Gold Fingers on a PCB: Design Rules for Edge Contacts

Gold fingers are the plated contacts on the edge of a board that slide into a socket. They are one of the few parts of a circuit board designed for mechanical wear, and they fail for mechanical reasons: worn plating, a deformed bevel, a contaminated surface or a board that is slightly too thick for the connector. This article covers what the fingers have to do and the design rules that keep them working for thousands of insertion cycles.

What Gold Fingers Are

A gold finger is a strip of plated copper on the board edge, usually on a 1.27 mm or 2.54 mm pitch for a card edge connector, with the copper running from inside the board to the edge itself. The edge is chamfered so that the finger enters the socket without catching, and the contact area is plated with a hard gold layer over a nickel barrier.

The assembly is inserted into a socket that contains spring contacts, and the electrical connection is made by the pressure of those springs against the plated surface. Nothing is soldered, so the entire connection depends on contact force, surface finish and geometry.

Two families exist. A standard card edge connector is meant to be inserted a few times during assembly and service, and a connector designed for frequent insertion is specified with a cycle life in the thousands or tens of thousands. The finish, the bevel and the thickness all follow from which of the two a design needs, and this is where an edge plated board and a gold finger board differ: the plating on an edge can exist for shielding or grounding, while gold fingers exist specifically to be slid into a mating part.

Why Hard Gold

Gold is used because it does not oxidise, so the contact resistance stays low even after long storage, and because a hard gold alloy resists the abrasive wear of a sliding contact far better than softer finishes.

The hardness comes from the alloy and the plating process. Pure soft gold is used for wire bonding, where the requirement is a clean surface with no contamination, and it wears quickly in a sliding contact. Hard gold contains a small amount of a hardening element, is deposited to a specified thickness over a nickel barrier, and is qualified by a wear test rather than by appearance.

The nickel layer between the copper and the gold is essential. Gold and copper diffuse into each other over time, and a gold layer directly on copper loses its barrier properties and its hardness. The nickel stops the diffusion, provides a hard base that supports the thin gold layer, and carries the contact force. Specifications for electroplated gold such as ASTM B488 define the thickness classes, and it is worth quoting a class rather than a thickness in microns alone.

Geometry and Layout Rules

The fingers are a mechanical interface, and each dimension is set by the socket rather than by the circuit.

• Pitch and width: matched to the connector, commonly 1.27 mm or 2.54 mm pitch. Finger width and the gap between fingers are specified together, since the socket contact has to land on the copper and not on the laminate.
• Length: long enough for the full contact travel plus the bevel plus a margin, and consistent across all fingers, since a short finger will make contact later than its neighbours.
• Board thickness: the socket is designed for a nominal thickness, typically 1.6 mm, with a tolerance that is tighter than a general board tolerance. A board that is thick at the high end of a normal tolerance range may be difficult to insert and may damage the contacts.
• Bevel: the edge is chamfered, usually 20 to 45 degrees, on both faces, so the finger tapers into the socket. The bevel must not cut into the plated contact area, which means the plating has to extend past the point where the chamfer begins.
• Keepout around the fingers: no vias, no component pads, no solder mask openings and no legend on the contact area. Mask between the fingers is normal and helps prevent bridging.
• Routing: traces enter the finger from inside the board, so the routing has to fan out from the finger pitch to whatever the rest of the design uses, and that fanout is where the width and spacing constraints are tightest.
• Keying: a polarising slot or a missing finger position prevents the board being inserted the wrong way round, and where it exists the mechanical drawing has to show it.

The plating bar is the process consideration that designers usually meet last. The fingers have to be connected electrically during plating, which is done with a bar on the panel that is later routed away. That operation leaves a small step on the board edge, and the clearance between the fingers and the point where the bar was removed has to be enough that the step does not affect the contact area. Where the board also has edge plating for grounding, the two features have to be specified separately because their tolerances and finishes differ.

Contact Force and Wear

The reliability of a gold finger connection depends on the contact force the socket applies and on the area over which it is applied.

A higher force gives a lower and more stable contact resistance, but it also increases the wear on the plating and the insertion force the user has to apply. Connectors are designed around that trade, and the board cannot change it, but the board can avoid making it worse. Copper that is too thin under the finger deforms under the contact pressure, and a bevel that is too sharp concentrates the insertion force on the contact edge.

Wear appears as a gradual loss of the gold at the point of contact, exposing the nickel beneath. A small amount of wear is normal, because nickel also forms a stable contact, but once the nickel is breached the resistance rises quickly. That is why the cycle life figure matters: it describes how many insertions the assembly survives before the gold is worn through at the contact point.

