Topic: Digital Cable Minimum length?

[Roving Sign]

I remember some high-end audio type of argument about making digital coax cables too short?

I was going to make some shorties for a stealth bag...any reason not to make some as short as 6 inches...?

[cshepherd]

Shorter is generally preferred when it comes to cable lengths, with the exception of a/c mains cables that employ filtering properties.  I've never heard of minimum length requirements for spdif cables.

Chris

[cshepherd]

OK, it seems I may be wrong on this one   .  UHF in Canada (http://www.UHFmag.com) aparently took the digital cable length issue to task.  Apparently, there is a reflection that gets sent down the wire when the first bit is transmitted from the source.  When the cable length reaches 1.5m, the first bit reaches its destination before the reflection...and this is good.  It was a big enough difference that UHF is going to phase out digital cables at lengths shorter than 1.5m.  Our Atlas sales rep told us about this today, who is also UHF's Atlas rep (they have a retail outlet and online store as well).  He took some Harmonix spdif cables in varying lengths home to check this out for himself.  These cables cost about $1,000 each.  He said he was completely shocked at how much better the 1.5m cable sounded over the 1m cable.  So much that May Audio (our distributor) is returning all of their spdif cables under 1.5m to the manufacturers.

Just thought I would pass along the info.

Chris

[JasonR]

It sounds like this is an argument against using optical cable shorter than 1.5m.  Is that the case?  I wouldn't imagine there's any such 'reflection' in the world of coaxial digital, unless you believe in the magic goblins that eat jitter and such.   

- Jason

[Roving Sign]

Thanks - that sounds very much like the info I was remembering...

1.5 M - thats pretty long! - like about 5ft right?

I wonder if this is as critical recording a signal as it seems to be feeding a DAC?

[cshepherd]

It sounds like this is an argument against using optical cable shorter than 1.5m.  Is that the case?  I wouldn't imagine there's any such 'reflection' in the world of coaxial digital, unless you believe in the magic goblins that eat jitter and such.   

- Jason

This was definitely about a spdif connection.  Feel free to keep using your imagination though.   

Chris

[rodeen]

This was definitely about a spdif connection.

So what you are saying is any SP/DIF connection whether it is optical or coax?
I guess it's time to get longer cables.  I should go read the article, but did it say
what the issues were?  You talked about reflections but I'm not sure what that means.
Do they cause jitter or is it an issue like standing waves in electromagnetic signal that
causes data loss? 

Interesting topic!

[JasonR]

This was definitely about a spdif connection.  Feel free to keep using your imagination though.   

Chris

SPDIF can be coaxial or optical.  My point was that if the article talked about "reflections" being a problem, that sounds like an optical issue.

- Jason

[MattD]

Reflections can occur in coaxial cable too. From my experience, it's caused by an impedance mismatch, but apparently there are other mechanisms.

[cshepherd]

The topic came up over Atlas cables, and they don't make optical cables.  UHF does a lot of business in Atlas so it was news when they asked to return all their 1m digital cables.

Chris

[cshepherd]

Here's the review.  The whole issue is in one pdf file, the review is on pg. 24.  UHF used two Atlas digital interconnects, the Compass ($93) and the Opus All Cu ($278), in both 1m and 1.5m lengths in a blind comparison.  The differences were more dramatic in the cheaper interconnect, but the Opus All Cu also benefitted from the extra length.  They weren't talking about splitting hairs either.  It's an interesting read.  It does a pretty good job of explaining how 1's and 0's can actually have trouble with signal integrity.

The theory was published some time ago in Positive Feedback, written by Steve Nugent of Empirical Audio.  The readers digest version is that the longer cable gives the DAC more time to sample the data ('transmission" as it's referred to in the article) before its reflection hits the DAC.  The reflection travels from the DAC to the transport and then bounces back to the DAC.

http://www.uhfmag.com/Issue74/UHF74.pdf

Chris

[SparkE!]

When cables are connected to a device whose input impedance is the same as the intrinsic impedance of the cable itself, there will be NO reflections from that end of the cable.  S/PDIF is supposed to be 75 ohms, so if you are using 50 ohm cable, you're going to get reflections.  If you are careful to use 75 ohm cable and 75 ohm connectors and your S/PDIF equipment is properly designed with 75 ohm port impedance, it doesn't matter how long the cable is, except that the cable will attenuate the signal.  Accepted design practice is to limit S/PDIF cables to less than 10 meters in length.  Misinformation like Nugent is spreading is usually based in some tiny morsel of truth, but the fact is  that if the bits get through the cable unaltered, a properly designed playback system will not be affected by the cable length of the S/PDIF cable that feeds it.  The only way that Nugent's argument has real life merit is if you are using cables of the wrong impedance.  In that case, length does matter.

