Goal: Perfect Transfer of Signal at Power Amp Output
Terminals to Input Terminals of Speaker

Conditions:
Very Low Output Z Voltage Source,
Output Voltages up to 100V,
Output Frequencies up to 21 kHz/70 kHz at close to full power,
8 ohm nominal load, 1.6 ohm dynamic load (or lower) with
reactive components at phase angles up to +/- 60 degrees.
Current up to ten's of amperes.

Minimum Performance Requirements:
Dynamic Range > 100 dB.
Simple Distortion < -80 dB.
Complex Distortion <-60 dB
(Preferably much better, of course)

Important Design Considerations:
Minimum Resistance (less than 0.1 ohms for run to speaker)
Minimal Cable Inductance (try to keep to 2X resistance @ 20 kHz,
this controlled through geometry)
Use Least Distorting Materials (bare copper/silver free from
oxidation and a dielectric with minimal capacitance aberrations
and the most linear capacitance)
Maintain Physical Geometry under signal conditions (minimum
"magnetostrictive effect" or motor action between hot and ground
strands causing motion and induced EMF's)
Long Term Stability (lack of oxidation/corrosion effects)

Relatively Unimportant Design Considerations: Capacitance (under 0.1 uF to avoid modern amplifier instability) Overall Impedance or load matching (a cable Z of 8 ohms) is not required. Appearance/Handling/Cosmetics (obviously, a retail product has to worry about those things!)

Specifics (my own theories and speculations): Tinned conductors: The question has been raised, why bother with bare copper if the components inside an amplifier, etc. are constructed with tinned leads, etc.?

My take on this is to consider that the speaker cable carries the most current of any single component for the longest distance (anywhere from 10 feet up to 30-50 feet as needed), so that any effects due to surface "contaminants" would be at a maximum. I consider "tinning", nickel plating, silver plating, or any other material on the surface of the conductor to be a contaminant.

How is a tinned conductor a contaminated surface? Pure metals have a crystalline internal organization molecularly, this is basic metallurgy. Platings or coatings do not get applied in a manner so as to totally avoid disrupting the surface of the base metal. And the plating/coating itself is highly disordered, with a large region where the two metals are mixed haphazardly, severely disrupting the normal crystalline structure of both materials. Any plating or worse, a coating, further disorders the copper/silver metals crystal structure at the surface, making worse whatever disorganization is already present due to wire drawing, extrusion into an insulator, bending, etc.

The skin effect/self-inductance pushes the current flow out towards the surface. Skin depth is defined as the depth at which the current flow is down to 37% of the total. But the distribution of the other 63% of the current flow is not linear in distribution towards the surface of the conductor. It is heavily skewed toward the outermost layer of conductive material, with a very high proportion of that 63% of the current, estimated at greater than 90%, forced toward the very outermost surface. Hence the surface purity/linearity of the conductor becomes important to passing a low distortion signal, especially at high frequencies where the skin effect/ self inductance comes into play even more.

One other fact about tin plated wires. For commercial wires, a tin plated wire has 6% higher DC resistance than a bare copper wire of the same gauge. The tin plating isn't as conductive as the copper, and is placed where it will do the most harm to conductivity at high frequency's.

Geometry, Inductance and Surface Coatings: The inductive reactance in a typical zip cord wire causes the AC resistance to begin to rise above the baseline DCR level somewhere in the midrange, usually by 1 to 3 kHz. It is at this point that the self-inductance/skin-effect comes into play and causes not only the cables effective resistance at this frequency to rise, but the forcing of most of the current towards the surface of the conductor. This aspect is often overlooked. It is implicit in the relationship between the inductance of the cable vs. the DCR of the cable.

Low-inductance cable designs not only minimize the variation in frequency response due to amp/cable interaction with the speaker system impedance, but helps reduce any dielectric nonlinearity effects. Inductance by itself shows up as a direct effect in the frequency response, these material problems are secondary, and would manifest in low-level nonlinearities.

Magnetostrictive or Motor Effects: The common reference to magnetostriction uses the term incorrectly, as true magnetostriction would have the conductor itself changing dimensions. This is not the case to any degree. What does occur is movement of the conductors relative to one another due to peak current flow, and the induced back EMF due to the motion while carrying current. This is one reason soft insulator/spacer materials should be avoided, stiff dielectrics like teflon or polypropylene or even polyethylene will allow less movement and reduce any magnetostrictive/motor effects.

My first introduction to this effect was with a pair of the old Discwasher Smoglifters, a very loosely braided cable that would show a pulse of motion with amplifier turn-on thumps.

