Gold Fingers and Edge Connector Design

A gold finger is a row of equally spaced rectangular pads along the edge of a board, plated with gold and left free of solder mask. The board slides into a slot connector and the pads make contact with the spring contacts inside it, which makes the gold finger an interface rather than a component.

This article covers the different finger layouts, why gold is used, what the fabrication drawing has to specify, and the details of bevelling and plating that decide whether the board mates reliably for thousands of insertions.

What a Gold Finger Is

The finger is formed at the edge of the board, on one or both sides, as a set of pads that extend to the board outline. The mechanical contact is made by the connector’s springs pressing against the gold surface, so the surface has to be hard enough to resist wear, conductive enough to keep the contact resistance low, and chemically stable enough not to oxidise in the environment it sits in.

The pads are usually wider and more widely spaced than the fine pitch features elsewhere on the board, because the connector is a mechanical part with its own tolerances. Registration to the board edge is more important than the pad-to-pad spacing in most cases, since the connector aligns to the edge and then expects the fingers to be where they should be relative to it.

Why Gold and Where It Is Used

Gold satisfies all three requirements unusually well: it does not form a significant oxide layer, its contact resistance is low and stable, and a hard gold deposit resists the abrasion of repeated insertion. It is also expensive, which is why it is applied selectively, only where a contact has to be made, rather than over the whole board.

The selective plating is what makes the process practical. The fingers are connected to a common plating bar during electroplating and then isolated when the bar is routed away and the board is profiled. A board with gold fingers therefore has a fabrication step that a board without them does not, and the layout has to make provision for the plating connection.

Flush, Segmented and Unequal-Length Fingers

The most common arrangement is a set of flush fingers of the same length and width, evenly spaced along the edge. This is the layout of a memory module or an add-in card, and its regularity makes both the connector and the board easy to design.

Segmented fingers break the row into groups with a gap between them, which allows a single connector to be keyed or to carry different signals in different regions. Unequal-length fingers are used where contacts have to engage in a defined order, such as a ground pin that connects before the signal pins and disconnects after them. That ordering has to be reflected in the connector as well as on the board, and the two have to be specified together.

Recognising an Edge Connector on a Layout

An edge connector board has a characteristic appearance. There is no silkscreen reference on the fingers themselves, because the mask is opened over the whole contact area rather than printed with a frame. The outline often has a notch or a key that prevents the board being inserted the wrong way round, and the fingers project towards the edge or sit immediately alongside it.

Some boards have fingers on both faces, others on one only, and a few use wider fingers for the power contacts. All of these variations are visible on the fabrication drawing if the drawing is complete, and the drawing is the document the connector specification has to agree with. A mismatch between the two is discovered when the board is inserted for the first time.

Bevelling and Insertion

The mating edge is bevelled, which means the edge of the board is chamfered so that it presents a taper to the connector. The taper guides the board into the slot and reduces the force needed to start the insertion, and it also protects the fingers from being lifted by the leading edge of the contact springs.

The bevel has an angle and a depth, and both belong on the drawing. Where the bevel depth is too small the fingers can still catch; where it is too large the effective length of the outermost finger is reduced, because part of it has been cut away. The tolerance on the bevel position relative to the fingers is one of the parameters that a connector specification will state, and it should be carried onto the fabrication drawing unchanged.

Plating: Hard Gold and Soft Gold

The gold on a finger is not the same as the gold used on a solderable pad. A contact needs a hard deposit, so the gold is alloyed with a small amount of cobalt or nickel to raise its hardness, and it is plated over a nickel barrier layer which prevents the copper beneath from diffusing into the gold. The nickel also provides the mechanical support that the thin gold layer needs.

The soft, pure gold used for wire bonding or for a solderable finish is unsuitable for a contact, because it wears away quickly. Where a board has both a gold finger and pads that will be soldered, the two finishes are applied in separate steps, and the plating thickness for each is quoted separately. The electroplating additives used in the plating bath are what determine the grain structure and therefore the hardness of the deposit.

Tolerances and the Fabrication Drawing

The drawing has to state four things about the fingers: their width and spacing, their position relative to the board edge, the bevel angle and depth, and the plating thickness for both the nickel and the gold. Where the board has uneven-length fingers, the length of each group is part of the same statement.

The position relative to the edge is the tolerance that causes the most trouble. A connector locates on the board edge and then relies on the fingers being at a defined distance from it, so the cumulative tolerance has to include the routing of the outline as well as the imaging of the fingers. Where the outline is routed after plating, as it usually is, the routing tolerance is part of the stack and should be accounted for.

Design Details That Affect Life

The number of insertions a connector is rated for is a property of the connector, but the life achieved depends on the board as well. A finger with a rough plated surface wears the connector springs faster, and a finger whose bevel is imperfect accelerates the wear further. The grain structure of the gold, which is set by the plating process, is what makes the difference between a smooth deposit and a rough one.

Contamination is the other factor. Solder flux, handling oils and ionic residues all increase contact resistance, and on a card-edge interface they cannot be cleaned after assembly. Keeping the fingers masked during soldering, handling the boards by the edges, and verifying cleanliness before packing are simple measures that preserve the surface. The protection applied to the rest of the assembly must be kept off the contact area for the same reason.

FAQ

Can a gold finger be soldered instead of mated? It can be soldered, but that wastes the finish. Where a permanent connection is intended, a normal pad with a solderable finish is cheaper and gives a better joint.

How thick should the gold be? The specification usually calls for a thin hard gold over a thicker nickel barrier, with the exact figure coming from the connector supplier and the expected insertion count. The plating requirements used elsewhere on the board do not transfer directly, because contacts have different needs from vias.

Why does the board need keying? Because inserting a card the wrong way round damages both the board and the connector. A notch in the outline or an asymmetrical finger pattern makes the error impossible rather than merely unlikely.