Issue link: https://iconnect007.uberflip.com/i/1491842
18 DESIGN007 MAGAZINE I FEBRUARY 2023 IC technology is described in nanometers, that traditional approach completely falls apart. An EM field, physics-based approach is essential. To make things worse, EMC standards have changed; we now have lower and higher fre- quency compliance requirements, much lower emissions levels allowed, and greater immu- nity required. e playing field and the equip- ment have completely changed. is really is a brand-new game. e challenges are not only about providing adequate power to the devices, but also managing the output signals. e smaller the transistor, the faster it turns on, and the bigger the impact it has on EMC and signal integrity. Shaughnessy: What is the relationship between smaller silicon and EM fields, and what can these designers do proactively to fight EMI? Beeker: e issue is driven by simple geom- etry. e smaller the transistor, the faster it switches, and the higher the frequency you must address. is is important both on the output side, which is generally acknowledged, and on the input power side as well. Most EMC issues I have seen are directly related to an improper power supply design. e problem gets even more difficult with today's technol- ogy, as the power supply requirements must first be addressed on the die. If the IC vendor has not done this properly, there is nothing you can do on the application side to correct it. You will find out the hard way if this is an issue once you try using the device. IC vendors do not tell you whether they have done this correctly. Shaughnessy: If we add increased rise time and faster signals to this mix, how does material and material selection figure into the equation? Should OEMs select the material, or allow the fab to do so? Beeker: Output signal characteristics are driven by the transistor geometries used to drive the pins. Assuming the input power is done cor- rectly, impedance matching becomes the chal- lenge. e material itself becomes a prime fac- tor for very fast data rates (not switching fre- quencies) and timing coherence. Propagation times become less predictable in the dielectric, because the mix of glass vs. epoxy is not homo- geneous, and the energy moves at a different relative velocity in each. As timing constraints become tighter, and data rates increase, this problem only gets more difficult to solve. Eyes close, and jitter reigns supreme. Shaughnessy: What advice would you give designers and EEs who are starting to have EMI issues because of tinier silicon and increased speeds? Beeker: e most important action for them is to go back to fundamental physics and make sure that the rules they use for design are based on the science. Current does not flow in a loop, and electrons do not carry the energy. EM field energy travels in the dielectric, and the job of the PCB designer is to create the pathways for controlling the fields. Layer counts cannot be driven by the trace density; they are driven by Dan Beeker