Null Difference Testing

[fdegrove]

THE NONSENSE I READ...

Hi,

I can't imagine how the music signal is creating RF oscillations in the power cord? If the cord was simply reducing the background noise, the RF argument might make more sense.

Not surprised there.
Suppose magnetic coupling is new to you too?

I know, you're going to ask for proof again?

And no, I don't think jitter reflects back on powercords, or more correctly AC powerlines.

Wondering why I bother educating this stubborn Rita,

 

[nw_avphile]

A revision

Anyway, The Alesis piece is close to the spec even with the regular cord. Keep in mind that we are reading DA and AD performance in series, so the error is multiplied.

I revise my earlier comment if you're talking about the THD spec. -75db would be .017% which is in the ballpark if we assume we can use the 10ms RMS average analysis value of the difference signal. Things get a little tricky when you start talking about distortion levels averaged over a clip of real music instead of sine waves.

I'm still intrigued by your results!

 

[jan.didden]

Re: The good and the bad guys....

My own tests show that a null of -50db is sufficient for most people not to hear a difference in blind listening tests. I extend that to -60db to allow for people with really exceptional hearing/listening skills. Bob Carver claimed he managed a -70db null in the Carver Challenge which fooled the ears of the Stereophile editors.

So somewhere between -50db and -70db null I believe we can safely say there are no audible differences to be heard. So, yes, if the amplifier is only managing -40db or -50db by itself, it may well mask differences in the cables. If it's closer to -70db, however, I'd have to say the null is sufficiently deep to argue the two things being compared will be indistinguishable in a blind listening test.

As was discussed a bit in the other thread, the more accurate way to do this would be to "weight" the null results (something like A weighting for S/N measurements) according to human hearing sensitivity. A poor null at 20khz is less likely to be audible than a poor null at 500hz.

We are in agreement then.

Jan Didden

 

[jan.didden]

Re: The good and the bad guys....

So for you interconects must be considered transmission lines ...so for a load impedance of 47k the cable feeding it must be a 47kOhms intrinsec impedance...

What about a digital cable for interconect??...it was built for to work as a transmission line...well...but only with 75 Ohms load!!!

It's amazing!!!


But who is in reality spreading nonsense here???

As I'm sure you all know, the difference between analog and digital interconnects is that the latter must faithfully transmit very high bandwith signals (perfect squarewave preferably). Therefore in digital interconnects transmissionline effects DO matter. In analog interconnects, the signal frequencies are way too low to cause transmissionline effects. Just calculate the characteristic impedance of a standard audio cable (signal or speaker) and try to find the resulting phase shift at anything less than a Mhz.

Jan Didden

 

[Circlotron]

Re: audio transmission lines

Except Tube_Dude is correct, the transmission line issues don't matter below 20khz! I challenge someone to show me a measurement of any cable artifact that can be reproduced with typical cables and equipment AND can be argued is audible (i.e. it's amplitude is greater than -70db below the signal).

"Typical cables" is probably the important part here. Back in the days before optic fibre, telephone conversations were routed all over a city in pure analogue form with the help of repeater amplifiers over copper wires with a characteristic impedance of 600 ohms for wires on poles. Propagation speed is about 150,000 miles per second for this open wire stuff. If you were talking to someone 20 or 30 miles away and you knew what you were listening for, you could hear reflections of occasional impulse type noise along the length of the line and back. Not many people have 30 mile long audio cables in their listening room but if they did then this audio transmission line thing would be a concern. As it is now, I don't perceive any real problem with "typical cables" and for that I am happy.

 

[EternaLightWith]

Like cellphones?

Is that like when talking on a cell phone and sometimes, with a bad connection, you can hear yourself speaking like 250ms delayed "echo".

If this is true then, we're talking about many miles right? Satellites are alot higher up than 30 miles.

David

 

[annex666]

With regards to the transmission line argument...

Do we all agree that electromagnetic waves exist?

It can be proven both mathematically and by measurement that moving charges create effects not only in their locality, but also very far away - this observation we describe as a field (i.e. the area within which one object has an effect on another object with which it has no physical contact).

These moving charges create both an oscillating electric and magnetic field that exist at 90 degrees to each other - if these are expressed as two vectors E and H a third vector can be described called the Poynting (not pointing) vector that is at 90 degrees to both fields (E X H = S - cross product) and describes the propagation of the electomagnetic wave through space (or infact any medium).

The "Transmission line" is a fairly misleading concept as it is often taught, "it's a few bits of wire that looks like coax or something" - infact any two conducting object can act to some degree as a "transmission line".

It is more prudent to think of a "transmission line" as an electromagnetic wave guide as this is infact the phenomenon people are as observing - an electromagnetic wave being guided by whatever physical structure is defined as the "transmission line".

When an electromagnetic wave reaches an impedance mismatch a reflection is observed - this is why we used a "matched load" system, to minimise such reflections. Although it is much less often explained that ANY electromagnetic wave will act in this manor - for example when light (an electromagnetic wave) reaches an impedance mismatch (air-glass interface) some of it will be absorbed (remember the absorbtion co-efficient tau) and some of it reflected (likewise the reflection co-efficient rho).

The impedance of a medium is not as simple as you expect - when measuring a simple resistor the impedance is simply the ratio of voltage and current (Volts / Amps), but in general the formula for impedance is the ratio of the electric field (E) and the magnetic field (H) - (volts per meter / amps per meter).

