Why Impedance Control Matters on a PCB

Impedance control is one of those requirements that is easy to accept and hard to justify in a design review, because the board works without it until it does not. The value is set by the geometry and the material, the process moves it, and the failure it produces is a signal that arrives with less margin than the timing budget assumed. Understanding what is being controlled, and what changes it, makes the requirement easier to specify and easier to defend.

What Impedance Actually Is

In a circuit containing resistance, inductance, and capacitance, the total opposition to alternating current is called impedance and is written as Z. It is a complex quantity: the real part is the resistance, and the imaginary part is the reactance. The opposition a capacitor presents to alternating current is its capacitive reactance, and the opposition an inductor presents is its inductive reactance. Together they form the reactance, and the unit of impedance is the ohm.

The useful point is that impedance is not a single number. It depends on frequency, because the reactance of both the capacitance and the inductance does. A trace that looks like a resistor at low frequency looks like a transmission line at high frequency, and the transition is exactly where digital design becomes difficult.

The Five Impedances That Matter

When a digital signal travels along a trace, the opposition it meets is the characteristic impedance. This is the value that the geometry and the material produce, and it is the one that must match the source and the load if reflections are to be avoided.

Differential links have their own set. Driving two identical signal shapes with opposite polarity onto two conductors and subtracting them at the receiver is differential signalling, and the impedance between the two lines is the differential impedance. Each line also has an impedance to ground on its own: the odd mode impedance is the impedance of one line of the pair to ground with the pair driven differentially, and the even mode impedance is the impedance of one line to ground with both lines driven in the same direction, which is also the common mode impedance and is normally the higher of the two. The five values are not independent, and the geometry that sets one sets the others. The way they are defined on a real stack is described in microstrip and stripline routing.

Why Control Is Necessary

A board has to accept components, and after assembly it has to conduct and to carry signals. At low frequencies that is a question of resistance: a low resistance means low loss and predictable behaviour. As the data rate rises, the frequency content of the signal rises with it, and the trace stops behaving like a resistor and starts behaving like a transmission line.

At that point the impedance value matters directly. If the geometry, the laminate thickness, or the trace width varies, the impedance varies with it, and the variation produces reflection, attenuation, and a change in the shape of the signal. At the receiver the distortion shows up as reduced margin, which is the point at which the design becomes sensitive to everything else: crosstalk, noise, and the tolerance of the components. Keeping the impedance inside a defined range is what keeps that margin from disappearing.

What Changes the Impedance

Several process variables move the finished impedance away from the value the layout intended. The laminate determines the dielectric constant and, for a given geometry, the impedance that results. The dielectric thickness between the trace and its reference plane is set by the stackup and is subject to lamination tolerance. The trace width is set by the artwork, and etching removes material from both sides, so the finished width is not the drawn width. Copper thickness adds its own effect, since heavier foil etches with more undercut.

The result is that the impedance of a finished board is a distributed quantity with a tolerance rather than a single number. That is why the target has to be specified with a tolerance band and with the layers it applies to, and why the fabricator is asked to verify it after production rather than infer it from the drawing.

The Surface Finish Connection

There is one more contributor that is easy to overlook, and it sits at the surface: the finish on the copper. The connection runs from the copper trace through the plated or immersion layer to the component, and the quality of that layer affects the impedance of the joint as much as its solderability.

Tin finishes illustrate the problem. A chemically deposited tin layer can oxidise or hydrolyse, which raises the resistance of the interface, and it can grow whiskers, which create shorts between adjacent features. Where the deposited layer is not pure tin but a mixture of tin compounds, the resistance is higher than the metal itself would give, and the effect varies with age and humidity. The consequence is an impedance that drifts with time and environment instead of staying where the design placed it, and it is invisible both to the eye and to any single measurement. The remedy is a process that produces a dense, pure layer verified by measurement rather than by appearance, and the broader set of finish choices is compared in PCB design and fabrication.

What to Specify

Three fields turn an impedance requirement into something a fabricator can deliver. The target value and its tolerance, stated per layer, because outer and inner layers need different geometries to reach the same impedance. The stackup, including dielectric thickness and copper weight, because those are the parameters the calculation assumed. And the verification method, so the delivered board can be shown to be inside the band rather than assumed to be.

Where the design uses differential pairs, the requirement should be given as differential impedance together with the single-ended geometry that supports it. The stack that carries the controlled lines has to be continuous across the board, without a local change of dielectric thickness that would shift the value while appearing nowhere in the documentation. How the alloy interacts with the assembly process belongs in the same conversation and is discussed in lead-free versus leaded solder, while the routing conventions the geometry depends on are covered in high-frequency trace and data bus routing.

FAQ

What is the difference between characteristic and differential impedance? Characteristic impedance is the value a single trace presents to a signal. Differential impedance is the value between the two conductors of a pair driven with opposite polarity.

Which impedance of a pair is higher? The even mode, which is also the common mode impedance, is normally higher than the odd mode impedance, because the two lines are driven in the same direction and their fields reinforce.

Why does the finished trace width differ from the drawn width? Because etching removes copper from both sides of the trace. The difference depends on copper thickness and process, and it has to be included in the impedance calculation.

How does surface finish affect impedance? It forms part of the connection between the copper and the component. Where the layer oxidises, hydrolyses, or is not pure metal, its resistance rises and the impedance drifts with age and humidity.