High-Frequency RF PCB: Materials, Impedance and Via Design

A high-frequency RF PCB is not simply a normal board specified for a higher clock rate. Once the signal wavelength approaches the dimensions of the routing, the conductor stops behaving like a wire and starts behaving like a transmission line, and every geometric decision turns into an electrical one. Boards in this category operate from around 500 MHz up to 10 GHz and beyond.

What Changes Above 500 MHz

The transition is gradual rather than sudden, but the practical consequence is that impedance replaces resistance as the property that matters. A trace with the correct impedance transfers energy from source to load; one with the wrong impedance reflects part of it, and the reflection shows up as ripple, loss or an unstable amplifier.

The second change is that the dielectric becomes part of the circuit. Its dielectric constant sets the geometry required for a given impedance, and its dissipation factor sets how much energy the material absorbs. Both properties vary with frequency and, in some materials, with moisture content, which is why material choice and design are decided together in RF work.

Base Materials

PTFE is the classic radio-frequency material. Its dielectric constant is around 2.1, its dissipation factor is very low, and it is thermally stable. Its drawbacks are equally real: it is soft, it is difficult to process, and it requires careful hole preparation and lamination control to produce reliable plated vias.

Ceramic-filled hydrocarbon laminates occupy the middle of the range. They offer a stable dielectric constant with low moisture absorption and a dissipation factor low enough for most radar and communication bands, and several grades process much like ordinary FR-4, which reduces fabrication risk and cost. Lower-loss ceramic grades trade a slightly higher dissipation factor for easier processing and lower price, and they are a reasonable choice when the loss budget allows it.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/electronics-manufacturing-pbc-assembly.jpg" alt="RF board with microstrip transmission lines and SMA connectors” />

Impedance Control

impedance control is the foundation of RF design. Most systems target 50 ohms single-ended, with 75 ohms appearing in video and broadcast interfaces and differential impedances of 90 or 100 ohms for high-speed digital links. The tolerance is usually plus or minus ten percent, and it is verified by time-domain reflectometry on test coupons.

Three variables set the impedance: trace width, the dielectric constant of the material, and the vertical distance between the trace and its reference plane. That third variable is the reason stackup is not an afterthought. Choosing a dielectric thickness without recalculating the trace geometry produces a board that is manufacturable and electrically wrong.

Transmission Line Structures

A microstrip runs on an outer layer above a reference plane. It is easy to fabricate and easy to probe, which makes it the default for many designs, but it is exposed to the environment and radiates slightly. A stripline sits between two reference planes, fully shielded and more stable, at the cost of a buried layer and a more complex stack.

Coplanar waveguide places ground copper on both sides of the signal trace on the same layer, with the reference plane below. It improves isolation and reduces radiation, and it allows the characteristic impedance to be tuned by adjusting the gap to the adjacent ground rather than only by changing the trace width. That flexibility makes it popular at millimetre-wave frequencies and for via transitions.

Stackup and Layer Planning

Symmetry matters. An asymmetric stack warps during lamination, and on a thin high-frequency build the warp is enough to move the effective dielectric constant and detune a filter. Dedicated ground layers keep return paths continuous and reduce radiated loss, and their separation from the signal layer is what sets the impedance.

Multi-layer RF designs should be simulated rather than calculated by hand, because the interaction between adjacent structures matters at these frequencies. A closed-form calculator gives a good starting geometry; a field solver shows how a neighbouring trace, a via or a cavity affects the result.

Via and Transition Design

Vias are usually the weakest element in a high-frequency channel. A through-hole via that passes through layers it does not need leaves a conductive stub that behaves as a resonant element, and at millimetre-wave frequencies that stub can sit inside the operating band.

Backdrilling removes the unused portion of the barrel after plating, which eliminates most of the stub and restores the transition performance. Blind and buried vias, whose selection rules are covered in the notes on blind and buried via structures, achieve a similar result by not passing through the whole board, and via-in-pad construction removes the short trace between the pad and the via, which matters most for ball grid array packages and millimetre-wave designs. Via-in-pad is a significant process step and typically adds twenty to forty percent to the cost of the affected area.

Hybrid Stackups

A hybrid stackup uses high-frequency material only where the performance is needed and ordinary laminate everywhere else. On a large board with one radio section and a substantial digital section, this keeps the material cost proportional to the area that actually requires it.

The complication is the boundary. Where a trace moves from one material to another, the dielectric constant changes and the impedance shifts locally, so the transition must be designed deliberately rather than drawn as a straight line. Lamination compatibility between the two materials, and the prepreg that bonds them, also have to be verified with the fabricator.

Manufacturing Challenges

PTFE and similar materials require plasma or chemical hole-wall preparation that works on a surface conventional desmear chemistry does not activate well, and they need a lamination cycle matched to their flow behaviour. Copper adhesion depends on a treated foil or a bonding layer, and over-etching changes a trace width that was calculated exactly.

Inspection closes the loop. Impedance coupons verify that the produced geometry matches the design, cross sections confirm plating quality through the high-aspect transitions, and electrical measurement on the finished board validates the performance the simulation predicted.

Applications

Radio-frequency boards appear wherever the signal must be transferred efficiently. Antenna feed networks in base stations, radar front ends for automotive and defence systems, satellite communication terminals, navigation and imaging equipment, and Wi-Fi routers at the upper end of the band all rely on them.

Each application sets a different priority. A phased array cares about phase stability across the array, a radar module cares about loss and thermal behaviour, and a consumer radio cares about cost. Those priorities determine the material grade and the process steps far more than the frequency alone does.

FAQ

Can FR-4 be used for RF work? At a few gigahertz on short traces, sometimes, with degraded and less predictable performance. As frequency or trace length rises, the loss and the variation in dielectric constant become the limiting factors and a radio-frequency material is effectively mandatory.

How accurate does impedance control need to be? Plus or minus ten percent is the common target, and tighter for critical filters. Achieving it requires the stackup, the trace geometry and the fabricator process to be agreed together rather than specified independently.

Is backdrilling always needed? Only where the stub is long enough to affect the band of interest. On thin boards with short stubs it adds cost for little benefit, and on thick boards at high frequency it is often the difference between a working channel and a failing one.