Managing PCB Design Rules Across Projects

Every layout is driven by a set of design rules, whether or not anyone has written them down. If the rule set is well built, the tool catches the mistakes as they are made and the layout proceeds without rework. If it is left at the defaults, the same mistakes are found later by the fabricator, at a point where the only options are a waiver or a new set of films.

This article looks at what a useful rule set contains, which rules actually prevent defects, and how to carry a proven set from one project to the next instead of rebuilding it from scratch every time.

What a Design Rule Set Actually Contains

A complete rule set covers four families. Electrical rules define the minimum clearance and the connectivity expectations. Routing rules define the allowed trace width, the via styles, the fanout strategy and the allowed routing angles. Plane rules define how copper connects to a net, which is where thermal relief settings live. Auxiliary rules cover placement constraints, manufacturing limits and high-speed requirements.

The important insight is that these are design intent, not tool settings. Writing down the intent first, in numbers that the fabricator can meet, turns a rule set into a document that can be reviewed and reused. Leaving the numbers at whatever the software shipped with means the intent is implicit and therefore unverifiable.

Electrical and Clearance Rules

Clearance is the rule that most often decides whether a design is manufacturable. Rather than one global value, the practical approach is a small number of classes: a default clearance for ordinary signals, a wider value for voltages above about thirty volts, a tighter value only where a fine-pitch device forces it, and a separate value between copper and the board edge.

The classes should be driven by the process, not by optimism. If the fabricator’s standard capability is six mils on inner layers and eight on outer, a rule set that allows four everywhere will pass the design rule check and fail the fabrication review. Setting the rule to the real limit, and using a tighter class only on the specific nets that need it, produces a check that means something.

Routing and Via Rules

Trace width rules should be expressed as a minimum and a preferred value per net class. A power class with a wider minimum protects the current capability, and a signal class with a narrower minimum keeps fine-pitch escape possible. Width is also a fabricator constraint, so the minimum in the rule set should never be smaller than the minimum the chosen process supports.

Via rules deserve the same treatment. Allowing a single default via size everywhere is convenient but wasteful, because it licenses a small via on a power net where a larger one would carry the current better. Defining two or three via classes, and assigning power nets to the larger one, removes a whole category of after-the-fact corrections. The design rules that govern vias and their annular rings are worth reading alongside the tool settings, because the two should agree.

Plane and Copper Rules

Plane rules control how a pad connects to a plane. Direct connection gives the lowest thermal resistance and the most difficult soldering; thermal relief makes hand soldering practical and slightly raises the impedance. The setting therefore depends on how the board will be assembled, and it belongs in the rule set rather than being decided pad by pad.

Copper pour rules are the related case. A pour that is set to the minimum width everywhere will fragment into slivers and leave islands that the tool connects to nothing. Setting a minimum pour width and a removal threshold for isolated islands keeps the plane continuous and keeps the design rule check quiet about fragments that are electrically meaningless.

Design Rule Check as a Release Gate

A design rule check is only useful if it is treated as a gate rather than as advice. The practical arrangement is to run it before every review, to resolve every violation or record it as an accepted deviation with a reason, and to re-run it after any late change. Violations that are left unresolved have a habit of being rediscovered by the fabricator, where the cost is an order of magnitude higher.

Two failure modes undermine the check. The first is a rule set so loose that nothing is ever flagged, which is common when the defaults are never examined. The second is a rule set so tight that it produces hundreds of warnings on a design that is correct, which teaches everyone to ignore the output. Both are avoided by tuning the rules to the process and to the design, and by keeping the exception list short.

Reusing a Proven Rule Set

The value of a good rule set is that it survives the project. Most CAD tools allow the whole set, or selected rules, to be exported to a file and imported into another design. That makes it possible to start a new layout from the accumulated experience of the previous one rather than from an empty dialog.

Reuse should be deliberate. Exporting the electrical, routing and plane rules together and importing them into the new project carries the intent in one step, and the imported set can then be adjusted for the specific design. Where only part of a set is relevant, importing individual rules keeps the new project from inheriting constraints that do not apply to it — a rule set copied wholesale from a fine-pitch board will fight a layout that has generous geometry.

Keeping Rules Aligned With the Fabricator

A rule set is a contract with the process, and it has to be updated when the process changes. Moving from a prototype house to a volume fabricator, or from one laminate to another, changes the achievable minimums. Exporting the rule file and sending it with the fabrication drawing makes the agreement explicit, so a discrepancy is found before the files are released.

The same discipline applies to the assembly side. Paste aperture ratios, courtyard clearances and component-to-component spacing are all expressed as rules in some tools, and those rules are what keep the package manufacturable at assembly rather than only at fabrication. Keeping fabrication and assembly constraints in one reviewed rule set is the simplest way to make sure that neither is forgotten and that the two do not contradict each other.

Rule Sets for High-Speed Nets

High-speed constraints are usually bolted on at the end of a project and cause the most friction when they are, because they arrive after the routing has already been started. Impedance targets, length matching tolerances, differential pair gaps and limits on the number of vias per net all belong in the rule set before the first net is routed, so that the tool guides the layout instead of objecting to it afterwards.

These rules work best as classes. One class for each interface group, each with its own impedance target, its own matching tolerance and its own reference layer, lets the router apply the right constraint without a manual instruction for every net. Where a matching tolerance exists, the rules for serpentine length matching describe how the adjustment should be placed so that the correction does not create a new problem.

FAQ

Should I start a new project from the tool defaults? No. Start from the exported rule set of the most similar previous project, then adjust. The defaults are permissive in some areas and restrictive in others, and neither matches a specific fabrication process.

How many clearance classes are reasonable? Three or four is usually enough: signals, higher-voltage nets, fine-pitch local exceptions and copper-to-edge. More classes than that tend to be applied inconsistently, and the inconsistency is harder to find than the original problem.

What should happen to an unresolved design rule check violation? It should be either fixed or recorded as an accepted deviation with the reason and the person who accepted it. A violation that is silently ignored tends to reappear during fabrication, when the choices have narrowed to a waiver or a re-spin.