High-Frequency Microwave PCB Manufacturing: Materials and Tolerances

Boards that carry radar front ends, satellite links or millimetre-wave radios are built on a different set of materials and a different set of tolerances from ordinary digital boards. A high-frequency microwave PCB is not simply a fine-line board with a faster clock; it is a transmission-line structure whose electrical performance depends on the dielectric constant of the laminate, the geometry of the copper, and the way the two are held together through lamination and plating.

What Makes a Microwave Board Different

At microwave frequencies the board is part of the circuit rather than a carrier for it. A microstrip line is defined by its width, the dielectric thickness beneath it, and the permittivity of that dielectric, and every departure from the nominal value shows up as a change in phase. Loss matters as well: conductor loss can be managed with surface finish and copper roughness, but dielectric loss depends almost entirely on the material chosen.

That is the reason microwave work concentrates on two parameters, dielectric constant and loss tangent, and treats mechanical tolerance as an electrical specification rather than a cosmetic one.

Substrate Choice: Dielectric Constant and Loss Tangent

Ordinary FR-4 has a dielectric constant that varies with frequency and with resin content, and a loss tangent in the region of 0.02, which is far too high for a low-loss microwave path. The alternative families are ceramic-filled composites, glass-reinforced polytetrafluoroethylene, and pure polytetrafluoroethylene with a controlled dielectric constant, typically between 2.2 and 10 depending on the grade.

The important distinction is not which material has the lowest loss, but how stable the dielectric constant is. A material with a dielectric constant that shifts by two percent across temperature will move the phase of a long line, and a phase shift in a phased array is a beam error. Ask for the specification across the operating temperature range, and design the line width for the worst case rather than the room-temperature value.

Polytetrafluoroethylene and Its Handling Problems

polytetrafluoroethylene has excellent electrical properties and a difficult mechanical personality. It is soft, it creeps under pressure, it has a high coefficient of thermal expansion, and its surface energy is so low that nothing wants to bond to it. Laminating it requires a higher temperature, a longer cycle and a careful ramp, and drilling it needs sharp tooling and conservative parameters because the material smears rather than cuts.

Because of the creep behaviour, plated through holes in a soft substrate need a generous annular ring and a barrel that can flex. Panel design should also keep the working area away from the edges, where the material is most likely to move during drilling and pressing.

Dimensional Tolerance and Etch Control

Since line width sets impedance, the etch process becomes an electrical process. Standard subtractive etching produces a trapezoidal cross-section, and the difference between the top and bottom width of a 0.2 mm line is significant. Fabricators compensate by adjusting the artwork, and the compensation factor has to be established on the actual material, not assumed from FR-4 experience.

Profile and registration tolerances matter too. Drilled holes have to land inside their pads on a material that moves more than FR-4 during lamination, so the drill registration budget should be agreed before the panel is laid out. High frequency traces and data bus routing covers the geometry side of the same problem.

Shielding and Isolation on a Microwave Board

Isolation between adjacent channels is often the parameter that limits a design, especially in a receiver front end where a strong transmit signal sits next to a sensitive input. The tools are the same as on any board, but the numbers are tighter: a wider ground strip between lines, a via fence on both sides of a sensitive trace, and a metal wall or a shielded cavity where the required isolation exceeds what the laminate can provide. EMI suppression design principles covers the layout techniques that carry over from digital work, and the microwave version of the same idea is simply applied with less margin.

Where two channels must share a substrate, keep their ground returns separate until they meet at a single point, and never let a high-level signal run parallel to a low-level one for any distance. Coupling that looks negligible at 100 MHz can dominate the isolation budget at 10 GHz.

Grounding, Via Metallization and Plating

Microwave boards rely on a solid ground, and the via fence that stitches the top ground to the bottom ground is a functional element, not a convenience. The spacing between vias has to be a fraction of the guided wavelength; at 10 GHz in a typical laminate, a spacing of a few millimetres is already too wide. plating those small holes in a low-surface-energy material is the difficult part, because the chemistry has to wet a surface that resists wetting, and the hole walls have to be prepared with a plasma or chemical treatment that does not damage the dielectric.

Where the board is aluminium-backed, the backing plate serves as the ground and as a heat spreader at the same time. Connecting to it reliably means a metallization step that is compatible with the aluminium oxide that forms naturally on the surface, and a drill and deburr process that does not leave a burr standing proud of the substrate.

Surface Finish and Protection

Surface finish affects loss as well as solderability. Nickel layers, including those under an immersion gold finish, are magnetic and add conductor loss at microwave frequencies, so a thin gold over nickel should not be used on critical transmission paths. Silver finishes and bare copper with an organic coating have lower loss but shorter shelf lives, and the choice usually comes down to whether the assembly line can handle the storage window.

Conformal coating is standard on these boards, but the coating has to be applied evenly. A coating that is thicker over one part of a coupled line than another changes the effective dielectric constant and introduces a phase error that is repeatable but wrong, and repeatable errors in a phased array are the hardest to find.

Inspection and Test

Inspection follows the electrical specification. Coupons on the panel, measured with a time-domain reflectometer, confirm impedance and line width control. Via continuity and isolation are checked on every board. Where the design includes filters or couplers, the only meaningful test is a measurement at the operating frequency, because a coupled structure can pass continuity testing and still be out of specification.

For the fabricator, the practical requirement is a documented measurement of the parameter that matters, taken on the same stackup that the production panels will use. Microstrip and stripline routing explains how the transmission-line geometry is defined on the panel.

FAQ

Can a microwave board be built on FR-4? Below a few gigahertz, a low-loss FR-4 variant can be adequate for short lines, and it is much cheaper. The limitation is the loss tangent and its variation between batches, so the decision should be based on measured loss at the working frequency rather than on the material name.

Why is dielectric constant stability more important than its value? Because the value can be designed around; the drift cannot. A line width calculated from a nominal dielectric constant is only correct if that value holds across temperature, humidity and frequency.

Why is nickel avoided on RF paths? Nickel is magnetic and adds loss, and it is normally present under an immersion gold finish. Where the finish must be gold, a thin or nickel-free process should be specified for the transmission lines.