[dvw]

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif ah, I just thought of another bit of possibly relevant data. my day job is computer networking. I first learned on 'thickwire' (as DEC used to call it) which is 10base5 ethernet. 10meg bits/sec with a real (!) collision domain. because of the idea of collisions on an ethernet and the fact that there are min cable lengths in the spec, perhaps -that- is another reason why I do believe there could be something to the 'too short' theory. if you think of a reflection as similar to another ethernet station randomly starting to transmit, the collsion causes a data error. fortunately, in ethernet they both detect this (and there was a JAM signal to extend it, too) and they both back off a random (diff) time and one gets to 'win' - in realtime spdif you have no such luxuries so you -have- to ensure that no reflection would be 'loud' enough to interfere with the decoding of that bit at the receiver side. is there any tests (anyone know?) that show when a transmitter sent data, that a receiver mistook the bit? surely those that say 'cables matter' would have demonstrated this with some kind of test config and even error injectors? anyway, in the old 'hub' based ethernet days, collisions did happen and having too short a cable -was- a violation of the spec. too short from end station to repeater hub and too short from hub to hub or hub to bridge and so on. spdif and ethernet share a lot of similar concepts, if you look at it. datacomm is datacomm, at the phy level, for quite a lot of schools

Ethernet is also my day job and I also worked on the standard. What is minimum distance you are talking about? The limitation of distance is actually have to do with collision domain. It is very different than reflection and jitter. There is also a jitter specification on Ethernet. In the jitter budget of the component design; jitter in cable is insignificant.

There are cases where you need a minimum cable length. This have to do with the pre-emphasis of the transmitter. This is usually in digital telephony or high speed backplane application.

I think I have gone too technical. Please PM me regarding the subject.

 

[Dan Lavry]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Thanks for taking the time for in depth responses. As for cable length, I know Jocko would disagree, so I guess there are differing opinions on that topic.

I saw it coming, your comments about different opinions. That is why I stated in the first post that the subject is not about opinions, but is about the rules of physics (I called it "Mother Nature").

I took the time to state and explain the physical principles.

Regards
Dan Lavry
Lavry engineering

 

[CodeToad]

I guess the ultimate stereo would be built on a single chip with telepathic transducers.

 

[IPodPJ]

Quote:

Originally Posted by wavoman /img/forum/go_quote.gif Some people say -- I have no idea if it is true -- that digital coax S/PDIF cables must be 1.5 meters otherwise they suffer a reflection (digital bounceback) problem than can muddy the sound. 4 feet is less than 1.5 meters, so maybe that's why optical sounds better? Or perhaps the adapter is causing a problem, but it's hard to see what that would be. But hey, if it sounds better to you, that's all that matters as long as you don't overspend.

Do you want to know why most cable manufacturers make those claims? Because they can't sell an optical cable for the price of your mortgage. There is no way for them to plausibly use different materials to aid in the filtering of the signal.

Yes, digital cables can and will make a difference in many systems but ever since I upgraded the clock in my DAC to the Audiocom Superclock 4-S (less than 2ps jitter) I cannot hear the difference between any coax or optical cable (and I can switch inputs on the fly). Prior to the upgrade the jitter of the stock clock was rated at < 50ps. Power cords still make an enormous difference to my ears, but this has nothing to do with S/PDIF.

So next time you get the urge to splurge on a digital coax cable, do some mods on your DAC first.

 

[linuxworks]

I found this, while searching for 'min cable length' in ethernet:

Ethernet: the definitive guide - Google Books

this one talks about thinwire (10base2). it says "min distance is 1/2 meter. this spacing acts to min the signal reflections caused by transceiver tap connections on the cable."

its been a long time since I've used old school repeated domains or even thick/thinwire; but I do remember it being said that the min distance was there for reflection purposes.

and while collisions aren't cable-caused reflections, in terms of data corruption they act just like it - its 'stuff' that is on top of actual data, possibly confusing the receiver circuit.

what would REALLY help is some scope photos of the reflection (and its bouncing back and forth, seeing the echos and seeing how they coincide with the data waves (proper) and also how the receivers do or do not pick the right bits out.

