*WHITE PAPER* The Sound of Music - How & Why the Speaker Cable Matters

Go to Deletraz Fig 10.

Yes let’s look at that, where the author uses 100 meters of cable to show 10000nsec reflections that no one denies exists, but those reflections settle out. The only one that uses 100 meter cables is studios and you may want to look into what they do about source / load impedance matching.

Then the author, who does no math makes the giant and technically ignorant leap that if this is an issue at 100 meters then surely it can be an audible issue at 1 meter. That was either ignorance or intentional misleading. Math would show this ignorance. Why use 1, let’s use 5 meters about the longest most people use. The issue is not now 1/20th as small, it is 20 times exponentially smaller.



At 100 meters and 100khz bandwidth 5x audio, there are 10 round trip reflections. At 40 kHz, 25. At 5 meters, 200 and 500.


A 100khz signal at 100 meters may only settle within 60db, a 40khz signal will be 150db down. But let’s talk real world at 5 meters, where at 100khz the settling error is 1200db below the signal and at 40khz the transmission line settling error will be 3000db below the signal.

I don’t know about you, but I can’t detect -1200db errors, can you?

I suggest quoting real engineering papers not effectively blogs from those who are not thorough in their analysis.  There is a reason way more educated and experienced people on transmission lines say it does not matter at audio frequencies and cable lengths.

The quoted author then says a 10Khz square wave is audio frequencies. That should be your first clue. The fundamental of the 10khz is audible , i.e. 10khz. All the harmonics are inaudible. A 10khz square wave does not represent real audio. It is shown sometimes to illustrate amplifier stability / damping but that's a different problem.

I would withdraw the white paper. It's based on a poorly written and technically weak paper to draw conclusions that are simply not remotely correct while including erroneous and irrelevant simulations, and other noted errors. Withdrawing it would be the most honorable thing to do.

We have measured a phenomenon and have postulated an explanation that has been peer-reviewed by two highly experienced electronic engineers who agree that the conclusions are correct.
You obviously disagree, so could you please explain why there is a difference in the error voltage, shown in fig 3, that correlates directly with the characteristic impedance of the cables?

"Highly experienced electronic engineers" ....

It is not a matter of whether I disagree or not. Anyone with a basic knowledge of transmission lines, knows that the speed of reflection is so fast, that the signal quickly settles to the final value, and since we are looking at frequencies of <=20Khz and lengths of 7 meters, the error is literally smaller than 1000db.

Let’s talk about your "figure 3". As I stated previously, your scale is not defined (which another noted above), so it is hard to properly interpret comparative results, especially since in addition, you have a 2db roll-off at 20Hz, which we don’t know if it is causes by amplifier saturation, measurement roll-off, amplifier roll-off, etc.


Let’s talk about other errors. Your impedance shown in Figure 2, dummy load does not match the circuit. The 330uH inductor is 2*pi*f*l at 20KHz = 41.4 ohms. That is in series with the 7.3 resistor putting a lower limit on impedance of 48.7 ohms. The 1.4mh in the other leg is almost an open circuit at 20KHz. Yet you show 8 ohms impedance at 20KHz. Gee .... what could be wrong here? Could it be that your model is missing the Zobel that is typically on amplifiers for stability. Was that actually in your load or not? Hard to make claims when even the most basic aspects of your article are wrong.


Your appendix has a column Cable Fig. 2, but figure 2 is the dummy load, and the numbers in this column don’t actually relate either the traces Figure 3, or Figures 4-10 as there is no way to correlate them. The one marked XXXX appear to the same cable in 6, 7, and 8 meter lengths (point of that was), but in the Cable Fig. 2 column, it is marked 5, 4, and 6 which has no meaning to anything else. And then another marked XXXX that is 10 meters but is a different cable. I don’t know who you had peer review this, but you should fire them. I am up to at least 4 glaring errors without even getting to the conclusion.


So as I said again, Figure 3 is explain by inductance -- measured with a proper inductance meter close to the actual frequency, and as necessary, also taking into account skin resistance.


