06-07-11: MadfloydCapacitance and shielding, although the significance of shielding is lessened if the interface is balanced.
So capacitance is the main parameter to consider for longer runs (where the lower the better)?
Shielding will affect susceptibility to both noise pickup caused by rfi/emi, and ground loop-induced hum and noise. The likelihood of both of those things becoming an issue increases if the cable length is long. Some cable designs achieve increased shielding effectiveness at the expense of increased capacitance, so there may be a tradeoff that has to be considered.
As far as lower capacitance being better is concerned, what matters (if you want neutral behavior, as opposed to a softened upper treble) is that the capacitance has to be low in relation to the output impedance (at high frequencies) of the component driving the cable. Reducing capacitance beyond that point would be overkill.
That can be calculated. Determine the capacitance C of the total length of cable, based on its specs. Calculate its capacitive reactance Xc (the capacitive form of impedance, measured in ohms) at 20kHz, based on the formula
Xc = 1/(2 x pi x F x C)
where Xc is capacitive reactance in ohms, pi = 3.14, F is frequency in Hertz, and C is the capacitance of the total length of cable in farads (1 farad = 1 trillion pf). As long as Xc is much higher (say 5 or 10 times or more) than the output impedance at high frequencies of the component driving the cable, you are good to go, as far as capacitance is concerned.
All of this applies, btw, just to line-level analog interconnects. Speaker cables, phono cables, digital cables, etc. are a different story altogether.
Regards,
-- Al