High Frequency PCB Basics: Materials, Loss and Design

Where Ordinary Design Rules Stop Working

Below a few hundred megahertz, a trace is a conductor and the laminate is an insulator, and the design rules that follow from that assumption work well. Above roughly 500 MHz the same trace starts behaving as a transmission line with distributed properties, and by the gigahertz range the substrate has become an electrical component whose loss and dielectric stability decide whether the link works. A high frequency board is simply a board designed with that in mind.

The practical definition used across the industry is a board operating at 500 MHz and above, and increasingly into the millimetre wave region for automotive radar and satellite links. What changes is not the artwork software but the choice of material and the attention paid to geometry.

High Frequency Board Compared With FR-4

• Usable frequency: 500 MHz to above 100 GHz for the low loss laminates, against below about 1 GHz for standard FR-4.
• Dielectric constant stability: very stable across frequency and temperature for high frequency materials, noticeably variable for FR-4.
• Loss: very low dissipation factor in the RF laminates against a relatively high one in FR-4.
• Signal integrity: excellent versus limited by the material rather than the layout.
• Cost: higher, often by a factor of two to five on the laminate alone.

The last row is the reason a hybrid stack is so common. Using a low loss laminate only on the layers that carry the RF path and standard FR-4 everywhere else delivers most of the performance at a fraction of the material cost.

The Physics That Matter

Dielectric constant (Dk). It sets how fast a signal propagates and, together with the geometry, defines the impedance of the line. A Dk that drifts with frequency or temperature shifts the impedance and the delay, which on a phased array or a matched filter is a performance problem rather than a tolerance issue.

Dissipation factor (Df). It describes how much energy the dielectric absorbs, and it translates directly into insertion loss that grows with frequency. This is the parameter that decides whether a long RF run is viable on a given material.

Skin effect. At high frequency the current crowds toward the surface of the conductor, so the effective resistance rises and the current flows through less copper than the cross section suggests. This is also why copper surface roughness matters: a rough foil makes the current path longer and adds loss that no amount of layout skill can remove.

Radiation and coupling. Traces behave as antennas as the frequency rises, which is why shielding, via fencing and ground plane integrity become design requirements rather than good practice.

Material Selection

• Standard FR-4: acceptable for the digital and low frequency analogue portions of the board, and for RF paths only at the low end of the range.
• Ceramic filled hydrocarbon laminates: a middle ground with better loss than FR-4, easier processing than PTFE, and good dimensional stability.
• PTFE based laminates: the lowest loss options, used for microwave and millimetre wave circuits. They require plasma treatment before plating and dedicated drill parameters, which limits the number of shops that handle them well.
• Hybrid stacks: low loss material on the RF layers and FR-4 on the rest, which is the standard commercial compromise.

Three secondary specifications accompany the laminate choice. The copper foil should be low profile, because roughness adds loss at microwave frequencies. The dielectric thickness tolerance determines how tightly the impedance can be held. And the coefficient of thermal expansion has to be reasonable, because the finished assembly will see temperature excursions during both assembly and operation.

Layout Rules

• Compute the impedance rather than estimate it. On a high frequency board, trace width is an output of the calculation, not an input.
• Keep the RF path short and direct, and keep it on one layer wherever possible. Every transition is a discontinuity.
• Maintain a continuous ground plane beneath the RF traces. A slot or a plane split forces the return current to detour and turns a matched line into a radiating structure.
• Fence the RF traces with stitching vias, spaced no more than a tenth of a wavelength apart, so the line behaves as a shielded structure.
• Separate the RF, analogue and digital domains physically, and join them at one defined point rather than everywhere.
• Design the connector transition. The launch from an edge connector or a coaxial connector into a microstrip is a common source of return loss and is rarely given the attention it deserves.
• Return a via to ground next to every signal via used in the RF path, and shorten or back drill any stub. A stub that is invisible at 100 MHz can dominate the response at 8 GHz.
• Filter and decouple locally. Supply network impedance matters as much as it does on a high speed digital board.

