PCB design basics: layout rules and prototyping routes

A printed circuit board is a laminate sandwich: a fibreglass substrate (almost always FR-4), copper foil etched into tracks, a solder mask that keeps molten solder where it belongs, and a silkscreen layer of printed labels. Designing one means deciding where the copper goes — and while the CAD tools have become free and genuinely good, the judgement about where the copper goes is what separates a board that works first time from one that needs bodge wires.

This guide covers the layer stack and the design flow, the layout rules that account for most first-revision failures, the honest prototyping ladder from breadboard to fabricated board, the through-hole versus surface-mount decision for hand assembly, and the mistakes that catch nearly every first-time designer — because they caught nearly every experienced one first.

What is a PCB actually made of?

The standard stack, bottom to top: FR-4 fibreglass for rigidity and insulation; copper foil, most commonly "1 oz" weight, which is a 35 µm layer — the tracks, pads and planes are what's left after the rest is etched away; solder mask, the green (or any colour) lacquer that stops solder bridging between pads; and silkscreen, the printed component outlines and labels that turn assembly from archaeology into following instructions. Simple boards use one or two copper layers; four layers buys dedicated power and ground planes, which quietly solves a whole class of noise problems for a modest cost premium.

How does the design flow work?

Three stages, each checking the last. The schematic captures what connects to what — pure logic, no geometry. The layout places footprints and routes copper to honour the schematic's netlist, which is where the electrical judgement lives. The output is a set of Gerber files — the universal manufacturing format describing each layer — plus a drill file, which any board house on earth can turn into physical boards. Free, professional-grade tools (KiCad being the established open-source choice) cover the whole flow, and modern fabrication is cheap enough that hobby and prototype quantities cost less than the components on them. The flow's great virtue is its paper trail: when a board misbehaves, you can ask whether the schematic was wrong, the layout betrayed it, or the assembly betrayed the layout.

Which layout rules actually matter?

A handful of rules prevent the majority of first-revision faults:

• Size tracks for their current. Signal tracks can be thin — 0.25 mm is comfortable everywhere — but power tracks need width: a conservative rule of thumb for 1 oz copper is around 1 mm of track width per amp on outer layers, which keeps temperature rise modest with margin to spare
• Decouple every supply pin. A 100 nF capacitor placed close to each IC's power pin — millimetres, not centimetres — supplies the fast current demands the power tracks can't. Missing or distant decoupling is the classic cause of circuits that work erratically and "fix themselves" when probed
• Respect the return path. Every signal's current returns to its source, and it wants to flow directly beneath the outgoing track. An unbroken ground plane gives it that path; a plane carved up by routing channels forces returns into long loops that radiate and pick up noise
• Separate noisy and sensitive. Keep switching supplies and motor drivers physically away from analogue inputs and amplifiers, and let their grounds meet at one point rather than sharing a path
• Think mechanically from the start. Connectors belong at edges, mounting holes belong in the first placement pass — retrofitting them into a routed board is misery — and standoffs need clearance around their holes
• Add test points before you need them. A labelled pad on every supply rail and key signal costs nothing at design time and pays for itself the first time the board doesn't boot

Breadboard, stripboard or etched board?

The prototyping ladder trades permanence against effort, and each rung has a legitimate job. A solderless breadboard proves a concept in minutes and falls apart if you look at it sternly — ideal for trying, wrong for keeping, and unreliable above modest frequencies. Stripboard (Veroboard) — the 0.1"-pitch perforated board with copper strips — makes a soldered, permanent one-off without any CAD at all, and remains the backbone of school electronics and quick workshop tools; the whole family lives under prototyping boards. Etching your own from copper clad board using the etching and developing chemistry produces a true custom PCB on the bench the same afternoon — a skill with a satisfying payoff and a place in education, though modern cheap fabrication has claimed most of its former territory. The fabricated board is the top rung: solder mask, silkscreen, plated holes and repeatability, for less than the etching chemicals once cost. The honest modern workflow is breadboard to prove it, fab to keep it — with stripboard still winning wherever one-off simplicity beats waiting for delivery.

Through-hole or surface mount?

For hand assembly, the answer is usually both. Surface-mount passives in the 0805 size are genuinely easy with ordinary soldering equipment — easier than through-hole once practised, with no legs to bend or trim — and they shrink boards usefully. Through-hole earns its place at the edges of the design: connectors and anything a human plugs, unplugs or leans on wants the mechanical strength of legs through the board, and headers and IC sockets keep expensive or experimental parts replaceable rather than committed. The parts to avoid by choice at the bench are the leadless packages designed for ovens rather than irons — specify hand-assembly-friendly footprints and the whole build stays pleasant.

What mistakes catch first-time designers?

The same ones, reliably, in roughly this order. The wrong footprint — a part whose symbol was right but whose physical land pattern belongs to a different package — is the champion, and the defence is ritual: print the layout 1:1 on paper and sit every component on it before ordering. Two minutes of paper checking has saved more board revisions than any design rule. Behind it come power tracks sized like signal tracks (see the 1 mm per amp rule), decoupling capacitors placed wherever there happened to be room, silkscreen printed over pads where it interferes with soldering, connectors labelled in the CAD and not on the board, and boards with no mounting holes at all — destined to float in their enclosure on a cushion of hope. None of these is a knowledge failure; they're checklist failures, which is why even short experience plus a printed checklist beats long experience without one.

Featured prototyping and PCB

Frequently asked questions

What is FR-4?

The standard PCB substrate: a flame-retardant glass-fibre and epoxy laminate. It's rigid, insulating, dimensionally stable and cheap, which is why virtually all general-purpose boards use it — alternatives exist mainly for flexibility, extreme temperatures or RF performance.

How wide should PCB tracks be?

Signals are comfortable at 0.25 mm. Power needs width for its current: a conservative rule for 1 oz outer-layer copper is about 1 mm per amp, which keeps heating modest. When in doubt, wider costs nothing — copper you didn't etch away is free.

What are Gerber files?

The universal manufacturing output of PCB design: one file per layer (copper, mask, silkscreen) plus a drill file, describing the board photographically rather than logically. Every fabrication house accepts them, which is what makes designs portable between manufacturers.

Is stripboard still worth using?

Yes — for permanent one-offs, education and anything needed today rather than after a delivery. It requires no CAD and no waiting. For anything to be built twice, or with surface-mount parts, a fabricated board wins.

Can I still etch PCBs at home?

Yes — copper clad board, resist and etchant produce a working board in an afternoon, and it remains a worthwhile skill and classroom exercise. For multi-layer boards, plated holes and solder mask, fabrication is now cheap enough to be the practical default.

What's the difference between 1 oz and 2 oz copper?

Copper thickness — 35 µm versus 70 µm. Heavier copper roughly halves track resistance and doubles current capacity for the same width, which suits power electronics; 1 oz remains the standard for general work and costs less to fabricate.

From idea to board

Breadboards to prove it, stripboard to keep it simple, copper clad and chemistry to make it yourself — and everything that mounts, connects and supports the finished board.