Editor’s note: This question was asked in response to Interference Technology’s recent webinar by Keith Armstrong. To view the webinar, click here.
Question: Grounding strategy of circuits can improve EMC. But you mentioned that connecting to safety ground has no effect on EMI. Please explain it more?
Answer: I think I covered this well enough in my webinar, when I said that all currents, including stray CM currents, always flow in closed loops. This is a Law of Nature (or Law of Physics, if you prefer) and one of Maxwell’s famous equations.
Given that all currents flow in closed loops, there can never be any such thing as a “noise sink”, so the common idea that noise can be “shunted away into the safety ground” is just plain wrong and always has been.
Any current that flows into a safety ground network, or even into a safety grounding rod stuck in the soil, must flow back out of that network or soil again at some point to complete its loop.
Because stray CM noise currents are ultimately generated by the electronic activities happening in semiconductors in our transistors and ICs, which cause our wanted signals and power, they have to flow 100% back in closed loops to those same semiconductors. They generally take multiple parallel paths (i.e. flow in multiple parallel loops), through air, PVC, fiberglass, and along copper and other types of conductors, the current in each loop being inversely proportional to the impedance of that loop.
Good EMC design for stray noise currents thus consists of raising the impedances of the loops that we don’t want them to take (e.g. by using skin-effect shielding, RF-bonding and CM choke filters), and lowering the impedance of the loops that we do want them to take (e.g. using skin-effect shielding, RF-bonding, and shunt capacitor filters).
In the PCB example I used in my webinar, the issue was the emissions from a PCB and its cables, caused by poor EMC design. Good EMC design provided small, local, low-impedance paths for the RF CM currents to flow back to the circuits that emitted them, so they didn’t flow along the cables and cause excessive conducted or radiated emissions.
The safety ground wire in the mains cable might play a part in providing local loops for some of the CM currents generated by the PCB’s circuits, because it is a conductor like any other. But my whole point was that because all currents always flow in closed loops, the path back through the ground conductor in the mains lead to a ground rod stuck in the soil under the building has nothing to do with reducing emissions.
There is no such thing as a sink for RF noise currents, and never has been. Neither the safety ground, nor anything else, has ever been somewhere that noisy unwanted currents could be dumped into and forgotten.
However, many designers who thought they were providing improved paths to ‘dump the noise into the ground’ were actually providing smaller, more local loops for CM currents to flow in, without realizing that this is what they were actually doing.
They saw some reductions in emissions and thought that this was because the noise was flowing into what they called “the ground” and being lost (they may even have thought that the noise was flowing into the safety ground rod that was stuck in the soil, and being lost in the mass of the planet) – but neither is possible, at least not in this universe.
In systems and installations where there are two or more interconnected electronic units, because of poor EMC design there were often high levels of CM currents flowing along their signal interconnections (the signal cables themselves convert some of the wanted DM currents into unwanted CM noise, which is measured as their Longitudinal Conversion Loss, LCL, which varies with frequency). Adding CM chokes to the signal cables generally helped, but was not a universal solution, because:
- if the cables were not of high-enough quality we sometimes reached a point where the wanted signal became too degraded before the emissions had been reduced by enough
- there were sometimes weight, cost or accessibility issues associated with adding a sufficient number of CM chokes, especially in large installations with hundreds of long cables which needed CM chokes adding every meter or two of their length.
The photograph at left is an extreme example of CM-choking – I’ve included it as a sort of a joke, really, but nevertheless it isa real installation (although not one of mine!).
The safety ground wires in the unit’s mains cables, and the metal structure of the building they were installed in, provided the return paths for the stray noise current loops, and so – when trying to reduce their noise emissions – we generally found that improving the RF bonding between all of the units’ chassis (and/or their PCBs’ ground/0V planes) and their local support metalwork would reduce emissions. This had the advantage of costing less than clipping dozens, sometimes hundreds, of split-core ferrite chokes to most/all of the cables.
What was really happening was that we were providing smaller, lower-impedance loops local to the units and their cables, using existing support metalwork that just happened, for safety reasons, to be connected to a ground rod that was stuck in the soil under the building. IEC 61000-5-2 calls this technique of RF-bonding support metalwork: “creating a Common Bonding Network, CBN, and is the point of slide number 30 in my webinar.
Having got hold of the wrong idea, that the mass of the planet would somehow absorb the noise currents, this was often wrongly assumed to be the reason why improved RF-bonding to the metalwork reduced the noise emissions.
But we also generally found that routing the interconnecting cables along the metal supports, once we had improved the units’ RF-bonding to them, would reduce emissions by significantly more. This could not be happening if the noise currents were really being shunted into the mass of the planet via the ground rod that was stuck in the soil. It could only work if what we were really doing is making the stray noise current loops smaller in area. This, of course, is what was really going on, and is the reason why IEC 61000-5-2 (for example) strongly recommends using metal cable trays, trunking, ducts and conduits as the return current paths for the stray currents of the cables they carry.
Safety grounding only works because currents always flow in loops, too. One terminal of the AC mains supply to a building is connected to a ground rod stuck in the soil at the building’s utility entrance. We call this grounded AC supply terminal the Neutral. All of the ground wires in all of the mains cords that have them are also connected back to this ground rod, and therefore so are all the chassis of the units powered by those cords. And all of the building’s metal structures are also connected back to that ground rod.
When an insulation fault occurs in a wire or cable carrying AC power, it either flows from the live wire to the neutral wire creating a closed current loop (that we call a short-circuit) that opens the line fuse, or it flows into the local metalwork. The safety-grounding of all of the metalwork then creates a closed current loop all the way back to the ground rod and the AC mains supply, creating a short-circuit that blows the line fuse, preventing electric shocks from being caused by live chassis.
The ground rod in the soil has nothing to do with the safety function I have just described. It is there so that if lightning strikes the building or near to it, the metalwork does not attain a voltage that is too high above the potential of the soil – to try to prevent serious electric shocks.
(Lightning currents also flow in loops, but although I am assured that this is so by lightning experts, they are not as obvious to electronic engineers like me so I won’t attempt to describe how they work.)
-Keith Armstrong
