Wednesday, January 09, 2013

Vodoo continued: my final Statement on USB cables

This article here is the final sequel to "Hot Vodoo: Audio differences between USB cables - yes, they do exist" and "More USB cables - more differences". So before you read this one go ahead and scroll through the other two.

Almost one year has passed since my last article concerning USB cables and their sonic signatures was published here. Truth is that I ran out of steam: I didn´t want to do it since I was afraid that I had been measuring the effects of ground loops only. But before I start with the embarassing truth I´ll tell you about the stir those two USB articles created. The articles I refer to (linked above) were written almost a year ago and since then they have been popping up constantly in several online forums. I´ve received visits from Russians, Germans, Americans, Norwegians, Japanese or Chinese citizens to name but a few. Sometimes the participants of these forums made fun of me, sometimes they tried to find out how I was doing things, sometimes I was accused of lying or imagining things and on some occasions people actually tried to help me. It fascinates me to this day that 99 % of all those people didn´t comment here on my blog but only in their respective habitat, forums I´ve never heard of before. Since I´m the most gleefully cunning person I occasionally registered there just to take part in their discussions... and boy, it was delightful to witness how fast participants there transformed from being bullies into polite, elaborate people upon myself unexpectedly taking part in those discussions. Without me it was 'chic' to mock me - so why did people stop when I suddenly entered? I love mockery, especially when it´s about me (I tend to be too self-centered sometimes). This wasn´t restricted to one country, no, it was a trait exhibited by people coming from all around the world. For all those who were doubting me throughout the last year I assume this article will be a field day which is perfectly alright - yet it saddens me that people generally seem to ignore transparency and truthfulness. But let´s start, eh?

USB cables revisited: only the Belkin and the Audioquest were used
In the last article I compared five different USB cables, for the article you are now reading I used only two of them: the Belkin which performed the best & the Audioquest which performed the worst. Since the measurment differences between those two were the strongest I assumed that they´d be the most likely candidates for also showing differences today. Likewise, I used my Creative Labs Soundblaster X-Fi HD USB again... BTW, that was the most peculiar thing... because in those forums I mentionend above I was attacked for measuring cheap hardware 'on my kitchen table'. Well, of course I did, that was the whole point. What good would my test have been if I would measure with expensive boutique hardware (which I don´t own, thank you very much)? I wanted to show results with everyday hardware, something people would actually buy. That way my results would (hopefully) be more comprehensible. Furthermore, assuming that something more expensive is behaving in a superior way is audiophile bullshit. One word about the 'kitchen table' comment... I will ignore it since it reveals opinions much too conservative for my taste. Just something to think about. Back to topic: I had to use the X-Fi HD USB and my ASUS Xonar Essence ST again since I wanted to compare possible differences between one year ago and today. My testing methodology may have changed / improved during the last year, yet I still have to make sure to repeat testing circumstances exactly as before to obtain reliable results.

Creative Labs Soundblaster X-Fi HD USB was used again
For that reason my testing chain then and now consists of the X-Fi HD USB connected via the Audioquest King Cobra RCA cable to the Xonar Essence ST. The last time both soundcards were powered by my desktop PC and since that obviously created the problematic grounding loop I now took a slightly different approach. This time I used my nine year old laptop, an FSC Amilo 1425 which was used to power my X-Fi HD. Naturally this laptop is able to run on batteries, it completely avoids grounding loops that way. Everything else was the same though: the Creative still was connected to the ASUS with exactly the same RCA cable. The laptop uses Windows XP as its OS so I had to utilize the Kernel Streaming plugin for foobar2000 to avoid potentially harmful resampling (one year ago: WASAPI direct connection on Windows 7). Another difference to the older test is that I decided to show you the results of ten test runs (five for every USB cable). I also did five test runs for the last test but I sadly failed to disclose their results. So, let´s have a look at the facts, shall we?

