The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
High End Cables
Old 5th December 2014
  #331
nms
Lives for gear
 
nms's Avatar
Quote:
Originally Posted by Yannick View Post
I did some tests above, and showed genuine differences between different cables.
The ONLY response I got was from the non believers, is that my test is not valid.
You had multiple responses from people who understand digital audio, informing you that the level of difference your test showed was far beyond the physical limitations of human hearing.

I'm not saying with absolute certainty that there are no differences, but we remain without any evidence that can be considered audible.
Old 5th December 2014
  #332
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by nms View Post
You had multiple responses from people who understand digital audio, informing you that the level of difference your test showed was far beyond the physical limitations of human hearing.

I'm not saying with absolute certainty that there are no differences, but we remain without any evidence that can be considered audible.
In my test, i created a difference file with an IR only 65 dB down from the test signal.
If you call that insignificant, I do not know what to do.

https://www.gearslutz.com/board/10331954-post34.html
Post nr 35 is even more interesting,noone ever responded to that.

You can continue saying what I posted was insignificant, but then you are ONLY talking about the wav files I posted, and somebody else's interpretation.

Nobody ever refuted my posts 34 and 35.

Install REW and try it for yourself.
Then post back with a negative or positive result.
But please stop using the word insignificant and thus unhearable without responding to posts 34 and 35.

What part about digital audio is it that you are implying I do not understand ?
The part about the difficulty of digital measurements showing small time domain anomalies ?
Old 5th December 2014
  #333
Lives for gear
 
karumba's Avatar
Quote:
Originally Posted by Yannick View Post
In my test, i created a difference file with an IR only 65 dB down from the test signal.
If you call that insignificant, I do not know what to do.
correct your test-scenario and evaluation. bogosort showed you here that there is just a simple timing difference:
https://www.gearslutz.com/board/10336993-post304.html

as already has been said above, one weak point is the converter clock - did you take into account, that the timing offset of two converted signals could also be a fraction of 1 sample?
Old 5th December 2014
  #334
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by karumba View Post
correct your test-scenario and evaluation. bogosort showed you here that there is just a simple timing difference:
https://www.gearslutz.com/board/10336993-post304.html

as already has been said above, one weak point is the converter clock - did you take into account, that the timing offset of two converted signals could also be a fraction of 1 sample?
two things I did not address properly at that time:

first: obviously I did not use Fabian's master sweep in REW, but of course REW's internal sweep. So I did not understand this remark at all.

Second: REW performs (if you want it to) a time shift of the IR to 0s, up to about 1/100 th of a sample. In this case, one sample was 1/96000th of a second.

So, on both accounts, bogosort was wrong.

Please read the posts below !

In short: the length difference between the two cables is 2-3 cm - how much timing difference can this create ?
If this creates a much bigger difference - how/why ?

Second (as I already responded): how can the difference file of several IDENTICAL cables show NO CONVERTER drift, while the difference file of DIFFERENT cables suddenly shows converter drift ????

As case of cables having consious thougt ?
That would explain a lot ...
Old 5th December 2014
  #335
Lives for gear
 
bogosort's Avatar
Quote:
Originally Posted by Yannick View Post
first: obviously I did not use Fabian's master sweep in REW, but of course REW's internal sweep.
Then we're not measuring the same things and REW itself becomes an extra variable. If you want your results to be verifiable by those of us not using REW, then you need to use Fabien's master sweep. Otherwise you're just shaking your fist at the air.
Old 5th December 2014
  #336
Lives for gear
 
karumba's Avatar
Quote:
Originally Posted by Yannick View Post
REW performs (if you want it to) a time shift of the IR to 0s, up to about 1/100 th of a sample.
so you substracted your two signals sub-sample time aligned? could you please show me that option in REW?

