Copper Trace Width Tolerance and Etching Control on PCBs

Trace width is one of the first dimensions a designer specifies and one of the last to be measured properly. The number on the drawing is a target, and what arrives on the board depends on the imaging, the etching, the plating and the measurement itself. Copper trace width tolerance is the difference between the two, and understanding where that difference comes from is what allows a design to be built repeatably rather than sorted afterwards.

What Determines the Finished Width

The trace is defined by the resist pattern and then formed by removing the copper that is not protected. The width of the resist, the thickness of the copper, the strength of the etchant and the time in the etch tank all contribute, and they interact rather than acting independently. A change in copper thickness alone changes the finished width.

Plating adds to the same chain. Where copper is plated before etching, the deposit sits on top of the trace as well as in the holes, and a thicker deposit has to be etched for longer, which removes more material from the sides. The result is that the same artwork produces different widths on different builds if the plating is not controlled.

Etch Factor and Sidewall Profile

Etching removes material downwards and sideways at the same time, so the finished cross section is a trapezoid rather than a rectangle. The etch factor describes the ratio between the depth removed and the sideways removal, and a low etch factor means a wide undercut and a top width that is narrower than the designer expected.

Where the sideways removal is severe the trace is described as underetch, and the profile becomes a narrow top over a wide base. The sidewall profile matters for high frequency work because the current travels near the surface and the shape of the conductor changes the impedance. It also matters for adhesion and for the formation of a fine line, because an over etched trace has a thin, fragile cross section that can lift during assembly. The relationship between the profile and the process is described in the guide to the PCB etching process.

Measurement Methods and Their Differences

Width can be measured optically from above, by sectioning the trace and examining the cross section, or by a profilometer that traces the surface. Each method reports a slightly different value: the optical measurement sees the top width, the section shows the profile and the average width, and the profilometer gives a shape without a single number.

Because the methods differ, the specification should state which one applies. A tolerance quoted without a method is an invitation to disagreement, and the disagreement usually appears at the end of the order when the panels have already been built. The measurement should be taken at defined positions on a defined coupon.

Design Tolerance and Current Carrying

The width tolerance is normally expressed as a percentage or as an absolute value, and both are used. A tolerance of plus or minus twenty percent on a hundred micrometre trace is twenty micrometres, while the same percentage on a five hundred micrometre trace is a hundred, so designs with a range of widths often need both forms.

The tolerance matters most where the trace carries current or controls impedance. A trace that is narrower than nominal has higher resistance and less current capacity, and the temperature rise at the reduced width may exceed the design assumption. The relationship between width, copper thickness and current is set out in the guide to current capacity.

Fine Lines and Process Capability

As the target width falls, the tolerance becomes harder to hold because the absolute variation does not shrink with it. A process that holds twenty micrometres of variation will hold a five percent tolerance on a four hundred micrometre line and fail it on a hundred micrometre line, even though nothing about the process has changed.

The practical approach is to ask the shop what it can hold before the design is fixed. Capability data from the actual process is the only reliable input, and it should be requested as a distribution rather than as a single minimum, because the tail of the distribution is what determines the yield on a design with millions of features.

Copper Thickness and Its Effect

Thicker copper is harder to etch to a fine line because the etchant has to remove more material and the sideways attack continues while it does. Heavy copper builds therefore need wider minimum features, and a design that pushes both the thickness and the line width at the same time will find the process window very narrow.

Where heavy copper and fine lines are both required, the usual solution is a different process sequence, such as pattern plating or a semi additive approach. The decision affects cost and lead time as well as capability, and it should be made with the fabricator rather than imposed on them after the design is complete.

Effect of the Surface Finish

The finish adds a thin layer on the copper but does not usually change the measured width by much. What it does change is the appearance under an optical system, because a shiny or a dark finish affects the contrast at the edge and can shift the apparent boundary by a micrometre or two.

Where the measurement is automated, the threshold has to be set for the specific finish, and it should be verified on a coupon that has been through the same process. A measurement recipe that was tuned on a bare copper panel may not be valid on a finished production board.

Verification and Records

Verification is normally done on a test coupon that is processed with the production panels, so that the coupon experiences the same plating and etching. Measuring the production panel directly is possible with an optical system, but the coupon remains the reference because it can be sectioned for the profile.

The results should be recorded with the lot, together with the copper thickness and the process parameters. That combination is what allows a width problem to be attributed to a cause, and it also provides the evidence needed to accept or reject a batch against the drawing. The same discipline of defining the criterion and recording the result is described in the guide to judging PCB quality.

Registration, Line Spacing and Yield

Width is only half of the story, because the space between two traces is what determines whether the board works. Etching widens the gap as it narrows the trace, so the same process that produces a thin trace also produces a generous space. The parameter that ties the two together is the conductor width and spacing capability of the process.

Layer to layer registration adds to the same problem. Where a trace on one layer has to line up with a pad or a via on another, a registration error moves the feature and consumes part of the allowance that the width tolerance had left. The design should therefore state the width and the spacing separately, and the shop should confirm that both can be held at the same time rather than one at the expense of the other.

FAQ

What trace width tolerance can a standard process hold? It depends on the copper thickness and on the target width, and the shop should be asked for its own capability data. A common starting point for standard production is plus or minus twenty percent on features above a hundred and fifty micrometres, tightening only where the design needs it.

Why is the measured width different from the artwork? Because etching removes material from the sides as well as the depth, and the amount removed depends on the copper thickness and the process. The difference is predictable within a process, which is why the process should be characterised rather than assumed.

Does the tolerance change with the finish? The physical width changes very little, but the apparent width measured optically can shift because the finish changes the contrast at the trace edge. The measurement recipe should be qualified for the finish being produced.