[SI-LIST] Re: How to connect vias to power ground plane?
- From: Tom Woodward <tomw@xxxxxxx>
- To: zhang_kun@xxxxxxxxxx
- Date: Thu, 19 Dec 2002 11:53:39 -0500
Zhangkun and SI-list colleagues -
I assume that "full-connected" means a direct tie to the plane, while
"orthogonal-connected" means a plane tie using a "spoked" or "thermal"
connection. The rest of this note is based on that presumption.
This is a question that has concerned me for some time. I believe that
most people would assume, as you did, that the direct tie presents lower
resistance and inductance. However, if you consider skin effects, I
think that the spoked connection offers better characteristics. This
is because the small plane openings that are caused by the spokes
create surface-conductive connections between the top and bottom
surfaces of the plane.
The skin effect in conductors means that the top and bottom surfaces
of a power or ground plane are effectively insulated from each other for
100+ MHz signals. Such signals cannot transit through the bulk copper
of the plane, but must find a surface-based path. The most likely
such path is around the edges of nearby anti-pad openings in the plane
or around the edges of spoked via openings. If you use directly tied
vias (no spokes), currents must find their way to the nearest anti-pad
in order to transit between the top and bottom surfaces of the plane.
This increased path length translates into increased resistance and
inductance.
To visualize the problem, imagine a decoupling capacitor on the top
surface of a board, tied by vias to two planes that are separated by
a thin dielectric. The inter-plane capacitance is effectively limited
to the interior surfaces of the two planes (because of skin effect),
and we would like our decoupler to tie to those interior surfaces with
as little inductance as possible. If we use directly tied vias, then
the via that connects to the topmost plane has no surface-based path
to the interior (i.e. the bottom) surface of that plane, except through
nearby plane openings. The length of this path is unpredictable because
it is dependent on how far away the nearest anti-pad happens to be; in
any event, the increased current path length means that the mounted
inductance of the decoupling capacitor is higher than it would be if
we used spoked vias.
I don't know if this effect is actually causing the results you are
seeing in your simulation, but I believe it is real, and that plane
connections from vias should use at least two spokes. I think that
the only time a direct connection is better is when heavy DC or low
frequency switching currents are being carried. In those cases, the
full peripheral via contact to the plane should produce a slightly
lower resistance.
Regards,
Tom Woodward
Woodward McCoach, Inc
Zhangkun wrote:
>
> Hi all:
>
> In PCB, which method to connect vias to power ground plane do you favour,
> full-connected or orthogonal-connected? I think full-connected will be better
> for less resistance and less inductance. Recently I have done some simulation
> by means of Ansoft. The result surprised me. The inductance and resistance of
> orthogonal-connected mode is less than that of full-connected mode.
>
> If anyone are interesting of my simulation result, please inform me. I will
> send it to him. Si-list does not permit to add attachment.
>
> Best Regards
> Zhangkun
> 2002.12.19
>
> ------------------------------------------------------------------
> To unsubscribe from si-list:
> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>
> or to administer your membership from a web page, go to:
> //www.freelists.org/webpage/si-list
>
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
> or at our remote archives:
> http://groups.yahoo.com/group/si-list/messages
> Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
