Gold Finger PCB Manufacturing: Hard Gold, Bevel and Thickness
The Interface That Decides Product Life
In any product where a board is inserted into a slot, the electrical connection is made across a set of plated contacts on the board edge. PCIe add-in cards, memory modules, GPU and FPGA boards, server backplanes, industrial control cards and automotive and medical modules all rely on this interface. Gold finger contacts are repeatedly mated and unmated, sometimes thousands of times over a product’s life, and the plating system on those contacts decides whether contact resistance stays low or climbs until the link becomes unreliable.
Building them properly is a specialised process rather than a variation on ordinary surface finishing, and this guide covers the design rules, the plating and bevel requirements, the manufacturing flow, inspection and cost structure.

What a Gold Finger Board Is
A gold finger board carries electroplated hard gold contacts along one or more edges, positioned so that an edge connector or slot makes contact when the board is inserted. Four features define the technology. The contacts sit on the board edge, which changes how they are handled and supported. They are built from electroplated hard gold rather than from an immersion or electroless finish. The board edge is bevelled, typically at 30 or 45 degrees, to guide insertion. And the whole design is intended for repeated mating cycles rather than for a single assembly operation.
In effect, the gold fingers are the mechanical and electrical interface between the board and the system, and they carry both the signal integrity and the mechanical wear burden of that interface.
Where They Are Used
PCIe expansion cards are the most familiar case, followed by memory modules, GPU and FPGA accelerator boards, server and networking backplanes, industrial control hardware, and automotive or medical modules that use cartridge style connection. Any design that combines edge insertion with a requirement for reliable contact over many cycles will land on gold fingers as the solution.
Design Rules
Design decisions made at layout time determine whether the plating can be deposited evenly and whether the interface will survive its service life. Finger length and pitch must match the connector specification, with tolerances that account for insertion depth and any connector float. Solder mask must keep a defined clearance from the fingers, because mask bleeding onto a contact area creates an insulation defect that cannot be repaired. Traces and vias should be kept back from the board edge so that routing does not compromise edge strength or expose features during beveling. Edge mechanical strength matters, particularly on thin boards where the finger region is a stress riser, so copper distribution and, where necessary, additional stiffening must be planned. Finally, finger geometry should be uniform across the row; uneven pad areas deposit unevenly, and uneven plating produces uneven wear.
Poor design in these areas shows up as uneven gold thickness, gold lifting from the nickel, or abnormal wear after relatively few cycles.

The Bevel Requirement
Beveling is mandatory rather than optional. The board edge is cut at a defined angle, usually 30 or 45 degrees, which removes the burr left by routing, prevents the sharp edge from damaging connector spring contacts, and makes insertion smooth rather than a scraping action. IPC guidance sets expectations for bevel depth and angle tolerance, and a bevelled edge that is out of tolerance will either fail to guide the board or will scrape the contacts it is supposed to protect.
Why Hard Gold, Not ENIG
This is the single point that most often gets confused. Electroless nickel immersion gold is a chemical deposition process, and the gold layer it deposits is thin and relatively soft. It is an excellent surface for soldering and for a single assembly pass, but it is not built for repeated mechanical contact. Gold fingers require electroplated hard gold over a nickel barrier: the nickel provides a hard, diffusion resistant underlayer and the gold layer provides low contact resistance and wear resistance, with thickness typically in the range of 10 to 50 microinches depending on the mating cycle requirement.
ENIG can never substitute for hard gold in this application. A board built with ENIG fingers will work when first inserted and will degrade as the plating wears through.
The Manufacturing Flow
Gold finger production diverges from the standard board flow at the point where surface finishing begins. Lamination and drilling proceed normally, followed by pattern plating and then the dedicated steps for the fingers.
- Lamination, drilling and copper pattern formation.
- Nickel plating of the finger areas.
- Electroplated hard gold onto the nickel, with thickness controlled by current density and time.
- Solder mask application, deliberately patterned to keep the fingers clear.
- A separate finish, ENIG, HASL or OSP, applied to the remainder of the board.
- Precision beveling of the board edge at the specified angle.
- Cleaning and protective packaging.