For comparison: Old results, one year ago, both interfaces run from the same PC

Belkin USB cable, two different PCs, one running on batteries,
five different test runs which have been averaged for this table

Audioquest Forest USB cable, two different PCs, one running on batteries,
five different test runs which have been averaged for this table
Shit! You now can see that one year ago I indeed measured the effects of a ground loop caused by using the same PC for both interfaces. With one interface running on batteries and by that removing the ground loop I also removed the measurment differences between the two USB cables, the lower resistance of the Audioquest is not important anymore. Yes, the X-Fi HD USB now measures EXACTLY the same with two different USB cables. Which probably means that 1. differences between USB cables don´t exist or 2. that I indeed can´t measure them with my limited hardware or 3. that I need more and different measurments. In any way, two things are now crystal clear: the X-Fi HD USB is a gorgeous performer on its line-out while the Xonar Essence ST boasts an equally impressive line-in. With measurment results as fantastic as these it´s hard to imagine that more expensive soundcards can do much better; you also have to keep in mind that both cards are relatively cheap and are far away from being hardware for demanding professionals. Just take the noise levels: they have improved by roughly 12 dB - Creative was indeed telling the truth by claiming a noisefloor as low as -114 dB for the X-Fi HD. Back to the USB cables: the only differences I was able to find were so small that they are truly insiginificant:

X-Fi HD USB, Belkin USB cable
X-Fi HD USB, Audioquest Forest USB cable
Click on the upper of those two pictures, that way you can scroll between them using your mouse wheel - only then you´ll see that the X-Fi HD USB has a very, very slightly improved high frequency performance with the Audioquest. That´s it. You could now say "Hey, there is a difference!" Erm... no, there isn´t. It is way too tiny to be of significance. Maybe distortions are able to reveal something else?

X-Fi HD USB, THD & Noise, Belkin USB cable
X-Fi HD USB, THD & Noise, Audioquest USB cable
Again, click on one of those two pictures and scroll between them with your mouse wheel and you´ll see that nothing changes - again. If at all, only the noisefloor seems to reveal a slight tendency in favor of the Belkin so I won´t bore you with additional non-existing differences between two pictures. Today I´ve learned two things: ground loops are evil & USB cable differences cannot be measured when using my approach. Phew, now I´m embarrased: I wrote two articles that were basically describing the effects of a ground loop! Stupid, stupid, stupid. But wait, has the sound changed? That´s where it gets interesting because it hasn´t. The X-Fi HD USB still sounds its best with the Audioquest Forest USB and the worst with the Belkin. Let me point out that the sonic differences are indeed small, I assume that they won´t be audible to 90 % of all people. But I started this article with an emphasis on being honest and truthful (which I always am - and which is painful here) and therefore I´ll tell you that I had doubts about my ability to hear things; I was afraid I might have encountered a prime example of a placebo. So I did what I usually detest, namely a Double Blind Testthe results of which you can see below:

foo_abx 1.3.4 report
foobar2000 v1.2
2013/01/09 02:54:00
File A: RAM-Disk:Audioquest.wav
File B: RAM-Disk:Belkin.wav
02:54:00 : Test started.
02:55:31 : 01/01  50.0%
02:56:05 : 02/02  25.0%
02:56:27 : 03/03  12.5%
02:57:05 : 04/04  6.3%
02:57:18 : 05/05  3.1%
02:57:28 : 05/06  10.9%
02:57:45 : 06/07  6.3%
02:58:01 : 06/08  14.5%
02:58:19 : 07/09  9.0%
02:58:33 : 08/10  5.5%
02:58:48 : 08/11  11.3%
02:58:59 : 09/12  7.3%
02:59:17 : 10/13  4.6%
03:00:01 : 11/14  2.9%
03:00:17 : 12/15  1.8%
03:00:35 : 13/16  1.1%
03:00:51 : 14/17  0.6%
03:00:54 : Test finished.
Total: 14/17 (0.6%)
I had occasional problems making out differences. It took me 17 trials during roughly seven minutes to find out if I was imagining things - but now I´m at least confident that my ears (and brain) are still fully functional. I´m no so-called 'Golden Ear', those are a myth. Audiophiles often claim that they are able to hear more than, say, skeptical people relying purely on measurments - which isn´t true. The only thing that´s valid to state is that there are people out there who have trained themselves to hear differently compared to others. I only hear this because I know what to listen for, nothing more. I´m also proud of my (perhaps imaginary) ability of not falling for typical placebo effects. "How can Marlene be so arrogant and full of herself?" you ask... maybe you´re right, look what being conceited brought me: grounding loops! All kidding aside, the situation of most of the time knowing what to listen for really isn´t the easiest thing to live with. I sometimes wish to have the gift of blissful ignorance because everytime I hear an album I immediately spot errors or defects & I cannot enjoy the music until they are gone. This might sound contrived to you but it prompts me to ignore bad sounding albums, even if the music is engaging. Imagine the time I could save by not being obsessed with these things. Enough with the cocky self pitying and back to topic!

