Q&A with Jim Thiel on Internal Wiring of Speakers

During our review of the THIEL Audio SCS3 system, we pinged designer Jim Thiel on some technical questions regarding the loudspeaker designs and the choice of internal wiring and crossover topologies he employed in the speaker designs.  With his permission, we captured the interview which oddly focused more on the topic of cables than it did loudspeakers as can be seen below.

Italics = Jim Thiel's commentary and responses

Normal text = Audioholics Responses

Jim Thiel:
For the internal wiring we use a custom-made wire that is a solid, 18 gauge, oxygen-free, Teflon insulated, twisted pair that I think sounds best and I will try to explain why by first explaining some things about amplifier cable. It is necessary that amplifier cable have very low electrical resistance for two reasons. The first is that any resistance will cause a loss of power, the degree of which is determined by the impedance of the speaker. If a speaker has lower impedance then the cable's resistance will be a higher proportion to it and therefore "use" a higher proportion of the amplifier's total power.

Audioholics:
I agree cable resistance is paramount and should be kept minimum. That being the case, why not use a lower gauge wire, say 10AWG to further reduce DCR? I don't think it will matter much for a few feet of internal wiring, but since you make a case for it, please refer to the AWG Table below. This is for Solid Core Wire, Stranded wire is slightly less because of higher cross sectional area.

Note that 18AWG wire has over 6 times the DC resistance of 10AWG. Thus for amplifier cable runs of say 20ft of 10AWG wire (not considering LC losses) we have about 40mohms of DC resistive losses (below 50kHz - as AC resistance does double due to Skin effect), thus total loss across a 4 ohm load would be 20*log(4/(4+.04)) = .086dB - hardly significant.

Jim Thiel:
I am afraid I was not sufficiently clear. In the above I was first explaining some things about amplifier (-to-speaker) cable, as contrasted to speaker (internal) wiring. I was discussing the importance of low DCR in amplifier cable so that when I later explained why very low DCR is not required in our speakers it would not be thought that I was making that claim for amplifier cable. (For example, if the cable has 4 ohms of resistance)

Audioholics:
What cable has 4 ohms of resistance? Even 20ft of 18AWG only has 260mohms of resistance!

Jim Thiel:
Of course no cable has 4 ohms resistance. It is just a convenient value to use for illustration. and the speaker has 4 ohms of impedance, then the cable would use up half the amplifier power.) Now, in practice, a bigger problem than the loss of power, which is usually not more than 1 dB, is that since the speaker's impedance is (maybe quite) different at different frequencies, the power used up by the cable will be different at different frequencies. This causes the frequency response from the speaker to be altered by the cable's resistance, with very little reduction at frequencies that have high impedance and greater loss at frequencies where the speaker's impedance is low.

Audioholics:
True but again even the few feet of 18AWG wire internal in your speakers adds about 40mohms of resistance. If the driver impedance low point is 4 ohms, this represents 20*log(4/4+.040) = .086dB add that to the 20ft of 10AWG speaker cable and you get about .17dB of resistive losses which are still insignificant and really not worthy of debate.

Jim Thiel:
Well, you may be right that .17dB is insignificant. I do however believe that even this small amount of inaccuracy is audible under some circumstances and in fact I spend a great deal of engineering development time reducing response errors that are of this degree.

(By the way, this is one of the reasons we use additional components in our crossover networks which are not in the signal path but are for the purpose of evening out the impedance so that cable losses will be uniform at all frequencies and the response will not be altered even if the cable's resistance is significant.)

Audioholics:
This is an interesting goal, one that would certainly be advantageous when using high resistance exotic speaker cables. I would like to see how this works if you would so kindly share a schematic for it.

Jim Thiel:
The sub circuits added are usually one or two of either a series connected capacitor and resistor or a series connected capacitor, inductor and resistor. These series circuits are added across the input terminals, in parallel with the speaker. The first circuit can correct an impedance that rises at high frequencies and the second can correct a hump in the impedance.