The key characteristic of this flow is that surface finishing is zoned: the finger region receives one treatment and the rest of the board receives another. That zoning is what makes the process demanding, because the mask and resist steps that separate the two regions must be perfect.
Common Defects
Five defects account for most trouble. Uneven gold thickness across the finger row, which produces uneven wear and unreliable contact resistance. Gold lifting or peeling, which usually traces to poor nickel adhesion or process contamination. Out-of-tolerance bevel angles. Burrs left along the edge. And inadequate nickel adhesion resulting from inadequate surface preparation before plating. These failures cluster in facilities that do not run a dedicated hard gold line, which is a useful signal when qualifying a supplier.
Inspection and Quality Control
Verification of a gold finger board is more specific than on a standard build. X-ray fluorescence measurement quantifies gold thickness, and it should be measured at several points across the finger row, not at a single location. Automated optical inspection and visual inspection check for mask bleed, scratches and contamination. Bevel angle measurement verifies the geometry against specification. Adhesion testing on coupons verifies that nickel and gold bond to copper as intended. Where the application demands it, mating cycle wear testing characterises how the contacts behave over the expected life.
Assembly Considerations
Most damage to gold fingers happens after fabrication, during assembly. Handling and processing equipment can scratch or contaminate the contacts. Solder paste, flux and cleaning chemistry can deposit residue on the exposed gold. Boards can be stacked or slid against one another. Protecting the fingers with a mask or tape through the assembly process, controlling handling, and inspecting and cleaning the contacts before final packaging are basic but effective measures. Equally important is using the correct mating connector: an incorrect or worn connector will damage a perfectly good board edge.
Assemblies that combine fine-pitch surface mount with edge contacts need an assembly partner who treats the interface as a controlled feature. The practices described in our PCB assembly notes on handling and cleanliness apply directly here.
Cost Structure
Gold finger boards cost noticeably more than equivalent boards without them, and the reasons are specific: hard gold electroplating consumes real gold and requires a dedicated line, beveling is an extra machining step, gold thickness directly scales material cost, and higher layer counts and greater board thickness add fabrication cost. Typical market pricing illustrates the spread. A two layer 1.6 mm board with 10 microinch fingers sits around 80 to 120 US dollars for ten pieces. A four layer board with 30 microinch plating runs about 180 to 260 dollars for ten pieces. A six layer server class board with 50 microinch hard gold runs roughly 400 to 650 dollars for ten pieces. Thickness is the dominant variable, followed by layer count; the same cost logic that applies to custom PCB pricing governs the rest of the quote.
Selecting a Manufacturer
Five capabilities define a capable supplier for this work: a dedicated hard gold plating line, precision beveling equipment, XRF thickness measurement in house, experience with PCIe, memory module and server backplane products, and a quality system that documents plating and bevel measurements per lot. Not every PCB factory can build gold fingers well, and the difference usually only becomes visible after a few hundred mating cycles. Buyers should ask for the plating thickness distribution across a panel, not a single nominal figure, and should review how the supplier protects the contacts between finishing and shipment under a documented quality management system.
For a wider look at how plating and surface finishing interact with the rest of the build, see our PCB manufacturing overview and the range of finishes covered under immersion silver and other surface finishes.
Questions Engineers Ask
Can ENIG replace hard gold on fingers? No. ENIG is too soft and too thin to survive repeated mating cycles.
What gold thickness is normal? Between 10 and 50 microinches, selected according to the number of mating cycles the product must survive.
Is beveling really necessary? Yes. It removes the burr, protects the connector contacts and makes insertion reliable.
Why is the board more expensive? Hard gold plating and beveling are specialised processes with their own tooling and inspection requirements.
Conclusion
Gold finger boards sit at the intersection of plating chemistry, mechanical finishing and connector mechanics. Design determines whether plating can be even, hard gold over nickel determines whether it survives repeated mating, beveling determines whether insertion damages the connector, and inspection determines whether any of that can be proven. For PCIe cards, memory modules, GPU boards and server backplanes, choosing a supplier with a genuine hard gold capability is the difference between an interface that lasts the life of the product and one that fails unexpectedly in the field.



