[sdiy] skin effect

[jhaible at debitel.net]

> >I have a question related to the following effect:
> >
> >We can assign some inductance and resistance to a piece
> >of wire. Also some capacitance (related to other wires
> >or planes).
> 
> 
> I think we have been over this before, but a piece of wire itself doesn't 
> really have an inductance.  As with capacitance, the inductance depends on 
> the relation of the piece of wire to the "other wires and planes", 
> specifically to the current loop it is part of.


Yes. So we could introduce things like "partial inductance"; a system
that can indeed work if you also indroduce mutual inductance values
to avery other piece of wire in proximity.

But that's too difficult. (at least for me)

To get back to Martin's problem: Yes, you can use the Transmission
line model for any kind of wire (including the return path).
Wheter this is a coax cable (as the symbol suggests), or any
other wire, doesn't matter. But if your line impedance
changes over the lenght of the wire, you may have to approximate
this by many short transmission lines with fixed impedance each
(don't know how fine you have to make the increment).

The important difference to an attempt to model it with tiny
L's and R's and C's and G's is that the PSpice Lossy Line
model allows frequency dependent R' and G'. But there is
no way to assign a frequency dependent (real valued!) resistance
to a simple resistor element: This would lead to a non-causal
impulse response! (simple conclusion from fourier transform
theory). I'm not entirely sure how it can be done (frequency
dependent real value for R' and G') as part of a cable, but
some clue would be that the cable also implies a delay, so
we might avoid the trap of defining a non-causal system. (;->)

Can anybody back this up, or does anybody have more details
about the theory of frequency dependent real values and causality
in transmission lines ?

JH.


-------------------------------------------------
debitel.net Webmail