one for the theorist

[Haible Juergen]

	>That's what I think, too. Nonsense.
	>People could think that they could get a equivalent concentrated
element >circuit
	>by just counting pcb line lenght and thus dropping a L here and
there.

Right. OTOH, the rule of thumb ("mm = nH") will remarkably well explain
some "unexpected" signal forms (where people thought it was "only a piece
of wire").

	>In the case of several connected current loops this will obviously
not work.
	>There will be problems doing the "large run" and smaller loops and
	>comparing the L.

If your wavelength is large compared to your structures, you can still make
a good approximation of your different loops using inductance and mutual
inductance models for the separate parts. And you might find that in a
simple
case such as GND plane and no neighboring loop the "wire" would indeed
make the most dominant contribution of the loop - therefore the rules
of thumb.

	>There is "inner" Inductance of a wire, ie. that part of the whole
	>inductance caused by current distribution in the wire only.

If it's the currents thru *other wires* you're after, this can be covered
with
mutual inductance. But if your frequency is too high (so high that
open wires will work as transformers ...err ... antennas), the 
epsilon*dE/dt in the space between your wires will cause significant
contributions to the magnetic field,  i.e. you need a full wave field
solution.

	>This makes sense, but is only a part.

Two completely different steps of simplification, from full wave Maxwell's
equation to various inductance and mutual inductance contributions,
and from the latter to something as "stupid" as nH/mm.

Each step of simplification has its uses.

JH.