Gold Finger: Design Rules and Process Limits
An edge connector uses the board itself as the plug. The contacts are copper fingers plated with a wear resistant metal, and they slide into a mating socket when the card is inserted. The finger geometry, the plating stack, and the bevel at the leading edge decide how many insertion cycles the card survives, and all three are decided at the design stage rather than at the plating line.
This article explains the geometry of the finger area, the plating stack and why a nickel barrier is used, how the bevel is produced, and the handling rules that keep the contacts clean.
The Geometry Of The Finger Area
A gold finger is a rectangular pad, usually between one and a half and three millimetres wide on a pitch between one and two and a half millimetres, running inwards from the board edge for the length that the mating contact slides. The length is set by the connector and by the insertion depth, and the finger should be long enough that the contact rests on gold over the whole tolerance range of the mating position, including wear and the tolerance on the board thickness.
The board thickness is part of the design, because the mating socket grips the card by its thickness, and a card at the low end of the tolerance range sits more loosely than one at the high end. The board edge itself has to be clean and square, and the area behind the fingers is usually kept clear of components, of vias, and of solder mask features that could interfere with the socket. Keying slots, which prevent the card from being inserted the wrong way, are routed in the same area and have to be positioned to suit the connector rather than the layout.
The Plating Stack
The standard stack is hard gold over nickel over copper. The nickel layer, typically two to five micrometres thick, is a diffusion barrier: without it the gold and the copper would interdiffuse, and the gold would lose its wear resistance as copper appeared at the surface. The gold layer is between about half a micrometre and one and a half, and the thickness is chosen from the number of insertion cycles the product requires, with a thicker deposit for a connector that will be mated many times.
Hard gold is an alloy of gold with a small amount of cobalt or nickel, which raises its hardness and its resistance to abrasion, and it is used for the mating surface rather than pure gold, which is softer. The gold is usually plated only on the connector area, with the remainder of the board carrying the surface finish used for soldering. That selective plating requires the fingers to be connected electrically during the plating process, which is done with a bus bar or a plating strip that is removed after plating, and the removal leaves an area that must be defined in the artwork with a clean edge and no exposed copper.
The Bevel And The Leading Edge
The bevel is the chamfer machined along the leading edge of the board so that the card enters the socket without catching and without damaging the contacts inside it. The angle is usually between twenty and forty five degrees, and the depth is set to remove the sharp corner while leaving enough of the finger to make contact. A bevel that is too shallow leaves a corner that can damage a contact spring, and one that is too deep removes the plating from part of the finger.
The bevel has to be plated rather than left as bare laminate. On most boards the bevel is machined after plating, which means the plating is then removed from the chamfered surface, and the copper or the laminate beneath it is exposed unless a second step restores the finish. Unless the process controls this precisely, the bevel should be machined before plating so that the gold covers the chamfer as well as the flat surface, because a card whose bevel is bare copper will corrode and can short adjacent contacts in the socket.
Assembly And Handling
The fingers must be kept free of solder, flux, and coating. A solder splash on a finger changes the thickness of the contact and can prevent the card from seating, and a film of flux residue raises the contact resistance and can be corrosive over time. The usual practice is to protect the area with a tape or a peelable mask during assembly, and to remove it before the card is packed, and the process instructions should state that the fingers are not to be handled with bare hands at any stage.
Tinning a finger is never acceptable. Where a wire has to be attached near the edge, it is attached to a separate pad, and the finger area is treated as a contact surface rather than as a solderable area. Where a conformal coating is applied, the fingers are masked so that no coating reaches them, because a coating on a contact surface is an insulator that the contact pressure may not displace.
Tin Fingers And Their Limits
A tin or tin lead finish is occasionally used on a low cost edge connector, and its limitation is cycle life. Tin is soft and oxidises, and the oxide film that forms on the surface raises the contact resistance after a small number of insertions, so a tin plated finger is suitable only for a connector that is mated once or twice during assembly and never again. A gold plated finger costs more and holds a stable contact resistance over hundreds of cycles, which is why it is the default for any card that will be inserted repeatedly.
The contact resistance also depends on the contact force and on the geometry of the mating spring, which are properties of the connector rather than of the board. The board contributes the surface, the flatness of the finger, and the thickness of the card, and a finger that is covered by a mask ledge or that is not flat changes the contact geometry in a way that no plating can compensate for.
Inspection And Test
Plating thickness is measured by X-ray fluorescence on the flat finger and, where the process produces it, on the bevel, and the measurement is taken on a sample from each panel or each lot. Adhesion is checked by a tape test, and wear resistance is verified by an abrasion or insertion cycle test on a coupon, which is a better predictor of field behaviour than a hardness measurement alone.
The visual criteria cover the defects that affect contact: scratches that penetrate the gold, nicks in the finger edge, exposed nickel or copper, plating that wraps onto an adjacent area, and burrs left by the bevel cut. A scratch in the middle of a finger is less serious than one that removes the gold, because the contact can slide over a shallow scratch but cannot make a reliable connection to an exposed layer. The pad geometry that the fingers follow is described under pad design standards, the plating sequence under electroplating additives, and the outline features under board outline and mounting design.
FAQ
Why is nickel used under the gold? It is a diffusion barrier that stops the copper from migrating into the gold. Without it the contact surface changes composition and loses its wear resistance and its stable resistance.
Should the bevel be cut before or after plating? So that the gold covers the chamfer. A bevel cut after plating leaves exposed copper or laminate on the leading edge, which corrodes and can short adjacent contacts.
How thick should the gold be? It depends on the number of insertion cycles the product requires. A connector that is mated once needs only a thin deposit, while a card that is inserted repeatedly needs the thicker end of the range.