Of course, to properly correlate anything to Figure 3., you would need to fix the measurement errors, define what the scale actually is, define the readings on the spectrum analyzer (rms? voltage average, peak), and provide accurate inductance and capacitance values for specific cables as they relate to figure 3, something you have not done at all.

However, as I have stated in my previous posts, using educated guesses, variances of 4-5db in the cable drop method you have used would be what happens between the inductances listed for the several different types.

audio2design235 posts11-21-2020 2:58pm

"Highly experienced electronic engineers". Actual engineers do not call themselves "electronic" engineers. They call themselves electrical engineers, perhaps electrical and electronic engineers. Please feel free to share the names and education and work qualifications of those that peer reviewed it.

If you're going to request this information, then it's only fair that you provide the same.

I had actually taken that out shortly after I wrote that Cleeds as I wanted to save them embarrassment after I tried to simulate the results in figure 3 and found more and more critical and fundamental errors in the paper.  I know it was not fully reviewed by engineers with the qualifications to analyze it.

However, I have my Bachelor's (EE) from Cornell before working in the recording industry for a major equipment manufacturer, before I left to start my own company supplying custom tools and services for equipment calibration and maintenance to the industry, that I sold before going back and doing a Master's (EE) and starting my PhD at Berkeley. Never finished that because I ended up running a group developing professional audio equipment for the recording industry. They were happy with their niche, I wanted to grow, so I put together a team that worked on software plug-ins and hardware that we eventually sold to a competitor. Since then been dabbling with a number of companies across a range of tech.  In other words, yes, I actually do have a clue what I am talking about, but that is probably evident from me picking up easily things like incorrect speaker models, simulations that can't possibly match actual transmission line effects, quick estimation of cable losses, and all the other errors in the paper.

It correlates because spreading wires apart increases inductive reactance that is dominant factor here. The idea that 7m speaker cable is a transmission line for audio signals is insane, IMHO. For 20kHz signal you will need about 1000m long speaker cable (1/10 of the 20kHz wavelength) to even start becoming transmission line. In such case reflections would be inaudible because they are in MHz range (and because speakers are 1000m away).  Why not to use 20kHz sinewave for the test? It is the highest audio frequency component of interest in the cable.  Please show me reflections of 20kHz sinewave in 7m cable.  Any cable.

Oh, and your "White Paper" shows 0.94uH/meter for Isolda, but your website shows 0.002uH but does not show a length.

I looked further into the Isolda, pictures and the dimension estimates I made.  I expect that 6.6uH is quite wrong, but that 0.002uH is quite wrong too (for 2 meters).


Given the lack of consistent numbers for Isolda, could be a function of the cheap meter, or measurement error, let's use the Isolda as a reference, use some of the inductance numbers in the chart, and the shown (but wrong) speaker simulator, using inductance only.

                    My Calcs    Value on  Graph
ISOLDA       2.6              2.6         (Used as ref)
                   8.5758         8.8
                   15.210        13.8
                   24.69          24.4

That's close enough to me, to show that inductance alone completely explains Chart 3, certainly within the framework of the obvious measurement errors.

Here is what you have done:
1) Showed a graph with expected change in frequency
2) Showed that order of magnitude characteristic impedance had a rough correlation to the measured results.

What you did not do:
a) Investigate other related effects ... like INDUCTANCE.b) Show a direct correlation via a measurement of impedance and measured error
c) Show a mathematical correlation between impedance and measured error.

See, I just did #3 above, and showed a very close correlation between inductance and the measured results.  That's science.

You also left out many necessary details so that your experiment could be corroborated.
And ...perhaps most of all, you did not relate your result to what the actual change in frequency response is.
On the very worst cable it is 0.4db at 20KHz.

Not IMHO kijanki, but absolute fact that at 7Mhz it will have no effect. At 1Mhz, worst case would be -170db down.  I would be careful with the 1000 meter example. In that case, the worse case would be -60db error at 20KHz, which could be argued as audible. That is worst case though. Odds are it will be much better.