When the cable length reaches 1.5m, the first bit reaches its destination before the reflection...and this is good.

Let's do a little math...  48kHz times 16 bits times 2 channels is 1.536 Mb/s.  Each bit, then, is about .65 us long. In .65 us, the pulse could travel about 195 meters at the speed of light.  The speed of propagation on a cable is roughly 2/3 the speed of light, so we're talking a cable that is about 130 meters long if the cable is as long enough to hold a whole bit at once.  The cable would have to be half that long for the bit to be reflected back to the input at the time the next bit starts to be transmitted.  So, from your description of the cable, we'd need a cable that was longer than 65 meters!

Remember, when someone is trying to justify selling $1000 S/PDIF cables, they're going to have to resort to fairy tales in order to sell any of them.

The bottom line is if you use the correct cable impedance, it does not matter how long the cable is, except that the cable cannot be so long as to unacceptably attenuate the signal it carries.

[cshepherd]

Nobody was selling $1000 digital cables.  The cables UHF used to confirm validity of Nugent's claim awere Atlas' entry-level and top-end digital ables ($93 and $278 for 1m).  Our rep didn't believe UHF either so he grabbed the most expensive cables they carried to do a similar comparison, which were from Harmonix.  Actually, the 1.5m cable he used sells for $1700.  Different cables, different ears, same results.

Nugent goes through some different math so I'm just going to copy the text from the article.

Chris

"There is a short but finite time between the initial rise of the signal voltage and its arrival at full voltage. The receiver circuit, at the front of the converter, must interpret the signal, and “decide” when it will consider that a “1” has arrived. That interpretation presents a potential for an error in timing, unless it is absolutely consistent from one data block to the next.  In fact, it may not be.  Because neither the circuits nor the connectors nor the wire are perfectly matched for impedence, some of the energy arriving at the converter will actually bounce back toward the transport, from which it will bounce back again. Thus the converter will receive the imperfect square wave, and its echo. Now we have potential for confusion. Here’s Nugent’s take:

When a transition is launched into the
transmission line, it takes a period of time
to propagate or transit to the other end.
This propagation time is somewhat slower
than the speed of light, usually around
2 nanoseconds per foot, but can be longer…
When the transition reaches the end of the
transmission line (in the DAC), a reflection
can occur that propagates back to the driver
in the transport. Small reflections can occur
in even well matched systems. When the
reflection reaches the driver, it can again
be reflected back towards the DAC. This
ping-pong effect can sustain itself for several
bounces depending on the losses in the cable. It
is not unusual to see 3 to 5 of these reflections
before they finally decay away.
So, how does this affect the jitter? When
the first reflection comes back to the DAC,
if the transition already in process at the
receiver has not completed, the reflection
voltage will superimpose itself on the transition
voltage, causing the transition to shift in
time. The DAC will sample the transition
in this time-shifted state and there you have
jitter.
If the rise-time is 25 nanoseconds and the
cable length is 3 feet, then the propagation
time is about 6 nanoseconds. Once the transition
has arrived at the receiver, the reflection
propagates back to the driver (6 nanoseconds)
and then the driver reflects this back to the
receiver (6 nanoseconds) = 12 nanoseconds).
So, as seen at the receiver, 12 nanoseconds
after the 25 nanosecond transition started,
we have a reflection superimposing on the
transition. This is right about the time that
the receiver will try to sample the transition,
right around 0 volts DC. Not good. Now if
the cable had been 1.5 metres, the reflection
would have arrived 18 nanoseconds after
the 25 nanosecond transition started at
the receiver. This is much better because
the receiver has likely already sampled the
transition by this time.

[Brian Skalinder]

Okay, so it seems this is an explanation of one potential cause of jitter.  It seems a simple buffer would solve the jitter problem.  Do not quality DACs have a buffer in which to store the 0s and 1s as they arrive, re-clocking them to the proper timing upon playback?  Seems (to me) like a simple solution, regardless of the cable used.

[Roving Sign]

It seems like they are counting on this:

"He said he was completely shocked at how much better the 1.5m cable sounded over the 1m cable.  So much that May Audio (our distributor) is returning all of their spdif cables under 1.5m to the manufacturers."

...to bolster their argument and make the sell - as much as any real eveidence.

I'm always super skeptical about the "sound" of digital cables...as longs as it locks - its working.

