[ctviggen]

This should be interesting:

Loudspeaker cables and their effects

[denjo]

Hi Bob

Thanks for the article and the interesting read. I recently bought some Belden coaxial wires (more commonly used by Radio HAM enthusiasts as antennae cable) and requested my audio cable man to terminate them for use as speaker cables. You should have seen the look on his face - and those of a group of audiophiles who were there in his shop! They rebuked me for using a coaxial cable, which they said was not suitable for use as interconnects, much less as speaker cables. I was a little crestfallen and wondered if I had made the right decision at all. It seems a gentleman named John Risch (who is well known for his home brew DIY cables, popular with DIY cable hobbyist) also feels that the coaxial design is inappropriate. Being non technically inclined or trained, but daring enough to try if such a coaxial design works, I went ahead with the objective of using just such a cable for my speakers. I should be able to report within the next few weeks whether it works or it does not.

The article seems to suggest that a coaxial design should work fine. If anyone has thoughts about coaxial designs as speaker cables, please feel free to share.

Best Regards
Dennis

[Marbles]

Kevin Haskins (DIYcables.com) made a nice set from Belden coax for me many years ago.  They worked fine.  I ended up with some nicer stuff, but those cables were pretty good.

[jneutron]

This should be interesting:

Loudspeaker cables and their effects

Quite.  Thanks for the link.

While most of it is reasonable, they messed up on the fundamentals of e/m theory.  To wit, they  don't talk correctly about inductance.  I guess that should be expected, as they are audio guys and not physics types.

I tried to register on their forum to outline their errors, but when I tried to submit, got a server busy message..sigh.

Cheers, John

[Daygloworange]

This portion of the article in regards to SC's intrigues me the most on the topic.

What is more, loudspeaker cables may be carrying 11 octaves of frequency range and not just the single frequency of an electrical supply, so what happens over the whole range of operation is of interest, and all frequencies must be passed as uniformly as possible.

I've talked to a number of people who explain to me that different frequencies travel along the surface of the wire, while others travel through the core of the wire.

Is this the case? And could that lead to certain frequencies being out of phase with others?

Cheers

[DaveC113]

I've had a couple of coax ICs, from Zu and Vampire...  they are ok, but other geometries do much better in my system. I just made some ICs with the same construction as the Speltz Anti-ICs, and they work very well. Theres a few pics in the Lab thread AWG of Anti-cables. They cost about $20 for a 1m set using the materials that I used. I think shielded ICs in general don't sound as good.

[Bob Reynolds]

This should be interesting:

Loudspeaker cables and their effects

Quite.  Thanks for the link.

While most of it is reasonable, they messed up on the fundamentals of e/m theory.  To wit, they  don't talk correctly about inductance.  I guess that should be expected, as they are audio guys and not physics types.

I tried to register on their forum to outline their errors, but when I tried to submit, got a server busy message..sigh.

Cheers, John

I'm sure we'd all like to see the corrections posted here.

[jneutron]

I'm sure we'd all like to see the corrections posted here.

No problem...  I've tried unsuccesfully to register on that site so I can post in the forum, but so far no good.  Some message about server busy or down, or sumptin like that.  I would have preferred to talk to them first, but what the heck.

Their statements are in blue..


The first way to combat the resistance problem is to shorten the cable; halving the length of the cable will halve all the impedance components. Another way to halve the resistance would be to double the cross-section of the cable, but whilst this may be effective on the resistive part of the impedance, the increased spacing between the centres of the conductors will increase the inductance. The effect may therefore be beneficial at low frequencies but detrimental at high frequencies.

Consider two #24awg wires that are uninsulated.  Put them side by side, seperated by an insulator a millionth of an inch thick. That is as close as they can reasonably get.  The inductance will have some value..

Now, do the same with a pair of #12 wires...they will have the same inductance as the #24's.  Do the same with two copper rods 1 inch in diameter, they will have the same inductance.   The key point here, is the wire size is unimportant in the inductance equation.

What matters is indeed the spacing, as was said..but it is the spacing in relation to the wire diameters.

If you take a picture of the cross section of a wire pair that has L nH per foot, no matter what size you make that picture, the inductance remains the same. Scaling doesn't change it.

