Digital Cable Differences

[lucey]

I'm not going to argue the concept of digital cable tone I'm just going to share these four files. Flip the phase and play with the four passes of file->cable->DA=>analog chain=>AD->cable->file, where the same two SPDIF cables were used on each pass. The four passes were done in quick succession to minimize any variables from time/temperature drift in the analog or converters.

The 1 and 2 passes are using the same cable and cancel around -78 (representing the differences from the analog chain, DA, AD and spdif cable). The A and B passes the same (other) cable and also cancel in that range. Both are quality designs with good connectors, and each costs around $140 for 1M of SPDIF.

Magic Garden Mastering Admin
USER: digicable (case sensitive)
PASS: 4321

[Lagerfeldt]

I'm not sure I get your explanation of the files.

Could you be more detailed or explain again.

[lucey]

Originally Posted by Lagerfeldt

I'm sure I get your explanation of the files.

Could you be more detailed or explain again.

Sorry....

Let's see.

4 files uploaded, same analog chain used on all four, all four files made in rapid succession of each other.

File names 1 and 2 use the same set of 2 SPDIF cables to go to and from the DA and AD.

File names A and B use another pair of SPDIF.

Files 1 and 2 cancel with a phase flip at around -78. As do files A and B cancel in that same range.


but


Files 1 and A, 2 and A, 1 and B, 2 and B ... do not cancel in that same range at all. And they sound a little different.

[Lagerfeldt]

Thanks, that's what I thought you meant but wanted to make sure.

I've only experienced a difference when using a 8 ch. ADAT signal and not 2 ch. S/PDIF via TOSLink, where the difference in bandwidth could have an effect perhaps.

Or perhaps the tip of the optical cable has a minor problem like dirt, is scratched or plastic from the mould partly obscuring the light.

Did you test the tip of the cables and see how far you could pull them out and still have a signal? Does that vary from the one set of cables to the other? E.g. you can pull one set of cables out 0,5 cm of the I/O module, and the and the other less? That's one of the tests I do.

[bob katz]

Originally Posted by lucey

I'm not going to argue the concept of digital cable tone I'm just going to share these four files. Flip the phase and play with the four passes of file->cable->DA=>analog chain=>AD->cable->file, where two SPIDIF cables were used on each pass. The four passes were done in quick succession to minimize any variables from time/temperature drift in the analog or converters.

The 1 and 2 passes are using the same cable and cancel around -78 (representing the differences from the analog chain, DA, AD and spdif cable). The A and B passes the same (other) cable and also cancel in that range. Both are quality designs with good connectors, and each costs around $140 for 1M of SPDIF.

Magic Garden Mastering Admin
USER: digicable (case sensitive)
PASS: 4321

Lucey, try the same thing without even changing the cables. A D/A/D loop adds enough phase shift and timing issues so that you are likely to never get complete cancellation on two different passes.

BK

[lucey]

Originally Posted by bob katz

Lucey, try the same thing without even changing the cables. A D/A/D loop adds enough phase shift and timing issues so that you are likely to never get complete cancellation on two different passes.

BK

As I said Bob, I did that. The same cables through the chain null against each other well below the different cables. And there is an audible difference here as well. Not as obvious in this track as a more dynamic acoustic track, but that one is not cleared for sharing.

[Spartacus]

Originally Posted by lucey

Flip the phase and play with the four passes of file->cable->DA=>analog chain=>AD->cable->file, where two SPIDIF cables were used on each pass.

Both are quality designs with good connectors, and each costs around $140 for 1M of SPDIF.

you used :
RCA coaxial S/PDIF ?
or
Toslink optical S/PDIF ?

also, are you clocking with those S/PDIF in this test?

[lucey]

Coax cable, not Toslink

Originally Posted by Spartacus

also, are you clocking with those S/PDIF in this test?

Yes.

[arf]

I think nulling is only a valid test when the clock is held constant for each pass, which lets out all D/A/D loops except where the DAC can sync to word-clock in. There have not been any studies of how deep a partial null has to go for files to sound the same. It's like being a little pregnant. There is a complete null or there isn't. Possibly the eye-pattern is different for each type of s/pdif cable and that could presumably cause the PLL in the DAC to behave differently with each cable causing a timing difference, but that should be audibly filtered out by a proper DAC. I guess the real question is can you identify 1/2 vs A/B in a blind test?

[lucey]

It's the same clock every time ... what do you mean by held constant?

Originally Posted by arf

I guess the real question is can you identify 1/2 vs A/B in a blind test?

Not so easy in this case, but yes.

[arf]

You'd have to clock the D/A/D chain from the source so that every sample in the source triggers the same clock edge in the DAC.

