In The Trenches
by Gerard Fonte
The Business of Electronics Through Practical Design and Lessons Learned
In The Trenches
Printed Circuit Board Layout
APrinted Circuit Board (PCB)
is more than just traces that
together. It is an integral part of any
design. A good PCB design is one
that you never notice. A bad design
can cause headaches for years.
Basically, a PCB is an insulating
base material containing copper traces.
The insulating material is typically
epoxy fiberglass. Less expensive
phenolic is sometimes used where
performance is not critical. For
special cases, teflon and ceramic
bases are employed. Occasionally,
there are other materials, but that’s
rare. The standard base thickness is
typically 1/16” (or 0.062”). However,
small boards (a few inches square or
less) may have a reduced thickness.
The vast majority of PCBs made
in the US start with a solid copper
side that is well bonded to the
base material. The thickness of the
copper is designated by “ounce.”
This refers to the weight of one
ounce of copper spread over one
square foot. “One ounce” copper
plating is about 0.003” thick. Two
ounce copper is 0.006” thick, etc.
NUTS & VOLTS
The desired traces are printed
onto the copper in some manner (to
be discussed in more detail later)
with a material called “resist.” The
board is then placed in a chemical
bath that dissolves — or etches — all
of the copper not covered by the
resist. The chemicals most often
used by hobbyists are ferric chloride
and ammonium persulphate.
Commercial PCB fabricators
generally use different chemicals that
are cheaper, like ammonium chloride.
They heat the bath and add catalytic
agents to improve performance.
After washing and cleaning the
etched board, the result is a copper
circuit pattern on an insulating base
— a printed circuit board.
Before Surface Mount Technology
(SMT) arrived, there were two
designated layers: the component side
and the solder side. These are
self-explanatory with standard
through-hole components. Since SMT
allows components to be placed on
two sides, these designations have
become “top” and “bottom,” but the
older terms are still widely used. The
convention is that all layouts are
viewed from the top side. This makes
the bottom layer similar to an X-ray.
You see it through the insulator. The
result is a mirror image with left and
right reversed. This can be very
confusing if you are not aware of it.
There are also two additional
types of “layers” that are not real
layers — these are the “silkscreen”
layer and the “solder mask” layers.
The silkscreen layer is an aid to
assembly and troubleshooting. It’s
just text (with occasional, simple
graphics) that identifies each
component and its orientation. A
good silkscreen can save tremendous
amounts of production and repair
time during manufacturing.
The solder mask is used to
control where the solder is applied to
the PCB traces. You only want solder
to connect the components to the
solder pads. You don’t want solder
on all of the other traces. Again, this
is very useful during manufacturing.
Hobbyists who make only one or two
boards with a simple design don’t
really need a solder mask or a
Okay, so most of you know all
that. Did you know that, at one time,
all PCBs had only a single side of
traces? Nowadays, the typical PCB
has two trace layers (called a double-sided board). Complex boards — like
motherboards — may have four to six
layers (where each layer is insulated
from the other by board material).
There are some multilayer boards that
have eight or more layers of traces.
How many layers are necessary?
In theory, if you use only
through-hole passive components
and discrete transistors, a single
layer is all that is necessary. Most RF
boards have traces on only one side,
although they use the second layer
as a ground plane to improve
performance. If you add
standard-footprint DIP (Dual-Inline
Package) opamps, you can still get
by with a single layer. (I’ve never had
to go to two sides.) This is because
there is plenty of room for traces to
pass between the leads of the
When you go to digital logic or
other designs that use high pin-count
chips, two trace layers are needed.
When you have a chip with lot of pins
in a fixed physical relationship, it is
sometimes impossible to connect
them to another chip with a different