On the other hand - I think DATs can sound REALLY wierd - especially on these old Sony D7/D8s...its like the signal is working but sounds unfocused. When you get them transfered on a full size machine - they seem to sound much better...

[SparkE!]

Okay, so it seems this is an explanation of one potential cause of jitter.  It seems a simple buffer would solve the jitter problem.  Do not quality DACs have a buffer in which to store the 0s and 1s as they arrive, re-clocking them to the proper timing upon playback?  Seems (to me) like a simple solution, regardless of the cable used.

Ding, ding, ding, ding! We have a winner!  Yes Brian, properly designed playback systems don't use their input clock do directly drive their sample output clock.  Jitter on the input does not directly translate to jitter on the output, though input clock jitter can affect your circuit's ability to create a low jitter output clock.  A well designed output clock circuit will average out most of the received jitter by using a phase locked loop whose loop bandwidth is many, many clock cycles.

Something else to remember... If reflections from an improperly terminated cable actually come back during the rise time, they will serve to hasten the rise of the signal.  Not only that, but they will hasten that rise by exactly the same amount of time every time.  (The cable is not changing lengths from bit to bit, nor is the speed of propagation on the cable changing.)  So even if there is an effect on the rise time, it's the same for each and every bit transmitted.  That simply cannot cause an increase in jitter.  Remember that jitter is a variation in bit timing.  Since the effect is always the same, there is no increase in the variation in bit interval that is due to cable reflections.

And again, my main point in my previous post is that THERE ARE NO REFLECTIONS if your cable impedance is the same as the input impedance of the electronics that are connected at the destination end of the cable.  It simply does not matter how long the cable is unless you have an impedance mismatch.

[cshepherd]

I don't think improperly designed playback systems or cheap dacs are really the issue here.  The claim is that it's impossible to have signal path components that are perfectly matched for impedance...and that reflections have to occur because of this. 

In the end, no matter which theories get debated, the 1.5 meter cables rendered a better sound when two lengths of the same cable were compared.  This same result happened with three different quality cables.  I think it's pretty interesting.

Chris

[Brian Skalinder]

I don't think improperly designed playback systems or cheap dacs are really the issue here.  The claim is that it's impossible to have signal path components that are perfectly matched for impedance...and that reflections have to occur because of this.

Agreed, that's not the point of the article.  But properly designed playback systems (mainly DAC) are a solution to jitter.  Properly designed DAC = doesn't matter whether your cables length / brand / composition minimizes jitter.  So why worry about cables (other than if you want to sell them), better to address the problem with a broader solution.

[JasonR]

In the end, no matter which theories get debated, the 1.5 meter cables rendered a better sound when two lengths of the same cable were compared.  This same result happened with three different quality cables.  I think it's pretty interesting.

Chris

Double-Blind A-B-X testing, of course, right?

- Jason

[cshepherd]

In the end, no matter which theories get debated, the 1.5 meter cables rendered a better sound when two lengths of the same cable were compared.  This same result happened with three different quality cables.  I think it's pretty interesting.

Chris

Double-Blind A-B-X testing, of course, right?

- Jason

Yeah, three people at UHF Magazine testing two lines of Atlas cables ($93 & $278 for 1m), testing both with 1m and 1.5m lengths.  One person in the know, two people are not.  The more expensive 1.5m Opus All Cu wound up staying in their reference system.

Our Atlas rep did a comparison with Harmonix digital cables (~ $1700 for 1.5m) at the same lengths upon hearing about the Atlas comparison.  He came to the same conclusion.  Both UHF and our Atlas distributor are returning all 1m digital cables.

Okay, so it seems this is an explanation of one potential cause of jitter.  It seems a simple buffer would solve the jitter problem.  Do not quality DACs have a buffer in which to store the 0s and 1s as they arrive, re-clocking them to the proper timing upon playback?  Seems (to me) like a simple solution, regardless of the cable used.

Ding, ding, ding, ding! We have a winner!  Yes Brian, properly designed playback systems don't use their input clock do directly drive their sample output clock.  Jitter on the input does not directly translate to jitter on the output, though input clock jitter can affect your circuit's ability to create a low jitter output clock.  A well designed output clock circuit will average out most of the received jitter by using a phase locked loop whose loop bandwidth is many, many clock cycles.

