The Null Testing Thread

[Prog Rock Man]

Null testing seems to be to be a very good way of finding out if there is an audible difference between different hifi kit. My understanding is that such tests invlove recording any sound or music with one setup, then the same sound or music with another and subtracting one recording from the other and seeing what, if anything is left. If nothing is left, there is no difference. If something is left it has to be listened to to find out if it is audible or not. As I said, I know little of the subject and how it works with measuring claims of audibility, but I started to google and  found this -
 
http://web.mac.com/davewronski/audio/null.html
 
A Dave Wronski plays guitar through different cables and measures the difference. The first two cables he used find a difference that can be heard. The second test and there is silence.
 
So, hopefully this thread can examine null testing and what it can tell us about audiophile claims of the likes of cables sounding different.

 

[JadeEast]

I think it's a pretty good idea it doesn't even have to be done in the digital domain. Turntable azimuth can be set by the nulling of out of phase signals: using a mono track and inverting the phase of one of the cable feeding the preamp. One might be able test for a difference in cables by producing a mono track, duplicate the track and invert the phase of the copy, and then create a stereo track with one of the mono tracks on the right and the phase inverted track on the left. In theory, feeding the signal into a summing amplifier would leave only the differences between the two signals and difference between cables would be there. Unfortunately, the differences between the sides of the stereo amp and thermal noise differences would also be there. 
 
One similar idea for testing amplifiers is here: http://sound.westhost.com/sim.htm
 

 

[Prog Rock Man]

I presume that the Audio Diffmaker would be another means of null testing.

 

[Prog Rock Man]

I thought it was time to have another go with this thread as a way of gathering together null tests. This is the Caver amp test on Stereophile
 
http://www.stereophile.com/content/carver-challenge
 
Any others appreciated along with coments and constructive criticism of null testing appreciated.

 

[Willakan]

It's kinda funny (at the expense of Stereophile probably), to see that the engineer of a solid state amp COULD add distortion/skew frequency response to make his amp sound like an "audiophile" tube amp, but ultimately decided to deliver what's on the recording.

 

[nick_charles]

Quote:

It's kinda funny (at the expense of Stereophile probably), to see that the engineer of a solid state amp COULD add distortion/skew frequency response to make his amp sound like an "audiophile" tube amp, but ultimately decided to deliver what's on the recording.

If you look at Stereophile's reviews of Amplifiers two things are apparent, one amps measure pretty badly compared to CD players , secondly SP are extremely forgiving of poor technical performance in tube amps accepting 1 or 2 % distortion and lousy FR and at the same time will tut-tut over a CD player that has 600ps jitter which is utterly inaudible...
 
 

 

[xnor]

^ Probably because they can't hear it.

 

[tmars78]

Quote:

I presume that the Audio Diffmaker would be another means of null testing.

Yes it is. I used it to check the difference between a WAV and a WMA lossless. There was no difference. I had a track full of silence. 

 

[Prog Rock Man]

Has anyone got any explanation for the first link to the two null tests, one of which appears to find an audible difference between cables?

 

[xnor]

The explanation is quite simple. If you just substract an inverted recording from the original then even tiny differences like clock drift can make the result look like there are huge differences between both signals. All the samples need to line up perfectly in order to null.

That's why DiffMaker was developed. It does not do a simple sample by sample comparison but accounts for clock drift, tiny phase differences etc.

Imo, a simple nulling test is more useful if you processed the files with some DSP only, i.e. when no D/A or A/D conversion was involved.

edit: installing Quicktime in my VM atm, to be able to view what's on that page, heh..
edit2: Since you can hear the guitar clearly in test #1 I think he ran into some sort of sample offset (a single sample is enough to cause huge differences) or volume difference problem.

 

[Funkuponatime]

nothing new from these guys to comment on?

 

[Roseval]

https://deltaw.org/

 

[gregorio]

nothing new from these guys to comment on?

“New” in what sense? A new response to the OP as to why guitar cables can sometimes make an audible difference, new results from null tests or new developments with null testing?

If it’s the latter, then @Roseval gave good advice.

G