Home › Guides › PCB Manufacturing Process

PCB Manufacturing Process: From Bare Board to Finished Assembly

A printed circuit board starts life as a design file and ends as a finished assembly covered in components. Between those two points it passes through two very different factories in spirit: fabrication, which makes the bare board, and assembly, which populates it. This guide walks through both stages in order, explains what each step actually does, and flags the decisions a designer makes that quietly determine how well the whole process goes.

Fabrication and assembly: two halves of one process

It helps to keep the two halves separate in your head. Fabrication takes your Gerber files, drill files and board specification and produces a bare printed circuit board: layers of copper on an insulating substrate, holes drilled and plated where needed, a protective solder mask over everything except the pads, and printed silkscreen markings. Assembly takes that bare board plus a bill of materials and a pick-and-place file, and turns it into a working circuit by placing components and soldering them. Problems in fabrication usually show up as board defects: broken traces, misaligned layers, poor plating. Problems in assembly usually show up as component defects: tombstoned parts, solder bridges, missing components. Knowing which half a fault belongs to is the first step in getting it fixed, and it is also why this guide treats them separately.

The handover between the two halves is the bare board itself, and its quality is checked before a single component is placed. A fabrication house will run electrical tests on the bare board, checking continuity and isolation, and will inspect it visually and often with automated optical inspection. If the bare board is wrong, assembling it only wastes components. Good designers therefore treat the fabrication output as a product in its own right, not as an intermediate step to rush past.

Step 1: from design files to a panel

Manufacturing starts with the design outputs. The Gerber files describe each copper layer, the solder mask layers, the silkscreen layers and the board outline. The drill file describes every hole: its position, diameter and whether it is plated. The board specification covers the material, the number of layers, the finished copper thickness, the board thickness and the surface finish. Before anything is cut, the manufacturer checks these files for problems: missing layers, drill sizes that do not match the specification, clearances that are too tight for their process, and features that cannot be made at all. This check is often called a design-for-manufacturing review, and it is where many first-time designs get their first corrections.

Boards are rarely made one at a time. The manufacturer groups several copies of your board onto a larger panel, with tooling holes and test coupons around the edges. Panelisation makes handling easier and lets the same machine steps process many boards at once. Small boards in particular are almost always panelised, and the designer usually specifies whether boards should be separated by routing, by V-scoring, or left in the panel for the assembler to break out later. That choice matters: V-scoring is cheap but leaves a rough edge, while routing gives a clean edge at the cost of panel space.

Step 2: imaging and etching the copper

The core of fabrication is turning a copper-clad laminate into a pattern of traces. For each copper layer, the manufacturer starts with a sheet of insulating material clad on one or both sides with copper foil. A light-sensitive film called photoresist is applied, and the layer pattern is transferred onto it by exposure through a photographic film or by direct laser imaging. The exposed resist is developed, leaving resist only where copper should remain. The sheet is then etched: a chemical bath dissolves the exposed copper, leaving only the traces, pads and planes protected by the resist. Finally the remaining resist is stripped away.

For a two-layer board this happens on both sides of one laminate sheet, with holes drilled and plated to connect the sides. For a multilayer board the process is repeated for each inner layer, and the layers are then laminated together under heat and pressure into a single stack, with prepreg — partially cured insulating material — bonding them. Alignment between layers is critical: if an inner layer is misaligned, a via can miss its target pad, and the board will fail electrical test. Registration marks on each layer let the manufacturer measure and correct alignment before lamination.

Etching quality is limited by physics. Very fine traces and very small clearances are harder to etch accurately, because the etchant undercuts the resist slightly and eats sideways as well as downwards. This is one reason manufacturers publish minimum trace widths and clearances: they are not being difficult, they are stating what their etching process can reliably reproduce. Designing comfortably inside those limits is cheaper than designing at the edge and paying for scrap.

Step 3: drilling and plating the holes

Holes come in two kinds. Non-plated holes are simply drilled for mounting or mechanical purposes. Plated holes have a thin layer of copper deposited on their walls, which connects copper on different layers or provides a solderable barrel for a through-hole component lead. Drilling is done on computer-controlled machines that can drill thousands of holes quickly, and the hole pattern comes directly from your drill file.

Plating is an electroless and then electrolytic copper process. First a chemical bath deposits a thin conductive seed layer on the hole walls, including the inside of the hole, which the laminate alone would not conduct. Then electroplating builds that seed up to a solid copper barrel. Plating thickness matters: too thin and the barrel can crack under thermal stress or during soldering; too thick and small holes can close up. Vias — the small holes that connect layers but carry no component lead — go through the same process, and on dense boards they are often filled or capped as an extra step.

Step 4: solder mask and silkscreen

With the copper pattern complete, the board gets its solder mask: a thin polymer coating applied over the whole surface except where solder is needed. The mask is usually green, but other colours are available, and its job is threefold. It prevents solder from bridging between adjacent traces during assembly, it protects the copper from oxidation and handling damage, and it insulates the traces from accidental contact. Openings in the mask, called mask openings, expose only the pads and test points. The alignment of those openings to the pads matters as much as the copper alignment: a mask opening shifted over a pad reduces the solderable area and can cause weak joints.

