Measured (!) speaker cable difference!?

[Raindog123]

I'm not sure what the fuss is about.

All true... Except the difference between various measured cables seems to be an order of magnitude higher (in dB's) than what others report...

 

[amirm]

As mentioned by @SIY, this type of analysis and measurement has already been done and published. Here is where they net out:

https://www.aes.org/e-lib/browse.cfm?elib=5975
Effects of Cable, Loudspeaker, and Amplifier Interactions
Author: Davis, Fred E.
JAES Volume 39 Issue 6 pp. 461-468; June 1991

 

[preload]

As mentioned by @SIY, this type of analysis and measurement has already been done and published. Here is where they net out:

https://www.aes.org/e-lib/browse.cfm?elib=5975
Effects of Cable, Loudspeaker, and Amplifier Interactions
Author: Davis, Fred E.
JAES Volume 39 Issue 6 pp. 461-468; June 1991

View attachment 148332

View attachment 148333

This.

 

[preload]

All true... Except the difference between various measured cables seems to be an order of magnitude higher (in dB's) than what others report...

Eh? No, the FR variation measured here is well within what has been reported elsewhere. -1.5dB in the highest octave is totally within range (see the Davis paper from Amir above). And 0.5dB at a lower frequencies is also perfectly reasonable depending on the particular impedance curve of the loudspeaker.

 

[Raindog123]

Thanks, @amirm!

-1.5dB in the highest octave is totally within range (see the Davis paper from Amir above).

Sorry, can't see 1.5dB. Which figure is it? I see a non-linearity 'dip', eg, in Fig. 14. But the curve 'bundle' is pretty tight everywhere (<0.5dB).

 

[preload]

Thanks, @amirm!





Sorry, can't see 1.5dB. Which figure is it? I see a non-linearity 'dip', eg, in Fig. 14. But the curve 'bundle' is pretty tight everywhere (<0.5dB).

Figs 13 and 14 in that paper report -1.5dB at 20khz across multiple cables. You'll have to look at other sources if you want to see 0.5db differences in other parts of the audio spectrum because again it varies based on the impedance curve of the speaker chosen for the measurements.

As a separate source look at the effects of simple dc resistance - it creates FR deviations on the order of 4-5 dB!
http://www.roger-russell.com/wire/wire.htm

 

[Raindog123]

Figs 13 and 14 in that paper report -1.5dB at 20khz across multiple cables. You'll have to look at other sources if you want to see 0.5db differences in other parts of the audio spectrum because again it varies based on the impedance curve of the speaker chosen for the measurements.

I think we still do not understand each other. We do not argue that there is variation with frequency. The conversation is that between various cables, the difference is not significant (thus audible):

 

[escksu]

Absolutely unsupportable. Cables can certainly affect sound reproduction. It's been documented that net impedance (both resistive and reactive) affect amplifier frequency response in many cases. And cables can vary significantly from one another in their reactive load, capacitance and more especially inductance.

Now, this doesn't mean that stupidly expensive cables ever make economic sense, but that's another issue altogether.

I think we still do not understand each other. We do not argue that there is variation with frequency. The conversation is that between various cables, the difference is not significant (thus audible):
View attachment 148336

OK, regarding the article, I don't have access so unable to read it. Could you describe a bit more on how the testing is done? Since most speakers have crossovers, I would say its entirely possible for cable + crossover measurements to be very different from just cables alone. This is why I always feel that measuring cables alone isn't going to be adequate, its has to be together with the speaker. Since it may not be esay to simulate an actual speaker, perhaps it might be easier to simply use a mic to measure the sound produce. If a cable does produce any difference, this should be also reflected in the sound produced.

OF course, there are always experimental errors involved since air temp/pressure and even equipment temp affects sound produced. However, this should not produce big variations once equipment is sufficient warm-up.

 

[solderdude]

To be sure, what are A’ and B’?

The important connections you didn't draw.
A' is the - side of the amp output and B' is D but at another point.

 

[amirm]

OK, regarding the article, I don't have access so unable to read it.

There is a copy of it online: http://tmr-audio.de/pdf/kabl_cap.pdf

 

[solderdude]

I think we still do not understand each other. We do not argue that there is variation with frequency. The conversation is that between various cables, the difference is not significant (thus audible):

Between the cables tested the difference is kind of insignificant.
Take a look at the delta at 10kHz (something most old times can still hear and those are the audiophile ones.
The difference is 0.15dB there. The worst cable/speaker combo seems to be 0.2dB.
You can consider this audible, I don't or at least it is highly questionable but certainly not night and day.
This is a 3m cable, no idea what diameter/geometry cables though nor speaker impedance.
Amp B seems to have a non close to 0 output resistance and given its FR response appears to be a tube amp with output transformer.
The 300b amp Clorin used seems to be superior in FR (but not in output resistance ..perhaps)
Also note that the 300b amp, aside from high output R has an output transformer so the ground path differs so that's why it has a different shape.