The idea that 7m speaker cable is a transmission line for audio signals is insane, IMHO. For 20kHz signal you will need about 1000m long speaker cable (1/10 of the 20kHz wavelength) to even start becoming transmission line. In such case reflections would be inaudible because they are in MHz range (and because speakers are 1000m away).  Why not to use 20kHz sinewave for the test? It is the highest audio frequency component of interest in the cable. Please show me reflections of 20kHz sinewave in 7m cable. Any cable.

Let’s forget nonsense about reflections at audio frequencies and concentrate on frequency response. Increased characteristic impedance is not the reason for the signal attenuation at the high frequencies - increased inductance is. Same attenuation can be achieved by increasing capacitance. Higher dielectric constant insulation (same geometry) will increase capacitance, but inductance will stay the same. That way we will get bigger attenuation at high frequencies at lower characteristic impedance. Finding correlation between characteristic impedance and frequency response is pretty much like saying that tattoos are causing motorcycle accidents. Correlation, a very dangerous tool, assumes that if B happens when A happens, then A has to be causing B. It completely ignores the fact that both can be caused by C. In our case increase in characteristic impedance and increased attenuation at high frequencies were both caused by increased inductance.

Damn!
This is one of the worst spankings I have ever witnessed on any board on any subject.....

The only way to get a square wave out is to match the cable impedance to the load impedance. This is a standard way of determining the characteristic impedance of unknown coax cable. The trick was shown to me when I was working on the ill-fated Blue Streak Rocket development at the Weapons Research Establishment in Adelaide in the early 60s.
See my videos Geometry Matters on youtube.

Max

Audio2design is suggesting that transmission line theory does not apply at low frequencies. Well, it does, even at DC. see

https://www.youtube.com/watch?v=ozeYaikI11g&t=136s

And at 50/60Hz bigtime. 

Max

How many square waves do we experience in audio at 10KHz. The correct answer would be 0. That has a fundamental sine-wave component at 10Khz, and every single harmonic is inaudible. It is a meaningless test.

Cool, what does Rocket Development have to do with understanding electrical theory and transmission lines?

townshend-audio OP303 posts11-22-2020 9:53amThe only way to get a square wave out is to match the cable impedance to the load impedance. This is a standard way of determining the characteristic impedance of unknown coax cable. The trick was shown to me when I was working on the ill-fated Blue Streak Rocket development at the Weapons Research Establishment in Adelaide in the early 60s.
See my videos Geometry Matters on youtube.

Max

townshend-audio OP303 posts11-22-2020 9:59amAudio2design is suggesting that transmission line theory does not apply at low frequencies. Well, it does, even at DC. see....

Now you are lying to discredit me. In fact, I have specifically pointed out to others in this thread that transmission line effects absolutely come into play even at 1KM with audio frequencies. It is readily available for anyone to see in my posts. I have copied them below.

The video you posted is not showing transmission line effects for DC, it is showing transmission line effects for the transient when a voltage is connected which is of indeterminate (due to parasitics) but very high frequency. It is not "DC" at the point of turn on, it is a transient.


Please don’t get angry at me for your lack of understanding of a topic.

audio2design245 posts11-18-2020 10:20pm

Technically there are always reflections, however, the impact on power transfer at audio frequencies is several hundred db down.

audio2design245 posts11-21-2020 10:55am

A 100khz signal at 100 meters may only settle within 60db, a 40khz signal will be 150db down.

audio2design245 posts11-21-2020 4:21pm

.... kijanki, ... I would be careful with the 1000 meter example. In that case, the worse case would be -60db error at 20KHz, which could be argued as audible. That is worst case though. Odds are it will be much better.

The main reason for the test in the first place was to find a simple measurement that quantified the difference between different geometries. The easiest way is to see what the voltage drop was across a single conductor. There is no sleight of hand or skullduggery going on here. I am an engineer and engineers only work with the truth, because everyone can see when your bridge falls down. (doctors bury their mistakes, lawyers suggest better luck next time, accountants can come up with any figures they like). What fascinates me, is the imagination of the trolls who don't want to know.

I know there is a difference between cables and the is not down to a slight frequency variation at 20kHz, there is something more going on and I have tried to identify it.

