Gold Finger RF PCB: Step Bevel & Edge Connectors

What Is a Gold Finger RF PCB?

A gold finger RF PCB is a board whose card-edge connector pads carry RF signals through hard-gold-plated contact fingers into a mating receptacle. The gold fingers must survive hundreds of insertion cycles while maintaining a controlled-impedance path—typically 50 Ω—right up to the connector interface. Unlike digital gold fingers that only need DC continuity, an RF gold finger must manage return-loss performance through the transition from trace to exposed pad, making bevel geometry, plating thickness, and copper-to-edge clearance all critical to signal integrity.

Why Hard Gold Matters for RF Contacts

Standard electroless nickel / immersion gold (ENIG) deposits only 2–6 µin (0.05–0.15 µm) of gold—too thin and too soft for a sliding contact. Hard gold plating for RF PCBs deposits electrolytic cobalt- or nickel-hardened gold to a minimum of 30 µin (0.76 µm), with 50 µin (1.27 µm) common for high-cycle applications. The nickel underplate is typically 100–200 µin (2.5–5.0 µm).

From an RF standpoint, the nickel underplate is the concern. Nickel is ferromagnetic with a relative permeability well above 1 and a skin depth of roughly 0.5 µm at 1 GHz. At frequencies above a few hundred megahertz, most of the current flows in the gold layer and the top of the nickel. Keeping the gold thickness at or above 30 µin ensures the majority of the current sees gold rather than lossy nickel, limiting insertion-loss degradation to a manageable 0.02–0.05 dB per finger at frequencies up to 6 GHz.

Gold Thickness Selection by Application

For defense and radar applications requiring IPC Class 3 RF PCB compliance, gold thickness and nickel underplate must meet IPC-4556 for electrolytic gold and the workmanship criteria in IPC-6012, Class 3.

Step Bevel Geometry for RF Edge Connectors

A bevel—the chamfer machined along the board’s insertion edge—guides the card into the receptacle without scraping the gold. RF edge connectors almost always require a step bevel (also called a dual-angle or compound bevel) rather than a simple 30° or 45° chamfer, because the board may carry a ground-signal-ground finger pattern at different layer depths in a multilayer stackup.

How a Step Bevel Works

A step bevel removes board material in two or more stages. The first, wider chamfer (typically 20°–30° over 1.0–1.5 mm depth) thins the board edge so it slides into the connector housing. The second, steeper step (30°–45° over 0.3–0.5 mm) exposes inner-layer fingers that contact a second row of receptacle springs. This lets a single card edge carry both signal and ground on separate layers, which is essential for maintaining a coplanar or stripline ground reference right up to the mating point.

Dimensional Tolerances

• Bevel angle: ±2° for standard, ±1° for RF-critical designs.
• Bevel depth: ±0.10 mm (±4 mil). Over-beveling exposes copper below the gold boundary; under-beveling prevents full insertion.
• Finger-to-finger pitch: 1.27 mm (50 mil) or 2.54 mm (100 mil) standard; RF designs often use wider pitch with ground fingers between every signal finger.
• Board thickness at insertion zone: must match the connector’s card-thickness specification, commonly 1.6 mm ± 0.10 mm (63 mil ± 4 mil).

Edge Connector Rule Checker

Enter your finger count, pitch, bevel angle, and board thickness below to verify your gold finger layout against standard edge-connector manufacturing rules. The tool flags violations in finger width, plating zone length, and bevel depth before you send files out for fabrication.

A typical result for a 20-finger, 1.27 mm pitch card on a 1.6 mm board with a 30° bevel will show a minimum finger width of 0.8 mm and a plating zone length of at least 4.0 mm. If the checker flags a “plating zone too short” warning, extend the hard-gold area by at least 0.5 mm beyond the bevel termination line—this ensures every contact spring lands on gold, not on bare nickel or solder mask. Run a DFM review for RF PCBs if you get more than one flag.

Copper-to-Edge Clearance for RF Gold Fingers

Copper-to-edge clearance is the gap between the nearest copper feature (trace, pour, or via pad) and the routed board edge. On a gold finger RF PCB this clearance serves two purposes: it prevents copper exposure after routing (which causes corrosion and shorts in the connector), and it keeps the impedance environment predictable where the controlled-impedance trace transitions to the exposed finger pad.

Recommended Clearances

Use the checker below to validate your copper-to-edge distances against your stackup and routing tolerance. Enter the layer, feature type, and nominal clearance.

