Re: [SI-LIST] : Guard banding

Andrew Ingraham (ingraham@wrksys.ENET.dec.com)
Tue, 11 Nov 97 17:29:26 EST

I did some rudimentary simulations with a 2D field solver some time
ago, looking at grounded guard traces. Even with ideally grounded
guards, the coupling (crosstalk) was unchanged or slightly worse when
the guard traces were added, keeping the other traces in place.

This was pretty much what I expected. This topic has come up here a
few times before. One of the responses, I think from a consultant or
someone at a board shop, was that they've seen cases where designers
eventually ripped out the guard traces to get their boards to work,
thus leaving bigger separations between signal traces. The guard
traces serve a purpose only by forcing the signals further apart.

Grounded or driven guard traces have a place in very high impedance,
low frequency applications where surface contamination is an issue.
For RF, if you have the room for guard traces, don't add them; just
leave bigger separations between your signals and you may be
better off.

There doesn't seem to be a lot of good research into guard traces
at RF. Don't hold your breath waiting for that article.

Regards,
Andy Ingraham