Please participate in the Zoom call at 6 PM GMT 5 Dec 2020. Details to follow.

What you are attempting to do is bury your mistake.

Townshend audio

Lets talk frankly. 
Any capacitance on a speaker cable (stand alone, just vs. a measurement equipment), is the fact there are two cables in parallel, for all the length, at a close distance.
Such a capacitance can be fixed easily by building such a cable, from two separates. Capacitance is gone!
Do you thing that now all cables will sound identical?
No!
The factor to build a speaker cables, is custom built, per the Amp's DF and length required. This would effect the cross section (in gauge or sq. mm).
Such a cable can be calculated. 
I did.
Not only that, but I conducted a test on multiple participants and found it works 100% of cases. 
I also figured, that if a cross section is calculated, adding thickness, will not make any improvement.
So I have the formula. 
It was tested.
None of the sound characteristics can be measured by any of the methods of the tests you conducted.
It is about controlling a speaker, that is a coil moving in an electromagnetic field. None of the AP tests apply.
 

BTW.

1.  Has anyone heard a difference between speaker cables?

2. Has anyone heard a difference in the sound on the clip https://youtu.be/v11hmOE1Vcc.

3. I have tried to explain it, but some are not happy. Is it just coincidence that the traces track characteristic impedance quite closely?

3. I have tried to explain it, but some are not happy. Is it just coincidence that the traces track characteristic impedance quite closely?

No, you basically started with a conclusion and then attempted to make the data match without correlating, or compensating for other variables. Keep in mind you have presented no mathematical correlation.

Zo = sqrt (L/C) .... of course impedance is going to have a relation to inductance, that value you continually ignore.

1. Has anyone heard a difference between speaker cables?

You are now appealing to audiophile emotions which is just wrong when you present what purports to be a scientific paper. This has nothing to do with whether people hear a difference between cables or not. It has to do with the fact that your paper is error filled and draws a conclusion without proving direct correlation, while investigating no correlating variables.


And still even though I have pointed out numerous errors in your paper, you have not addressed them, or retracted until you could fix them. Instead you are blaming others here. We did not write the paper.


Keep in mind, you are not just presenting an error filled paper, this is also meant to market your product.

Many people more knowledgeable about transmission lines and electricity have pointed out all the issues with this "white paper". No change has been made. That says a lot.

Exactly, what errors? Please clarify.

Have you duplicated the test?

We are doing a Zoom session 6 PM GMT 5 Dec, and we will show you the test with the same cables.

Is it just coincidence that the traces track characteristic impedance quite closely?

Characteristic impedance always depends on inductance but taking false conclusions and writing papers on it is pretty bad.  I'm sure your cables sound wonderful, but please stop this "scientific" nonsense.

US spending on science, space and technology has 99.79% correlation with suicides by hanging, strangulation and suffocation.

http://www.tylervigen.com/spurious-correlations

Exactly what errors? Really. Now that just comes across as insincere attempt at deflection. I have very clearly my posts highlighted what 5+ errors

- Speaker electrical model does not match impedance curve shown.

- I will add that as well this is I believe, as there really is not a full standard model, an electroacoustic model, but transmission lines don’t care about mechanical effects

- your model does not include parasitic capacitance or inductance in your elements and transmission lines do care about those

- no scales on oscilloscope plots

- your method that YOU used of measuring characteristic impedance is wrong, and at audio frequencies grossly wrong. You can’t use a transmitted square wave to measure characteristic impedance at audio frequencies

- your simulation is wrong. You have not simulated characteristic impedance at all in your simulation nor transmission line effects. You would need a complex model of transmission line characteristics across audio frequencies and you must include bulk capacitance and inductance and resistance and even skin effects

- you ignored skin effects though this is likely <=1db of error

- Your table of wires and their characteristic values does not correlate to the graph and confuses readers.

