I recently attended one of Lee Hill’s classes on EMC. The class had an large portion devoted to “ground”. Hill pointed out that the word “ground” is not clearly defined. There are two common meanings of ground.
- Power or Signal Return Path
- Safety Ground - An example is the connection between a product’s chassis to the ground prong on an electrical outlet. Current only flows during a fault condition in which the chassis is shorted to something.
When I say “ground” I usually think of the return path or the plane on a PCB, but Lee Hill discourages this. If disconnecting such a wire would make the system not work right, he calls it a return. If the ground is protection against a fault condition and could be removed, it should be called ground.
Some people think of earth ground as being an especially good place to sink noise, but this is a case of the word “ground” fooling us. Devices like battery-powered laptop computers, cars, and airplanes have no connection to earth ground and do not require special EMC considerations because of that fact.
Occasionally grounding causes a noise problem in the form of ground loops (also known as “common impedance noise coupling”). Often people are vaguely aware that ground loops involve noise from the ground resulting from multiple ground connections. Here is a model to understand ground loops.
Suppose you have these two systems: The red system and the green system
Notice the red system has its return connected to ground at both ends. On the green system the signal return line is connected to ground at the load but not at the source. The resistance of the ground is not zero, so currents on the ground will develop voltages. Current on the ground could affect the red system by making the return reference different between the source and the load. Such current will not affect the green system because its return path is grounded at a single point. If the green system’s return path were connected at both ends, return currents from the red system could interfere with the green system.
A schematic drawing of this shows why it is called “common impedance coupling”:
Currents are shared among the desired return path and ground according to their impedances. The return wire has a much lower impedance at high frequencies, so the ground does not carry high-frequency currents, and therefore high-frequency noise does not get into a system by ground loops. Crosstalk that is worse at lower frequencies points to a ground loop.
(I learned about Lee Hill's class through an IEEE member. I paid full price and have never spoken to Lee Hill or his company outside the class. A good portion of the class will be review for experienced engineers, but it's a worthwhile review. It's a good place to learn more about the grounding issues in this post.)