Choosing the Thickness

The gold thickness follows from the number of insertion cycles the product has to survive, and the relationship is roughly linear until the plating is thick enough that wear is not the limit.

A board inserted a handful of times during assembly and service can use a thin gold layer, in the region of 0.5 to 1 micrometre. A product inserted hundreds of times, such as a plug in module in a test rack, typically uses 1.5 to 3 micrometres. A connector specified for tens of thousands of cycles uses more, often 5 micrometres or beyond, and at that point the socket design and the insertion force matter as much as the plating.

The nickel barrier is usually 2 to 5 micrometres, and it should be specified with the gold because a thick gold layer over a thin or porous nickel layer will fail through the pores rather than through the wear surface. The supplier should be able to state both, and to demonstrate them by cross section or by X-ray fluorescence measurement.

Where the requirement is modest, selective gold over nickel on the fingers alone keeps the cost of the precious metal confined to the contact area.

Process, Cost and Handling

Three process factors drive the price of a board with gold fingers.

The selective plating operation requires masking, a plating bar and an additional process step, and it is usually priced per panel rather than per finger. The gold itself is a material cost that follows the market, so a quote can move with the metal price rather than with the manufacturing cost. And the bevel is a separate mechanical operation, performed after plating, that needs its own tooling and a tolerance that keeps the chamfer off the contact area. Our guide to PCB fabrication sets out how these extra steps are handled in the process flow.

Handling requirements follow from the surface. Gold fingers should not be touched, since skin oils and salts contaminate the contact area, and the boards should be packed so the fingers do not rub against each other or against the packaging. A protective tape over the finger area is common on high value assemblies, and it has to be removed without leaving adhesive residue, which is a specification for the tape rather than for the board.

Soldering near the fingers should also be avoided. Solder wicking onto a contact area changes its geometry and surface, and rework there is hard to reverse. Where a component has to be close to the fingers, the layout should keep the paste and the reflow away from the contact area, and the assembly drawing should say so explicitly.

Verification

The plating on gold fingers is verified by measurement rather than by inspection, and three checks cover most of the requirement.

Thickness is measured by X-ray fluorescence, which gives the gold and nickel thickness without destroying the sample. A cross section on a first article confirms the layer structure and shows whether the nickel is continuous. Adhesion and porosity are checked by a tape test and by a corrosion test, where a porous coating shows as a stain pattern that reveals the nickel and copper underneath.

The mechanical properties are verified on the assembly rather than on the board. The insertion and withdrawal force confirms that the board thickness and the bevel are correct for the socket, and the cycle life is established by repeatedly inserting and removing a sample while measuring the contact resistance.

A reading in the low milliohms is normal for a gold to gold contact with reasonable force, and a reading that drifts with each insertion indicates wear or contamination rather than a design error. Where the fingers are also used as test points, the probing adds wear that the connector cycle count does not include, which is a reason to keep test points away from the contact area. Our article on in-circuit test covers how test access is normally provided without touching a wear surface.

FAQ

• Can soft gold be used on a card edge? No. Soft gold wears through quickly in a sliding contact. Wire bonding and sliding contact require different gold specifications.
• How many insertion cycles can gold fingers survive? It depends on the gold thickness, the contact force and the finish of the socket contact. A thin plating suits a few insertions and a thick plating suits thousands.
• Is a bevel always required? For a card edge connector, yes. It guides the board into the socket and prevents the contact from catching on the edge.
• Can gold fingers be soldered? They can be soldered in an emergency, but the joint is poor and the process destroys the contact surface for later insertion. It should not be part of a design.
• Does the board thickness tolerance really matter? Yes. A socket is designed for a nominal thickness with a narrow tolerance, and a board outside it either cannot be inserted or damages the contacts.

Summary

Gold fingers are a mechanical interface built on a circuit board. Hard gold over a nickel barrier resists the wear of a sliding contact, the bevel guides the board into the socket, and the finger geometry, the board thickness and the plating thickness are all set by the connector rather than by the circuit.

The design work is therefore mostly about constraints: match the pitch and thickness to the socket, keep the contact area clear of vias, mask and legend, specify the gold and nickel thickness together, and confirm the achievable tolerance with the fabricator and the insertion force with a sample. Done properly, the fingers last for the life of the product; done carelessly, they are the reason a module fails after a few insertions, and the failure is in the layout rather than in the connector.