Multilayer PCB Design Tolerances Explained

Every dimension on a fabrication drawing carries a tolerance, and every tolerance costs money to hold. The skill in a multilayer design is not to demand the tightest value everywhere, but to identify the few dimensions where accuracy actually matters and to leave the rest at the fabricator’s standard.

Where the Error Comes From

Several independent sources contribute to the final position of a feature. The artwork is imaged with its own tolerance, the material moves during lamination, and the drill has a positional tolerance of its own. The errors add rather than cancel, and the total is what the design must survive.

That is why registration is the central concern in a multilayer board. Layers that shift relative to each other reduce the annular ring on one side of a hole and increase it on the other, and the smallest ring on the panel is what decides whether the board passes.

Registration Tolerance

Registration describes how well the layers line up with each other and with the drilled holes. It is quoted as a figure the fabricator can hold, and it tightens as the layer count rises because there are more interfaces to align.

The design influence on registration is the pad geometry. A generous annular ring absorbs a larger registration error, while a tight ring demands a tighter process and a higher price. Where a design has only a few critical holes, keeping the rest generous and tightening only those is cheaper than tightening the whole board. The same interaction shapes the stackup, as described in our layer assignment article.

Drill Tolerance and Finished Hole Size

Drill tolerance covers both the diameter and the position. The finished hole is smaller than the drilled hole because plating adds copper to the wall, and the difference must be accounted for by the fabricator rather than by the designer.

Positional tolerance is the more consequential of the two for a dense board, because it combines with registration to place the hole relative to the pad. The figure a fabricator quotes usually includes both effects, which is why the requirement should be expressed in terms of the finished feature rather than the tooling.

Annular Ring and the Minimum Value

The annular ring is the copper remaining around the hole after all the errors have been accounted for. Its minimum value is a fabrication limit, and it is one of the figures a designer should know before drawing a padstack.

A ring that is too thin can crack during thermal cycling, breaking the connection to the inner layer. The failure may not appear at assembly at all, which makes it a reliability issue rather than a manufacturing one. Our component tolerance and reliability notes describe how that failure develops.

Impedance Tolerance

Impedance controlled traces carry a tolerance of their own, typically expressed as a percentage of the target. Holding it requires the trace width, the dielectric thickness and the dielectric constant to each stay within their own limits.

That means the stackup must be specified with the impedance requirement in mind, and the fabricator must be able to measure the result. Test coupons on the panel are how that verification is done without cutting a customer board, and the coupon should be designed for the impedances that matter rather than generically.

Tolerances That Are Usually Unnecessary

Some dimensions are routinely over specified. The absolute position of a mounting hole relative to the board edge is rarely critical, and neither is the thickness of the solder mask over a finished trace.

Specifying a tight tolerance where the function does not need it guarantees a higher price and, worse, may cause the fabricator to reject panels that would have worked. The useful discipline is to ask what happens if the dimension drifts, and to tighten only where the answer is that something breaks.

Documenting the Tolerances

Tolerances belong on the fabrication drawing, where they can be discussed at quotation. A number buried in the CAD file is not visible to the person pricing the work, and the first time it is noticed may be when the board fails inspection.

The drawing should state the tolerances that matter functionally, the standard to be followed for everything else, and the measurement method where the value is difficult to verify. That last point matters more than it appears: a tolerance that cannot be measured cannot be held.

Tolerances Introduced by Assembly

Fabrication tolerances are only part of the picture. Placement accuracy, stencil registration and reflow all introduce their own variation, and a component placed at the edge of its pad tolerance has less margin when the board itself has drifted.

Considering the two together is what produces a robust design. A fine pitch part with a generous pad and a well designed stencil tolerates fabrication variation, while the same part on a minimal pad does not. The cumulative picture is described in our component tolerance article.

Cost of Tightening a Tolerance

Every tightening has a price, and the price is not linear. Moving from a standard tolerance to a moderately tight one costs a little. Moving from moderate to very tight can double the cost or make the board unbuildable at any price.

That is why the requirement should be derived from function. A designer who can say what breaks when the dimension drifts has the information needed to set the tolerance, and a fabricator who receives that information can propose the cheapest process that meets it.

Verifying Tolerances in Production

A tolerance that is stated but never measured is a hope rather than a requirement. Verification is normally done on a sample using the coupons on the panel, with the results recorded against the panel and the build.

Where a value sits close to its limit, the record shows whether the process is drifting before the boards fail inspection. That early warning is what makes the difference between adjusting a process and scrapping a batch, and it is the reason a fabricator with good records can quote a tighter tolerance than one without.

Standard Values and Where to Find Them

Most fabricators publish a capability sheet that states the default tolerances for outline, hole position, hole diameter, registration, annular ring and impedance. Those figures are the starting point for any design, and using them removes the need to negotiate every dimension.

Where a design needs something outside the published range, the conversation should happen before the layout is complete rather than after. A tolerance that is impossible to hold is not a specification, and a design that depends on one has a problem that no amount of inspection will solve.

FAQ

What tolerance should be specified for a board outline? The fabricator standard is usually adequate unless the board fits a groove or a close fitting enclosure. Stating the functional requirement is more useful than stating a number.

How tight can impedance be held? It depends on the stackup and the process, and it is quoted as a percentage. Tighter control requires tighter material and process limits, which costs more.

Should every hole be tolerance controlled? No. Only those whose position or size affects assembly or function. The rest should be left at the standard so that the fabricator has room to work.