Quote:
Second (as I already responded): how can the difference file of several IDENTICAL cables show NO CONVERTER drift, while the difference file of DIFFERENT cables suddenly shows converter drift ????
all i've seen up to now is a timing difference. this might come from the converter, but also from the cable or something else, it doesn't matter for now. in the end it is just a timing shift & before doing phase difference evaluation the timing needs to be corrected. i've done a lot of such tests with cables in different price ranges before. i've never seen such very high differences (-65dBFS) that you are showing.
Old 5th December 2014
  #337
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by karumba View Post
so you substracted your two signals sub-sample time aligned? could you please show me that option in REW?
In REW go to [preferences] - [preferences] - [analysis]
On the left side, in the middle, you can V Sub-sample timing adjustment.
I contacted the devloper, he answered that this is to around 1/100th of a sample.
In his opinion not enough to adjust for cable timing differences, as he says these are several orders of magnitude smaller (3 nanoseconds for 1 meter of cable, in my case the difference was a few cm !) But those differences should become trivial. Point is, REW can compensate to 1/100th of a sample - which should more than correct any ADDA loop timing problem (my Benchmarkmedia is a resampling converter)

Quote:
Originally Posted by karumba View Post
all i've seen up to now is a timing difference. this might come from the converter, but also from the cable or something else, it doesn't matter for now. in the end it is just a timing shift & before doing phase difference evaluation the timing needs to be corrected. i've done a lot of such tests with cables in different price ranges before. i've never seen such very high differences (-65dBFS) that you are showing.
That is why I was amazed.
Instead of people saying what I showed is unhearable, I would very much appreciate somebody pointing out what I did wrong - or not.
Old 5th December 2014
  #338
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by bogosort View Post
Then we're not measuring the same things and REW itself becomes an extra variable. If you want your results to be verifiable by those of us not using REW, then you need to use Fabien's master sweep. Otherwise you're just shaking your fist at the air.
I do not see the problem in just sampling the cable with a standard sweep.
That is just how REW works.

Next thing is you are going to say I used a PC, not a MAC
Old 5th December 2014
  #339
Lives for gear
 
bogosort's Avatar
Quote:
Originally Posted by Yannick View Post
I do not see the problem in just sampling the cable with a standard sweep.
The idea is that everyone has access to the same materials, in this case the input (the sweep itself) and the outputs (the sweep recorded through the cables). In this way the data can be analyzed by other people using other tools.

If this doesn't make sense to you, then nevermind.
Old 5th December 2014
  #340
Lives for gear
 
karumba's Avatar
Quote:
Originally Posted by Yannick View Post
In REW go to [preferences] - [preferences] - [analysis]
On the left side, in the middle, you can V Sub-sample timing adjustment.
I contacted the devloper, he answered that this is to around 1/100th of a sample.
In his opinion not enough to adjust for cable timing differences, as he says these are several orders of magnitude smaller (3 nanoseconds for 1 meter of cable, in my case the difference was a few cm !) But those differences should become trivial. Point is, REW can compensate to 1/100th of a sample - which should more than correct any ADDA loop timing problem (my Benchmarkmedia is a resampling converter)
thanks, i'll check that out.
could you please provide your mdat file?
Old 6th December 2014
  #341
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by karumba View Post
thanks, i'll check that out.
could you please provide your mdat file?
here is the .mdat
Two types of cable.
Velox (= solid core silver cable), two measurements with single sweep. One with 8x averaged sweep.
AES = Gotham audio aes cable. Also 2 single sweeps and one 8x averaged.
The last file is Velox-AES.

You can create your own difference files in REW of any combination to see what happens. There is always some residual, but nothing in the order of different cables with the same sweep.

I have yet to try (first construct) my newest favorite cable, to see what happens.
Three cables under test could give some clearer info.

As I wrote before, the problem is my AD interfaces are all with D25 connectors, so it is not so straightforward to connect any cable without some soldering work.
Old 6th December 2014
  #342
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by bogosort View Post
The idea is that everyone has access to the same materials, in this case the input (the sweep itself) and the outputs (the sweep recorded through the cables). In this way the data can be analyzed by other people using other tools.

If this doesn't make sense to you, then nevermind.
If it does not make sense to you that I cannot do anything sensible with the sweep that was provided, except for relying on utter scepticism to prove nothing, then nevermind.

