[zoiberg]

Its easy to see why people get confused

 

[mmerrill99]

You have tried to work your way through the many twists, turns, backtracks, self-contradictions & down right mis-information in Lavry's posts, I see

But all this is just Ohm's law & so "FACT", you know, not opinion, so you should just accept it because the laws of physics are incontrovertible !

I know let's make it a sticky

 

[haloxt]

Because of all your silly long vs short debate I tried to put my h120 optical out directly into my dac by using a cannibalized plug to mate them together but I cut the plug too short to reach into the dac. But I still bet you're both wrong, no cable is best!

 

[mmerrill99]

Quote:

Originally Posted by haloxt /img/forum/go_quote.gif Because of all your silly long vs short debate I tried to put my h120 optical out directly into my dac by using a cannibalized plug to mate them together but I cut the plug too short to reach into the dac. But I still bet you're both wrong, no cable is best!

All of this "silly" debate referred to SPDIF co-axial 75 ohm cable not to Toslink optical cable. But you're correct in one thing - no cable is the best i.e internal DAC (But then we would probably be using I2S signalling, not SPDIF

)

 

[mmerrill99]

At the risk of boring everybody, I want to post my findings, somewhat as a nod to Dan Lavry:

- short cables below a certain length work if the cable is of a length such that the reflections don't return to the receiver at or during the time of the decision point (so for a particular cable, termination resistance, rise time, cable delay time there is a length of cable below which there will be no transmission line (TL) effect)
- long cables also work if you calculate the cable length(s) that adheres to the above condition (so again for a particular cable, etc there are particular cable lengths for which there will be no TL effect - outside of these specific lengths, you will get TL effects)
- Both Lavry's & Jocko's views are wrong in that they are half-truths but Lavry's statement is more correct than Jocko's, I believe, as Lavry's works unconditionally below a certain length (for a particular cable) but Jocko's contention only works conditionally for longer cable i.e when the reflections don't fall on a receiver's decision point. At other long lengths it will be bad and cause jitter!

 

[Dan Lavry]

Quote:

Originally Posted by jkeny /img/forum/go_quote.gif At the risk of boring everybody, I want to post my findings, somewhat as a nod to Dan Lavry: - short cables below a certain length work if the cable is of a length such that the reflections don't return to the receiver at or during the time of the decision point...

You miss the point in 2 ways:

1. When the cable is short, the delay is short. If the cable delay is significantly shorter then the signal rise time, there is NO REFLECTIONS to worry about, so talking about when they happen is a non issue.

When the cable is long, so that the cable delay is around the same time as the signal rise time, you start considering reflections. In fact, when the cable delay (length) is much longer than the rise time of the signal, you must consider reflections and use good cable termination.

2. You seem to be stuck on the notion that a reflection time should not be during the detection time. But a reflection is not some "short blip" that only happens for a short time. The reflections ARE THERE DURING THE DETECTION TIME. It takes a very long time for the reflections to die out. They go back and forth between the cable input and cable output, and the effect of the reflection is not over for a while. The detection happens DURING the time when the reflections impact the signal.

In fact, with a longer cable, you are making the reflection round trips longer, so they will take longer to die out.

You have been sticking to those 2 wrong assumptions, I have tried to explain, taking much time and energy, and at the end you said I am trying to mislead.

It would be good if you can read this post carefully, realize that you were wrong, and remedy your attacks about my intentions.

And BTW, reflections is not an audio thing. It is about digital transmission of data over cables, and there is a a whole world out there doing so. And no where will you find that false claim that there is some minimum cable length that is better then using a shorter cable. Does it not make you wonder?

Dan Lavry
Lavry Engineering

 

[mmerrill99]

First let me say I PM'd Dan & apologised for being harsh in my comments & my only excuse was that it was born out of a frustration with his posts which I didn't understand & still don't after re-reading them a number of times. Also remember, Dan, was the one that started this whole discussion by attacking a post I made about long cables. I copied my last post above to him to explain what I've learned in the meantime.

But, I'm afraid, Dan, you are being just as confusing as before & I have to take issue with you again. I have a long post prepared to go but I thought I might post this short piece first & so avoid a point by point response.

Can I make a statement to see if you agree with this:
The problem arises when a reflection arrives at the receiver during it's decision window. This can be avoided by using either short cables (below a certain calculated length) or long cables (that are a multiple of a calculated length). Short cables below a calculated length will work. As we add length to this short cable the reflection will begin to interfere with the decision window until a cable length is reached where the reflection now doesn't arrive during the decision window? So this is a good cable! Keep adding length and this will occur again at another length & so on - so a good/bad pattern emerges.

 

[b0dhi]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif In fact, with a longer cable, you are making the reflection round trips longer, so they will take longer to die out.

The question seems to be: - is it more effective to use short cable lengths and attempt to minimise the reflections before the point of detection at the receiver, or use long cable lengths and attempt to delay the reflections until after the receiver has detected the signal transition, while hoping that the trailing reflections have died down before the next signal cycle.

 

[mmerrill99]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif The question seems to be: - is it more effective to use short cable lengths and attempt to minimise the reflections before the point of detection at the receiver, or use long cable lengths and attempt to delay the reflections until after the receiver has detected the signal transition, while hoping that the trailing reflections have died down before the next signal cycle.

