Do speaker cables make any difference?

[jneutron]

Originally posted by keladrin
I'm sorry but electrons do not see the amp to input connectors as a separate part of the circuit

Why are you sorry? It is not your fault an electron sees only voltage gradients and magnetic fields..


Do a mathematical comparison between the resistive losses incurred in a biwire setup vs a monowire one, for the case of a two way system and orthogonal signal components.

Not integrated loss, that is equivalent..but instantaneous.

Then, do the analysis for the inductive energy storage within the wire. I caution you, that is a more difficult analysis..

Cheers, John

[Johan Potgieter]

Yes, let us do a calculation of the resistive losses in the leads between the amplifier and the loudspeaker. Also the inductive energy stored therein. And let us also measure to correlate how our mathematics agree with practice.

"Instantaneous" will of course vary from moment to moment depending on the frequency content (never mind the loudspeaker system impedance at every frequency contained).

I believe I have an average domestic system. In my case the resistive losses amount to about 0,15dB; the inductive losses to less than 0,07 dB. (Are they losses - depends on the angle .....)

Is this instantaneous, average or r.m.s.? Who cares!
(As an EE, I have the disadvantage to be limited to practice. )

I believe in my contribution no. #1037 I said that we have then gone round the circle 4 times. If that was correct, then this makes it 5 times (to the nearest decimal). But it is an open discussion .....

Yawningly yours (it is 01:51 in the AM here, and I am going to bed).

[jneutron]

Originally posted by Johan Potgieter
Yes, let us do a calculation of the resistive losses in the leads between the amplifier and the loudspeaker. Also the inductive energy stored therein. And let us also measure to correlate how our mathematics agree with practice.

First things first. It is first necessary to understand mathematically the issue. I find that most do not first achieve this, but yet still attempt to waffle onward.

Measurement is second, understanding of course, the limitations of measurement imposed by the FFT algorithms. Recall, the error component is a zero integral entitiy which I believe cannot be seen via FFT analysis..

Correlation is third. This is the desired end goal.

As I said, understanding is first..let us not proceed beyond that until it is accomplished. For this reason, KISS is the best path (a concept I live by).

Originally posted by Johan Potgieter
"Instantaneous" will of course vary from moment to moment depending on the frequency content (never mind the loudspeaker system impedance at every frequency contained)..

That is the definition of instantaneous. As I specified, use two orthogonal signals which travel different load branches. Until what I speak of is understood, the introduction of far more complex components serves no useful purpose.

Originally posted by Johan Potgieter
"I believe I have an average domestic system. In my case the resistive losses amount to about 0,15dB; the inductive losses to less than 0,07 dB. (Are they losses - depends on the angle .....)

Resistive is losses..Inductive is simply lagged energy..it gets there, but not specifically when it was supposed to..

What is an "average" system. What is the total loop resistance to the crossover nodes?

If you peruse some very good sites for wire size recommendations, you find the recommendations are based either on overall power loss, damping factor, or some combination of both. Nowhere does anyone consider also the connector contact resistance, btw.

Consider a power loss based decision. If one says 5% is an acceptable loss, the loop resistance is simply defined. For a single load entity (no passive crossover). The loss within the cable is exactly 5% of the delivered INSTANTANEOUS power provided by the amp, regardless of frequency content.

Now, make the load a two way. ALL orthogonal signals which go through separate branches will modulate the resistive losses incurred within the wire. NOW, the power loss is no longer exactly 5% at all instants in time.....IT VARIES FROM ZERO TO 10%....yet, the rms value is still 5% loss. IT IS A DIFFERENCE WHICH MUST BE CONSIDERED first, then dismissed afterward if found to not achieve a level of effect worthy of consideration.

Originally posted by Johan Potgieter
Is this instantaneous, average or r.m.s.? Who cares!

I do, as it requires modification of wire guage recommendations based on the load configuration..and it lends credence to anecdotal accounts of differences incurred when the changeover is made from monowire to biwire...(note, I did not say it makes the stupid explanations given as legitimate, those are still hogwash).

Originally posted by Johan Potgieter
(As an EE, I have the disadvantage to be limited to practice. )

Strangely, as an EE, I do not have that limitation. The bulk of my work advances the SOTA in several fields....tasks to which I am assigned do not have previous work upon which to draw, which has some definite advantages (fun being one of them).