But, since you are able to ID the files blind, my point is mute. (moot.)

[lucey]

Originally Posted by arf

You'd have to clock the D/A/D chain from the source so that every sample in the source triggers the same clock edge in the DAC.

Well the AD is the master clock, and it's clock runs the DAW and the DAs clocks. So the files are sample accurate. Is this not what you're meaning?

[Darius van H]

I don't see the need for the analog loop in this test. If the 1's & 0's are being shuffled around, it should be detectable with a straight digital transfer.

The only logical and scientific conclusion to make is that (all other things working correctly) there is something wrong with your DA or AD converter (provided the cables null with the digi to digi transfer).

I'm not denying your results, only taking a scientific approach.

[lucey]

Originally Posted by Darius van H

I don't see the need for the analog loop in this test. If the 1's & 0's are being shuffled around, it should be detectable with a straight digital transfer.

The only logical and scientific conclusion to make is that (all other things working correctly) there is something wrong with your DA or AD converter (provided the cables null with the digi to digi transfer).

I'm not denying your results, only taking a scientific approach.

Sure .... and if this was only about science I'd have done it that way, but my concern was both music and science, so I did an actual processing pass, and tested the cables in action. I'm sure there would be the same results either way, 100% sure, as the cables do sound different and that's simply reflected in the RMS readings after phase flip.

[Nishmaster]

Brian, much respect, but,

By using an analog stage in this test the way you have done, you have not isolated the digital cable step as the offending problem. The only thing this test truly proves is that different cables OR different D/A passes will not null completely. You need to isolate down to the only variable you wish to test, and that is the cable only.

I would expect that the cable is not the factor here that you are thinking it is.

[lucey]

Originally Posted by Nishmaster

Brian, much respect, but,

By using an analog stage in this test the way you have done, you have not isolated the digital cable step as the offending problem. The only thing this test truly proves is that different cables OR different D/A passes will not null completely. You need to isolate down to the only variable you wish to test, and that is the cable only.

I would expect that the cable is not the factor here that you are thinking it is.

Sure it is, the digital cable is the only changed part of the chain ... the analog stage was unchanged (including temperature as it was all done so fast) and the 'same cable' passes null to each other far more than they null with a combo of cable passes. Any jitter artifacts are perhaps amplified by the analog chain, but all the more reason to use it.

Why would a test without the chain even matter? This is a real world thing. I'm looking for practical improvement, not a theory. I'm showing the numbers here to back up what's audible. Admittedly it's barely audible in this messy example, but it's audible. It was more audible on the acoustic/vocal record I was doing the other day when I installed the new cable but that's a label release and not out yet.

Anyway, the numbers don't lie.

[Schnert]

Originally Posted by lucey

Anyway, the numbers don't lie.

But the conclusion may do so

[lucey]

Originally Posted by Schnert

But the conclusion may do so

Okay... so step up and make your argument.

If the only change is the two digital cables, and each cable set nulls equally well against itself in successive passes (negating the effect of the analog chain) but not as well against the other set, then what is your conclusion? What else even could be making this result?

[Bluelake]

Originally Posted by lucey

Okay... so step up and make your argument.

If the only change is two digital cables, and each cable set nulls equally well against itself (negating the effect of the analog chain) but not as well against the other, then what is your conclusion?

Back in the second part of the 80's, it was a must among audiophiles to listen to CDs with a CD transport separated from the converter (DAC).
In that configuration it was notorious that the cable inbetween was of great importance. Different cables would sound different ...and those differences were really easy to hear in blind tests... I experienced it myself when I bought a Micromega transport and an audio alchemy converter.....I had 3 RCA / coaxial cables to test. It was easy to hear that one of the cables didn't sound good at all, while there was a slight difference between the two others...and the one sounding the best wasn't the most expensive....

So something is going on with digital cables even if it's supposed to be just 0 and 1..... So I'm not surprised that the test doesn't null....

Juan

[Cellotron]

Originally Posted by lucey

Okay... so step up and make your argument.

If the only change is the two digital cables, and each cable set nulls equally well against itself in successive passes (negating the effect of the analog chain) but not as well against the other, then what is your conclusion?

Personally I would with hold a conclusion until I isolated one variable fully and did just a few D to D transfers using the various cables to establish a "control" group for the experiment. But that's just me.

Best regards,
Steve Berson

[lucey]

Juan, I appreciate your experience but I do want to avoid the standard arguments about cable tone that get ugly fast and provide nothing to discuss.

