Re: +3.3,5-board stackup problem

Ravinder Ajmani (ajmani@us.ibm.com)
Mon, 7 Jul 1997 13:18:42 -0400

I have one question about your stackup. If 70% of your board uses 3.3V supply
then why do you need a 5V plane. If the logic can be segmented into 5V and
3.3V sections, you can split one power plane between 3.3V and 5V, and use the
extra plane for Ground.
About the effectiveness of high-frequency capacitance between 5V and GND, I
have seen that this buried capacitance is very useful for frequencies above 200
MHz. While it is not quite possible to compensate for this capacitance, one
can try by flooding the PCB with low value (470 - 680 pF), high-quality (NPO)
capacitors. A very good paper on this subject was presented at the IEEE 1996
International Symposium on EMC, Santa Clara, titled "Investigation of EMI on
Multilayer PCBs: Radiated Emissions" by S. Caniggia, V. Costa, and L. Vitucci.
I have no experience of using mesh planes but, in my opinion, mesh planes are
not suitable for the present day high-frequency boards.
Regards, Ravinder

spm@net2net.com
07/07/97 07:48 AM
Please respond to spm@net2net.com @ internet

To: si-list@silab.Eng.Sun.COM @ internet
cc:
Subject: +3.3,5-board stackup problem

hi ,

Currently we are spinning a set of boards with +3.3v, +5v and gnd.
for planes.
We are looking at 8 layers with a stackup as follows;

1 ----- top or top
2 ----- signal gnd
3 ----- gnd signal
4 ----- +5 +5
5 ----- signal signal
6 ----- +3.3v signal
7 ----- signal +3.3
8 ----- bottom bottom

A manufacturability problem exists with a non-symetrical plane stackup
as shown. In our system , most power runs off the +3.3v plane. (70%)

I have
suggested a mesh for the +5v plane. Does anyone have experience with
using
mesh type power planes ?

How tight does the mesh have to be ? (for low inductance, good high
freq returns,ect.)

I understand I will be reducing the high frequency capacitance
between the +5v
and the gnd plane, but does anybody have any data indicating what affect
this will have ?, and what can I do to compensate, if anything.
Any papers out there on this subject? I have read some in the past
, but I looking
for current information/references.

thanks,
stuart