- you only showed a rough correlation to impedance not an actual mathematical correlation. Standard line, correlation does NOT equal causation

- you have ignored other relevant variables, namely inductance

- you called a graph the frequency response of the cable when it is clearly not. It is the cable voltage drop versus frequency which is lot the same thing

- you don’t explain or correct for the 2db measurement drop at high frequencies which without knowing why puts unbounded error on the measurements at high frequencies

- the scale of the spectrum graph is not listed, only db and this is a near meaningless term without the required scale.

- the settings for the spectrum analyzer are not listed so we don’t know if it is average, RMS average, peak, etc

I could go on but what would be the point. You fail to acknowledge the obvious and substantial flaws.

We are doing a Zoom session 6 PM GMT 5 Dec, and we will show you the test with the same cables.

Great so you are going to do a zoom call that shows that lower inductance cables have less voltage drop than higher inductance cables, something almost no on disputes and you are going to claim it is because of transmission line effects.

Townshend-audio, It appears that you joined our audio forum only in attempt to place free advertisement for your product. It is dishonest IMO and I wouldn't buy anything from you. Audio2design attempted to explain to you why using square waves in audio is nonsense, but you don't get it.

Administrator please remove townshend-audio posting. We don't want it here. You want to advertise? - pay like everybody else.

Oops forgot about that whole square wave error as well. I lost count.

Square waves have been universally used in audio since the invention of the oscilloscope and the square wave oscillator. It is an essential tool in our industry because it allows you to analyse all audio frequencies at once.

As far as accuracy is concerned, some measurements may be out by +/-20%, but that is not the point.

Join in the Zoom session 6 PM GMT 5 Dec and I will show you the experiment and you can ask any question you like.

The process of the scientific method involves making conjectures (hypotheses), deriving predictions from them as logical consequences, and then carrying out experiments or empirical observations based on those predictions.[5][6] A hypothesis is a conjecture, based on knowledge obtained while seeking answers to the question. The hypothesis might be very specific, or it might be broad. Scientists then test hypotheses by conducting experiments or studies.

We use low frequency square waves if we want to cover the audio band. A 10Khz wave may be used to show amplifier extended frequency response which is only important for showing phase shift.

A 10Khz square wave has exactly one frequency component, 1, uno in the audio band, namely 10KHz. Every other harmonic is inaudible. A 10Khz square wave does not contain all audio frequencies it contains 1, 10KHz, no more, no less.

Measurements out 20% can be important because that can be the difference between causation and correlation but you have shown no causation between impedance (measured wrong anyway) and your results. That literally does not exist in your paper.

Again telling that your ignore 10+clear errors in your paper, but try to attack one small item (wrongly).

But you go ahead and spam your zoom call. I expect you won’t allow the likes of me or Kijanki an open forum to point out to the audience how woefully flawed the paper is including the conclusion.
Scientists don’t ignore all previous knowledge and it in general is why scientists make new discoveries after absorbing all the knowledge and discoveries of those that came before them, never ignoring it, even if they don’t agree with it.

You have had lots of opportunity to fix your errors and clear up unclear info as indicated. What would I ask in a call that you have not already ignored, not addressed or got wrong already? 

Is this the same cable? 

https://audiosciencereview.com/forum/index.php?threads/townsend-isolda-cable.7109/

That’s the one @djones51 , the same thread that includes this little ditty, apparently from the starter of this topic

"Note, there is no frequency component in characteristic impedance and it applies for DC attached."

Hey note to the op, your equation for Zo is the simplified one that only applies at high frequency. The full equation for characteristic impedance ABSOLUTELY has a frequency component!

Leave the thread as a monument. I would not erase this history.

Leave the thread.

When discussing or simulating characteristic impedance and transmission lines:One must be careful not to use the short form formula for ’Radio Frequency Characteristic Impedance’ which is only valid for frequencies near one megahertz and higher.Below 1 MHz down to low audio frequencies you need to use the long form complete formula.The four important parameters (which should be measured at the frequency in question) are:
[G] Conductance ( siemens ) (was mho)
[R] AC resistance [L] Inductance
[C] Capacitance
* * * * * * * * * * * * * *
at radio frequencies the [L] & [C] values are large so the [R] & [G] values are not important.
at audio frequencies the [L] & [C] values are small so they are not important. It’s the [R] & [G] values that are important.