Originally posted by Johan Potgieter
(I believe in my contribution no. #1037 I said that we have then gone round the circle 4 times. If that was correct, then this makes it 5 times (to the nearest decimal). But it is an open discussion .....

I did not restart the topic, Keladrin did.

What strikes me as odd is the fact that despite a clearly defined difference (which should not exist as defined by our educational background), you (and oh so many others) have resisted continuing the logical path I have laid down in front of you. Non scientific types can be forgiven that failure to launch...

Originally posted by Johan Potgieter
(Yawningly yours (it is 01:51 in the AM here, and I am going to bed).

Sheesh..I thought I was bad...

Cheers, John

[SY]

John, in this context, "orthogonal" has a different meaning than I'm familiar with (in my world, orthogonal means that the dot product is zero). Can you define explicitly what you mean by "orthogonal signals" for us non-engineer types?

[keladrin]

I would be grateful if some of the explanations here were in plainer English - we don't all have degrees in advance electronic field theory. I have a higher degree and physics A level and am a published author but the last few posts seem to read like some kind of techno-babble from of Star-Trek to me - very impressive but where are the Klingons?

[jneutron]

Originally posted by SY
John, in this context, "orthogonal" has a different meaning than I'm familiar with (in my world, orthogonal means that the dot product is zero). Can you define explicitly what you mean by "orthogonal signals" for us non-engineer types?

Actually, it is the same meaning. To keep it simple, I am considering two sine signals. Orthogonality guarantees that the product of the two signals averages zero over time.

Originally posted by keladrin
I would be grateful if some of the explanations here were in plainer English - we don't all have degrees in advance electronic field theory. I have a higher degree and physics A level and am a published author but the last few posts seem to read like some kind of techno-babble from of Star-Trek to me - very impressive but where are the Klingons?

Sorry bout that chief..

The gist is: when one wire is carrying two independent signals which go to the same simple load, the losses that occur in the wire are exactly the same (albeit smaller) as the dissipation within the load. (for ease of understanding, I am considering only a resistive load for now).

When one wire carries two independent signals which go to a load which has a frequency divider (crossover), the math changes a little.

Think about when one signal is +1, the other is -1...with separate wires, each wire will dissipate power at that instant, wheereas if they are both in one wire, they cancel by addition and there is no dissipation within that one wire..


Cheers, John

[anatech]

Hi John,
I'm still of the opinion that, given reasonable wire gauge and terminations for the power level, a passive crossover and voice coil heating effects dwarf particular termination (connector types) and losses in the cable.

Put simply, my belief is that people worry far to much about speaker wire. The other considerations and loses will generally exceed those in the wire. Indeed, special wire constructions that increase capacitance may cause amplifiers to mis-behave even though the value of that capacitance will not even come close to having an effect near the audio band.

Marketing people have figured out that the average human is very much a "follower" and personal pride would allow most of us to accept marketing nonsense. This is to avoid being seen as ill informed. The very same effect allows listening groups to hear what is not there. Expectations you know.

Wiring threads. They are as bad as religion for some.

-Chris

[jneutron]

Originally posted by anatech
Hi John,
I'm still of the opinion that, given reasonable wire gauge and terminations for the power level, a passive crossover and voice coil heating effects dwarf particular termination (connector types) and losses in the cable.

We are in agreement there. The ramifications of my analysis, if duplicated, tested, and accepted, would be the modification of accepted wire losses based upon the load topology. In other words, where previously a 2 or 5 or whatever percent powerloss was acceptable for the system, now, one must consider the modified powerloss and distortion due to load branching, as that powerloss no longer tracks the delivered loss, but introduces an error.. Tracking powerloss is simple, turn up the gain. Error is not that simple.

So the solution can be via three different methods..throw more copper at the problem, split the runs and conserve copper, or another technique to be named later..

Originally posted by anatech
Put simply, my belief is that people worry far to much about speaker wire. The other considerations and loses will generally exceed those in the wire. Indeed, special wire constructions that increase capacitance may cause amplifiers to mis-behave even though the value of that capacitance will not even come close to having an effect near the audio band.

I personally don't worry about it..