Inductance can be calculated using the Terman equation.  This is comprised of three parts...the first is the external inductance of the dipole field, which is proportional to the term:  Log(D/d), where D is the wire spacing, and d is the wire diameter.  As long as the ratio of diameter to distance remains the same, the total dipole field inductance will be exactly the same.

It depends heavily on the insulation thickness which defines that ratio.

What is important is to always keep the pairs of loudspeaker wires as close and parallel as possible. This enables the magnetic fields around each core to cancel as much as possible of the inductance. Twisting the wires is another way to achieve this,

Twisting only helps keep the conductors together.  It doesn't change the inductance per se.  If you twist a zip cord, there will be no change in the inductance.

Coupling to other fields is a different entity however..

Skin effect is the tendency for high frequencies to travel through the outer skin of a conductor, and not through the centre of the core. The whole cross-section of the conductor is therefore not used, so the resistance rises as the conducting section of the cable reduces, introducing a high-frequency roll-off. Once again, the shorter the cable, the less the problem.

Some manufactures have opted to address the problem by plating the outside of the conductors with a lower resistance metal. Another approach is to use Litz-wire, where multiple, individually insulated, hair-like wires are twisted together. They thus have a much greater ratio of surface area to volume.

While litz does indeed have a greater ratio of surface area to volume, that is not the reason it works better at higher frequencies.

Skin effect is the result of the creation of potential voltage loops within a solid conductor.  When this happens, current goes toward the outside.  Litz breaks that radial conduction path, this makes the current stay within the wire strand it's been in all along.

Their figure 6 is in need of drastic repair..but I'm sure that's not a misconception on their part, but rather a limitation of their drawing tool.

Cheers, John

[Steve Eddy]

Hey John!

Skin effect is the result of the creation of potential voltage loops within a solid conductor.  When this happens, current goes toward the outside.  Litz breaks that radial conduction path, this makes the current stay within the wire strand it's been in all along.

Mmmm. That's not quite how I understand it.

Let's consider a typical, radially arranged bundle of wire. In this configuration, even when the wires are twisted, they still retain the same radial position from end to end. And even if the individual wires are insulated, eddy currents should still be greater in the wires located toward the center of the bundle and the the signal current density should be higher in the conductors toward the outside of the bundle as frequency increases.

It's my understanding that litz wire is woven in such a way so that on average, each individual wire occupies every position radially along the length.

For example, take four strands of wire and twist them together. Then take four of those bundles and twist them together in what's called a rope lay. In such a configuration each individual wire within the pitch of the twist is both in the center of the bundle and the outside of the bundle and all points in between. And if the pitch of the twisting is small compared to the electrical wavelength, the current density should be uniform in all of the conductors.

se

[jneutron]

Hey Steve, how's it goin??

Hey John!

Skin effect is the result of the creation of potential voltage loops within a solid conductor.  When this happens, current goes toward the outside.  Litz breaks that radial conduction path, this makes the current stay within the wire strand it's been in all along.

Mmmm. That's not quite how I understand it.

Let's consider a typical, radially arranged bundle of wire. In this configuration, even when the wires are twisted, they still retain the same radial position from end to end. And even if the individual wires are insulated, eddy currents should still be greater in the wires located toward the center of the bundle and the the signal current density should be higher in the conductors toward the outside of the bundle as frequency increases.

Two effects come into play in the standard use of litz.  Micro and macro effects.

Remember, the pseudocurrents which are skin effect, are toroidally formed currents which go against the primary current of the bundle.  These pseudocurrents in a bundle which can communicate strand to strand, set up the gradients which make the current tend towards the outside.

With non random litz, there is no way for the current to jump to the outer conductors..in fact, the pseudocurrents cannot be established simply because there is no radial conductivity.

Each conductor within the litz will suffer from macro effects, which can cause the current within every strand to redistribute, but each strand will not have the same current density redistribution.  That will depend heavily on the position within the bundle.  This is the effect which comes into play for hf transformers, where the higher magfields will cause higher dissipation in the wires.

It's my understanding that litz wire is woven in such a way so that on average, each individual wire occupies every position radially along the length.

Yep.  That is to prevent the macro effect of one wire being in the highfield vs others in lowfield.  That prevents the individual strands from carrying different currents.

Cheers, John

[Bob Reynolds]

Their statements are in blue..


