majenko:
High speed cabled communications is a rather involved subject...At higher frequencies you have lots of things to consider:
- The resistance of the cable
- The capacitance of the cable
- The inductance of the cable
All these add up to basically form a low-pass filter, which limits the maximum frequency that can be transmitted down the line with minimum attenuation.
Well this was the state of understanding before Oliver Heaviside solved the telegraph equations. So long as a cable has a constant cross-sectional geometry it can be modelled with a single characteristic impedance. Capacitance and inductance effectively cancel across all frequencies - provided you drive and terminate the cable at the correct impedance.
There is attenuation due to series resistance (which increases at very high frequencies due to the skin-effect) and dielectric losses in the insulation (air or foam insulation is often used to reduce the latter). CAT5 ethernet cable is designed as a transmission line (well 4 transmission lines actually). 100Mbit signalling over 100m works nicely over CAT5 because its driven at its characteristic impedance of 100 ohms. Try driving it at 1k and it simply will not work at all at that speed or anything like it.
At 100Mbit there are 50 or so bits in transit along the 100m length at any one time(!) - a transmission line carries a wave, it cannot be modelled as a simple RC filter. There is still complexity, but its to do with skin-effect, radiation losses, dielectric losses, cross-talk, impedance mismatching and bending radius.