I do hear a lot of theory being talked about. where's the data/experiments to prove it?

saying "I can hear the diff" does not impress me in the least.

as the kids today say 'pics or it didn't happen' (lol).

surely, someone who believes this stuff has DATA on his side, not just formulas for ideas that MAY or MAY NOT matter in the real world.

 

[linuxworks]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif Do you want to know why most cable manufacturers make those claims? Because they can't sell an optical cable for the price of your mortgage. There is no way for them to plausibly use different materials to aid in the filtering of the signal.

when a vendor makes claims, its even MORE important that he show his hand and give backup data to support his claim.

a seller has an agenda to convince you that their secret sauce is what solves a given problem. the burden of proof is on them to SHOW (with data) that their before/after solution really does work.

most vendors simply wave their hands and leave it at that. because they have a financial interest, I really need to have MORE data from them, in fact, before I believe them.

 

[Pars]

Quote:

Originally Posted by Dan Lavry /img/forum/go_quote.gif I saw it coming, your comments about different opinions. That is why I stated in the first post that the subject is not about opinions, but is about the rules of physics (I called it "Mother Nature"). I took the time to state and explain the physical principles. Regards Dan Lavry Lavry engineering

DIYHiFi.org &bull; View topic - Parallel SPDIF outputs
I guess opinion may be the wrong word. I'm not an RF engineer, but when I see more than one come to the same conclusion, I tend to believe it.

The referenced link to TimesMicrowave is most likely this:
http://www.timesmicrowave.com/conten...mr/194-205.pdf

Regards,

Chris

 

[Dan Lavry]

Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif what would REALLY help is some scope photos of the reflection (and its bouncing back and forth, seeing the echos and seeing how they coincide with the data waves (proper) and also how the receivers do or do not pick the right bits out. I do hear a lot of theory being talked about. where's the data/experiments to prove it? surely, someone who believes this stuff has DATA on his side, not just formulas for ideas that MAY or MAY NOT matter in the real world.

Hi again

The data is available "all over the place". In fact, we all use a lot of hardware that utilizes those principles.

The first time I learned about cable termination and reflections was around 1973. I was a young engineer doing design work. The company was EH Research in California, and the product lines were state of the art high speed pulse generators. My work involved the state of the art high speed IC test systems (for IBM and Toshiba), and later time delay generators, all at fast rise times, all requiring understanding of reflections (and yes fast scopes)

The digital high speed technology of the time was ECL logic (Emitter Coupled Logic) made Motorola. The rise time of MCELIII IC family was around 1nsec, and I also used a lot of MECL10K which was around 3nsec.

Motorola had an application book. I think it was called "Designing with MECL" (or similar). Try to get hold of that book, it has all the material, and it is well explained. Also look for information about strip lines and microstrips, there is a lot of information regarding high speed layout that deals with transmission line reflections.

Say you want to generate a 1V 3nsec pulse; you can take a 1.5nsec delay line (such as a coax). Say the cable is 75Ohms impedance, then drive a 2V fast step into the cable through a series 75 Ohms resistor, and be sure that the far end of the cable is shorted.

(Voltage step) --- RESISTOR -- XXX..CABLE...XXX(SHORTED END)

The output is at the connection of the resistor and cable.

When the 2V step occurs, the cable initially looks like a 75Ohms pure resistor, so the circuit behaves as a voltage divider (by factor of 2). The step value at the output is 1V (half of 2V).

The step current wave will propagate to the end, and then the current will reflect back. The reflection coefficient is -100%. When the reflection gets back to the source, it sees a perfect termination (75 Ohms resistor to an AC ground - a voltage source is near 0 Ohms). The cable short at the end is seen at the cable beginning, so the voltage drops to 0V. That happens at 2 time cable the delay (a round trip).

Such circuits are used in gear very often. There are numerous other such similar circuits. Delay lines and gates are used a lot to generate pulses. Those are some of the most basic tools that designers use. And when we deal with high speed, coax cables and also traces on a printed board become very handy, both are transmission lines.

