Reducing EMI in PCBs December 6, 2006
I got the following question by email:
When a data sheet says “minimize the trace area for lowest EMI”, do they mean thin trace? Does higher reactance minimize EMI?
It is a good question, so I though I would post something on this.
Minimizing EMI requires minimizing the electric and magnetic fields around the trace. If a trace has a lot of voltage that is changing fast, it will tend to emit a strong electric field. In such a case, making the trace have less surface area can help limit the emissions. Better yet, shielding the noisy trace with a trace that lets the energy return to where it came from is very effective - the shorter the return path, the better. In fact, if the noisy trace is completely surrounded by the shield, the electric field lines won’t get out and the noise will be contained. Once shielded, the area of the trace is not so critical in terms of noise, however more area also means more capacitance, which your circuit may not like. Therefore, yes, it is usually a good idea to minimize the area of an E-field emitting trace, but it must be able to carry the needed current of course.
Next is the magnetic field, which comes from a changing current. The higher the current and the faster it changes, the worse the EMI can be. The magnetic field can be somewhat canceled by routing the returning current as close as possible to the outgoing current. In other words, the loop area traveled by the current needs to be small. We are talking about AC current here, so it can be contained to a small area by effective use of decoupling capacitors. These should be placed near the source of the noise current.
Ground planes can help contain both electric and magnetic fields. They give the electric field lines a place to terminate, and those high frequency return currents will automatically flow near the outgoing currents due to an imaging effect where the ground plane will actually mirror the trace currents with currents flowing in the opposite direction.
- Posted in : Uncategorized
- Author : Brad
Comments
They meant to say that you should minimize the LOOP area.
The signal has a source, destination, and a return.
Obviously you have to do what’s necessary to get from source to destination, but don’t forget every milliamp of current in that trace has a milliamp in the ground. If you don’t provide a solid, low impedance, definite path back to the source for it, it will find A path.
An indicating symptom of this kind of problem is when your noise emission gets worse with shielded cables attached, as opposed to not attached. This tells you that there’s high frequency currents running around the board, trying to find their way home.
Usually, this comes down to trusting the ground planing to “fix” everything… Make solid ground paths first, then flood.
Dave wrote: “They meant to say that you should minimize the LOOP area.”
Yes, loop area is of major importance. However, excessive surface area of a trace can certainly contribute to noise by increasing the capacitive coupling of the noisy signal to the surroundings. Maybe they should have said to decrease both the loop area AND the surface area…
Brad
Not to beat a dead horse, but regarding loop area only (all other things being equal): Loop area can be determined by the fact that the return current will tend to try to run along EXACTLY the same path as the source current (in the opposite direction), in accordance with Ampere’s law. The loop size tries to reach zero. If the ground plane is broken up or forces the return current to follow a significantly different path back to the ground terminal the loop will obviously be larger. Designing the board to minimize loop size from the start will greatly help EMI.
As for beating a dead horse, EMI is one of those dead horses that just can not be over-beaten. I agree with your comments about loop area. Keep it small and your electrons will thank you. A cut plane can totally mess up the return path, causing electrons to take the long way home. Larger loop area means larger inductance which means larger energy goes into the magnetic field associated with that inductance. Some of that energy radiates as EMI, some of it dumps back into the circuit and can cause spikes, voltage stress on components, higher losses, signal distortions, etc. Always visualize the complete round-trip path of currents and keep it short. The preceding discussion pertains to AC (changing) currents.
All the comments are ture. But it is better to think about power transmission systems in USA. In this country there are three wires ( Hot, Neutral and Ground) in the power transmission system. We can act like this system and define three potential levels Hot and Neutral wire transfer power to bord and Ground which is connected to case can avoid the emissions. The problem is that how e can creat second ground. It is easy by add some filters.