Need some advice to calculate PCB stray inductance

[neel_c@xxxxxxxxxxxxxxxx]

Hello,

I would like to calculate the stray inductance of a multilayer 
printed circuit board (PCB).

Basically, I want to modelize a large PCB (50 x 40 cm). It contains 
at least 2 layers of 17.5 micron thickness copper (that covers the 
entire area of the PCB), separated by a standard FR4 (epsilon = 4.5) 
insulating layer, thickness around 0.2 mm. 

I have a power supply, located say at X = X0 and Y= Y0), (X, Y)
coordinates related to the position in the 50 x 40 cm plane. Each 
lead of the power supply connected to a different copper plane. 

I have a load, located say at X1, Y1 in the PCB plane.Each lead of 
the load connected to the corresponding copper plane. 

My goal is to calculate the inductance between the source and load 
current path.

I know the problem is basically a tridimensionnal one (Z being the 
axis orthogonal to the PCB. I did a quick 2D simulation with femm, of 
a (X,Z) cross-section (two layers with equal and opposite currents, 
basically like two 17.5µm x 10 cm copper sections, separated by 0.2 
mm air gap). I get around 36 nH/m for a 10 cm wide strip, wich I find 
acceptable as an order of magnitude.

Now, I need to get further by transposing the results on the real 
current path in the PCB. I imagine that the current density in XY 
plane is similar to a 2 dipole electrostatic E field problem...
(probably could be obtained by hand ?) Can I get the results with 
femm ? When I get the current density on each XY coordinate, can I 
sum each inductance contributions using the previous femm analysis ?

Probably this kind of problem is trivial for some of you and I would 
be very gratefull for any hint (using femm or any combinaison of 
tools, or a link) that allows to get closer to the solution.


Thank you in advance

Christian Néel

France