REW is a (acoustic) measurement tool. It makes its own sweeps. I used that tool to try and show some difference. What is wrong with that ?

As I asked several times, stop saying I showed nothing or did not do what some expected. Start answering the questions I asked or commenting the test results I provided.

It could very well be my test is flawed, but not because I used another sweep.
Surely ???
Old 6th December 2014
  #343
Lives for gear
 
karumba's Avatar
Quote:
Originally Posted by Yannick View Post
here is the .mdat
i'm sorry, but i can't see it.
Old 6th December 2014
  #344
Lives for gear
 
DSD_Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by Yannick View Post
If you want to bury your head in the sand, be my guest.
In the meantime I am making better recordings with my new mic cables.

If you cannot hear it in your setup, you seriously need to consider upgrading your monitoring chain...
We're mastering engineers..... I don't do recordings.
Besides... my monitoring chain can't get any better.
Old 6th December 2014
  #345
Lives for gear
 
bogosort's Avatar
Quote:
Originally Posted by Yannick View Post
If it does not make sense to you that I cannot do anything sensible with the sweep that was provided, except for relying on utter scepticism to prove nothing, then nevermind.
I don't understand why you cannot use Fabien's sweep. The reason it is preferable to the REW sweep is that Fabien's sweep is accessible to everyone, whereas REW is not. And without both the input and the output, the transfer function cannot be determined independently -- we're left relying on your pictures of REW's analysis. While it may be the case that REW is the right tool for this task, it may not be.

If you want to make it easier for others to verify your data, use Fabien's sweep.
Old 6th December 2014
  #346
Lives for gear
 
Yannick's Avatar
 

Here are the .mdats
(had trouble with the file limit, no .rar, no .zip splits allowed...)
Attached Files
File Type: mdat Velox01.mdat (2.99 MB, 115 views) File Type: mdat Velox02.mdat (2.99 MB, 111 views) File Type: mdat Velox 8sweeps.mdat (2.97 MB, 89 views) File Type: mdat AES01.mdat (2.99 MB, 86 views) File Type: mdat AES02.mdat (2.99 MB, 92 views) File Type: mdat AES 8sweeps.mdat (2.97 MB, 90 views) File Type: mdat Velox8x minus AES8x.mdat (2.72 MB, 97 views)
Old 6th December 2014
  #347
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by bogosort View Post
I don't understand why you cannot use Fabien's sweep. The reason it is preferable to the REW sweep is that Fabien's sweep is accessible to everyone, whereas REW is not. And without both the input and the output, the transfer function cannot be determined independently -- we're left relying on your pictures of REW's analysis. While it may be the case that REW is the right tool for this task, it may not be.

If you want to make it easier for others to verify your data, use Fabien's sweep.
REW cannot use external sweeps.
I used Fabians, the discussion is between post235 and post248.
In post253 is a wrong conclusion: the IRs are clearly NOT identical.
Just swap quickly between both of them. You will see one point near of the peak has a different amplitude.

Afterwards I tried to look for a more meaningful way to visualise potential differences.

What is the point in continuing to hammer on using Fabian's sweep, which I cannot use, which we already did, presented findings in a non meaningful way AND already misinterpreted "unvisible" differences (with the naked eye) between the IRs ???

Again, what is so special about Fabian's sweep ?

If you want I can try to produce the same sweep later, just straight digital in-out. If only I could remember which setting I used.
Old 6th December 2014
  #348
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by DSD_Mastering View Post
We're mastering engineers..... I don't do recordings.
Besides... my monitoring chain can't get any better.
You clearly have a state of the art studio and monitoring chain !
But still, you call yourself "audiophile" on your website ?

Do you really have a 100K + monitoring chain an use one dollar a piece interconnects ?
Is still do not get in what camp you are...

Anyway, those amps are great. But still, I would dare to ask, did you try the newest Devialet amps ?

Never say it cannot get better
Old 6th December 2014
  #349
Lives for gear
 
DSD_Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by Yannick View Post
You clearly have a state of the art studio and monitoring chain !
But still, you call yourself "audiophile" on your website ?

