DIYHiFi.org

With a proper terminated cable, the reflections are insemnificative. I don't think they won't change the jitter since they are a low level and they are happen after the main transition. The inputs of the SPDIF digital receivers are front-triggered, not plain level triggered. Typicaly what is 1ns out of that edge is ignored (at least that's what it is spec for DIR9001). For a regular RG59, propagation speed is 66-85% so the distance traveled in 1nS is some .7 feet. Both ways that means that a cable shorter than .35feet wold be MAYBE a problem - if is not adapted and reflection is really high.

Sure, you can calculate it.

SPDIF has a NRZ frequency of 2.8224MHz at 44.1KHz sample rate, which means a 0-1 or 1-0 transition can occur every 1/5.6448MHz. If your cable has a velocity factor of 0.9, the wavelength of the cable is 47.83 meters. Or lets say 48 for this example to make the numbers easy.

If your cable is a half wavelength (24m) then that's bad. A reflection will bounce off the receiver, then the source, and hit the receiver again right when the next transition is hitting the receiver. We'll call this the "least optimal length".

If you shorten your cable to 12m (1/4 wave), then the 1st reflection will hit the receiver exactly between the initial transition and the next. Which is optimal for reducing the effect of the 1st transition. However, the 2nd reflection will hit the receiver at the beginning of the next transition. The 2nd reflection should be considerably less in magnitude than the 1st, and much better than the fundamental, but you can argue that this isn't optimal either.

Shorten your cable to 8m (1/6 wave), the 1st reflection will hit 1/3 through the receiver interval, the 2nd will hit 2/3 through, and the 3rd will hit with the next transition. We're getting better once again... but we're completely missing the point. Which is to pick a cable electrical length that's relatively prime to the wavelength.

All of this being said, make the return loss on both source and transmitter better before you do anything else like this.

Useless prattle from idiots who have no conception of return loss.

Try measuring it sometime, before you go and run off to ask the same question at some other forum, populated by ignorant know-it-alls like this Lavry clown.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Which it usually is. How many pieces of gear out there have "proper termination"? Just sticking a 75R on the end isn't sufficient.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

70% is a good ballpark figure. Do GM's math now with that velocity.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

You see, this is why I said I needed the details because that's where the Devil is . So, let me summarise, one needs to select the SPDIF cable length based on it's velocity factor. Is it practical to so? This is much different to saying "longer cable is better for jitter reduction" as it patently isn't.

Jocko, that's a completely unhelpful answer, as you know I can't measure this, I don't own the equipment!!

Jocko, it would seem to me to be foolish & crazy to use a ball park figure & then hope you were avoiding the transition clash ??

_________________
There is a crack in everything, that's how the light gets in!

Most coax is in the 68% or so to 73% or so range.

Real companies put that in the spec sheet. It isn't hard to look up.

You may consider it a "completely unhelpful answer", but I can assure you that before the day is over, someone will spread this on some other forum, and the Google hits on my site will spike. It never fails. The same questions, over and over, on every forum. Even more "unhelpful" answers from people who write white papers and have crap to peddle, muddying things up with their misconceptions.

This is just like all the prattle about jitter: everyone talks about it, yet none of them mention if it is word clock, bit clock, whatever frequency. And zero mention of jitter frequency range, and whether it is Gaussian or data correlated. Without those stated, all the "You can't hear less than x nSec is crap!" statements are bogus.

So, you don't have the equipment, eh? I bet that you have most of it.

OK...........remember the "Make your own TDR" thread?

"No."

Well, to summarise;

Get a 'scope, '465 or better. Hook up your SPDIF transport thing to it with a short 75R cable, and a BNC "T". (It has to be short, because the "T" and the capacitance of the 'scope will muck up the waveform.) Get a long piece of the coax that you use for your SPDIF cable. You can use some RG-59 crap from Rat Shack for this. No excuses. Hook this up to the free end of the "T".

Now, here is the one part some won't have: a 75R BNC termination. Assuming that you do..........