Audioholics:
What you are describing here is a classic Zobel network usually used to flatten out impedance of a speaker to counter the rising impedance of a tweeter voice coil, or in some cases to restore amplifier phase margin and increase stability for heavy reactive loading due to an exotic cable or complex impedance of a speakers crossover. This is a fairly standard practice but I fail to see how this will uniformly counter resistive losses due to cabling. My advice here, if you think it is of concern, is to use heavier gauge wire internally in the speaker cabinets and recommend consumers to use heavy gauge speaker cabling (say 10-12AWG).

Jim Thiel:
The second reason amplifier cable resistance needs to be low is to avoid significant intermodulation. Intermodulation is when the signal to the woofer affects the signal to the tweeter, or vice-versa. Since current that will be directed to the woofer by the crossover network will cause a loss in the cable by its resistance, the tweeter will "see" this reduced input also. So at frequencies that are reproduced by both drivers, there will be distortion produced in one driver by the current used by the other driver.

Audioholics:
Again I don't see how cable resistance in a decent 10AWG speaker cable would be detrimental. What you seem to be describing is back EMF, which can be more problematic with series crossovers than with parallel networks, see:

http://www.sound.westhost.com/parallel-series.htm
Series vs Parallel Crossover Types

Jim Thiel:
Well, again, I agree that this will not be a problem with amplifier cable of sufficiently low resistance. And again I am just explaining why amplifier cable resistance needs to be low; why, say, 4 ohms is not sufficiently low.

So for both these reasons we need amplifier cable to have low resistance and the most straightforward way to get low resistance is to use a large gauge wire, but this causes a practical problem in that the wire gets very stiff and unbendable. Therefore stranded wire is used but this causes sonic problems because the current will pass from strand to strand and encounter non-linear, distortion producing extra resistance.

Audioholics:
Non linear distortion in a cable? We haven't found this to be true after extensive testing and calculations. This has been an unfortunate fallacy promoted by many exotic cable vendors for years, whom incidentally offer no measurable or analytical proof to justify these types of claims. I realize Thiel supports many high end dealers who sell exotic cables that market this fictitious claim, but you may wish to reconsider making such a declarative statement about this, especially in light of the inarguable proofs we offer rebuking this in the following articles:

Cable Distortion Audio Cable Vendor Skineffect hjohnson

How does strand interaction produce extra resistance? I don't understand.

We realize many exotic cable vendors buy solid core wire in bulk from manufacturers such as Belden and therefore attempt to rationalize its superiority by introducing pseudo sciences, but please realize over the past two years we have written extensive articles about cables, while demystifying many of the snake oil claims.

Please see: Cable articles

As a result, consumer awareness about cable snake oil is at an all time high. It would serve all manufacturers best interests to be more objective about cables and not support baseless claims to server dealer sales of exotic cabling. We have nothing against solid core wire, or expensive cables for that matter, providing that the manufacturer claims aren't without merit. As a result, many legitimate cable vendors advertise on our site and others have even updated their literature to be more factual-based on provable cable theory.

Please note our contributing staff (in addition to myself) includes a list of field experts such as Dr. Howard Johnson, Henry Ott and engineers from Wayne Kerr - a Leading Test Equipment Manufacturer

Jim Thiel:
A "fix" for this is to separately insulate each strand but then you can have the problems caused by poor dielectrics (insulators) since good dielectrics cannot be made very thin.

Audioholics:
What defines a good/poor dielectric for speaker cables? You may wish to review our article on this topic. Other than serving the role of insulation and controlling cable capacitance,cable dielectrics really serve no other role on speaker cables, which will influence measurements or sonics.

Cable Dielectric Absorption

Jim Thiel:
I agree that much of the promotional claims about amplifier cable is baseless and/or misleading and also that much high end cable is overpriced (sometimes drastically) in terms of performance. However, I believe that some cables do have detrimental effects on sound quality that are not explainable by resistance or inductance. Whether these almost always subtle distortions are caused by dielectrics, oxidation, bi-metal plating, etc are not determinable by me. But of course that something cannot be proved doesn't mean it does not exist.

Audioholics:
Again I encourage you to review the article we authored on cable distortion. It clearly demonstrates there is no mechanism responsible for causing non linear distortions in wires. If there were, don't you think a credible source such as IEEE or experts in Electronics Fields and Waves experts would have authored a paper on this? The only folks touting cable distortion are exotic cable manufacturers and supplementary cable forum cult hobbyists.

Jim Thiel:
So, with this background, we come to internal speaker wire. What we do is, first, internally bi- or tri- wire all our speakers. This means that we have completely isolated the electrical current in the woofer circuit from that of the tweeter; there are no wires, even ground wires, in common between drivers, beginning right at the input terminals. The benefit of this isolation is that there cannot be any intermodulation, no matter how high the resistance of the internal wiring is. The next thing is to realize that, since we have eliminated intermodulation as a problem, low wire resistance is no longer necessary at all because the losses caused by resistance of the short lengths of internal wire is low and because any such losses can be corrected by adjusting other elements of the speaker's design, like crossover component values, driver magnet size, etc. In other words, we can engineer the speaker to have the correct response taking into consideration the small amount of resistance of the internal wiring.

But why would we bother having to take this into account? Why not just use larger, stranded wire? The reason is that by using smallish solid wire we can completely eliminate any distortions produced by stranding and/ or their insulation.

Audioholics:
Again you may wish to reconsider this viewpoint since it lacks any scientific proof, merit or credibility. There is no proven distortion mechanism for wires. Using the best test gear on the market (Audio Precision 2) we have demonstrated no nonlinear mechanism exists for this to occur. If it did, how would engineers design sensitive test equipment, space probes etc if their wires were causing so much distortion? Even if measurable wire distortion did exist, it would not even be a 1/100 th of the magnitude of distortion induced by the drivers or circuit components. Wouldn't you agree?

Jim Thiel:
Yes, I would agree that wire distortion would be 40 dB lower than usual driver distortion. (But we have reduced that by a factor of 10 also.) But even so this would not ensure that such distortion was inaudible. I don't care much if something is proven or not since my solution is of low cost. As I mentioned earlier I do believe there are sonic effects that I don't know how to measure and since I can eliminate this possibility for very little cost, I don't see why I would not do so.

Audioholics:
Wire distortion 40dB lower than driver distortion!?! That's news to me. If this were true surely a simple measurement would demonstrate this. Again I defer you to an article we already conducted thorough testing of wire distortion for a wide assortment of cables and found nothing.

Cable Distortion

Let's figure a usual driver distortion of 1% or -40dB. If we look -40dB lower than that for cable distortion, we would need a resolution of -80dB. Please note that the accuracy of our test contained resolution of greater than -120dB (.0001%) or over 100 times more precision than required, and certainly more resolution than even the best audio systems could provide, yet no presence of wire distortion was found.

Jim Thiel:
We need only one layer of insulation which can be of the best type, Teflon, without worry that the thickness of the insulation will be a practical problem. So we can choose the best sounding wire, which is smallish solid wire, without having to accept any compromises caused by (a little) higher resistance.

Audioholics:
Again I don't see how you can conclude the "best sounding" wire for a few feet of internal cabling of even the worst (18AWG wire). I assume this is merely a statement of opinion.

Jim Thiel:
Well, I have tried to explain why, in our speakers, even ten ohms of internal resistance would not cause any sonic problems other than sensitivity loss, how we wire our speakers to eliminate the possibility of other subtle factors having a negative effect on the sound, and how we can achieve our objectives without spending a lot of money that I feel is better spent on better drivers or cabinet.

Audioholics:
Ok so what you are saying is that the consumer can use any type of wire, stranded or solid core, 10AWG or 22AWG and still expect excellent performance with perhaps some loss in sensitivity when higher gauge wires are used. So why are we having such an involved discussion on wires? :-)

Jim Thiel:
Well, this had been a very long answer, and probably more than you wanted to know, but I hope it answers your question.

Sincerely,

Jim Thiel