Here we are, December 3, and no details have "followed" about the December 5 zoom call. When has a vendor given less than 2 days notice of a presentation? That is just odd .....

That's because he's afraid you'll attend.......

Townshend is inviting you to a scheduled Zoom meeting.

Time: Dec 5, 2020 18:00 GMT

Topic: Townshend cable measurements Demo + Q&A.

To demonstrate that characteristic impedance matters in the case of speaker cables.

Forum questions regarding voltage and oscilloscope timebase settings will be addressed.

The session will have the following format:

A brief overview of the paper.

Introduction of the test set-up.

A live repeat of the experiment.

An opportunity to discuss the theory and test practices.

To facilitate preparedness, should you have any specific questions, please feel free to pre-submit them to us. This is not a requirement.

Open section for ad hoc or any further questions concerning cable testing.

Join Zoom Meeting

https://zoom.us/j/93569703283?pwd=Q3hHOStzU0dENnArY0lGV2F3ZGJFZz09

Meeting ID: 935 6970 3283

Passcode: JYzH2K

The session will be recorded.

Please come prepared to discuss the following questions, as a start.


1) Why does your speaker electrical model does not match impedance curve shown.

2) Why are you using an electroacoustic speaker model that is not accurate for electrical transmission line effects.

3) Why are you not incorporating parasitic capacitance which is essential for proper transmission line modelling.

4) What are the scales on the oscilloscope plots.

5) Why do you use the wrong formula for impedance? The formula you use only works at high frequencies, not audio frequencies.

6) Why do you use an incorrect method of measuring characteristic impedance in yoru paper.

7) Why do you insist that a 10KHz square wave is "audio" frequencies, when only the fundamental 10KHz sine wave in that square wave is in the audible band.

8) Why are you passing off a simulation that is obviously not incorporating properly characteristic impedance as a simulation of characteristic impedance in a transmission line circumstance.

9) Why are you passing off a simulation as accurate that does not take into account the very significant change in characteristic impedance of cables at audio frequencies.

10) Why have you ignored skin effect.

11) You have shown no CAUSATION between characteristic impedance and cable voltage drop.  You have only shown at best some correlation, but no causation.  Why do you feel, given the lack of proven causation, that you can say definitively it is due to transmission line effects?

12)  Given the posted impedance curve of your simulated load, why do the cable voltage drops NOT follow the impedance mismatching that must occur as the impedance of the load changes with frequency. Given your hypothesis, the voltage drop w.r.t. frequency MUST follow the cable impedance / load impedance graph. It does not.

13) You have completely ignored inductance as a causation. Why?


14) Why does your table of wires and their characteristic values does not correlate to the graph and confuses readers.


15) Why did you label a graph the frequency response of the cable when it is clearly not. It is the cable voltage drop versus frequency which is not the same thing. This is deceptive to an uneducated reader.


16) Why is there a 2db measurement drop at high frequencies which without knowing why puts unbounded error on the measurements at high frequencies.


17) What is the scale of the spectrum graph is not listed, only db and this is a near meaningless term without the required scale.


18) What are the settings for the spectrum analyzer as they are not listed so we don’t know if it is average, RMS average, peak, etc

Well Max showed, and continued to ignore 100+ years of settled engineering. He was joined by David from Wireworld who communicated his own brand of lack of engineering knowledge. Altogether it was as expected.

Oh ... get this, the source impedance was not an amplifier with say close to 0, or at least 0.1 ohm, it was a 400 ohm source impedance. According to Max, that does not matter cause it acts as a current source.

I feel sorry for those potential customers who will be mislead by someone who passes themselves off as knowledgeable, but clearly, on this topic, is not.

kfscoll,

The construction of his wire, will result in low inductance which is good for high frequency transmission. I will not fault it on that, though some amplifiers may not like it.

However, that does not negate the paper which is poppy cock.

Zoom meeting with Max Townshend here: https://www.youtube.com/channel/UCKruAdjBL1lxmYMZiXe9P2A

White noise 😳
uh, ok then...happy trails!