It certainly is not 1.5dB for double the wire resistance as measured here. Given the measured speaker impedance it makes no sense either.
It is very clear the measurements are incorrect. They simply can't be correct. It is pointless to find out why unless Clorin wants to make more measurements using the same test setup. I don't recommend it the setup is clearly incorrect.

Look at PMA test results for reference to reality.

In short:
The measurements seem to be incorrect due to the execution of the test setup. No one can replicate it as info to do so is missing.
Measuring cables using suitable gear and connections is not that hard but just like any other measurement must be executed well. No point in figuring out exactly what went wrong (maybe more than 1 aspect), If someone wants to measure cables the same way result (no accurate description present) it simply is not possible unless the exact same wiring and gear is used. Then you get the same (but wrong) result.

Better to redo the test but this time with a proper setup.

 

[KSTR]

Let's look at the original "frequency response" again:

It is blatantly obvious (and was mentioned already) that this is NOT a frequency response of the voltage at the speaker terminals. We can see it scales with current (as implied by the speaker impedance), more current gives higher reading, and the higher the cable resistance the larger --> this is a current** measurement with the cable as the sense resistor/impedance. And of course the differences seen have nothing to do with actual voltage...

EDIT
**) current is dominating but some part of voltage is also present, this is why the "no cable" measurement doesn't give an even lower values than the thickest cable. Also it seems to be subtractive as at some points the cables drop below the "no cable" line which can only be the result of cancelling (and note current and voltage are in phase exactly at the high points of the speaker impedance and those points render lowest here --> so what was measured seem to be a*(speaker voltage) + b*(voltage drop along the cable), with a and b having opposite signs.

 

[pma]

Look at PMA test results for reference to reality.

This is my typical measurement of speaker cable transfer function when loaded with a real speaker (more in the link already posted). It is a voltage transfer of

A(f) = V(speaker terminals)/V(amplifier terminals)

In my case the ripple due to interaction of cable impedance with speaker impedance is <0.2dB. It will differ a bit with another speakers but will not get as high as 2.5dB. As you have already written, the test of the system has to be performed first to exclude possible loop voltage induction. Both measurements of speaker voltage and amp voltage must be made with balanced inputs. Purely resistive load may be used instead of the speaker to verify the measurement accuracy. R+L to R calculations are trivial.

 

[solderdude]

6m of 0.75mm extension cord will be a lot worse of course but won't be reaching the levels Clorin got at 4m using a rather 'friendly' 8 ohm speaker.
It's a bit silly to connect speakers with 6m of 0.75mm2 extension cord though. This is asking for audible differences.
Not one person serious about audio and highe end speakers will ever use such cable. Fun to test but pointless.
There are nice tables showing what diameter wire is needed for which length and which speaker.
In the end its mostly voltage division due to the ohmic component in the cable and for longer lengths at frequencies above 10kHz other aspects are at play.

 

[escksu]

There is a copy of it online: http://tmr-audio.de/pdf/kabl_cap.pdf

Cheers!! Thanks!!

 

[audio2design]

And with s 8.4ohm resistor at the cable end it looks like this.
View attachment 147639

With the 1.5mm cable versus 6mm cable into 8.4 ohms there should be a very noticeable level difference due to the resistive losses. There is almost not difference in your graphs. This tells me there is a flaw in the measurement.

 

[audio2design]

The Audio Critic had an interesting feature (I've tried unsuccessfully to find it again) in which the effect of cable impedance (and of that, largely inductance) on amplifier frequency response. I think this was back in the 1980s, so amp design varied from today. Bottom line is that high-inductance cables could have pretty sizeable impact on the FR of the power amp.

That's not news.

 

[audio2design]

You say the amp output impedance is .1 ohm. I don't think it stays there at high freqs. or the plots into a restive load would be flatter.

Cable inductance. Wide spaced cables can have high inductance. Rise seems a bit large but the no cable is not flat so add the two.

 

[audio2design]

Does anyone else have a problem with a signal being 3db hotter with a cable than without? Somethings not right with these measurements.

Obviously flawed measurement.

 

[Thomas_A]

I use very long thin cables for my surrounds. A WAF thing.

Nice to see the classic Davis paper. I have paper copy in my “audio research” binder.