Re: gEDA-user: Blind and buried vias?

[Gabriel Paubert <paubert@xxxxxxx>]

On Tue, Sep 29, 2009 at 05:21:16PM +0000, Kai-Martin Knaak wrote:
> On Tue, 29 Sep 2009 17:01:23 +0000, Michael Sokolov wrote:
> 
> >> I'm told that the OMAP3430's Package-on-Package configuration requires
> >> at least six layers to get all the signals out.  Ugh.
> > 
> > OK, that explains the need for a lot of layers.  But how does the need
> > for blind/buried vias arise?
> 
> The balls of the BGA occupy most of the real estate available on top 
> layer. If all vias penetrate the hole stack, this occupation maps to all 
> other layers too.

Actually most often the top layer looks like a two interlaced grids:
- one grid of pads to solder the balls,
- one grid of vias that go allow to route the signals to other layers

That's as long as you don't use exotic techniques like via-in-pad.

Basically the two outermost rings of the BGA can be routed on the top
layer (there is room for one trace between pads). The next two rings
can be routed to the second copper layer with simple vias. But then
every additional ring needs one layer if you don't have blind vias
(buried vias are not necessary for BGA exit patterns), since there is
only room for one trace between vias. With blind vias it is possible
to route 2 rings per layer, and this adds up quite fast for large BGA.

Of course there are also sometimes stupid packaging decisions by
manufacturers, one example is the Spartan3A in BGA256 package where
they decided to put some power connections on the second ring. The worst
case: on the first ring it is easy to put a via just outside without
blocking the exit for the second ring, from the third ring on you
need a via anyway. Spartan2/2E in the same package did not have
this stupid pinout.

Therefore realistically blind vias for BGA exit patterns are only
necessary for 20x20 matrices and larger (the center is often taken
by power supplies which are easier to route).

	Gabriel


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user