Something else to remember... If reflections from an improperly terminated cable actually come back during the rise time, they will serve to hasten the rise of the signal.  Not only that, but they will hasten that rise by exactly the same amount of time every time.  (The cable is not changing lengths from bit to bit, nor is the speed of propagation on the cable changing.)  So even if there is an effect on the rise time, it's the same for each and every bit transmitted.  That simply cannot cause an increase in jitter.  Remember that jitter is a variation in bit timing.  Since the effect is always the same, there is no increase in the variation in bit interval that is due to cable reflections.

And again, my main point in my previous post is that THERE ARE NO REFLECTIONS if your cable impedance is the same as the input impedance of the electronics that are connected at the destination end of the cable.  It simply does not matter how long the cable is unless you have an impedance mismatch.

Brian, I understand what you are saying about the buffer solution in dacs.  It sounds like a logical solution.  If dacs with buffers were available to handle this issue, they would be in the caliber of dacs in the reference systems being used to make the comparisons.  A lot of of money, time and effort goes into these systems to make sure they are properly set up and have the best gear available to them. 

SparkE, your assessment of  improperly terminated cables isn't very realistic.  Testing six different digital cables from Atlas and Harmonix (ranging from $93 - $1700) that are all assembled improperly would be statistically challenging to say the least.  Atlas and Harmonix know how to terminate cables.  I'll take that statement to the bank.

The article stated that the rise-time is not consistent from one data block to the next.  It did not go into detail on the issue, but the issue was briefly addressed.  The inconsistency was blamed on the less-than-perfect impedance matching.  If their reasoning doesn't hold water, what other issues could be causing the sound quality to be better on the 1.5 meter cables?

Chris

[JasonR]

... If their reasoning doesn't hold water, what other issues could be causing the sound quality to be better on the 1.5 meter cables?

Psychology for one.

If the reasoning doesn't hold water, the next step is to look at the testing methodology for any holes, and the ones described are far too significant to conclude that one cable really does sound any different (forget better) than another.  At least two problems are that the sample size is too small to reach meaningful conclusions, and one person in the room was aware of which setup was which.

Maybe coaxial digital cables over some length really do sound different, or even better, but it doesn't seem to be supported by the evidence thusfar.

- Jason

[rodeen]

... If their reasoning doesn't hold water, what other issues could be causing the sound quality to be better on the 1.5 meter cables?

Psychology for one.

Wouldn't you capture the bits and diff them?  If the data coming across the wire is the same, wouldn't it sound the same?  I have to imaging if we are dealing in the digital domain that there is some way to measure the differences and explain them. 

[Brian Skalinder]

Wouldn't you capture the bits and diff them?  If the data coming across the wire is the same, wouldn't it sound the same?  I have to imaging if we are dealing in the digital domain that there is some way to measure the differences and explain them.

It's not so much that the bits are different, but the jitter that causes arrival time differences of the bits at the DAC.

[Roving Sign]

Ok - backing up a bit - a lot of this addresses sending a stream to a DAC.

Does any/all of this mean anything when it comes to cables between ADs and Recorders?

If a bit comes out of the AD and is recorded...isnt is still the same bits - no matter when it arrived.

[SparkE!]

Wouldn't you capture the bits and diff them?  If the data coming across the wire is the same, wouldn't it sound the same?  I have to imaging if we are dealing in the digital domain that there is some way to measure the differences and explain them.

It's not so much that the bits are different, but the jitter that causes arrival time differences of the bits at the DAC.

Brian, the speed of propagation on the cable is constant and the reflections are the same size every time.  Nothing about the reflections will change from bit to bit.   As such, there is no increase in jitter, even if there are measurable reflections.

The size of the reflection is given by Vr = Vi * G, where G is the reflection coefficient, Vr is the reflected voltage and Vi is the incident voltage at the mismatched end of the cable.  G may be computed as (Zcable - Zinput) / (Zcable + Zinput) where Zcable is the intrinsic impedance of the cable and Zinput is the effective input impedance at end of the cable that is mismatched.  Notice that if Zcable = Zinput, G is 0 and there is no reflected voltage.  Nothing in any of these parameters changes from bit to bit, so the reflection is the same size for every bit.

The reflections may speed the rise time up a bit, but the rise time of the signals on the installed cable will be exactly the same time for each and every bit transmitted.  Mismatch reflections simply do not and cannot affect signal jitter unless the cable significantly limits the signal bandwidth (which is not the case for S/PDIF signals on commonly available S/PDIF cables).

[Brian Skalinder]

Brian, the speed of propagation on the cable is constant and the reflections are the same size every time.  Nothing about the reflections will change from bit to bit.   As such, there is no increase in jitter, even if there are measurable reflections.

