Ten Golden Rules of PCB Design That Still Apply

Semiconductor integration keeps improving and complete systems on a chip are available off the shelf, yet most products still need a board of their own. That board is the physical platform on which the rest of the design is built, and the rules that govern it have changed surprisingly little since commercial layout began. The ten golden rules that follow are the ones worth keeping in front of you: they apply to a one-off prototype and to a product built in thousands, and none of them has become obsolete with the arrival of better tools.

1. Choose the Grid Deliberately

Set up a grid that matches the pitch of most of the components in the design and use it consistently. Multiple grids look convenient and cause trouble, because objects placed on different grids do not align, and the errors appear as small offsets that are invisible at the zoom level where the work is done. The grid also determines how copper pours behave: a pour generated on a single grid is more regular than one generated across several, and the difference shows up in the finished artwork.

2. Keep the Path as Short and Direct as Possible

This is obvious enough to be overlooked. Every stage of the layout should be checked against it, up to and including moving parts to shorten a route. It matters most in analog and high-speed digital circuits, where the impedance and the parasitic elements of the path are part of the circuit rather than a nuisance attached to it.

3. Use Planes to Distribute Power and Ground

Pouring a plane is one of the quickest operations in a layout tool and one of the most valuable. A plane ties many connections together with the lowest practical impedance and voltage drop, and it gives the return current a path immediately under every trace on the adjacent layer. Where several supply rails exist, they can be poured in the same region on different layers, and the extent of the ground plane should be checked so that it covers most of the area beneath the signals it references.

4. Group Related Components Together, With Their Test Points

Place the parts that work together close together. The bypass capacitor, the feedback resistor and the device they serve belong in one group, because the grouping is what keeps the connections short and what makes the circuit measurable. Include the test points in the group at the same time, so that fault finding on the finished board does not require a probe balanced on a component lead.

5. Panelise the Board for the Manufacturer

Repeat the design across a larger panel rather than submitting a single small board. Ask the fabricator which panel size suits their equipment and work within it, because the panel the shop prefers is the one that costs least. The same exercise sets up the assembly panel, so the decision is useful twice.

6. Consolidate Component Values

Where two values are functionally equivalent, use the same one. Fewer distinct values means a shorter parts list, simpler procurement and a lower risk of a substitution error at assembly. It also makes inventory management possible: a product family built from a standard set of values can be stocked and planned instead of bought per project.

7. Run the Design Rule Check Often

The design rule check takes seconds, and running it only at the end of the layout turns a series of small corrections into a rework session. Run it after every significant editing pass, and read the results rather than accepting them, because a rule that fires repeatedly is usually a rule that needs adjusting rather than a set of mistakes.

<img src="https://www.gopcba.com/wp-content/uploads/2026/05/医疗电子可持续性.png" alt="silkscreen labels and test points on a board” />

8. Use the Silkscreen

Silkscreen is free documentation printed on the product, and it is read by the fabricator, the assembly operator, the service engineer and the installation technician. Label the function of each connector and the location of each test point, and mark the orientation of the components and the connectors whose rotation matters. Where a feature is only visible after assembly, print the note on the side that will remain visible. Clear markings on both faces reduce the repeated questions that a poorly marked board generates.

9. Always Decouple

A decoupling capacitor is the cheapest part in the design and the one most often omitted in the hope that the device data sheet limits will be met without it. Place one beside every integrated circuit, add a bulk capacitor at the supply entry point, and keep the connections short. Consolidate the values used, as rule six suggests, so the parts list stays short while the coverage stays complete.

10. Generate and Check the Fabrication Data

Produce the manufacturing parameters yourself and verify them before release. Most fabricators will happily check a data package, but a check performed by the designer with a viewer is faster and catches misunderstandings before they reach a quotation. Opening the artwork layer by layer has found more than one design with a missing outline, an omitted plane layer or a silkscreen block printed over a pad.

Why These Rules Survive

Reference designs are more widely shared than ever, and the result is that a great deal of layout is done by adapting something that already works. That makes the rules more valuable rather than less, because the reasoning behind a reference design is rarely written down with it, and a change made without understanding the reasoning is where the design breaks. The rules above are the reasoning in compressed form.

Panelization, the silkscreen conventions and the fabrication data check are also the points where the designer and the factory have to agree. A layout that follows rules five and ten is one that transfers without a round of questions, and a board whose silkscreen follows rule eight is one that assembles without an operator having to guess. Those are the parts of the order process where our engineering team spends the most time, and the rules that make the transfer short are the same rules that make the board quiet. For designs that are intended for a specific application, the same ten rules are applied with the constraints of that application on top, which is what our industrial PCBA and medical PCBA programmes do.

Where the Rules Pull Against Each Other

Two of the rules conflict often enough to be worth naming. Rule three asks for extensive planes while rule two asks for short paths, and on a dense board a plane that has to be interrupted for a route stops being a plane for the signal that wanted to cross it. The resolution is to give the plane priority and move the signal, because a signal can be routed around an obstacle while a return current cannot. The same tension appears between rule five and rule one: the panel size the manufacturer prefers may not suit the grid the components were placed on, and the panel usually wins, because its cost is visible on every order while the cost of a grid change is a one-off inconvenience.

FAQ

Are these rules in any order of importance? Not really. They are usually applied together, and the ones that matter most depend on the design: rule three dominates a high-current board and rule two dominates a high-speed one.

Does rule nine mean one capacitor per device? It means at least one per device, placed beside the supply pins, plus a bulk capacitor where the supply enters the board.

Why does rule ten matter if the fabricator checks the data? Because the fabricator checks it against what they were asked to build, not against what you intended.