The electronic world has gone faster, and there is tons of literature about reflections and cable terminations, as well as gear (TDR) to measure such things.

Try to look for material about reflections, not about standing waves. It really is the same physical properties, but standing waves is more oriented towards analog signals, transmission, antennas... Reflections are more about digital transmission (voltage and current steps).

I have been designing gear for nearly 40 years, and what I said about reflections is an everyday thing for me. I do try to educate. In fact, I have lectured about cables and reflections, and I do have a document that generates "pictures" of the signal waveform given 4 parameters:
1. Cable delay time
2. Cable characteristic impedance
3. Source impedance
4. Termination resistance.

The document is prepared on MathCAD software, and the variables are controlled by means of sliders, thus the whole thing is not suitable for web and email. One needs to have the software.

Regards
Dan lavry
Lavry Enginering

 

[mmerrill99]

Somebody has pointed here from another thread when I was off-topic posting because Dan Lavry attacked my statement about longer SPDIF cables. I see what I said, quoting Jocko, was already referred to here as well.

Can anybody explain to me what he is saying here about SPDIF cable length - it doesn't make any sense to me - a lot of bold statements but I can't see the proof or logic of what's being said & some of it seems contradictory? Quote:

Quote: : Originally Posted by jkeny It is reported by Radio frequency Engineers that SPDIF cables needs to be 2 meters or longer on a properly terminated 75 ohm line, shorter than this leads to all sorts of cable reflection problems..... Edit Again: I should have also said that making the cable as short as possible will also work with non-ideal SPDIF (RCA connections etc.) - 12" is possible short enough That is not so, and no self respecting radio frequency engineer or any other electrical engineer will come up with such false claim. In fact, the shorter the cable, the better you are. I am not suggesting to use 3 inches cables, but a 3 foot is better then 10 foot, and at over 30 feet you are certainly asking for trouble. You said the reason for keeping the length at least 2 feet had to do with reflections. Reflections have to do with MORE LENGTH, not with less length! Reflection becomes an issue when the cable becomes LONG, making the signal propagation delay longer (the signal travel time from the “driver” end of the cable to the destination “end”). What does longer time mean? Longer with respect to the digital signal rise (and fall) time. A typical cable delay is around 1.5 nano second (nsec) per foot. The velocity is slower then the speed of light, in the range of 1/3 to 2/3 of the speed of light, and it depends almost entirely on one factor - the cable inner material isolation (the dielectric). The rise time for the digital signal is between 5nsec and 30 nsec. 30nsec is slow but still within the specifications. 5-15 nsec is nice, and the reason that faster is not allowed has to do with setting a limit on the electromagnetic radiation (transmission of interference). At say 10 feet, the cable delay is around 15 nsec, and a 5nsec rise time is 3 times faster then the delay, so one DOES NEED to terminate the cable and do so properly. But at say 8 inches length, the delay is around 1nsec and even a fast 5nsec rise is 5 times slower then the cable delay, and the signal will have virtually no reflections at all. The shorter the cable, the better it is from reflections stand point as well as from many other standpoints. I am not suggesting 8 inch cables. I am not suggesting not terminating. In fact, as a rule the termination is built into the receiver side. The issue here is cable length, and the notion that there is a minimal cable length one should keep is just plain wrong.

Now, I respect his standing in the audio community & the products he has developed so I presume he has experience in this area but I can't figure what he is saying or why the statement about longer cables is wrong in his eyes.

 

[mmerrill99]

Like other posters here, I can understand the reason for longer cables - the initial reflection arrives back after the digital signal transition is completed so there is no interference of the reflected wave on the digital signal.

I can't see him addressing this or disproving it in any way. In fact he seems to go to a lot of trouble to state the opposite but it doesn't make sense

He says this Quote:

At say 10 feet, the cable delay is around 15 nsec, and a 5nsec rise time is 3 times faster then the delay, so one DOES NEED to terminate the cable and do so properly. But at say 8 inches length, the delay is around 1nsec and even a fast 5nsec rise is 5 times slower then the cable delay, and the signal will have virtually no reflections at all. The shorter the cable, the better it is from reflections stand point as well as from many other standpoints.

but I thought reflections are caused by impedance mismatches (in the cable or at terminations) and NOT by the length of the cable? He also seems to be saying that longer cables need to be terminated properly but shorter cables don't?

