I thought I'd wait a while to see what list members came up with. You may be
getting the opinion there is more than one way to look at grounds: there is!
My world is mainly EMC, sadly few schools teach this topic, it's good to hear
you taking an interest and talking to folks that deal with it every day.
I try to think of ground as nothing other than a reference. I make allowance
for the fact that ANY connection has impedance (Z not just R), that grounds
in particular have different current circuits sharing the same impedances, and
that ground most likely will not be at the same potential at different
locations in my system. From this simple concept most of the EMC grounding
concerns can be developed...
1) Current flowing around a loop ideally has almost zero impedance. In
reality it doesn't, and the intended signal gets modified as it flows around
the loop: I think of this as the signal integrity aspect. Even though EMC
engineers typically don't like fast devices, I hold the opinion that speed
forces a designer to pay more attention to layout or their circuit just does
not work. To counteract this try to make grounds have the lowest impedance
2) Current sharing the same path as a wire that is capable of radiating,
either by making a loop or forming a dipole etc. is a problem when trying to
meet EMC Radiated Emission requirements. I keep digital circuits seperate
from my I/O circuits....
3) The reverse of 2). Where current enters your system from the outside
world, and flows in those common impedances, it causes your system to behave,
interestingly.... Again, isolation is perhaps the best starting point here
In a nutshell, teach that grounds have impedance, they are not ideal, that by
determining where circuit current flows ( including parasytics(?) )you can get
reliable operation and, knowing that noise WILL come from outside, provide a
path away from your circuit. Sounds easy doesn't it... Hmmm.
Grounding is a huge topic, I hope the collection of answers help you.
Owner L F Research EMC Design and Test Facility