Estimated reading time: 4 minutes

Elementary, Mr. Watson: Controlled Impedance—When ‘Connecting the Dots’ Stops Working

All designers should be familiar with the 1993 engineering masterpiece, High-Speed Digital Design: A Handbook of Black Magic, by Howard Johnson and Martin Graham. So, is the discussion on high-speed or signal integrity really talking about black magic? Yes, and at the same time, absolutely not.

I agree that Johnson and Graham nailed the experience every first-time designer goes through: You follow the rules, the schematic looks clean, layout passes checks, and then the signal misbehaves. It reflects, rings, and does things that seem completely disconnected from the lines you just drew. That disconnect is what creates the “black magic” feeling.

But nothing mystical is happening. Signal integrity follows physics that are consistent, predictable, and fully explainable. But those rules don’t show up in the schematic. Instead, they live in the electromagnetic fields around the traces, in the stackup, and in the relationship between conductors and their return paths. The authors weren’t claiming black magic; they were calling out how it feels until you learn what’s going on. The fact that they said it more than 30 years ago only reinforces how fundamental and enduring these concepts really are.

If it sometimes feels like black magic, we should be asking: How do we make the invisible visible? How do we take something that behaves in fields we can’t see and turn it into something intuitive, something we can design with confidence instead of guesswork? Because once you can see it, even conceptually, the mystery disappears, the reflections make sense, the behavior becomes predictable, and what once felt like magic becomes something you can control.

Just the term “controlled impedance” sends shivers down a designer’s spine, the kind where a perfectly clean layout suddenly feels haunted. Nothing’s broken, but everything’s reacting, and for a moment, you understand exactly why it was called “black magic.”

Why do so many designers struggle with controlled impedance? It’s not because the math is difficult; it’s because the concept is counterintuitive. You see, we’re taught to think of traces as simple connections. Early on in my career, I was told routing was “connecting the dots,” and that worked right up until my first high-speed design, when reality said otherwise. The problem now lies in physics at higher speeds, where traces stop being simple paths and become part of the system itself. You can see a connection, but you can’t see the impedance that lives in the relationship between the trace, the plane, and the surrounding material. That ma it harder to grasp.

Additionally, most designers encounter it only when something goes wrong, so it feels like a fix rather than a foundation. The real shift in the quality of my designs came when I stopped thinking of a signal as just voltage and started seeing it as energy moving through a controlled environment. Once that clicks, controlled impedance starts to make sense.

Imagine driving down a highway at a steady speed, with a smooth road, consistent lanes, and everything is predictable. You don’t even have to think about moving. Suddenly, the lane narrows, then widens, and the pavement shifts from smooth asphalt to gravel and back again. You’re still moving forward, but now you’re reacting—slowing down, adjusting, and hesitating.

A properly designed PCB trace is like that smooth highway, providing a consistent environment so the signal moves cleanly from one point to another. But when impedance changes, whether from variations in trace width, distance to the reference plane, or inconsistencies in the surrounding environment, it’s like the changing road conditions. The signal doesn’t just continue unaffected; it reacts. Part of it keeps moving forward, but part of it reflects, as if the signal momentarily hits the brakes and questions what just changed. Those reflections then interfere with the original signal, distorting its shape and timing. The more inconsistent the path, the more the signal is forced to respond rather than travel. Bottom line: The smoother and more consistent the path, the less you notice it because everything works.

The $64,000 question: How do we make the PCB “road” smoother and more consistent?

You do it by following a few fundamental principles and removing surprises from the signal’s path. It’s that simple. In PCB terms, that starts with consistency, keeping the trace width uniform, maintaining a solid, continuous reference plane beneath the route, and preserving a constant spacing between the trace and that plane. Each element defines the environment the signal experiences. The moment one of them changes even slightly, the signal notices and hits the brakes hard.

Think about what happens when you neck a trace down to get around a component, let it wander over a split plane, or change layers without providing a clean return path. From a layout perspective, those might feel like small, practical decisions. From the signal’s perspective, they are abrupt changes in its environment, and when the environment changes, the signal reacts. Part of it continues forward, and part of it reflects. That’s physics doing exactly what it’s supposed to do.

Furthermore, the stackup matters because it’s not just a manufacturing detail; it’s the foundation of your electrical design, the controlled environment for the energy to live. The dielectric thickness, material properties, and copper geometry all contribute to defining the impedance. If you treat the stackup as an afterthought, you’re essentially designing a road without knowing its width, surface, or even its makeup. But when you define it intentionally, you’re setting the rules of the environment before a single trace is routed.

Most assume this is about perfection. It’s not. The goal is predictability. You don’t need every trace to be identical, but you do need every critical signal to see a consistent environment along its path. That means minimizing unnecessary width changes, avoiding breaks in the reference plane, controlling layer transitions, and being deliberate about how signals move from one part of the board to another.

When you do that, something interesting happens: The signal stops reacting, reflecting, distorting, and fighting the path you’ve created. It simply moves and flows.

That’s really the point of controlled impedance design: not to perform engineering “black magic,” but to create a path so consistent that the signal has no reason to do anything unexpected.

This column originally appeared in the May 2026 issue of I-Connect007 Magazine.