A simple demonstration of the presence of this effect is to take a loop of thin limp wire several feet long, lay the wire out in a long skinny "U", with the spacing about 1/2" and pulse some current through it (a car battery or such), with the usual precautions and disclaimers, i.e., don't put the wire in your pants and do this, always wear safety glasses, and have a fire extinguisher handy. A pulse of current should cause the wires to twitch or move noticeably.

Vinyl Insulation (dielectric) Concerns: Vinyl as an insulator/dielectric has a high dielectric coefficient and a large amount of dielectric absorption. The dielectric coefficient of vinyl varies with frequency, that is, the coefficient does not remain constant. While the dominant action of concern in a speaker cable is the current flow, and related inductive effects, there is still a fairly high voltage present too, one of the highest in the system (unless the power amp is a tube amp). This voltage can be corrupted by the delayed discharge involved with dielectric absorption.

A simple demonstration of D.A. can be had with a hand full of parts: a power supply (could be a battery), an electrolytic capacitor of several hundred uF, a pair of (alligator) clip leads, and a high impedance voltmeter. If you don't own these or know anyone who does, you could probably talk the local Radio Shack counter dude into a demo with store parts. Charge the electrolytic capacitor up with the power source (observe proper polarity and stay within the voltage rating of the cap, easy if your using a 9V battery) and then remove the source. Measure the voltage stored on the cap with the voltmeter. It should be close to the voltage of the power supply/battery. Now discharge the cap with one of the test leads, holding the wire on the cap terminals for a good couple of seconds, while the voltmeter is still connected . The cap will have been completely discharged to some extremely small voltage, perhaps 0.001 volt. After removing the shorting wire, watch the voltage reading. Not only will it start rising in spite of the load from the voltmeter, but it should rise to an appreciable % of the original applied voltage, if 9 volts, it could ultimately reach a volt with some caps. Now that's distortion, something for nothing.

Most electrolytic caps have at least several % D.A., some even higher. Vinyl has so many different formulations, that it is difficult to even state a range, but generally in the one to several % region.

>From the results of controlled listening tests, the following materials preferences have been established.

-Speaker Cable Materials-
================================
-Preferred Conductor Materials-

In descending order of preference:
Bare copper
Enameled copper
Tinned copper
Silver plated copper

NOT RECOMMENDED FOR SERIOUS AUDIO USE AT ALL:
Cadmium copper, beryllium copper and other copper alloys.
Nickel plated copper
Silver plated copper clad steel
Copper clad steel (Also called Copperweld)
Tinned steel
Bare steel

-Insulator Materials-
================================
-Inner Conductor Insulation-

In descending order of preference:
Foamed Teflon (TFE)
Solid Teflon (TFE)
Foamed FEP Teflon
Solid FEP Teflon \
Foamed Polypropylene / These two are real close
Solid Polypropylene
Foamed Polyethylene

NOT RECOMMENDED FOR SERIOUS AUDIO USE AT ALL:
Solid Polyethylene
Rubber
PVC (Polyvinylchoride) \ These two actually attack most conductors
Polyurethane / over a period of time, the severity
depending on the exact formulation.

-Filler or molded insulation-

Fillers with lots of air are best.
SEE "Inner Conductor Insulation" for order of plastic style fillers Cotton
Rayon
Nylon
The above fibers would all be placed just below foamed polyethylene, but above solid polyethylene.

Summary
My ideal speaker cable would have low series resistance across the entire audio band in order to minimize the variation in frequency response due to the amplifier/speaker cable impedance interactions with the speaker impedance. This would be achieved by utilizing any one of a number of low-inductance geometry's with a sufficient equivalent AWG. It would have the most linear materials available in terms of conductors and insulators, in order to avoid any materials related non-linearities. It would have the rigidity to minimize self-motor action "magnetostriction", to avoid any potential for generating it's own motion related distortions. Finally, it would have the proper material formulations to avoid long term degradation effects from reducing the performance over time.

High Performance DIY Speaker Cables from Commercial Wire:

It is possible to obtain high-end speaker cable performance from commercially available wires/cables. The information for these home-made cable recipes was obtained through subjective listening tests, and is based in part on earlier work done on line-level interconnects. Materials used in the recommended wires/cables are of the highest quality with regard to conveying an audio signal. Some people may feel that the cables presented here are wild overkill, I feel that they are barely adequate. You have been warned!

Cable #1 - Highest Audio Performance

This is a difficult cable to assemble/solder, it is extremely stiff, and the cost is somewhat high. But it sounds great, and could be a contender against ANY retail high-end speaker cable at any price!

