RE: [SI-LIST] : Earth Ground

Andrew Ingraham (Andrew.Ingraham@digital.com)
Tue, 17 Feb 1998 18:07:11 -0500

> I am teaching a class out of "High Speed Digital Design" to first year
> grad students. We recently discussed grounding issues and there was a
> lot of confusion about "digital" ground vs "analog" ground vs chassis
> / earth ground. Grounding in a cellular system .. etc..
>
> If anyone has clever ways to explain these concepts intuitively I (and
> my students) would appreciate the help. How do you view these issues
> from a practical everyday point of view?

I encourage you to steer your students away from use of the word
"ground". We all misuse it (I too am guilty), and it would help to get
them in the habit of avoiding it.

In the old days of radio, I suppose that the term "ground" came about
because AM radio receivers required connections to an antenna and to
earth. To this day, people continue to use the term "ground" as if it
is somehow magical.

But it is nothing more than a reference. One of several possible
references.

Einstein tells us that there is no such thing as "standing still."
Everything is relative. An observer picks an arbitrary world or
platform from which to observe the universe around him; but he cannot
claim that his world is the one that isn't moving.

Likewise, there is no such thing as "ground". I can pick one point and
call that MY reference, and you can pick your point and call it yours,
and both of us will probably see a non-zero voltage on the other one's
reference (relative to our own). This is true even if we both have
connections to earth ... even two connections to the same hunk of metal!

Earth ground is not some sort of giant unipotential electron sink.
There can be significant voltages from point A to point B, often both AC
and DC. Yet the earth's surface, containing water, is a useful (if
poor) electrical conductor that becomes important near antennas and
such. In that context it is significant simply because it is there ...
a large conductive boundary, which affects the electromagnetic waves
around it in various ways; bending them, reflecting them, conducting
them. Direct connections to earth ground are essential at low radio
frequencies (AM and down), where the ground is an integral part of the
antenna system. In cellular, connections to earth may be for safety
only.

In circuit design we use "ground" to denote one wire or connection point
in our circuit. It may have a DC path back to chassis and from there to
safety ground and earth ground, but that is not necessary for the
circuit to work. It's just a local reference. Nothing more.

The mistake one gets into, is thinking that ground is ground is ground.
Using "ground" as a reference only works locally. As soon as you look a
few inches away, the "ground" over there isn't the same. The "ground"
in another circuit board, or another box, can be much different. There
might be negligible DC voltage between them, but a huge AC voltage may
be present.

The other mistake is thinking that currents that go into ground,
disappear into it with no effect. As if the "ground planes" in our PC
boards are perfect superconductors, and the laws of inductance don't
apply to them.

Fortunately many folks realize that their "grounds" get dirty.
Sometimes you can't accept a noisy ground in parts of your circuits. So
you may use more than one reference or "ground". They are just
different, isolated conductors, which may eventually connect together at
some common point.

Typically, "digital ground" is noisy because lots of things are
switching fast. Digital logic is somewhat immune to it. Critical
analog circuits aren't, so one or more "analog grounds" may be isolated
from "digital ground" in some fashion to keep the noise out. But what
you have to keep in mind is where the currents go. Not only do the
currents cause the noise, but separating analog and digital "grounds"
has consequences on where the currents need to go. So it must be done
right. For example, using parallel analog and digital "ground planes"
in your PC board provides lots of coupling from one into the other; but
cutting one "ground plane" into digital and analog portions creates a
split in the plane and you'd better not run active signal traces across
that split.

Regards,
Andy Ingraham