The first way to combat the resistance problem is to shorten the cable; halving the length of the cable will halve all the impedance components. Another way to halve the resistance would be to double the cross-section of the cable, but whilst this may be effective on the resistive part of the impedance, the increased spacing between the centres of the conductors will increase the inductance. The effect may therefore be beneficial at low frequencies but detrimental at high frequencies.

Consider two #24awg wires that are uninsulated.  Put them side by side, seperated by an insulator a millionth of an inch thick. That is as close as they can reasonably get.  The inductance will have some value..

Now, do the same with a pair of #12 wires...they will have the same inductance as the #24's.  Do the same with two copper rods 1 inch in diameter, they will have the same inductance.   The key point here, is the wire size is unimportant in the inductance equation.

What matters is indeed the spacing, as was said..but it is the spacing in relation to the wire diameters.

If you take a picture of the cross section of a wire pair that has L nH per foot, no matter what size you make that picture, the inductance remains the same. Scaling doesn't change it.

Inductance can be calculated using the Terman equation.  This is comprised of three parts...the first is the external inductance of the dipole field, which is proportional to the term:  Log(D/d), where D is the wire spacing, and d is the wire diameter.  As long as the ratio of diameter to distance remains the same, the total dipole field inductance will be exactly the same.

It depends heavily on the insulation thickness which defines that ratio.

Well, part of what you said makes sense, but some doesn't follow from the equation you gave.

If we look at the log term, rewritten with S as spacing and d as diameter (so I don't get confused ), we get I ~ log(S/d). First, I assume that inductance is >= 0, so that implies that S >= d. So, I don't see how this equation can be applied to typical zip cord speaker cable, since the spacing is less than the conductor's diameter. But, if we were to keep the constraint of S >= d, then we can still do the arithmetic.

I ~ log(S/d) = log(S) - log(d). Thus, as the diameter of the wire increases the inductance decreases. In fact, as d approaches S inductance approaches zero.

What did I miss?

[jneutron]

What did I miss?

What you missed was something that should have been obvious by inspection..(I negected something)

I forgot to say that the spacing is from wire center to wire center, not from the outer surfaces of the conductors.  So when the conductors touch, the ratio is 1.  For a solid conductor, it can never go below 1.

Sorry about that, I should have been clearer..

Oh, almost forgot..here's that equation played out for some different wire guages. The inductance is per foot, btw.
http://www.audiocircle.com/index.php?action=gallery;area=browse;album=1191&pos=4

http://www.audiocircle.com/index.php?action=gallery;area=browse;album=1191&pos=3




Cheers, John

[Bob Reynolds]

Thanks for the clarification John. The formula didn't make sense to me based on the "incorrect" image I had in my head.

Funny thing about obvious -- it is only when it is.

[george_k]

Skin effect is the result of the creation of potential voltage loops within a solid conductor.  When this happens, current goes toward the outside.  Litz breaks that radial conduction path, this makes the current stay within the wire strand it's been in all along.

Granted, it's been a while since my undergrad course in EM but, if I remember correctly skin effect was only relevant when discussing high frequencies (microwave frequencies and beyond). While there may be some degree of skin effect with audio signals, I'd be surprised if would produce an audible difference, especially if it were a blind listening test.

[jneutron]

Granted, it's been a while since my undergrad course in EM but, if I remember correctly skin effect was only relevant when discussing high frequencies (microwave frequencies and beyond). While there may be some degree of skin effect with audio signals, I'd be surprised if would produce an audible difference, especially if it were a blind listening test.

At most, it accounts for 15 nH per foot, that is due to the magfield within a cylindrical conductor.  Whether or not that amount of inductive change could possibly be heard, I haven't said.  I also think it unlikely, but cannot offer  proof of that.

The skin approximation equation is off by about a factor of 3 to 5 for normal wire sizes at audio freqs, but that error is not very large given the amount of skinning that does occur.

If the goal was to make a cable with a characteristic impedance of 8 ohms using tefzel dielectric, as an example, the endgoal L would be about 10 nH per foot, C at about 300 pf per foot.  Clearly, that is not possible if the conductors are both cylindrical, as that would add 15nH per foot per wire in the audio range, the cable would only be 8 ohms in the rf regime.  I've made 8 ohm cables, but it required a double braid construction to avoid the internal inductance.

Cheers, John