This is a cross-connected dual-coaxial cable with foamed/solid Teflon insulation, and bare copper conductors, using heat shrink tubing with an inner layer of adhesive to bind the two individual coaxes together. Cross-connecting a pair of side-by-side coaxial cables involves electrically connecting the center conductor of one cable (cable A) to the shield braid of the other cable (cable B), and the center conductor of the other cable (cable B) to the shield braid of the one cable (cable A) at both ends. Cross-connection is used to reduce the inductance to an absolute minimum. Merely paralleling the center wire and shield would create two separated different polarity composite conductors with an inductance much higher than the cross-connected pair.

The coaxial cable used is Belden #89259, the heat shrink either SPC Technologies type PHD-032 1/2" adhesive-lined polyolefin shrink tubing, or 3-M EPS-200 1/2" adhesive lined, flexible polyolefin heat shrink tubing. The cable and heat shrink are available from electronics distributors all across the country.

If you don't need all the length, cut the coaxial cables appropriately shorter, and use the leftover 89259 for interconnects; it makes very high quality line-level cables when used with a nice teflon/gold-plated RCA plug. The heat shrink tubing runs around $44 for just barely enough to do the job (a package of 5 four foot lengths, for a total of 20 feet), and it might be worth it to get twice that. Use of Belden 82259 is not recommended for this application due to it's use of a modified PVC jacket (Flamarrest) which will be exposed to the cable's electric field due to the cross-connection. 89259 has a teflon jacket which will not compromise performance as the PVC might.

Precise wiring instructions follow. The coaxial cables are stripped back 2.5 inches of the outer jacket, the braid combed and dressed to one side, and the center conductor cut off for 1", and then stripped off for 1". The entire center conductor from the shield braid point is bent at a right angle to then wrap the exposed center conductor wire around the base of the adjacent cables braid. The adjacent cables entire center conductor is then bent toward the base of the original cables dressed out shield braid, and the exposed center conductor wrapped around the dressed braid. Make sure that each center/opposite braid pair of conductors does not touch or make contact with the other pair. Dress out the leads and color code one coax braid/cable to be the hot side with a slice of heat shrink (1/4") or using a red marker pen. The two wrapped areas are now soldered, using plenty of solder and heat. Solder quickly, so as to not melt the inner insulation very much. (My favorite solder is Kester "44" in the 63/37% ratio)

Now, before completing the other end of the composite dual coaxial cable, spiral the two cables around one another, being careful not to twist them or stress the coax unduly. Spiral them as tight and neatly as you can. Two spirals per foot is enough if it stays tight. Then slide as many of the sections of heat shrink tubing as you decided to use over the unfinished end, and space out accordingly.

Pre-form the dual cables in the shape/layout you desire, making gradual bends, no sharp sudden radical angles, and make sure you have each bend covered with a piece of heat shrink (or two, as needed). A minimum radius would be 6 inches, and let the cables "slide" into a top-bottom configuration for bends made in the horizontal plane & vice versa.

Once the cable has been pre-formed to the shape for it to connect to your power amp, and run to the speakers, carefully melt the heatshrink onto the cable pair, keeping the pair as close and snug as possible all throughout the length. Heat shrink should cover at least 1/3 of the distance along the cables, and be at every bend. With 3-4 inch length pieces, spaced every foot or so, this will keep the cable pair from being able to move relative to one another. This is important to reduce any possible motor action effects.

Once the cables have been heat shrinked or secured together, the other end can be similarly terminated, and be sure that the same coaxial cable is coded at this end as the first end.

Once completed, check for shorts with an ohmmeter.

You now have two flexible copper braid hookup leads that should fit readily into most amplifier posts. If you place a termination on them, do it as well as you can: if a crimp, use the correct tools, if you solder it, use plenty of heat and flux. A good crimp should NOT be soldered! It could actually degrade the connection.

Measured specifications: Equivalent to 10 Ga. wire, capacitance per foot approx. 48 pF/ft. Due to the low capacitance, this is excellent for longer speaker cable runs. The extremely low resistance and low capacitance do not come with a penalty of high inductance, the inductance is very low at 0.067 uH/foot. Only the ribbon cables are lower, and they have a much higher capacitance.

Speaker Cable #2 Another option is Belden's Vari-Twist cables. They make a Teflon insulated version, but unfortunately, it has silver plated copper conductors. Now I find it to sound a bit bright or "silvery", but many who have heard it liked it very much, as the brightness is very clean and has no trace of harshness. Belden part #'s 8V28040/50/60 (40 wires, etc.), and they are expensive. Measured parameters on these make them the lowest inductance (with a very reasonable capacitance for the inductance) cables available, every other pair has a reverse twist, and they are spaced apart on the ribbon substrate. Again, all the tan "ground" wires would be connected together, and all the colored wires connected together for the "hot" lead.

