Several readers have made comments regarding multipoint grounding (MPG) systems. The concern is that this arrangement creates multiple ground loops. As a result, we need to address that. As some of you know, I make a lot of presentations and teach a lot of EMC seminars. A long, long time ago, I was discussing ground loops in one of my training classes when a student interrupted me and told me that I obviously didn’t know what a ground loop was, so I let him tell me.
As it turned out, he was a member of a Civil Air Patrol (CAP) squadron flying an antique Piper J-3. This is a little wooden and metal framed tail-dragger covered with canvas, which looks a lot like a large version of a model airplane that we—with the help of our kids— might build at the kitchen table. That said, in the world of aviation, a ground loop is an on-the-ground spin-out of the airplane, where the pilot is suddenly looking back at where he just came from. This usually only happens with conventional landing gear (not tricycle), where the center of gravity is behind the wings. I suspect that this kind of ground loop is very scary . . . especially if going fast.
We all had a good laugh at the airplane definition of ground loop, but ever since then I have tried to use the more technical term: common impedance coupling. However, that term is hard to visualize. Regardless, the MPG creates a lot of ground loops and they are just as scary to the EMC engineer as they are to the airplane pilot.
Figure 1 illustrates the issue of loop creation within a MPG system. This figure shows five cases of grounding connections, i.e. none (0), 1, 2, 3 and 4.
None (0) and 1 (which is the single point ground) do not create a loop. Two ground connections (one loop) might be all right if there are no ground currents and no radiated electric or magnetic fields. Yeah right! With three or more intentional or unintentional ground connections, there are progressively more loops than ground connections, and the loop complexity gets worse and worse. This is a graph theory problem, with the number of loops (L) related to the number of grounding (G) connections as follows:
If G = n, then L = n (n-1) / 2.
So, four interconnected cabinets might have 6 loops, but a PCB with 20 devices might have 190 loops. A similar situation occurs when the number of interconnecting cables is considered, but that is another problem for another day.
Unfortunately, as frequencies increase to where wavelengths approach the dimensions of boxes, cables and grounding connections, single point ground systems fail and capacitive coupling creates MPGs whether they are wanted or not. Since many of the high frequency ground connections (if we want to call them that) are the result of capacitive coupling, they are not visually identifiable. We can’t see where they are or determine what loops were formed. Consequently, the grounding design for high speed systems must be adapted to work at these higher frequencies. Individual grounding wires cannot be used because their resistance and inductance is too high. Instead, ground planes are used. This may be a dedicated PCB layer, the chassis or a grid underlying a complete installation. It all depends on the frequencies of the device.
Although there are many loops created by MPGs, that doesn’t mean that they should be avoided. MPG ground systems have connections that are shorter and more direct. In addition, they are easier to build, their loop areas are smaller and they can operate to higher frequencies before standing waves become a problem. Nevertheless, like anything else (even SPGs), MPGs can’t be ignored, especially the large-scale grounding systems. There needs to be routine maintenance to counter the effects of corrosion, shock and vibration, and mechanical damage to prevent the introduction of unwanted high impedances.
– Ron Brewer