Re: [SI-LIST] : Modeling BGAs with floating ground plane

Raghu (raghu@apsimtech.com)
Mon, 27 Apr 1998 16:28:08 -0700

The effect of a floating ground plane is to reduce the inductance. Eddy
currents are induced in the floating plane and this reduces the
inductance. Of course, the closer the floating plane is to the paths or
filaments, the lower the inductance.

As you have said, the inductance can be calculated relatively easily
without the plane. The introduction of the plane increases the
computation significantly. A common approach is to grid the plane and
find the currents in the grids and hence the current distribution in the
plane.

A possible reference for this is

T.S. Smith and C.R. Paul, Effect of Grid Spacing on the Inductance of
Ground Grids, Proc. IEEE Symposium on Electromagnetic Comaptibility, pp
72-77, Aug. 1991

There is a BGA example with ground plane containing lots of holes,
etc.in

Raj Raghuram, et. al. , Inductance Computation of Multiple Arbitrarily
Shaped Planes, 3rd Topical Meeting on EPEP, pp 216-218, Monterey, 1994.

The best way, in my opinion, to model SSN noise from this would be to
make a n-port circuit at the terminals or pins. Use both a static 3D
inductance solver (as above) and a 3D capacitance solver. Once you have
the n-port (after a few days of computation!) you can do very fast SPICE
runs by connecting any kind of drivers and loads you want. This is
because the package is linear. In principle (and practice) you can find
the currents in every part of your structure, including the floating
planes. Once you have the currents, you can attempt to calculate the
radiation.

-- 

************************************************************************ Raj Raghuram Home address: Applied Simulation Technology, 1483 Eddington Place 2188 Bering Drive, San Jose, CA-95129 San Jose, CA-95131. Tel: (408) 252-1285 Tel: (408) 434-0967 ext.101 Fax: (408) 434-1003 e-mail: raghu@apsimtech.com ************************************************************************