Originally Posted by Cellotron

Personally I would with hold a conclusion until I isolated one variable fully and did just a few D to D transfers using the various cables to establish a "control" group for the experiment. But that's just me.

Again, there is a control group.

The successive passes with the same cables null near -80, in each case. Crossing up the cable passes changes that null audibly (try it and see) and on the RMS meter, actually by a huge percentage as compared to deviation in the same cable nulls.

[minister]

For multiple D/A-->A/D passes, which is the best number to chose for Mastering on a budget??

[Cellotron]

Originally Posted by lucey

Juan, I appreciate your experience but I do want to avoid the standard arguments about cable tone that get ugly fast and provide nothing to discuss.


Again, there is a control group.

The successive passes with the same cables null near -80, in each case. Crossing up the cable passes changes that null audibly (try it and see) and on the RMS meter, actually by a huge percentage as compared to deviation in the same cable nulls.

Brian -
Again - you are dealing with 2 variables. What's so tough about acknowledging it would be valuable to eliminate 1 of them prior to establishing conclusions as to what is going on?

Best regards,
Steve Berson

[lucey]

Originally Posted by Cellotron

Brian -
Again - you are dealing with 2 variables. What's so tough about acknowledging it would be valuable to eliminate 1 of them prior to establishing conclusions as to what is going on?

Best regards,
Steve Berson

Eliminating the practical nature of the test I've already done twice, and posted once? Sorry, not going to do that for your denial as to what I'm showing here. Let's focus on what I'm showing, not what you think I'm not showing.

There is one major element changing (well two cables per pass, but one variable) and everything else is the same more or less. What makes the analog chain an important "variable" when both of the 'same cable' files null to around -80 but the different cable passes do not get anywhere near there?

And there are far more than two variables, but the others are not significant. There is "analog" in the DA and the AD for example. My AD is Class A and changes with time and temp The temperature was not exactly the same to a certain number of decimal points! These things are all taken care of by the nulling vs. same cables number that you can see for yourself.

Bottom line ... the analog processing is always in the chain, so it's important. And this testing method, imperfect as you may see it, has shown there are significant differences when the SPDIFs are swapped.

[Cellotron]

Originally Posted by lucey

Eliminating the practical nature of the test I've already done twice, and posted once? Sorry, not going to do that for your denial as to what I'm showing.

Since you've established that you're not getting same results between them through da->ad aren't you curious yourself whether these cables are nulling on a digital loopback also?

There is one element changing (well two cables per pass, but one variable) and everything else is the same. What makes the analog chain an important "variable" when the same cable files null around -80 but not the different cable passes?

Because passes through any DA -> AD almost always do not null against each other.

There are far more than 2 variables, but the others are not significant.

How do you know?

Best regards,
Steve Berson

[lucey]

They're not supposed to null against each other perfectly Steve, that's moot. BobK made that moot point as well. We're looking at the degree of change from different pair of coax in a chain.

We know the other variables are not significant because the 'same cable' tests of 1 and 2, or A and B show a roughly -80 null, phase flipped between both. So everything in the chain is handled in that comparison. Then we compare different cable pairs going to and from the DA and AD, and flip phase with same ... very different. It's clear as a bell.

I'm not curious at all about the D to D test. I'd expect nothing to change as would be no noise introduced to the analog via jitter or whatever makes these noises, but I don't care one way or the other. I leave that to Mythbusters, I'm interested in the sound of the whole in a working app.

[Schnert]

Originally Posted by lucey

I'm not curious at all about the D to D test. I'd expect nothing to change, but I don't care one way or the other. I leave that to Mythbusters, I'm interested in the sound of the whole.

So the fact that the different cables carrying data actually transfer the excact same numbers are of no interest? Instead you add more variables of which you have no control, and rather draw conclusions based on that?? I'm lost.

I've transferred digital with a variety of cables. They all come out excactly the same. Clones, no less. Adding more variables doesn't change that.

[dcollins]

Originally Posted by minister

For multiple D/A-->A/D passes, which is the best number to chose for Mastering on a budget??

Ban Tom now, and lock this thread!


DC

[Susceptor]

When you entered analog domain the test became worthless.

Testing S/PDIF cables is like testing if different SATA, IDE, FW, USB or WHATEVER else you may think of transmits the data differently.

When you transfer your documents from one HDD to another, do you lose your data or get other documents?

That's what I thought.

[lucey]

Originally Posted by Susceptor

When you entered analog domain the test became worthless.

Maybe it's worthless for the test you think I'm doing, but the test I'm actually doing is not about the numbers or the D to D, it's about the result of changing digital cables to and from the DAW. The numbers are supporting what's audible.

Did you listen to the files and flip the phase and check them out?