Therefore any medium can have a characteristic impedance and is subject to what you may call "transmission line theory".

In short any two wires WILL exibit "transmission line" characteristics, but the measurable effects may well be negligable at the frequencies you observe. Therefore it is often not of interest to simulate wires as "transmission lines" as this would require a large amount of unnecessary calculation.



I appologise if none of that makes sence - it's late and I just stumbled across this thread and thought I'd throw in what I know.

 

[SY]

You have to be careful about scale; at audio frequencies, the wavelengths are so long that cabling has negligible transmission line characteristics. In order to demonstrate any tl-like effects, you have to go up to frequencies where the cable is an appreciable fraction of the wavelength. This is important for video signals and high speed digital, far less so for audio.

 

[michaelab]

Just a quick note to say that the article by Douglas Self about subjectivism linked to by nw_avphile on the first post of this thread has moved. It's new location is:

http://www.dself.dsl.pipex.com/ampins/pseudo/subjectv.htm

A very interesting article it is too

Michael.

 

[audio-kraut]

@michaelab - thanks for this excellent article.

But it is like this - whoever wants to believe will not be swayed from the path of righteous subjectivism, anymore than a believer will be deterred from his religion.

I for some time build my audio chain with pro equipment - cheaper and mor reliable than any high falutin burmysteries...

annex666 - ithe question is - what are the effects on my ears after that electrical information has been converted to air pressure differentials.

 

[DrDyna]

I'm gonna necro this old thread, as I've become quite interested in actually trying to build a portable null test device specifically for cables.

I would like to make a box that's got a selection of plugs on it, and perhaps a headphone jack so that the nulled output from cable vs cable can be listened to while a live source is playing.

This sounds like a fairly trivial thing, right? I only know enough about electronics to be dangerous, so if anyone has any guidance on the subject, I'll take everything I can get.

 

[DF96]

Almost all the difference between audio cables is due to connector problems, and cable capacitance, and possibly cable resistance. You will need to eliminate these as issues in order to compare the cables themselves. The actual differences between cables are negligibly small. I would regard this as being a highly non-trivial thing to test. It will be easy to get 'results' and almost impossible to get any meaningful results other than 'null'.

 

[DrDyna]

Almost all the difference between audio cables is due to connector problems, and cable capacitance, and possibly cable resistance. You will need to eliminate these as issues in order to compare the cables themselves.

I wouldn't want to eliminate anything that would be present if the cable was simply used normally.

Maybe I'll just pay someone to design a 2 channel buffer one inverted that outputs a summed signal, it seems trivial. 4 rca jacks for 2 loops, a mono input, and a headphone jack for listening to the residual.

 

[DF96]

I wouldn't want to eliminate anything that would be present if the cable was simply used normally.

In that case what you will be listening to is the low pass filter formed by the output impedance of the source and the cable capacitance. As capacitance for normal cables does not vary much (except according to cable length) what you will hear is largely down to the source, not the cable. As I said, getting a meaningful result about the cable is non-trivial. A much simpler way of measuring cable capacitance is to - wait for it - simply measure the capacitance of a length of cable with a capacitance meter - or calculate it from the cable datasheet.

 

[DrDyna]

Right, I'm not expecting any meaningful output on the summed side, that's the whole idea, build a box that can dismiss stupid cable claims audibly.

 

[DF96]

The problem you will have is that an HF rolloff at, say, 200kHz will not be audible when the cable is simply used as a cable but if you null off the difference you will be left with an audible signal. You may be accidentally making a cable troll feeding station!

 

[DrDyna]

Wouldn't it just be the 200khz part left though? I might be thinking about this wrongly, lol

 

[DF96]

A 200kHz rolloff will start doing some phase shift from 20kHz. Almost certainly inaudible, but when you do the subtraction you get some 20kHz left because the original and filtered version are different phase although almost identical amplitude. That is why any null test has to equalise frequency response on both paths - the very thing you say you are not going to do. As I said, getting a result is easy; getting a meaningful and useful result is hard.

 

[DrDyna]

Ahh, ok. That makes sense.

So, it would have to be filtered...and that would give everyone something to point at.

At the very least would have to LCR the cable to see if it was going to be problematic.

 

[jcx]

just amplifying, listening to a Linear difference doesn't signify

this really is well known tech - and the concerns in null difference testing do require some EE understanding of signals, amplitude and phase

and small Linear differences simply don't excite most knowing both EE and some Psychoacoustics, Sound reproduction realities such as Transduction Physics and variability of source, room, speakers, head position...


an example of the possible concerns:

measurements just show some measurable skin/proximity effect in the last octave with even zip cord

http://www.firstwatt.com/pdf/art_spkr_cable.pdf

Pass measured 14 mOhm DC, rising to <50 mOhm/ft in 18 gage zipcord by 20 kHz - that would need 20 ft to get 1 dB rise with 8 Ohm speakers

and you should compare to known Psychoacoustic thresholds too when planning a listening test:

Clark's ABX frequency response jnd threshold curves

Quote:
...
Clark, David L., "High-Resolution Subjective Testing Using a Double-Blind Comparator", Journal of the Audio Engineering Society, Vol. 30 No. 5, May 1982, pp. 330-338

ABX Amplitude vs. Frequency Matching Criteria

it seems like in most home audio speaker cable situations skin/proximity effects will not be expected to reach audibility

try reading what has been done, and how to understand the calcs without which you can't really plan a measurement, interpret the results