Do you really have a 100K + monitoring chain an use one dollar a piece interconnects ?
Is still do not get in what camp you are...

Anyway, those amps are great. But still, I would dare to ask, did you try the newest Devialet amps ?

Never say it cannot get better
Let's define "audiophile". audiophile/n/
noun
1.a person who has a great interest in high-fidelity sound reproduction

Yes, I can be considered an audiophile then. I use cables that cost $2k/foot as well as $1/foot. If it does not harm or degrade the signal flow, then I'm all in. These are tools just like anything else. Can my monitoring chain get any better? maybe. I've tweaked it over the past 8yr. to what it is today. I've taken measurements of the purpose built room and equipment til I'm blue in the face and I'm satisfied that I'm getting the best there is available.
Old 6th December 2014
  #350
Lives for gear
 
bogosort's Avatar
Quote:
Originally Posted by Yannick View Post
REW cannot use external sweeps.
I used Fabians, the discussion is between post235 and post248.
In post253 is a wrong conclusion: the IRs are clearly NOT identical.
Just swap quickly between both of them. You will see one point near of the peak has a different amplitude.

Afterwards I tried to look for a more meaningful way to visualise potential differences.

What is the point in continuing to hammer on using Fabian's sweep, which I cannot use, which we already did, presented findings in a non meaningful way AND already misinterpreted "unvisible" differences (with the naked eye) between the IRs ???

Again, what is so special about Fabian's sweep ?

If you want I can try to produce the same sweep later, just straight digital in-out. If only I could remember which setting I used.
I'm not trying to hammer on anything; as I've said, Fabien's sweep is special only in that it is readily available -- I don't have nor plan to install REW, so I can't analyze your data, I can only look at your graphs and wonder what I'm actually seeing.
Old 6th December 2014
  #351
Lives for gear
 
FabienTDR's Avatar
 

Verified Member
btw, there's also a set of impulses in the test file. These can be easily transformed to frequency/phase magnitude plots.

No need to use the sweep (the original intention behind the sweep in the test file is detection of frequency dependent nonlinearity and/or noise). Given all other tests show no sign of nonlinearity (new partials), there's really no reason for using anything else that the impulse response of the system. This has been the case for all 3 samples.

Again, I cut the resulting impulses out, windowed them and ran a simple FFT on them. I can't see any mistake in this arrangement. This returned 3 magnitude plots, all varying by the amount of noise inherently produced by the test-system. Put differently, there where no measurable differences between all 3 cables. They neither have any form of linear behaviour nor any trace of nonlinearity.

As far as I remember, all made sure to perfectly align all three impulses. You're free to run a simple null test, I bet the differences are near -105dB.
Old 7th December 2014
  #352
Lives for gear
 
Yannick's Avatar
 

That is my point.
The null test in REW gave an impulse only 65 dB down.
Old 7th December 2014
  #353
Lives for gear
 
esldude's Avatar
Quote:
Originally Posted by Yannick View Post
That is my point.
The null test in REW gave an impulse only 65 dB down.
Since other people using software suited for the purpose don't get that result, the most likely thing is REW is handling this in some manner different than the other people doing such testing. Which why several times people have asked you to use something else. Or do a simple null test, and see what happens.
Old 7th December 2014
  #354
Lives for gear
 
JP__'s Avatar
 

Verified Member
anyone who can proof that REW works unproper? all I read here is the question about who has the 'biggest sweep' and some theoretical science.
Im waiting for karumba to cross-analyse the mdat...
Old 7th December 2014
  #355
Lives for gear
 
JP__'s Avatar
 

Verified Member
the last time I did a cable test I receive nulls down to -85db. but I think you never know what produces the difference in reality. what about smallest clockdrifts, dither noise etc.
but there where only one loop per cable. so no possibility to crosscheck this. but yannick has done so, doesnt he?
Old 7th December 2014
  #356
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by esldude View Post
Since other people using software suited for the purpose don't get that result, the most likely thing is REW is handling this in some manner different than the other people doing such testing. Which why several times people have asked you to use something else. Or do a simple null test, and see what happens.
I do not know. Most here are looking at FR and phase graphs, maybe distortion.
Itry to look at the time domain.

