Who cares about EMI?
Brian Bailey - February 7, 2013
I have to admit that PCB design is not my area of excellence and the only involvement I ever had with it was when I used to etch my own PCBs to make my first computer. The tracks had to be thick enough such that when I placed it in the acid bath, the copper I wanted removed would be removed before too much etching happened under the area I had masked. I guess that issue is very similar to ones they face in chip polishing where consistent fill densities are required; otherwise you finish up with problems. So when I was given a PCB book to review, and particularly one about signal integrity and electromagnetic interference, I had to do a little homework to make sure I understood the basic concepts and to find out how different this is to the problems on-chip.
For those who are perhaps not experts in the area, I thought a little primer on fields and EMI might be a good idea – especially since it makes sure that I fully understand it – or at least remember it from my physics lessons so many years ago.
The most basic concept is that as current flows in a wire, electric and magnetic fields are created. The speed at which they travel along the wire is determined by the conductor and the relative dielectric constant of the medium surrounding the conductor.
The next concept is that of induction. When a magnetic field changes, a current can be induced in an adjacent conductor. This means that when an alternating current is fed down one wire it will induce a similar, albeit much smaller, current in another wire. Without this we would not have radio or that old fashioned form of video called TV transmission. But when this happens on a chip or board it is generally unwanted and the effects of it have to be minimized. It is often called crosstalk.
When I used to work in the avionics industry, there were a lot of wires that connected all of the flight computers together. We always used twisted pairs so that the signal path and its return path kept together or in other cases differential signals were sent. I never really thought about the implications of this in terms of EMI. What happens in a twisted pair is that the same currents are flowing in opposite directions and so establish equal but opposite fields. Because of the twisted nature of them, any other conductor that is nearby would see both fields equally and they would cancel each other out leaving no residual interference. Any noise picked up by a differential pair would be eliminated when it was fed into a differential amp.
I also remember the many problems I used to have when setting up stereo systems long ago. Most of those problems were caused by either not having a proper ground (I hate that term) between the various components, or from having ground loops. As a digital designer, we almost never think about the current return paths. They just – well, happen. But if those return paths are not properly thought about by someone, then all kinds of problems result from that.So, while I know that was very basic, perhaps it might provide some food for thought to those who do not work in this field (pun intended) every day and tend to forget about some of the physics behind what they create. Are there other very basic concepts that you wish digital designs were more aware of?
Brian Bailey – keeping you covered
If you liked this feature, and would like to see a weekly or bi-weekly collection of related features delivered directly to your inbox, sign up for the IC Design newsletter.