Measures extremely well on its line-out: Creative Labs Soundblaster X-Fi HD USB

I´m afraid there isn´t that much to conclude. I still hear the effects of different USB cables, yet I´m unable to measure them; the centuries old dilemma of audiophiles. It all leads to you and the decisions you make, dear reader. Are you a believer or a skeptic? I hate to exploit this common prejudice but I started measuring USB cables through external soundcards as a means of diminishing troubles between those two groups - I thoroughly failed. The only thing I did was creating a lot of hot air. Shame on me. I did however learn a few things: more expensive cables have less resistance causing ground loops to leak through more easily, that I cannot measure everything, that ground loops hamper effective resolution and that the X-Fi HD USB is a great performer on its line-out. But where is the benefit for you?


  1. Good day! In this entry did you use the information from any researches or these are totally your private thought? Waiting forward to hear your answer.

    1. These are my personal thought processes. I took inspiration of course from others I mentionend in this article (members of some forums) who advised me that I might have been measuring ground loops whose effects I described in the two related articles before.

      Why do you ask?

  2. "It took more than one cable to change my name to USB Billy".

    "So you're USB Billy."

    "The notorious sighted audio tester of Germany. You've heard of me, and you always believe what you've heard."

    "And I still do, because I abx'd it. You see, I haven't changed at all."

  3. Nice writeup, I was looking for some information about this. I was a skeptic, but couldn't find any facts to confirm my suspicions. Kudos to you for revisiting a subject you'd already covered and finding something you hadn't accounted for, as well. That's something that we need more of in society.

    1. Thanks... but do you really think so? I contributed nothing worthwile to the discussion, in the end I wrote insane loads of hot Air. If something happened it´s that I´m even more skeptical than before.

  4. Question: How did you create those wav files you used for the ABX?

    1. I recorded the same piece of music twice; once with the Audioquest, another time with the Belkin (playback software: foobar / recording software: Sound Forge, using ASIO). After recording I edited them with sampleprecision in order for them to start and end at the same points. Level matching was unnecessary since both had the same gain level (except for the USB cable everything else was the same of course).

    2. I am curious about the sonic differences in the files. As they are so obvious to hear (acc to the ABX test results) there must be substantial differences IMO. Could you try to 'null' both files (as good as can be done) and post the differences ?
      If there are sonic differences they must be audible with nulling.

      The used playback soundcard must be really susceptible to timing problems in order to create audible differences.

      Maybe even compared to the original file to see what file changed most ?
      Maybe the files were different because of post processing ?

      Have you performed 'blind' tests where someone else changed or did not change a USB cable and you can reliably pick them out ?

      If I were in your shoes (could reliably hear yet not measure) I would not rest until the reasons were clear.
      If only to ease my own mind.

    3. You can listen to the files yourself: on the first USB-cable article you can find an archive, containing samples. They are edited with sample precision so a DBT would be easily possible.

      And when I did those articles I didn´t use the AudioDiffmaker which would have enabled me to create a "null" test. But I won´t return to the matter, editing those articles. I´m done with USB cables. And don´t worry, I can sleep very well knowing this :)

    4. Downloaded and synchronised the 2 files (by deleting a lot of sample at the beginning of one track, I used 'car trouble').
      Problem is there is a timing issue as the files vary + and - 1 sample over the entire recording, after synchronising the middle of the files.
      This is due to variances/drift in the clock frequencies over time.
      This could be in the playback and or recording end.
      It makes nulling the complete file impossible but with shifting the files +/- one sample can find 3 'spots' where nulling is quite good.