The size of the reflection is given by Vr = Vi * G, where G is the reflection coefficient, Vr is the reflected voltage and Vi is the incident voltage at the mismatched end of the cable.  G may be computed as (Zcable - Zinput) / (Zcable + Zinput) where Zcable is the intrinsic impedance of the cable and Zinput is the effective input impedance at end of the cable that is mismatched.  Notice that if Zcable = Zinput, G is 0 and there is no reflected voltage.  Nothing in any of these parameters changes from bit to bit, so the reflection is the same size for every bit.

The reflections may speed the rise time up a bit, but the rise time of the signals on the installed cable will be exactly the same time for each and every bit transmitted.  Mismatch reflections simply do not and cannot affect signal jitter unless the cable significantly limits the signal bandwidth (which is not the case for S/PDIF signals on commonly available S/PDIF cables).

I think I actually followed that.    Thanks for the correction, and explanation, SparkE!

[SparkE!]

Guys, I'm really not trying to be a jerk about this.  It just bothers me when I see gross misinformation being peddled as the truth.

That said, I could still be wrong in my opinion on the topic, but so far no one has given a plausible explanation as to why short data cables are problematic.  In fact in every other application that I've worked on, there was a concerted effort to limit the length of data cables in order to avoid problems associated with reflections.  Here's why:

The size of the reflection is determined by the impedance mismatch and the size of the incident signal.  For all practical purposes, the reflection happens instantaneously when an coming voltage transition arrives.  If there are mismatches at both ends of the cable, the reflection will bounce back and forth and reduce in amplitude at each reflection.  The amount of time that the reflections can affect things is determined by how long it takes for pulses to traverse the cable from one end to the other and back.  (Remember that the reflections are instantaneous, but each reflection reduces the amplitude of the reflected signal.)  So, the shorter the cable, the less time that the reflections can actually affect the signal that is detected at the destination end of the cable.  In that respect, shorter is good.  I have never seen an application that required longer cable for better performance unless the additional cable was being used to reduce the signal level of an rf signal that's so large that it will overdrive the front end of the device that will receive it.  (For example, GPS receivers don't like too much signal at their input because it can cause intermodulation distortion in their front end amplifiers.  So, when hunting for antennas for your GPS receiver, you need to make sure that you don't overdrive the front end by choosing an antenna that has too much gain.  On the other hand, if you want to drive a longer cable, you may actually want to use a higher gain antenna to compensate for the additional cable losses that are associated with the increased cable length.)

[cshepherd]

What a conversation!  I don't think anybody's being a jerk here.  I knew this topic would stir it up.  If longer digital cables really do sound better than shorter lengths, that's worthy news around this forum. 

SparkE has made a solid case against their explanation.  I believe the reviewer's results though.  I don't see anything faulty with their revieweing process.  As part of the reviewing process at UHF, the person controlling the review comments last to avoid influence.  The fact that they didn't review 400 different cables doesn't change what they heard with four...and what our distributor heard with the two Harmonix digital cables.

I'm inclined to believe the debate lies with the explanation and not with the comparison.  I'm going to email Steve Nugent and ask for his explanation on this subject.  I'm hoping he'll respond.  If he does, I'll post it here.

 

Chris

[SparkE!]

I have no problem believing that 3 people can agree on a single conclusion.  Heck, I can toss 3 coins in the air, they'll all land the same side up about 1/4 of the time on the average.  It's not at all hard for me to believe something like that happened when you are talking a single trial with 3 people.  On the other hand, if I was the one deciding which cables to stock in my store, I certainly wouldn't base my stocking choices on the opinion of 3 people unless they could demonstrate to me that they can actually tell the difference by running the test blind over multiple trials.  I seriously doubt that they could identify which digital cable was in use in random trials any better than this 1980 penny in my pocket can.  I'll flip you to see who goes first!

[ghellquist]

It is very interesting that people trying to make you buy $1000 cables cannot be bothered to measure jitter. There are only two error modes for this kind of digital cables, lost bits or time jitter. Only these errors could influence the sound of a digital interconnect, if the rest is the same.

Lost bits are very unusual and is a sign of a really inferior design. If you need to, this can be easily measured.

Time jitter, or short jitter, is inherent in every transmission of a clock signal. The audible effects of jitter can easily be measured if you know how to do it and has acess to some fairly inexpensive measurement instruments. None, none, none, none of these producers of ultra expensive stuff can be bothered to measure it.

My guess is that there is no difference, only marketing. And as long as people buy it, they make their money and laugh all the way to the bank. Guess the world is full of suckers waiting to get it.

Gunnar