Difficult Loads
While it is true that reasonable quality twin cables (figure eight or zip cord) are adequate for nominal 8 ohm loads over short distances, there are a number of popular loudspeakers that are anything but nominal at high frequencies.
Two that a reader advised me about are the AR11 and the Quad ESL (old model). Both of these drop below 2 ohms in the treble frequencies. The AR bottoming out at 5kHz and the Quad at 18Khz (although anything from 15kHz to 18kHz is common). The dips are fairly sharp and so the load impedance is highly capacitive on the way down and inductive on the way up. The frequencies are high enough to not worry good amplifiers but what about the response at these dip frequencies?
Twin wire cables all have significant inductance which increases in proportion to length. With 10 amp rated twin flex over only 5 metres the response was down by 2.5 dB into one Quad ESL at 18kHz, and 3.5 dB into the other speaker which had 8 metres. This was audible and unacceptable.
The only way to reduce cable linear inductance is to make the two wires talk to each other. Running in close parallel is a start, tight twisting is better but only by using multiple wires for each and interweaving can you really get the inductance down. Several cable makers have done this and sell them as low impedance cables, which is exactly what they are. There are several different cables that use this method, and twin coaxial cable is also used to achieve a similar result.
One construction uses two groups of 72 strands of enamelled wire plaited around a solid plastic core. Using these cables with difficult loads, the droop at either 5 or 18 kHz disappeared and the sound was distinctly better. There would be virtually no other way to solve the problem short of mono amplifiers sited next to each loudspeaker.
One (potentially major) drawback occurs if you own certain amplifiers that are unstable with capacitive loads. Typical multiple twisted pair cable has about 9nF per metre of capacitance with little resistance or inductance, which causes many amplifiers to go into parasitic oscillation. The fix is simple, wind twelve turns of wire around a pen and put it in series with the beginning of the cable. This tiny coil has far less inductance than even one metre of twin flex. The other alternative is to connect a 10 ohm resistor and 100nF capacitor in series, and connect this Zobel network at the speaker end of the cable. Wiring should be kept very short.
This possible issue with speaker cables is one of very few that makes some sense from a technical perspective. There is sufficient evidence from my own measurements and those of many writers that there are indeed some detectable (and measurable) differences. With this in mind, and wanting to provide all the information I can, I have included this information - and this is the one area where properly sized and well made cables really does make a difference. If you own speakers that present a highly capacitive load, or have deep 'notches' in the impedance curve, I would take this information seriously.