Silkscreen comes next: white or contrasting ink printed on top of the mask, showing component outlines, reference designators, polarity marks, logos and assembly notes. Good silkscreen is a practical aid, not decoration. A clear polarity mark next to a diode or electrolytic capacitor prevents a whole class of assembly errors, and readable reference designators make debugging and rework far faster. Silkscreen that overlaps pads, however, is a defect: ink on a pad prevents solder from wetting it, so manufacturers keep silkscreen clear of all solderable surfaces.

Step 5: surface finish

The exposed copper pads left open by the solder mask would oxidise within days if left bare, so they receive a surface finish. The common finishes each trade cost, shelf life, flatness and compatibility. HASL, hot air solder levelling, coats the pads with solder and levels it with hot air; it is cheap and robust but not perfectly flat, which matters for fine-pitch parts. ENIG, electroless nickel immersion gold, deposits nickel then a thin gold layer; it is very flat, has a long shelf life and solders well, but costs more and needs care with the gold thickness. Immersion tin and immersion silver are flat and solder well but have shorter shelf lives and need handling care. OSP, organic solderability preservative, is a thin organic coating that is cheap and flat but has the shortest shelf life of all and is easily damaged by handling.

The finish choice is part of the board specification, and it should match the assembly plan. A board with fine-pitch surface-mount parts benefits from a flat finish such as ENIG. A simple board assembled soon after fabrication can use HASL or OSP without penalty. What matters most is that the finish is applied evenly and that the board is assembled within its shelf life, because a finish that has degraded is one of the quietest causes of widespread soldering problems.

Step 6: testing and routing the bare board

Before the bare board leaves the fabricator, it is electrically tested. A flying-probe tester or a bed-of-nails fixture checks every net for continuity and for shorts to other nets, comparing the result against the design netlist. This catches etching defects, drill errors and layer misalignments that visual inspection would miss. Boards that fail are usually scrapped or, if the defect is minor and repairable, reworked — but rework on a bare board is limited, and most manufacturers prefer to remake a failed board than to ship a repaired one.

Finally the panel is routed or scored into individual boards, the tooling edges are removed, and the boards are cleaned, inspected and packed, usually in moisture-resistant packaging. Boards with a moisture-sensitive finish or with components planned later are packed with desiccant and a humidity indicator, because moisture absorbed into the laminate can cause delamination during reflow soldering. That packing is not cosmetic: it protects the board all the way to the assembly line.

Step 7: SMT assembly

Assembly usually starts with surface-mount components. Solder paste — a mixture of tiny solder spheres and flux — is printed onto the pads through a stencil, a thin metal sheet with apertures cut exactly where paste is needed. A pick-and-place machine then places each component onto its paste, at high speed and with vision systems checking alignment. The board then passes through a reflow oven, where a controlled temperature profile melts the paste, lets surface tension pull components into alignment, and then cools to form the joints. The profile matters: too cool and joints do not form; too hot or too long and components or the board itself can be damaged.

After reflow, the board is inspected. Automated optical inspection checks component presence, orientation, polarity and obvious solder defects. For hidden joints, such as ball grid arrays where the solder balls sit underneath the package, X-ray inspection is used instead, because no camera can see those joints. Defects found here are reworked by hand or by selective rework stations before the board moves on.

Step 8: through-hole assembly and final inspection

Through-hole components are inserted next, either by hand or by machine, and soldered. Wave soldering passes the board over a standing wave of molten solder, which wets the leads and fills the plated holes. Selective soldering solders individual joints for boards that cannot tolerate a full wave, such as boards already carrying sensitive surface-mount parts on the underside. Hand soldering remains common for connectors, large components and low-volume work, and it is also where rework happens.

The finished assembly then goes through final inspection and test. Visual and automated inspection catch physical defects. Functional test powers the board and checks that it does what it should, using either a dedicated test fixture or the product's own firmware. In-circuit test, where it is used, checks individual component values and solder joints electrically before functional test. Boards that pass are cleaned, conformally coated if the design requires it, and packed. Boards that fail go back for diagnosis — which is where the troubleshooting discipline in our companion guide comes in.

What the designer controls

Almost every manufacturing problem has a design decision behind it. Minimum trace widths and clearances inside the manufacturer's capability, a sensible stack-up, correct drill sizes, a surface finish matched to the parts, a clear silkscreen, and a complete and accurate bill of materials together remove most of the common causes of delay, scrap and rework. The cheapest time to fix a manufacturing problem is before the files are sent, when it costs nothing but a few minutes of checking. The most expensive time is after assembly, when components, boards and time have all been spent. Designing for manufacture is not a separate skill from circuit design; it is the part of circuit design that decides whether the circuit ever gets built at all.

Frequently asked questions

For the design side of this workflow, see PCB design with AI. For what to do when a finished board misbehaves, see PCB troubleshooting guide, and for the wider picture of AI in engineering work, see AI predictive maintenance and AI in power systems.