Re: Shielding/ Guardbanding

Heinz Blennemann (heinz@live.engr.sgi.com)
Tue, 1 Jul 1997 16:57:31 -0700

I too have heard arguments for and against guardbanding
critical traces (analog or digital).

The main reasons for guardbanding I have heard are
1. reduced coupling to neighboring lines
2. reduced EMI
3. guardbands provide return current for a
signal line crossing a slotted groundplane

The counter-arguments are
1. to reduce coupling, it is more space efficient
to simply space the signal lines further apart
(in the cases I have analyzed, for constant Zo
and crosstalk, "spacing further apart"
has indeed allowed higher densities than "guardbanding")
2. I have been told that EMI due to vias is more
significant than EMI due to a stripline that
does not change layers. So, for non-adjacent
signal-layer-jumps (see Prof. Merkelo's or
Dr. Cheung-Wei Lam's work), for guardbanding to be effective,
one would also need to "guardband vias."
In order to complete the return current path,
a signal via would be paired with a ground via.
Such doubling the number of vias may
be undesirable in routing-constrained layouts.
3. as for using guardbands for crossing slots,
even though it is a good idea to avoid the slots
in the first place, guardbands have an application here.

So, my vote is, in general, avoid guardbands.

Any feedback, particularly from someone who has modeled
the EMI effects?

Heinz

On Jul 1, 3:26pm, Sean Murray wrote:
> Subject: Shielding
> I'm a little confused on the subject of shielding (guard traces). I'm
> currently designing a board for a customer, in which the analog & digital
> traces are on seperate layers. Both layers are shielded by their respective
> planes (i.e. Digital by Dgnd; Analog by Agnd). My question is; do the
> analog traces still need to have guard traces to protect against any noise
> that might be coming from other analog traces on that layer? (and likewise
> for Digital). The customer has stated that he wants the traces shielded,
> but when asked why he could not give me a scientific answer. I was under
> the understanding that you only needed to shield the traces if they were
> very high speed and were not utilizing proper ground shielding (planes)
> techniques.
>
> Any help?
>
> Regards,
> Sean Murray
> M&M Specialties, Inc.
>-- End of excerpt from Sean Murray

-- 

------------------------------------------------------------------------------- HEINZ BLENNEMANN Silicon Graphics heinz@sgi.com (415) 933-4236 -------------------------------------------------------------------------------