RF Board Fabrication Control for Repeatable Performance

A microwave board can pass every electrical check in the design office and still miss its performance target after fabrication. The difference between a simulation and a finished board comes from tolerances that matter little at low frequency but dominate at gigahertz: line width, dielectric thickness, copper roughness and via geometry. RF board fabrication is therefore a process control problem as much as a design problem.

Why RF Fabrication Is Different

At low frequency, a five percent variation in trace width changes the impedance by a fraction of a percent and nobody notices. On a controlled impedance line at microwave frequencies, the same variation shifts the impedance enough to raise return loss, and the accumulated effect of several tolerances can move a filter’s center frequency outside its passband. The process window that produces acceptable digital boards is much wider than the one required here.

RF laminates also behave differently from standard FR-4. They are softer, more sensitive to drilling parameters and more prone to dimensional movement during lamination. A fabricator who builds them weekly has the process knowledge; one who builds them occasionally will produce boards that meet the drawing and fail the electrical requirement. The practical test is to ask for measured coupon data from a previous build in the same grade, not for a statement of capability, because the numbers reveal whether the process is actually under control.

Material Handling and Storage

RF materials absorb moisture, and moisture changes the dielectric constant. Panels should be stored sealed with desiccant and used within the manufacturer’s recommended window once opened. Where a panel has been exposed, a bake before lamination restores the intended properties, but the bake schedule must come from the material supplier rather than from habit.

Handling matters as much as storage. The copper foil on a low profile material is thin and easily scratched, and a scratch on a controlled impedance line is a permanent defect. Keep panels in sleeves between process steps, avoid stacking them face to face, and specify the copper surface treatment deliberately rather than accepting whatever the standard process provides.

Etching and Line Width Control

Line width control is the single most influential fabrication variable for impedance. Etch compensation must be applied from measured data on the actual material and copper thickness, because the etch factor depends on both. A fabricator who applies a generic compensation produces lines that are consistently narrow or consistently wide, and the error appears as a uniform impedance offset across the whole board.

Measure the result rather than assuming it, because an impedance tolerance that is not verified is only an intention. Include a test coupon on every panel with lines of the same width as the design, and check the finished width under magnification. Review the data with high precision PCB material selection in mind, because some laminates hold a tighter feature tolerance than others and the choice should be made before the stackup is frozen.

Drill Quality and Via Transitions

Vias in an RF board are not just connections; they are part of the transmission structure. A via stub left in a thick backplane adds a resonance that appears as a notch in the insertion loss, and a rough barrel adds loss that no design change can remove. Specify the drill size, the finished hole size and the back-drill requirement on the drawing, and confirm the fabricator’s capability against those numbers.

Drilling parameters also affect the laminate. Some RF materials smear during drilling, and the desmear process required to remove the residue can erode the resin and change the dielectric properties near the hole. A fabricator experienced with the material will adjust the drill speed, the feed rate and the desmear chemistry together rather than following a generic recipe.

Plating and Surface Finish Control

Plating thickness must be uniform across the panel, because an uneven deposit changes the impedance of the plated line and the loss of the barrel. Copper plating defects such as nodules, voids or thin deposits are more damaging here than on a digital board, since a single defect can shift the electrical performance of a whole filter section.

Surface finish selection affects loss as well. Thin, smooth finishes preserve the conductor profile, while thick finishes with a rough interface add loss. For high frequency boards, electroless nickel immersion gold with a controlled nickel thickness, or an immersion finish that leaves a thin and uniform layer, is usually preferred. Review the plating quality control practices described in copper plating defect prevention when specifying the requirement.

Panel Handling and Dimensional Stability

Dimensional movement during lamination shifts every feature on the panel and changes the impedance of every line. RF laminates with low expansion are selected precisely because they keep that movement small, but the handling that follows matters too: a panel that is bowed during imaging or drilling will not register correctly, and the resulting misalignment shows up as a variation in line width across the board.

Specify flatness and confirm that the fabricator stores panels flat rather than in bundles. Where the design uses mixed materials, the expansion difference between them creates additional stress, and the panel layout should place the high frequency material symmetrically where the stack allows. The same principles apply to PCB dimensional stability and expansion management on any high reliability build.

Inspection and Coupon Testing

Electrical verification belongs on the panel as well as in the laboratory. Impedance coupons with the same geometry as the product lines, measured with time domain reflectometry, confirm that the fabrication produced what the stackup intended. Coupons also provide a record that can be compared between lots when something changes.

Combine the coupon data with cross sections of representative vias and a visual check of the finished traces. A cross section reveals plating thickness, barrel quality and resin smear that no electrical measurement can show. Keep the measurements with the lot number so that a field failure can be traced back to the panel that produced it.

Documentation and Supplier Selection

Give the fabricator a stackup drawing that states the material, the thickness of each layer, the copper weight and the target impedance for each controlled line. Add the tolerance that the measurement must hold, since a target without a tolerance is not a specification. Note any requirement that affects the process, such as back drilling or a specific finish, in the same document.

Select the supplier on demonstrated experience with the material rather than on price alone. gopcb fabricates high frequency and mixed material boards and works from the stackup and coupon data as one package, which is what makes the electrical performance repeatable from the first panel to the thousandth.

FAQ

How much impedance variation is acceptable? Many RF programs work to within five percent and some to within two percent. The right figure comes from the sensitivity of the circuit to impedance, which is easy to evaluate in simulation and expensive to discover by measurement.

Can a standard FR-4 process build an RF board? It can build the geometry, but it will not hold the tolerances or the material properties without process adjustments. Ask for measured data from previous builds in the same material, not for a general assurance of capability.

Why do vias matter so much at high frequency? Because a via is a discontinuity with inductance and, in a thick board, a resonant stub. Both add loss and reflection that appear directly in the measured performance, and neither can be corrected after the board is etched.