Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!lll-crg!lll-lcc!well!stevef
From: stevef@well.UUCP (Steven Robert Fordyce)
Newsgroups: net.micro.amiga
Subject: Four-layer PC Boards
Message-ID: <1653@well.UUCP>
Date: Tue, 19-Aug-86 11:25:32 EDT
Article-I.D.: well.1653
Posted: Tue Aug 19 11:25:32 1986
Date-Received: Wed, 20-Aug-86 08:24:17 EDT
Reply-To: stevef@well.UUCP (Steven Robert Fordyce)
Distribution: net.micro.amiga
Organization: Whole Earth Lectronic Link, Sausalito CA
Lines: 50
Keywords: Attn Perry S. Kivolowitz.

I know this is an old subject, but I was a bit behind in my reading, and it
took a while for me to get the courage to write--this is my first posting.

Some people on this net have implied or stated that four-layer PC boards
are of higher "quality" in general than double-sided boards.  This is
not true.  I am an electrical engineer with some experience with both double
sided and multi-layer boards.  If a board works, it works--there is no
reason to prefer four-layer boards over one or two layer boards.

As a designer I like four-layer boards because they allow greater density,
reduce EMI, and *tend* to reduce cross-talk and noise.  Note that I said
tend to reduce; four-layer boards can have noise problems too.  I find that
with two layer boards I have to sit with the layout person daily to see
that the board is laid out properly with adequate ground and power traces,
separation between potentially noise sensitive and noise generating traces, 
and that the grounds for power, digital, and low level analog are kept 
separate and are connected at only one place (of course I realize the last
one doesn't apply to memory expansion boards).  With four-layer boards I
can, and do pay less attention to the layout of the board.  After all,
power and ground, the inner two layers are as wide as they can be.

There are however important disadvantages to four-layer boards.  They are
about twice as expensive, they are more difficult to rework (you can get to
the +5v and ground planes), and there is the potential for shorts between
the inner layers and feed-throughs.  On one batch of boards we received
there was a intermittent short between the ground plane and one trace. 
These were tested boards and the problem didn't show up until after the
assembly was tested and in use.  I don't want to over state this sort of
problem--it is probably rare, and shorts can occur on a two layer board,
but it is less likely and much less difficult to deal with.

On the last board I had a hand in, the way we justified the expense of a
four-layer board was that that alone was enough to get us past the FCC
without further shielding.  That saved us money over all, but if that
hadn't been true we never would have gone to the trouble of a four-layer
board.  And even so, I did have a problem with cross-talk, and ringing that
took a second rev of the board to correct.  So regardless of the number of
layers time and care must be taken do a proper layout.  And once that has
been done, cross-talk and ringing also won't be a problem irrespective of
the number of layers.

As a consumer I see no reason to prefer four-layer boards.  The Amiga
itself laid out on a two layer board.  Would it be in any sense better if
it was a four layer board?  Would it be faster, use less power, last
longer, or be more reliable?  The answer to all of these is no.  At higher
frequencies four-layer become necessary to stop excessive ringing in
switching, but the Amiga is safely below where it becomes necessary.

                                            Steve Fordyce
                                         hplabs!well!stevef