Essentially, what I did is a null test, but better, because the two files have been time aligned by REW.

How can a simple null test be better, and show something useful.
Please explain !
Old 7th December 2014
  #357
Lives for gear
 
esldude's Avatar
Quote:
Originally Posted by Yannick View Post
I do not know. Most here are looking at FR and phase graphs, maybe distortion.
Itry to look at the time domain.

Essentially, what I did is a null test, but better, because the two files have been time aligned by REW.

How can a simple null test be better, and show something useful.
Please explain !
If you have two files that null pretty well, and then you "time align" them to obtain a less deep null, you can be sure they were aligned better before you changed the alignment.

I have a suggestion for checking time alignment. Run a sawtooth wave through them. Do a null of that, and see what the FFT of the null looks like. A sawtooth is a fundamental and all harmonics with a decrease in level that follows a fourier series. If the FFT of the null has that same declining level with higher harmonics you are good. If instead the fundamental and all the harmonics are equal in level you have a time alignment issue. You get this pattern whether the timing shift is large or small until you get into the noise floor of the system.
Old 7th December 2014
  #358
nms
Lives for gear
 
nms's Avatar
Quote:
Originally Posted by Yannick View Post
Essentially, what I did is a null test, but better, because the two files have been time aligned by REW.
Better??

By what logic do you come to suppose that different cables would require or benefit from any amount of time alignment? You have added unnecessary processing and room for error.

That said, while REW isn't the ideal tool here, I don't even recall any difference that appeared to be within the threshold of human hearing.
Old 7th December 2014
  #359
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by nms View Post
By what logic do you come to suppose that different cables would require or benefit from any amount of time alignment? You have added unnecessary processing and room for error.
So you are implying the resampling done in REW introduces a -65 dB error ?
If will contact the developer again and ask him this question, in your name, not mine... It seems quite impolite, but anyway.

I will retest this week, as the original files are timeshifted already.
Yes, it is those very same files that in some software show no differences, but magically, when it fits your argumentation, have a BIG error introduced by unnecessary processing, when they appear different.


Quote:
Originally Posted by nms View Post
That said, while REW isn't the ideal tool here, I don't even recall any difference that appeared to be within the threshold of human hearing.
I won't repeat it again. A few posts back I just commented on the fact that some of you refuse to read, view and respond to my posts where I show big differences.

You are still maintaining that a -65 db error is below our hearing threshold ?
Or are you going to read over this sentences once more ?

-65 db error is below our hearing threshold ?
Old 7th December 2014
  #360
Lives for gear
 
karumba's Avatar
thanks for the mdat files. i've just had a look into it - please find attached figure (click to enlarge):



the most interesting curves are the last 4 where the difference between various constellations is shown.

the first thing that looks strange to me is the increasing difference for high frequencies. it is even more strange, since this effect is also shown when the same cable is evaluated! that alone make the evaluation questionable in my eyes.

yannik, you always argued with a difference of -65dBFS (whatever that means in REW). but why is the difference between:

AES01-AES02 (blue) or Velox01-Velox02 (red)
vs.
AES01-Velox01 (darker green) or AES02-Velox02 (purple)

only 10dB absolute? i guess its safe to say, that the difference between the same cable measured twice is in the range of, lets say, below -140dBFS, right? based on your argumentation, i could now say, that the green & purple curves are only 10dB "worse" compared to if the difference of the same cable is calculated. in that case we are still at -130dBFS. this is much below the human hearing threshold for sure!

all together, for me this leads to the strong assumption that REW is not suitable for such evaluations - at least the interpretation is very questionable! i recommend to use self made test signals (e.g. the one of fabien) & evaluation e.g. with matlab or any other digital signal processing library as others also alreay suggested. only then we can really say for sure what is evaluated.

//edit: mysterious side aspect: the difference spectrum of the Velox cable is less "wavy".
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump