Yup, Roger Russell is a good one, and his site has been up for a while.
The ability for speaker wires to change the net RLC characteristics of the wire-speaker system is not controversial. However, some people actually seem to prefer the sound caused by higher impedance wires, something I discovered when I started messing around with vacuum tubes. Can't find it at the moment, but there are anecdotes of homebrewers actually putting 1 Ohm power resistors in series with the speakers, and liking the sound...
Vacuum tube amps usually have a higher output impedance than do most solid-state amps. This is due to two things:
Tube output stages are usually common-cathode (whether single-ended or push-pull), so the impedance seen looking back into the amp output is the output tubes' plate resistance divided by the square of the output transformer turns ratio. Just as a rule of thumb, this will often be of the same order of magnitude as the speaker's nominal impedance for triode circuits and an order of magnitude higher for power pentodes or beam power tubes.
The output stages of solid-state amps, OTOH, are most commonly emitter followers (or source followers in the case of power MOSFETs), which have much lower source impedances than would a common-emitter circuit or a common-cathode tube stage.
In addition, solid-state amps usually use much more overall negative feedback than do tube amps. Since the feedback is voltage-derived, it reduces the output impedance of the ampby a factor equal to the loop gain. Higher loop gains, as in the case of most solid-state amps, means a greater reduction of an open-loop output impedance which was already lower than that of a comparable tube amp.
Some manufacturers of solid-state musical instrument amplifiers use current-derived negative feedback around the output stage to
increase the amp's output impedance, to get the lower speaker damping factor characteristic of tube amps. This works as long as the amp is within its range of reasonably linear operation, but kinda goes to Halifax when the amp clips and the feedback loop is effectively broken during the clipped intervals of the output waveform.
By way of numbers, some measurements I took from a Marshall 1986 50W head showed that with the output transformer on the 4 ohm tap, the output impedance at 1 kHz was just about 4 ohms- a damping factor of unity. With the "presence" control cranked (which works by rolling off the negative feedback at higher frequencies), this increased to around 40 ohms, which was probably pretty close to the amp's open-loop output Z.
Hi-fi amps aren't likely to be that extreme, but the general principle that tube amps have lower damping factors than solid state amps is pretty sound.
There have also been amps which use both voltage and current-derived negative feedback from the output, in adjustable proportion, to permit the amp's output impedance to be adjusted- even to the point of having a
negative output resistance, which, by "cancelling" the resistance of the speaker wiring and voice coil winding, can theoretically achieve an infinite damping factor. Crown manufactured such a model in the '90s, and the principle goes back to the '50s, when Bogen manufactured tube amps with variable output impedance.
Incidentally, using a source with a negative output resistance to cancel errors due to wiring and winding resistance is a common trick in speed-control circuits for small PM motors- if you've ever used a dictating machine you were probably using just such a circuit.
So basically there's two rational schools of thought -- run fat cables that lower series impedance as much as practical (I even heard of one guy on the AVS Forum who uses 00 arc-welding leads!), or run skinny cables and/or add passive elements to soften to taste. Neither requires any significant expense.
Well, back when I was running a large-club PA system, the cables I made up for my bass bins (double 18" drivers in ported cabinets) were two 30-foot lengths of 10AWG cable in parallel (effectively 7AWG), which gave me a wiring resistance of 0.035 ohm for each cabinet.
Since the speaker's input impedance (which will be a complex function of frequency) and the series combination of wiring resistance and amp output impedance form a voltage divider, the effects of excessive wiring resistance on the magnitude-versus-frequency response at the speaker terminals is measurable. I have measured peaks and dips of as much as 2 dB at the terminals of near-field monitors in a poorly-wired studio control room, while the same measurement taken at the amp output terminals was ruler-flat.
(unidirectional oxygen free copper! .... erm!)