L and C are inverse. The best one can do is make the cable Z equal to the load. And, keep the LC product low. Unfortunately, as you say, some amps do not like the C.[/B][/QUOTE]

Originally posted by anatech
Wiring threads. They are as bad as religion for some.

-Chris

Yah, you be right bout that..If I were approaching this from a belief instead of via mathematics and physics, I'd certainly be more emotional about it.

Cheers, John

[Johan Potgieter]

Originally posted by Johan Potgieter
[B]In my case the resistive losses amount to about 0,15dB; the inductive losses to less than 0,07 dB. [B]

John,

This is the part that summed up everything, yet it is about the
only part you did not comment on.

Your analysis was OK, sure, but in the field of physics. If something is clearly of no consequence (and I believe few would quibble about the fact that 0,2 dB is inaudible, is by comparison vanishingly small in an environment where movement of people and even temperature hardly have less influence than 1 dB, and often up to 3dB, and same model loudspeakers are not even near that consistent, etc. etc.) - haa - breathe ....., then I cannot see why in a practical discussion, where apparently the title is about what would make an (audible, I hope) difference, I must spend time on such a degree of calculation. (All breathe now.)

That was what I meant by "limited" by practice. Sure, one needs to understand this, particularly those of us who design, but then one needs to go on. Do we need to analyse basic semiconductor physics every time a circuit is posted? You and I have passed the necessary exams, but I wonder what percentage of readers here are interested to such a degree. And as said this was all posted before, as was explained that the differences are vanishingly small. Surely taking notice of that is also the duty of an EE?

No, no "accusation" that you brought this up, John - neither was it wrong, Keladrin.

And I am happy to see that you are also human, John! (i.e. working at odd hours/getting sleepy. )

Regards!

[jneutron]

Originally posted by Johan Potgieter


John,

This is the part that summed up everything, yet it is about the
only part you did not comment on.

Ah, I did, but in a rather circuitous way.

The first solution I proposed was to increase the wire guage (toss copper at the problem). This is the easiest to do, requiring only the calculation of the acceptable loss error within the system.

The second is to avoid it by separating the load wire runs. This is a better engineered solution.

The third solution, which is the most elegant, is to put the feedback node where it belongs when driving a branch load..at the load. That extends the topology of the amplifier to the load itself..removing the issue in it's entirety.

Originally posted by Johan Potgieter
Your analysis was OK, sure, but in the field of physics. If something is clearly of no consequence (and I believe few would quibble about the fact that 0,2 dB is inaudible, is by comparison vanishingly small in an environment where movement of people and even temperature hardly have less influence than 1 dB, and often up to 3dB, and same model loudspeakers are not even near that consistent, etc. etc.) - haa - breathe ....., then I cannot see why in a practical discussion, where apparently the title is about what would make an (audible, I hope) difference, I must spend time on such a degree of calculation. (All breathe now.)

As I stated prior, the historical basis for the calculation of conductor size requirements is made entirely on either simple loss, or damping. Not as a result of a complex branch load. If the dissipative error component due to branch loads is at the 1% level, is that audible? Or 2%?

I could easily just use a pair of 500 MCM cables and calculate the error as too small to worry about (which would be rather correct)...but in the real world, people use 12, 14, 16, 18 guage runs to loads that drop to 4 ohms..situations where the error component does indeed climb into 5 and 10 percent..

Originally posted by Johan Potgieter
That was what I meant by "limited" by practice. Sure, one needs to understand this, particularly those of us who design, but then one needs to go on. Do we need to analyse basic semiconductor physics every time a circuit is posted? You and I have passed the necessary exams, but I wonder what percentage of readers here are interested to such a degree. And as said this was all posted before, as was explained that the differences are vanishingly small. Surely taking notice of that is also the duty of an EE?

We are discussing ramifications which can very likely end up being part of those exams in the future. And once there, the end customers STILL will not care about the details. It will simply modify the guidelines which are used to select the appropriate wire guage or feedback topology for an application.

Originally posted by Johan Potgieter
And I am happy to see that you are also human, John! (i.e. working at odd hours/getting sleepy. )

Regards!

You should see me use a cheeze grater...in the hands of the right (wrong) individual, it can be considered a weapon of "skin" destruction..my kids rank me no end because of my callus disregard for my knuckles..

Cheers, John