If the tool reports a clearance violation on an inner-layer ground pour, pull the pour back by adding a board-edge keepout of 0.50 mm in your EDA tool. For signal traces approaching the card edge, taper the trace width to match the finger pad width over a length of at least 2× the substrate height (2H rule) to minimize the impedance step. Overlooking this taper is one of the most common causes of return loss exceeding −15 dB at the connector interface. Careful etch compensation on RF traces in this taper zone prevents the final etched width from falling short of the target.

Impedance Control Through the Gold Finger Transition

The transition from a 50 Ω microstrip or stripline on the board to an exposed gold finger in air is an impedance discontinuity. The finger sits in free space (εr ≈ 1) until the card is inserted, then it sits between the connector’s dielectric housing (εr ≈ 2.5–4 for LCP or glass-filled nylon). This swing in effective dielectric constant can shift the finger impedance from well above 50 Ω (unmated) to near 50 Ω (mated).

Design Strategies

• Widen the finger pad: A wider pad lowers impedance in the unmated state, centering it closer to 50 Ω when mated. Typical RF finger widths are 0.8–1.2 mm versus 0.5–0.6 mm for digital.
• Ground-signal-ground pattern: Place a ground finger on each side of every signal finger. This creates a coplanar-waveguide-like structure that stabilizes impedance and improves isolation.
• Short exposed length: Keep the exposed (unbeveled) finger length as short as the connector allows—often 3.0–4.0 mm. Every extra millimeter adds phase length at the mismatched impedance.
• Backplane ground stitching: Place ground vias in a fence along both sides of the signal trace as it approaches the card edge, spaced at λ/20 or tighter at the highest operating frequency.

For connectors that launch from the board edge rather than through gold fingers—such as SMA and 2.92 mm end-launch types—see the guide on edge launch connector PCBs, which covers footprint geometry and launch via optimization.

Fabrication Sequence for Gold Finger RF PCBs

1. Inner-layer imaging and etch: Controlled-impedance traces formed with etch compensation.
2. Lamination: Stackup pressed with RF laminates (e.g., RO4350B, Dk 3.48 ± 0.05) or hybrid FR-4/PTFE constructions.
3. Drilling and plating: Through-hole and blind/buried vias plated; barrel copper ≥ 25 µm for Class 3.
4. Outer-layer imaging: Finger pads and signal traces patterned.
5. Selective hard gold plating: Tape-mask or photoresist-mask the finger area; plate nickel underplate then hard gold to spec.
6. Solder mask and legend: Solder mask pulled back 0.50 mm minimum from the gold finger boundary to prevent mask creep onto contact surfaces.
7. Bevel routing: CNC bevel cutter machines the step bevel to the specified angles and depths.
8. Profile routing and inspection: Final board outline routed; gold finger area inspected under 10× magnification for plating voids, mask encroachment, and bevel dimensional accuracy.

Prototype quantities—even as few as five pieces—can be processed with selective gold plating if the fab house supports it. For initial validation runs, consider a rapid RF PCB prototype service with hard-gold capability to avoid tooling delays.

Common Defects and How to Avoid Them

Frequently Asked Questions

How thick should gold fingers be on an RF PCB?

A minimum of 30 µin (0.76 µm) of hard electrolytic gold over 100–200 µin of nickel underplate. For connectors rated above 200 insertion cycles or operating above 3 GHz, specify 50 µin to keep RF losses low and wear life adequate.

Can I use ENIG instead of hard gold for RF edge connectors?

No. ENIG gold is too thin (2–6 µin) and too soft for sliding contacts. It will wear through within a few insertions, exposing the nickel barrier and degrading both contact resistance and RF performance.

What bevel angle is standard for gold finger PCBs?

30° is the most common single-angle bevel per industry design guidelines. Step bevels combine a 20°–30° primary chamfer with a 30°–45° secondary step to expose inner-layer fingers in multilayer RF boards.

Does the gold finger area need impedance control?

Yes. The finger is part of the RF signal path. Its width, spacing to adjacent ground fingers, and the dielectric environment of the mating connector all influence impedance. Design the finger geometry for the mated condition and verify with a 3D EM simulation or TDR measurement.

How far should solder mask be pulled back from gold fingers?

At least 0.50 mm (20 mil) from the edge of the gold plating zone. This prevents mask from creeping onto the contact surface during curing and avoids contamination that increases contact resistance.