Fabrication Requirements

Manufacturing a high frequency board is a different exercise from ordinary multilayer work, and the differences show up in the sample data.

• Impedance tolerance of plus or minus five percent rather than the usual ten, achieved through tighter line width control and a specified dielectric thickness.
• Layer registration tight enough to keep the RF traces aligned with their reference planes.
• Plasma treatment before plating on PTFE materials, without which the copper adhesion is unreliable.
• Dedicated drill parameters, because PTFE behaves differently under a drill and smears easily.
• Low pressure lamination for the microwave laminates, and a controlled press cycle for hybrid stacks where two materials expand differently.
• A flat, stable finish. ENIG or an immersion finish rather than a thick solder coating, with the choice discussed under immersion silver PCB finishing.

Verification is as important as the process. An impedance coupon measured on the finished panel is the evidence that the delivered board matches the model the RF chain was designed around, and the measurement approach described under TDR impedance testing is the standard method. Coupon data should ship with the boards rather than being available on request.

Typical Applications

5G and radio access equipment, radar and automotive ADAS sensors, satellite communication terminals, aerospace microwave assemblies, high performance test instruments and medical RF imaging equipment. The common requirement is that a signal has to survive a path through connectors, filters and amplifiers without losing more energy than the link budget allows. Where the equipment also has to meet an operator’s acceptance criteria, the surrounding requirements are covered under telecommunications PCB manufacturing.

Cost Factors

• Laminate grade: the largest single item, with the lowest loss materials costing several times more than FR-4.
• Layer count and stackup: whether the board is a hybrid or all high frequency material.
• Impedance tolerance: tighter specification means more control and more measurement.
• Via structures: blind and buried vias, and back drilling, each add a process step.
• Finish and coating: flat finishes and conformal coating for outdoor equipment.
• Quantity: setup dominates at prototype quantity, and the material remains expensive in volume but the process cost falls.

A small RF board can cost anywhere from a few tens of dollars to several hundred depending on those factors, and the difference between a hybrid stack and an all RF stack on the same board is often the difference between a viable product and an expensive one. Getting a comparable quotation across two stackup options is the practical way to make that trade, since the material cost is visible but the process cost is not. That comparison belongs in a formal custom PCB pricing request with both stackups written down.

Working With the Fabricator

On a high frequency board the fabricator is a design participant rather than a supplier. Four conversations should happen before the layout is released: which nets are controlled and at what impedance, which laminate and thickness will be used and whether it is stocked, how the coupon will be measured and what tolerance is acceptable, and how the surface finish interacts with the assembly process. A shop that can answer those four questions with engineering reasoning is the right partner for the next board as well, and the process discipline behind the evaluation is the same one described under PCB manufacturing.

FAQ

At what frequency does a board become high frequency? From about 500 MHz upward, with material choice becoming critical above roughly 6 GHz.

Can FR-4 be used? For the digital and low frequency sections, yes. For the RF path above about 1 GHz it adds loss that the link budget may not tolerate.

What are the two most important material parameters? Dielectric constant, which sets impedance and delay, and dissipation factor, which sets insertion loss.

Why does copper roughness matter? Because of the skin effect. At high frequency the current travels near the surface, and a rough surface lengthens the path and adds loss.

Is the extra cost justified? Where the application needs low loss or tight impedance control, yes. Where it does not, a hybrid stack usually delivers the required performance for much less.

Summary

A high frequency board is one designed above roughly 500 MHz, where the laminate becomes part of the circuit and the trace becomes a transmission line. The controlling material parameters are the dielectric constant, which sets impedance and propagation delay, and the dissipation factor, which sets how much signal the material absorbs. Skin effect, copper roughness, via stubs and ground plane continuity then decide whether the finished board meets the budget. Practical design means computing the impedance, keeping the RF path short and matched, fencing it with vias, returning every via transition to ground, and verifying the finished result with a measured coupon rather than an assumption.