Solder Mask Opening Tolerance and Registration
Every solder mask opening is placed relative to copper that was patterned in a different process, and the two never align perfectly. The misalignment is small, usually tens of micrometres, but on a fine pitch board those tens of micrometres decide whether a pad is fully exposed or partly covered. Registration is therefore a specification, not a manufacturing detail.
What Registration Means
Registration is the positional agreement between the mask opening and the copper feature it exposes. It is expressed as a maximum displacement in any direction, and it is measured on the finished board rather than assumed from the artwork.
The displacement has two parts: a systematic offset, which shifts every opening on the panel in the same direction, and a random component, which varies from opening to opening. The systematic part can often be compensated during imaging, while the random part sets the practical limit on how small an expansion can be.
Expansion Rules
The expansion is the amount by which the mask opening exceeds the copper pad on each side. It exists to absorb registration error, and it should be large enough that the pad remains fully exposed under worst case misalignment.
The usual value is around 50 to 100 micrometres per side for general work, and it can be reduced where the process is capable and the pads are small. The rule that matters is that the expansion must be at least as large as the registration tolerance, or the exposure is not guaranteed. The pad constructions that this interacts with are described in our note on pad definition.
Effects on Fine Pitch Pads
On a fine pitch device the gap between adjacent pads may be little more than the expansion itself. A large expansion leaves no mask between the pads, so there is no dam to prevent bridging, while a small expansion risks covering part of a pad and reducing the joint area.
The resolution is usually to accept a mask defined pad, where the opening is deliberately smaller than the copper so that the mask always overlaps and the joint area is set by the opening. That trades joint area for predictability, and it is the standard answer at the finest pitches.
How Registration Is Achieved
The mask is imaged using targets that are referenced to the copper pattern, so the alignment depends on the targets being visible and unambiguous. Panels that have been distorted by lamination make exact alignment impossible, and the imaging equipment compensates by scaling and rotating locally.
Material behaviour also matters. A mask that shrinks during cure moves every opening, and the shrinkage depends on the cure schedule and on the copper density beneath. Uniform copper distribution across a panel reduces the variation, which is one of the practical benefits of copper balancing. The thickness aspects are covered in our article on solder mask thickness.
Measuring and Specifying
Registration is measured by comparing the position of mask edges to copper edges at a sample of locations across the panel, and reporting the maximum displacement. The measurement should cover the extremes of the panel, because registration error is rarely uniform across it.
The specification should state the maximum displacement, the sample size and the measurement method. It should also state the expansion, so that the two can be checked against each other rather than treated as independent numbers.
Interaction with Silkscreen and Finish
Silkscreen is registered to the mask, so a silkscreen feature placed on a mask opening boundary may end up partly on the copper and partly on the mask, which looks inconsistent and can affect inspection. Keeping silkscreen clear of mask boundaries removes the problem.
The finish also interacts, because it is applied to the exposed copper. An opening that shifts far enough to expose copper that was meant to be covered creates a place where the finish deposits on a surface that will later be soldered, with unpredictable results. Our notes on silkscreen rules cover the marking side of the same tolerance.
Working with the Fabricator
The practical question is what the process can hold rather than what the designer would prefer. A capable supplier will state a registration figure for the materials and the feature sizes involved, and the expansion should be set from that figure rather than from a template.
Where the design needs an expansion smaller than the stated capability, the answer is usually a change of construction or of pad geometry rather than pressure on the fabricator. The evidence-based approach used in PCB quality assessment applies to the discussion as much as to the finished board.
Checks Before Release
Where a process is at the edge of its capability, the margin should be bought deliberately rather than discovered during production. A record that identifies the operator, the date and the settings is worth more than a record that identifies only the result.
A measurement taken at the wrong point of the process describes the wrong thing, however carefully it is made. Where the supplier and the user both measure the same property, they should agree on the method before the first delivery.
A parameter that is set once and never re verified drifts, and the drift is usually discovered by a defect rather than by the record. The first article confirms that the setup matches the intent, and it is the cheapest point at which a wrong setup can still be corrected.
Points to Confirm at First Article
The sequence of operations is part of the specification, because a different order produces a different result from the same steps.
FAQ
Can registration be compensated entirely in the artwork? A systematic offset can, and it is often worth doing. The random component remains and it is what sets the minimum expansion.
Is a larger expansion always safer? No. Beyond a point it eliminates the mask dam between fine pitch pads, which trades a covered pad for a bridging risk.
How is registration verified on production? By sampling the panel and measuring mask edge positions, usually on a coupon or on a dedicated registration pattern.
