Not all common currents are bad
While common currents on external cables are the dominant source of EMC failures, not all common currents radiate, and some common currents are getting a bum rap.
In the EMC world, common currents are referred to as “common mode currents” or CM currents. As anyone who has read my books or columns or been to one of my classes knows, I am not a fan of adding the word “mode.” There is no additional value added by the word mode, and it only contributes to confusion. There are really just common currents. The word “mode” generally refers to the properties of the interconnect, independent of the signals on the interconnect.
A common current is a net current on a cable, usually as excess current flowing on the shield of the cable. The return for the common current on a cable is by displacement current through fringe electric fields to any nearby metal, like the chassis, the floor, any other wires or even a table top. Since the loop area is generally large, the common currents in external cables are very efficient at radiating.
In fact, it only takes 3 uA of common current on a 1 m long cable, at 100 MHz, to fail an FCC part 15 Class B test. This is a tiny amount of common current, really easy to generate. This is why common currents are so deadly. Just the words, “common mode current” will provoke an adrenaline rush of fear in any EMC engineer.
The problem is, not all common currents are bad. Some common currents are getting a bum wrap. This is why it is important to keep in mind what conditions are needed for common currents to radiate and when they are bad and when they are a “who cares.”
Generally, if the return path of the common current is far from the signal path, then these common currents will radiate. These are bad common currents.
But if you can engineer the return path of the common current to be in proximity to the source current, the loop area will be small and the common current will not radiate.
This is the value of adding a shield to unshielded twisted pair cables. The shield isn’t a shield, it’s a return path for any common currents on the twisted pair. The shield has no impact on the differential impedance of the twisted pair. It just turns the twisted pair into a coax geometry. By bringing the common return current in proximity to the source current and making it coaxial, this current, common to the twisted pair, does not radiate.
A microstrip transmission line is a signal trace with a return plane directly underneath and air above. Every board has microstrip traces on the top and bottom surfaces. There is a misconception that common currents on microstrip transmission lines will radiate. This misconception is compounded by the mistaken belief that even the differential signals in a microstrip pair will radiate and why it is necessary to use tightly coupled differential pairs. These two conceptions are totally wrong, for the same reason.
A single-ended signal in a microstrip transmission line has a current flowing down the signal line and returning in the plane directly beneath the signal. In the language of EMC engineers, this single-ended signal would be called a differential mode signal, and not radiate. A differential signal has a current distribution very similar to a single-ended signal, as shown in Figure 1.
Last year, about 0.5 billion square feet of printed circuit boards were shipped worldwide. This is a total of 1 billion square feet of microstrip surface traces. Every one of these 1 billion square feet of microstrip surface traces passed an EMC certification test and did not radiate.
Do microstrip traces radiate? 1 billion square feet of them did not in 2012. Of course, screw up their return path, and they will radiate, but so will stripline traces.
So why would you think two adjacent microstrip traces making up a differential pair need to be tightly coupled to prevent them from radiating? This is tied to the misconception that when you see a surface trace, all you see with your eyes is the signal path. You think all the currents associated with the microstrip are tied to the signal and forget about the adjacent plane. Out of sight out of mind.
You have to look with your mind’s eye to see the adjacent return plane, which is the second half of the microstrip transmission line. The signal and its close proximity return path make up the microstrip transmission line.
There are those in the industry propagating the myth that the return current of one line in a microstrip differential pair is carried by the other line. This is just not true, as seen in the figure above.
Where the return currents overlap in the plane, they cancel out, but there still remains about 90% of the return current for the differential pair in the plane.
Sending a common signal down a microstrip differential pair is just like sending two single-ended signals down the pair. Where is the return for the common signal? It’s in the adjacent plane. The close proximity of the return current of the common current prevents far field radiated emissions. Common currents on microstrip differential pairs do not radiate and tight coupling of microstrip differential pairs has no impact on radiated emissions.
In general, if you can engineer the common current’s return path in close proximity to the source current, it will not radiate.
Not all common currents are bad. Give microstrip traces a break.