      It did not null perfectly but did drop to -80dB (peak level) which is actually a quite decent nulling result.Archimago got about the same values while nulling.

      At these 3 'nulled' files (beginning in one testfile, middle in another and near the end in the third file) nulling showed the actual waveforms that came out of the DAC are VERY closely the same.
      Perhaps if you would make 2 recordings of the same DAC/cable combo and null them you would possibly get the same nulling result.

      It still has me baffled you can tell the files apart reliably (didn't try) as the actual recorded waveforms only appear to differ in speed (drift in the clock) and not in actual sound/content.

      Thanks for uploading those files.
      Personally I won't bother toying with USB cables and concur with Archimago's findings and nulling results.

      I have no problems sleeping either ;-)

      Do enjoy the music though !

    5. You don´t need to worry, I´ve stopped worrying about USB cables. If these articles showed me one thing it´s that we (audiophiles) spend too much time on things that are in hindsight unimportant.

      But the samplerate drift will happen in any case, simply because the crystal used within the combo isn´t precise enough (that would apply to 99,9% of all hardware out there). You could try a reliable nulling with Audio DiffMaker, that program can compensate for samplerate drift.

      But I wouldn´t bother. I still buy fancy cables (without a reason, just to be sure)... but I have had very suspicious results. The tested NuForce cable broke down after one and a half year - without being moved. A few months ago I bought a fancy Oehlbach cable (German company) which broke down after two weeks (the connector seemed to be the culprit). So this puts a rather shady light onto expensive USB cables with the result that I wouldn´t bother spending more than 20,- Euros for a cable. I now use whatever cable I have at hand, fancy or not.

      But I won´t remove these articles, they may be embarrasing but they also generate lots of traffic (I like to stay in the limelight). So they stay.

  5. USB cables transfer digital signal and that is the exact reason every USB cable is same. Analog signal is susceptible to changes, digital either works or it doesn't.

  6. "USB cables transfer digital signal and that is the exact reason every USB cable is same"
    The uneducated professor.
    If digital was identical, would the results above vary?
    No- digital does not work on the principal of either it works or not- digital works on the principal of correcting errors, thats why it has a limitation of 15feet- beyond that the errors are too great to correct. What did you think digital cant travel over 15 feet of cable? how do you explain Ethernet moron?

    1. "USB cables transfer digital signal and that is the exact reason every USB cable is same" - I didn´t write that. You did. An Ad Hominem. Google it.

      Furthermore, error correction is employed by USB with some protocols only. The isochronous mode used for audio (DACs and soundcards) doesn´t use any error correction (that includes adaptive and asynchronous modes). Bulk mode for printers and such does indeed use error correction. For interrupt modes (mouse or keyboard) there is only error checking.

      Nothing you write makes any sense.

  7. Dear Jörn, please, could you correct the errors below mentioned in your previous articles ?
    Thanks for your time and patience !

    • Jan 24, 2012 [Marlene's Missing] Hot Voodoo: Audio differences between USB cables - yes, they do exist !, by Jörn Druhmann
    ==> Result general performance, Audioquest cable, SoundBlaster to ASUS: much worse.
    Noise floor increased for +6 dB, more distortions, more crosstalk.
    [the picture doesn't appear !]
    < >

    • Jan 26, 2012 [Marlene's Missing] More USB cables - more differences, by Jörn Druhmann
    ==> THD NuForce Impulse. [the picture doesn't appear !]
    < >

    1. I cannot correct these errors. Those pictures are hosted by Imageshack... and Imageshack lost some of them. Besides, differences between these cables were either caused by a grounding loop or by measurment variance.

  8. Thanks for sharing this information

  9. Thanks for share this information.


Related Posts Plugin for WordPress, Blogger...

The Socials