This is one area where I believe Mr. Lavry is wrong - reflections are caused by impedance mismatches usually at both ends of the cable (where transmitter & receiver reside) so BOTH short & long cables will have reflections (it has nothing to do with length). That is the definition you will find everywhere!

So that quote Quote:

In fact, with a longer cable, you are making the reflection round trips longer, so they will take longer to die out.

is totally misleading. The reflections will die out based on the exact same number of round journeys they make up & down the cable so after 5 round trips (let's say) the reflections will have died out on both long & short cables. Whether those reflections affect the decision window at the receiver is dependent on whether they arrive there during the decision window.

So bodhi there aren't two different ways to do this - very high quality cables with well controlled impedances would have a reflection coefficient of -40db meaning only 1% of the signal is reflected at each cable end. So for a round trip the reflection would be 0.01% of the signal when it arrived at the receiver, the first time. The next time it would be 0.0001%. How many round trips before this level is insignificant? Perhaps Mr Lavry can tell us? This would apply to any cable long or short - completely irrespective of the length!!

Most of our cables aren't this good or the impedances inside the equipment that well controlled (crappy ones 10% to 40% reflection of signal) so the approach that is available to end-users is to try to avoid the reflection clashing with the decision window by using short or long cables (of a specific calculated length or lengths) as I tried to say above! How many round trips for this worst case 40% reflection coefficient cable before it's reflection level becomes insignificant?

 

[Dan Lavry]

Quote:

Originally Posted by jkeny /img/forum/go_quote.gif This is one area where I believe Mr. Lavry is wrong - reflections are caused by impedance mismatches usually at both ends of the cable (where transmitter & receiver reside) so BOTH short & long cables will have reflections (it has nothing to do with length). That is the definition you will find everywhere! So that quote is totally misleading. The reflections will die out based on the exact same number of round journeys they make up & down the cable so after 5 round trips (let's say) the reflections will have died out on both long & short cables. Whether those reflections affect the decision window at the receiver is dependent on whether they arrive there during the decision window. So bodhi there aren't two different ways to do this - very high quality cables with well controlled impedances would have a reflection coefficient of -40db meaning only 1% of the signal is reflected at each cable end. So for a round trip the reflection would be 0.01% of the signal when it arrived at the receiver, the first time. The next time it would be 0.0001%. How many round trips before this level is insignificant? Perhaps Mr Lavry can tell us? This would apply to any cable long or short - completely irrespective of the length!! Most of our cables aren't this good or the impedances inside the equipment that well controlled (crappy ones 10% to 40% reflection of signal) so the approach that is available to end-users is to try to avoid the reflection clashing with the decision window by using short or long cables (of a specific calculated length or lengths) as I tried to say above! How many round trips for this worst case 40% reflection coefficient cable before it's reflection level becomes insignificant?

You really should stop typing so much nonsense. This is pretty extreme!
Making up B.S. as you go along is not ok!

Dan Lavry

 

[Dan Lavry]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif The question seems to be: - is it more effective to use short cable lengths and attempt to minimise the reflections before the point of detection at the receiver, or use long cable lengths and attempt to delay the reflections until after the receiver has detected the signal transition, while hoping that the trailing reflections have died down before the next signal cycle.

You get around 177nsec between transitions of an SPDIF signal. The reflections (if there) go back and forth until they die out. If the cable is longer, it takes more time for the reflections to dies out, and a new transition shows up while the reflections are still active, and that will cause problems. The new transition is "riding" on reflections for previous transitions.

If you want to have the reflections die sooner, with a shorter cable, the travel back and forth is faster. Given that each time the reflection gets to the cable ends it becomes smaller, with a shorter cable the decay happens faster.

You too did not get the other point - if the cable is much shrter then the signal rise time, the reflection issue is not even there, because when the cble is real short, the signal gets there as it rises. That is why you do not terminate lines where the rise time is much slower then the delay. Examples: analog audio signals, DC, supply lines and much more.

Regards
Dan Lavry
Lavry Engineering

 

[mmerrill99]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif You really should stop typing so much nonsense. This is pretty extreme! Making up B.S. as you go along is not ok! Dan Lavry

Why don't you try to pick apart the points rather than make blanket statements like this - it would serve you better!

 

[mmerrill99]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif You get around 177nsec between transitions of an SPDIF signal. The reflections (if there) go back and forth until they die out. If the cable is longer, it takes more time for the reflections to dies out, and a new transition shows up while the reflections are still active, and that will cause problems. The new transition is "riding" on reflections for previous transitions. If you want to have the reflections die sooner, with a shorter cable, the travel back and forth is faster. Given that each time the reflection gets to the cable ends it becomes smaller, with a shorter cable the decay happens faster.

I agree with this but how many round trips before the rfelection strength becomes insignificant - you haven't answered this!

Quote:

You too did not get the other point - if the cable is much shrter then the signal rise time, the reflection issue is not even there, because when the cble is real short, the signal gets there as it rises. That is why you do not terminate lines where the rise time is much slower then the delay. Examples: analog audio signals, DC, supply lines and much more.

We are talking about digital cables not analog or even pcb traces so address the topic that's being discussed & don't veer off please!

 

[haloxt]

Well both of you agree no cable is better so can't we just agree on that?

 

[DanielCox]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif You really should stop typing so much nonsense. This is pretty extreme! Making up B.S. as you go along is not ok! Dan Lavry

What are you talking about? That IS the reality of it.