Issue link: https://iconnect007.uberflip.com/i/1210212
22 DESIGN007 MAGAZINE I FEBRUARY 2020 band. The total power radiated all around from that radio station is no more than 10 nanowatts, and if more than 10 nanowatts, it will fail the FCC test. And you say, "Wait, it doesn't take much radiated power at all to fail an FCC test. You really have to engineer your product so that it doesn't radiate very much." And that's why one of the fixes for passing a test is adding the spread spectrum clock gen- erator. If you have a clock system, you're ra- diating at the harmonics, so the spread spec- trum clocking just dithers that harmonic fre- quency range across that 120-kilohertz band- width. Because it only takes a tiny amount of radiated energy to fail an FCC test, EMI has been and always will be a problem in many systems. It's a tough problem to fix, and you have to be a clever signal integrity (SI) engineer, pow- er integrity (PI) engineer, and EMI engineer. You need everybody playing together in order to eliminate that much radiated emissions and pass an FCC test. Shaughnessy: Is EMI mainly caused by an im- properly designed power distribution network, or is that just one possible cause? Bogatin: There's a lot of noise with PI due to three things. One is the power system. What ra- diates are currents, and it's in the power deliv- ery system where you have the high currents. That's where you have the potential of hav- ing very high currents that can radiate. Then, the way those currents turn into radio emis- sions is related to where their return paths are, and you have components in the power distri- bution network that are big whopping compo- nents; there's an inductor, as well as the ICs that are on the board, and oftentimes, the way the board is laid out for the power delivery part doesn't keep the power currents and the return currents close enough together, so they have a tendency of radiating. The third piece is that part of the power distribution path is the re- turn path for signals. It's the discontinuities in the power and ground paths that cause any currents flow- ing into its path to see the discontinuity, to see a voltage drop, and create patch antennas. And it's because the same conductors that the planes used for power delivery are also used for return path for signals. If their discontinu- ity is there, the signals can radiate, and the power currents can radiate because of the dis- continuity. It doesn't mean all sources of radio emissions are from the PI, but if you fix those PI issues, that will go a long way toward help- ing reduce the radiated emissions. Shaughnessy: So, it's a system thing. Bogatin: It's always a system thing. On the oth- er hand, even if you have fixed those issues and have good SI, it can still radiate because, again, it doesn't take very much radiated pow- er to fail an FCC test. Matties: How would a designer or engineer de- termine if this is going to be a problem? Bogatin: There are a few approaches. There are a lot of known root causes for radiated emissions, and there are a number of these design rule checkers. Mentor has one, and I think Altium might have one, but what Men- tor does is look for the known problems in the layout that will cause radiated emissions. They look through your Gerber files, and will highlight potential problems, noting, "Here you have a signal over a gap," or "Here you have a split plane, and you have no path for return current." They will identify them and let you know what areas to check. That's one way ahead of time, but it's difficult because systems are so complicated now. It's challeng- ing to put the whole thing into a 3D full-wave solver and then predict where the radiated You really have to engineer your product so that it doesn't radiate very much.