PCB Design Workflow: From Netlist to Final Layout

Schematic capture exists for a practical reason: it produces the netlist that the board layout is built from and establishes the basis for everything that follows. Once the netlist is available, the layout work itself is a sequence of decisions that run in a defined order, and doing them out of order is what causes most of the rework. The workflow below applies to a double-sided board and to a multilayer board alike; the differences appear in the middle layers, where signal routing and plane splitting add a step.

Step one: plan the board

Planning sets the physical dimensions of the board, the package type for each component, the mounting method and the layer structure, meaning whether the design will be single-sided, double-sided or multilayer. These decisions constrain everything downstream. A board that is 10 mm too small cannot be fixed later by better routing, and a layer count chosen by habit rather than by the number of nets will either waste cost or force the routing into a corner. Confirm the board outline against the enclosure and the mounting holes before proceeding, because moving an outline after placement has begun invalidates the placement.

Step two: configure the working parameters

Working parameters cover the design environment and the layer definitions. Setting them correctly at the start saves a great deal of effort later, because grid, units, design rules and layer names propagate into every subsequent operation. In a multilayer design this is also where the layer count and the plane structure are declared in the layer stack manager, including the number of mid signal layers and internal planes and their order in the stack.

Step three: component placement

With the preparation complete, import the netlist into the layout tool, either from a netlist file or by updating the board from the schematic. Automatic placement is available in most tools, but its results are rarely good enough, and manual placement is the normal choice, especially for complex circuits and for components with special requirements. Placement is one of the most important steps in the whole process, because it determines how difficult the routing will be and how the inner planes can be divided. It is worth the time. Work from the connections rather than from the schematic grouping, keep the high-current and high-frequency paths short, and check that the layout leaves room for rework and test access. The rules summarised under placement order and pad positioning apply here.

Step four: define the middle layers

In a multilayer board, define the mid signal layers and the internal planes in the stack-up manager before routing begins. The number of layers, their type and their order determine the reference plane available to each signal layer and therefore the impedance of every trace. Changing the stack after routing has started means abandoning the routing, which is why this step belongs before the first net is connected.

Step five: split the internal planes

An internal plane layer usually carries more than one power net, so it has to be divided into separate, isolated regions, each connected to a specific supply. This is the biggest difference between a multilayer board and a simple double-sided one, and it is a significant part of the layout work. The shape of the split affects the routing of power and ground and is itself affected by placement and by the signal routing, so it cannot be finalised in isolation. An inner layer split that follows the placement is straightforward to route around; one that cuts across the middle of the placement forces both the routing and the plane to be reworked. Keep the return path in mind: a signal that crosses a split in its reference plane loses the continuous return directly beneath it. The reasoning is set out under power plane splitting rules.

Step six: set the routing rules

Routing rules define the trace width, the spacing between parallel traces, the clearance between a trace and a pad, and the via size. Whatever routing method is used, this step is not optional. Well-chosen rules keep the routing electrically safe, keep it within the fabrication process capability and control cost, because a design that demands tighter geometry than the process can hold will be quoted higher or rejected. Load the rules from the fabricator capability rather than from the tool defaults.

Step seven: route and adjust

Automatic routing exists in every tool and satisfies almost no real design. In practice, either manual routing or a combination of automatic routing and interactive manual routing is used. The important point is that placement, routing and plane splitting are not sequential and final; they interact. Routing may show that a component has to move, and a plane split may show that a net should be rerouted. Expect to iterate between the three rather than treating each as complete before the next begins.

Step eight: auxiliary operations and documentation

The remaining work includes copper pouring, adding teardrops, generating reports and saving and printing the documentation. Those outputs are used to check and modify the board, and the bill of materials extracted from the design becomes the purchase list for components. Treat the reports as part of the deliverable rather than as an afterthought, because a design that cannot be documented cannot be manufactured or maintained. Confirm the whole package against the design and fabrication requirements before release. gopcb reviews the netlist, stack-up and rules together, because a mismatch between them is the most common reason a layout has to be reworked after submission.

Step nine: verification and release

Before the design is released, run the checks that catch the mistakes the routing itself cannot reveal. An electrical rule check compares the board against the schematic and catches a net that was accidentally deleted, a component that was never placed and a pin that is connected to the wrong net. A design rule check compares the geometry against the process capability and catches a trace that is too close to a pad, an annular ring that is too thin and a hole that is outside the drill range. Compare the netlist extracted from the finished board with the netlist from the schematic, because the two diverge whenever a change was made on one side only. Review the silkscreen for overlapping text and for designators that sit on pads, and check the assembly in three dimensions for collisions with the enclosure and between tall components.

Then generate the outputs as a complete set rather than one file at a time: the Gerber layers, the drill file with its tool table, the bill of materials and the pick-and-place data. Add fabrication notes covering the stack-up, the controlled impedance nets, the surface finish and any via plugging requirement. The documentation is what allows the board to be built as designed, and a missing note is the most common reason a shop has to ask a question that delays the order.

FAQ

Why is automatic placement usually avoided? It optimises for a generic objective such as total connection length and cannot account for thermal paths, connector access, rework clearance or plane splitting. Manual placement guided by the circuit’s actual current paths produces a better board.

When should the internal planes be split? After placement is broadly settled and before routing is finalised. The split depends on where the components sit, and the routing depends on the split, so the two are adjusted together rather than fixed in sequence.

Can routing rules be left at the tool defaults? No. Default rules are usually looser than any real process and give a false sense of security. Load the fabricator’s capability, run the check, and review every remaining violation with a reason.