The Wrong Answer Fast is Not the Fast Answer. It’s the Wrong Answer!

James Clerk Maxwell’s derivation of his Maxwell’s equations is one of the great feats of classical physics. Many beautiful, and at the time unknown, insights fell out of their formulation such as the need for the electric displacement current, the free space propagation of electromagnetic waves (and subsequent realization that light was a form of electromagnetic energy) and the concept of the continuity principle. In the utilization of HFSS which solves the fully coupled, or full wave, set of Maxwell’s equation, it is this last concept that can often be troubling when setting up a simulation. Ultimately it says electric charge or current cannot appear out of nowhere, nor disappear therein and in the end means that if a current departs one location, it must return. In other words, you need a groundplane to return the current.??

It all means the user must consider how an HFSS port or excitation is defined and make sure it creates a connection between the “signal” conductor and the “groundplane” conductor. This also means that however careful you may be about designing your signal line to, for example, make sure it has the right impedance, the right spacing or the right cross-section, the same care and consideration must be made with respect to the geometry of the groundplane return. Now, twenty years ago groundplanes were simple and straight forward, large continuous sheets of copper providing very low loss paths for the return current. As a result, some efficient simulation approaches, while still full wave, could deliver fast results under the assumption of these large, assumed infinite, groundplanes. But with higher data rates and frequencies, smaller form factors and modern IC design processes such as HBM and 3D-IC, those days of big beautiful groundplanes are gone. Now these groundplanes, these returns paths, are complex, and their complexity is very important to understanding how your device will work.?

But it has come to my attention that some of our competitors are making some alarming suggestions about these very important structures and that is to simplify them aggressively, all in the drive to get a faster “answer”. But an answer to what? Maxwell’s equations tell us that these structures are very important to the electromagnetics, so you better be careful about what assumptions and simplifications you make and if I am interested in understanding how two signal lines interact one thing I surely do not want to do is over-simplify the groundplane between. It’s a change that is going to have a big impact. But I have seen exactly that type of suggestion coming from our competition. Sure, that may get them to an answer faster, at least faster for their tool, but at what cost to the design? Can you design with confidence? Can you design for the tightest margins, the highest performance? Do you dare tape-out based on this “answer”??

So, if there is a simulation solution out there telling you it will get you a fast “answer” by aggressively simplifying your groundplane, you may want to think twice about that. It’s actually the wrong design and the wrong answer. And worse, it’s likely to lead you to make the wrong decisions. It’s your design. It deserves the best tools.?