance.

As for optical cables, it would be interesting to see the effect of dispersion on jitter. Clocking is the big weak point with consumer-level external DACs. (Typical ways of solving that problem are either having a master clock output on the DAC and input on the source so the source can sync, or hitting it with a hammer, i.e. throwing the most fancy clock recovery circuitry at it.)

Given a PLL for clock recovery, I'd really consider reflections, significant losses and low bandwidth to be more critical though, all of which may readily appear with optical connections. The LED-based TOTX-whatever transmitters alone don't go arbitrarily fast, and I don't think that the common plastic fibers are loss and dispersion champs. (Low bandwidth can also appear with coaxial outputs if the wideband transformer used is no good. Happened with some CD player back in the late '80s, there the optical out actually measured better.)

About the best thing one can do is look at actual recovered signal quality in the lab. Knowing the phenomena which may appear is one thing, but putting things into perspective is quite another.

 

[world-mart]

very inspiring.i want to buy a AKG headphone,but i don't know if it is worth for this priceoriginal AKG K512 DJ Headphones AKG stereo headphone...USD38.74 Wholesale Price at DHgate