I kinda think the problem lies in that my understanding of what are reflections is different from his definition? I believe what he means by reflection is when the back wave interferes with the digital signal transition? Might be wrong though?

In another post he defines reflections: Quote:

A reflection IS when a signal super imposes itself on a transition. In fact, when the cable is long and the rise time is short, you are likely to have a lot of reflections. The first reflection is a signal traveling back from the load to the source. The second reflection is from the source to the load. The third reflection is back from the load to the source…. And so on.

I did say this but got no reply Quote:

I believe this is contrary to the accepted understanding of a reflection - as a portion of the signal that bounces back along the transmission line (multiple times) & one of the main causes is when a signal encounters a variation in impedance on the transmission line. Is this not what you understand by a reflection?

Edit: Sorry he did post a reply: Quote:

Yes and that is what I said, and in greater detail. You have a reflection coefficient at the load end and a different one at the driver end, and the reflections bounce back and forth. For a single rise or fall, the reflections decay exponentially, and of course the driver may send a new rise or fall way before the previous reflections decayed, which causes a mess. The cure is to terminate the line with a resistor equal to the line impedance (such as 75 Ohms.

That is correct but not what he said originally

 

[mmerrill99]

Then he goes on to say exactly what I said Quote:

Reflection is a good view to adopt for the cases where the cable is long enough, and the signal rise time is fast enough to have the complete transition happen at one cable end BEFORE it ever got to the other end.

but this is the very thing he disagreed with in his first post

He didn't reply when I pointed this out to him

So rather than adding clarity as he stated, I believe he has added confusion.

Can anybody understand & state clearly what he said or is saying? The original post starts here http://www.head-fi.org/forums/6160539-post194.html

 

[mmerrill99]

Brought over from this thread: http://www.head-fi.org/forums/6167686-post242.html

Dan, I've done some research into this now (not one to let such contradictory viewpoints just co-exist - I have to find my own position) & my evidence indicates that you are wrong - you don't take reflection coefficient into account which is the loss in signal that occurs at each end of the cable. So even if you have a really bad cable with 10-20% of the energy reflected, you cannot have anything like the number of reflections you state - i you have a good cable where 1% is reflected then after 1 round trip, your reflection from 1 volt is down to 0.01volt!

Here's my reference from Wenzel which gives a non-mathematical explanation of a lot of RF issues SWR, Return Loss, and Reflection Coefficient

 

[zoiberg]

This is interesting reading, I'm glad it got moved over to here rather than just forgotten.
I hope Dan finds the time to answer these questions, theres quite a lot to chew on

 

[mmerrill99]

I don't think he will, there are just too many holes in his argument & self contradicting inconsistencies to defend. It only took me a minute to find reflective coefficient which he completely leaves out of any of his many different explanations and without taking this into account his argument is complete bunkum. I don't know what his game is but as someone who has worked on this for 40 years he sure misleads.

And I got castigated for challenging this over on the other thread because who am I afterall

. In fact slim.a has put Lavry's post on the first page of his review as an worthwhile explanation of digital cables

Some people, more than one, seem to suspend their critical faculties for whatever reason & don't seem to ask questions if they don't understand something. Easier, I guess to just blindly follow along without question.

 

[mmerrill99]

Ah I just see that he has posted on the other thread & wimped out saying that I can believe what I want - yes Dan I believe in Science, what do you use for your logic?

You wasted a lot of my time in reading your mis-information & wasted a lot of your time in writing it. I don't understand why - surely you knew that it was incorrect?

Here's a link to a Motorola paper (you mentioned Motorola in one of your earlier posts) you might want to read about transmission Line Theory - you'll see lots of explanation of termination & Reflection Theory