I briefly mentioned some of the elements of grounding in the last post. Technically, there are three types of grounding systems that work: (1) the single point (a.k.a. star) ground (SPG), which uses wires; (2) the multipoint ground (MPG), which requires a ground plane or mesh; and (3) the floating (no) point ground (FPG). All three utilize some variety of RF conductor—a wire, a plane or a capacitor. If conductors had no resistance and no inductance, implementation would be easy. Unfortunately they do have these properties and that forces the grounding design to change based on frequency.
Even though there are three grounding configurations, there’s about six different flavors depending on which industry and what problem is being considered. Each system has a name— power, lightning, safety, technical, signal, maintenance or maybe static, equipment, systems, lightning, electronic and maintenance. The one that may not be familiar to everyone is the maintenance ground. This is the installation of an intentional (but temporary) ground or short across non-energized circuits to protect technicians and maintenance personnel from inadvertent TURN-ON. Sometimes a lock-out tag just doesn’t give enough warm fuzzies—like when you’re working in a 15 MW RADAR transmitter cabinet. Since we don’t want to fry anyone, safety is extremely important and we do what we must in order to be safe.
Grounding is used to minimize voltages that are being developed between two or more locations and are caused by errant currents. One location is generally an active part of a system, while the other is a reference. The power industry, NFPA, and NEC people have standards and procedures in place to cover things like power, fault, lightning and maintenance grounding. These are primarily low frequency systems in which lengths are long and impedance is dominated by conductor resistance. For a conductor Zo = R + jωL and at really low frequencies for large wire sizes R < jωL, wiring sizes that are used are based on current carrying ability and can be quite large for systems with large fault current.
Generally, at the signal frequencies (F > 1 MHz), inductive reactance is greater than the conductor resistance (jωL > R), and it dominates the impedance. Because of that, a lot of design time is spent trying to reduce the inductance of electronic/signal grounding systems. For insight on what’s needed to do that, consider a flat conductor/trace with thickness (T), width (W), and length (L). Making the conductor 10x thicker reduces the inductance by 10 percent. Making the conductor 10x wider reduces the inductance by 30 percent, but reducing its length to 1/10thof the original length reduces the inductance by 93 percent. That’s 93 percent! Whoopee! Yahoo! Eureka!
We just learned to keep conductors as short as possible. How short should it be? Short enough to minimize resonance effects. At a quarter wavelength (λ/4), the impedance is very high, so most design guides have a rule of thumb limiting grounding/bonding conductor lengths (L) to no more than λ/20 at the highest frequency that causes concern. If the highest frequency of concern is 3 MHz, then L < (100m * 40 in/m) / 20 = 200 inches = 16.7 feet. Unless we are in a really big system, this is not a problem. If it’s 100 MHz, then L < (3m * 40 in/m) / 20 = 6 inches and it’s getting to be a problem.. At 3 GHz, L < (0.1 * 40) / 20 = 0.1 inches and unless we are working at the chip level, this is most definitely a problem. Even at the chip level this isn’t easy.
The goal is to design an RF short circuit, not an antenna. Stay tuned! There’s more to it than this.
-Ron Brewer