Look at the wiggles on top of your waveform. Yeah, you have to calculate x ft. times z % velocity of propagation to know where to look. Try 73% if you don't know.

OK, now hook up your crappy RX to the cable, and look now.

Looks much worse, doesn't it? You know why?

RCA connector, instead of BNC. 35 or so ohms vs 75. Mismatch.

Mystery wire, twisted if you are lucky, to the PCB. Mismatch, >100R to 75 R.

75R termination resistor, maybe, at PCB. Mystery PCB traces to crappy Crystal RX chip. More mismatch. RX chip using Schmitt trigger inputs. More mismatch still.

Or, you can just take my word that your DAC measures like crap, and accept the fact that it has lots of reflections. Regardless of what some "expert" on some other forum will tell you.

I don't know of anyone, outside of Phred, who has spent as much time as I have measuring and quantifying this stuff. Probably includes all the manufacturers out there all put together. If any of them actually spent 30 seconds measuring their crap, then a I bet a lot of them would have done it right. But they don't, because they design audio stuff, and bits is bits.

What do you think all of those wiggles do to the waveform when the cable is 14" long, as opposed to 14' long? You don't need math to get the answer.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

I looked it up & the range for real cables seems to be 0.66 to 0.89 which would give a cable wavelength from 35.05 TO 47.27.

I posted here looking for some clarification about cable length - Gmarsh posted some useful info which makes sense & it looks like that from a jitter perspective. the statement that "longer cables are better" is just as mis-informing as Lavry's statement that "shorter cables are better"!

Jocko, you have done real world measurements - does your experience disagree with this? How?

_________________
There is a crack in everything, that's how the light gets in!

A 75 ohm coax, with 89% velocity. Show me one, before you make those claims. 93 ohms and the like does not count. Air-core, waveguide-like stuff does not count. Look up a typical RG-6, RG-59 type.

What part of "I don't know of anyone, outside of Phred, who has spent as much time as I have measuring and quantifying this stuff. Probably includes all the manufacturers out there all put together." did you need seem to understand?

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Jocko, yes you're correct 0.66 to 0.75 is the 75 ohm range.

I'll ask again in a different way - I'm not disputing that you've done a lot of measuring - are you still maintaining that "longer cables are better"?

_________________
There is a crack in everything, that's how the light gets in!

OK, let's do this in a way that you will actually learn something. After all, that is the point of this place.

'Splain to me any reason why either:

A.) Shorter is better, or
B.) Longer is worse.

Your choice; go at solving it from whatever perspective seems easier to defend.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

Did you read Gmarsh's post & my summary of it? It's all there - neither is better!

_________________
There is a crack in everything, that's how the light gets in!

Yes, I did, and you didn't say anything of merit.

No one has addressed the magnitude of the reflections. If you have a return loss of -40 dB on each end, then who cares how long it is?

Show me something............not made by me or any of my weird associates (including Phred)............ that comes close to that number.

Just like arguing jitter with no regard to its nature.

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"

But that is the best possible cable & impedance matching - is it not? How about dealing with real considerations rather than best possible cases? For the rest of us mere mortals cable reflections are a reality that don't become insignificant after 1 round trip! You said as much yourself a couple of posts back - most equipment (that you haven't designed ) has reflection issues that have to be dealt with!

What's this mean - you & your associates make the best cables possible - what does this add to the discussion?

_________________
There is a crack in everything, that's how the light gets in!

No, it means that no one who isn't an RF engineer can make stuff with a low reflection coefficient.

Since you aren't..............none of the folks who make the stuff you are likely to buy from are............., then, yes, you need to worry about reflections.

But you still miss the friggin' point:

If you have a 10% reflection on both ends, what is the magnitude of the first reflection? And the second?

You have to understand this before you get all worked up over cable length. In fact, you can forget about the cable impedance for the time being. Just concentrate on how the ends match each other. (Because they don't!)

Jocko

_________________
"Because, I hate you guys. I hate you guys so very, very much.

Yours,
General Cartman Lee"