Re: [SI-LIST] : GND on the outer layers

Douglas McKean (dmckean@corp.auspex.com)
Wed, 14 Jul 1999 10:32:37 -0700

Jon,

I worked with a vendor who just did this recently.
The answer is - it depends.

Situation I worked with started as a 6-layer
stackup such as

signal
signal
gnd
pwr
signal
signal

Much to my shagrin, simulation (I'm not sure which
package they used) showed more than 6dB reduction
in the far field by a restack such as

signal
gnd
signal
signal
pwr
signal

Respin of a new board and testing emi at 3 meters
proved no where near 6dB reduction. At best more
like 3dB. And that was unacceptable to us.

My recommendation to them prior to this little
experiment in order to keep rerouting costs and
time to market to a minimum was to ADD gnd planes
as you suggested for a 8-layer stackup such as

gnd
signal
signal
gnd
pwr
signal
signal
gnd

respecting the fact that all signal planes should be
in reference to gnd planes and not the pwr plane.
Plus, doing this all without creating an asymmetrical
board. They could not do this due to a few thru hole
devices that were already at the limit of the thickness
of the board. Adding any more planes would increase
the thickness making reliable connections impossible.

I have in fact at another company done exactly this
type of fix and easily attained a minimum 10db reduction
in the far field. But, individual anecdotes are probably
not what you're looking for. Therefore, why my answer
is - it depends.

But, as I see more and more simulators for the far field,
I'm more and more skeptical of them. There's been quite a
good debate going on within the EMC community regarding
near field measurements extrapolated into the far field
with any degree of certainty. I've argued against it
as I've seen in my own measurements in a lab.

Regards, Doug McKean

At 12:40 PM 7/14/99 +1000, Jon Keeble wrote:
>I'd be interested in opinions on whether placing tracks "inside" and
>putting GND planes on the outer layers reduces raditation.
>
>It seems to me that there is little difference regarding the high
>frequency magnetic component: the return current in an outer plane runs
>parallel with the track, and the materials have little effect on the
>field.
>
>The E field is different in that the conductivity of the outer GND
>planes would confine this component completely within the PCB.
>
>Has anybody modelled this?

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****