Measured specs: 20 pair - 255 pF/ft., less than .009uH/ft, equivalent to 15 Ga. wire in resistance. 25 pair - 318 pF/ft., less than .007 uH/ft, equiv. to 14 Ga. wire in resistance, 30 pair - 382 pF/ft., less than .006 uH/ft, equiv. to 13 1/2 Ga. wire in resistance.

Note: This is available only directly from Belden on special order, and there is a minimum and a long wait for it to be manufactured.

Speaker Cable #3

Another option is to utilize plenum type Category 4 or 5 computer wiring. Readily available, and relatively inexpensive, this type of wire comes in many different configurations. The main things to look for are in these Category 4 & 5 cables are: FEP Teflon insulation (plenum version), unshielded and bare copper conductors. The shielded versions shouldn't be used, as well as the plain PVC insulated versions or those with tinned wire. Excellent cost per foot for the performance.

The most common type has 4 pairs of 24 Ga. wires, when the hots and grounds are paralleled, the 4 pairs of wire equal a 18 Ga. single pair of wires. Two pair cable is available, but I would not recommend buying it for use, as the equivalent Ga. is only about 21 Ga., and doubling or quadrupling up on it is not as tidy. OK if you have a spool laying around though and don't want to buy the 4 pair and up.

If you are going to buy it, this is the preferred part #: Four pair 22 Ga. Belden #1557A, aprrox. $400 per 1,000 foot spool, less resistance than 16 Ga. wire, equivalent to 13 Ga. when doubled up (two four pair cables in parallel). Runs approx. 72 pF/ft. for a single 4 pair channel, and approx. 160 pF/ft for the double run heat shrunk together. Belden #1561A is an 8 pair version in one jacket, more convenient than bonding two together, runs about $820/1,000 feet.

Four pair 24 Ga. Belden part #'s: 1457A or B, 1585A or B, 1661A, 1669A Mainstream availability makes these easy to find. Double up for an equivalent 15 Ga. wire, and a quadruple run is better than 13 Ga.

Eight pair 24 Ga. Belden #1573A, roughly equivalent to 15 Ga. wire, a double run is better than 12 Ga. More convenient than two 4 pair cables doubled up.

Twelve pair 24 Ga. Belden #1577A, roughly equivalent to 13 Ga.

Two pair 24 Ga. (if you have some laying around) Belden 1590A, 1657A and 1665A. Must be used in multiple pairings to get the effective wire gauge up to a decent level.

Construction is straightforward, solder all the solid colored wires together for ground, and solder all the striped wires together for hot. If doubling up on a cable, it is recommended that adhesive style heatshrink of the appropriate diameter be used to lash the two/four cable sets together (remember the recommended heat shrink is a 2:1 shrink ratio, do not get too small of a diameter, or it could split when shrunk). See construction details for Speaker Cable #1 re heatshrink usage.

Other than these, there are very few options for low-cost DIY high- performance speaker wire. If the thought of soldering all those connections is a put off, then my personal opinion is that Kimber Kable has the best low-cost speaker wire with the most universal application to a wide diversity of stereo systems.

WARNING! These cables all have higher capacitance than ordinary zipcord, enough that some poorly designed amplifiers may oscillate into these cables! Most all modern amplifiers should not have any problems, but there may be some exceptions. Use at your own risk!

ADDENDUM

Different Cable Configurations:

Belden 82248, is a foamed teflon insulated coaxial which can have the 18 gauge solid bare copper center used for twisted pair interconnect or multiple conductor speaker cable.

For multiple strand speaker wire, use in multiple pairs. Twist the pairs around in a neat pattern, and secure with heatshrink every foot or less. For a four wire (two pair) version, connect the two opposite wires together for one polarity, and the other two for the other polarity, or the classic star-quad wiring. Four conductors will be equivalent to 15 Ga. overall. For more than two pairs, interleave the plus and minus polarity to minimize inductance.

Belden 89292, is a foamed teflon insulated coaxial cable which can have the 14 gauge solid bare copper center used for speaker cable, as above. Two pairs (four wires) will be equivalent to 11 Ga. overall.

Cable Costs and Availability:

The various Belden cables I recommend are not run of the mill Ham radio coaxial cable. In most cases, the local Belden distributor will have to order a spool from Belden to get it for you. Many local Anixter distributors in major metropolitan areas will cut to length, or at least sell a leftover portion of a cut roll.

Belden 89259, My highest recommended multi-purpose audio cable, coaxial. Can be used for interconnects, speaker cable and digital cables.

Belden 82259, 89259 without the teflon jacket, OK for interconects,
not recommended for use as speaker cable. Available only in 1000 foot spools

Belden 89207, twisted pair, 100 foot spool

Belden 89272, twisted pair, avaialble in 500 foot spool

Belden 1506A, used for digital coax 75 ohm, available in 500 foot spool