Common Problems in PCB Production and How to Solve Them

A board shop that is losing yield rarely has one problem. It has a symptom, a set of plausible explanations and a production schedule that makes it tempting to correct the most visible symptom rather than investigate the cause. The discipline that separates a stable line from a struggling one is the habit of treating every defect as a measurement, and of tracing it back to the process step that created it before changing any set point.

Symptoms and Causes Are Different Things

A short circuit between two traces is a symptom. The cause could be a resist sliver left in the gap during PCB production, a screen printing error, a plating nodule, or copper debris transferred from a previous panel. Changing the etching time in response to the short might fix that batch and create a new problem with line width on the next.

The same logic applies throughout the process. Delamination is a symptom of moisture, of a poor bond line or of thermal stress. A registration error is a symptom of a dimensional change in the laminate, of tooling wear or of an exposure problem. Naming the symptom correctly is the first step, and it is done by looking at the defect rather than at the yield report.

Reading a Scrap Pareto

Scrap data is most useful when it is grouped in more than one way. Grouping by defect type shows what is going wrong, grouping by panel position shows where, and grouping by shift or by machine shows when. Three views of the same data will often point at a cause that none of them reveals individually.

A defect cluster on one edge of a panel suggests a process with a gradient, such as spray etching or uneven lamination pressure. A defect that appears at the same point on every panel suggests tooling or artwork. A defect that appears only on the first panels after a changeover suggests that the line has not stabilised, and the fix is a process step rather than a chemistry adjustment.

Process Drift and Control Charts

Most PCB production problems are not sudden failures but slow drifts. Etchant concentration falls, developer strength changes, plating bath additives deplete, and each moves the process a little further from its centre. Without measurement, the drift is invisible until the defect rate crosses a threshold, at which point the cause is far from where it started.

Control charts turn that drift into a visible trend. Plotting a measurable parameter, such as etch rate, plating thickness or registration offset, against time shows whether the process is stable and how much it varies naturally. Two consecutive points outside the expected range are a signal to investigate before the scrap appears, which is the entire point of statistical process control.

Plating and Etching Problems

Plating defects usually trace back to the bath or to the current distribution. Thin plating in the centre of a panel suggests a current density problem, while nodules suggest particulate contamination or a filtration issue. Our notes on copper plating defects and prevention describe the common failure modes and the measurements that identify them.

Etching problems follow the same pattern. A width gradient across the panel points at spray or conveyor issues, while a sliver between fine features points at resist thickness or gap geometry. The chemistry and equipment interactions are described in our overview of PCB etching process control, and the important habit is to measure the result on a coupon rather than judge it by eye.

Lamination, Registration and Dimensional Change

Layer to layer registration depends on how much the material moves during lamination and on how accurately the tooling holds the layers in alignment. Material movement is a function of the resin system, the press cycle and the copper distribution, and it is one reason that an unbalanced stackup tends to register poorly. Our notes on the PCB lamination process describe how press parameters affect both bond quality and dimensional stability.

Where registration problems appear on a design that was previously stable, check what changed: a different laminate supplier, a different press recipe, a larger panel, or a denser copper pattern. Answering that question before touching the exposure unit usually saves a wasted shift of investigation.

Material and Storage Problems

Some of the most persistent problems come from the material store rather than the production line. Laminate and prepreg absorb moisture, and a panel that has been out of its packaging for too long will blister or delaminate during lamination or reflow. Baking schedules exist for a reason, and they should be driven by measured moisture content rather than by how long the material has been open.

Chemical storage matters as well. Etchant and plating baths age, additives separate, and a drum that has been open for months may no longer match its certificate. Rotating stock and recording batch numbers on the traveller gives the traceability needed when a problem later has to be attributed to a specific delivery.

Corrective Action and Verification

Once a cause is identified, the corrective action should be specific and testable. Changing a parameter without a prediction of what it will do is not a corrective action but a guess, and it makes the process history harder to interpret. State what will change, what measurement will confirm the change worked, and when that measurement will be taken.

Verification then closes the loop. Run the affected product, measure the result, and record it in the process history so the next person inherits the knowledge rather than rediscovering it. Yield problems recur mainly because the fix was never written down, and the same investigation is repeated by a different engineer six months later.

Operator Practice and Training

Not every cause is technical. Handling damage, incorrect panel loading, a stencil wiped with the wrong solvent or a bath topped up with the wrong additive all produce defects that look like process failures. Recording who performed each step, and reviewing that record when a defect cluster appears, often identifies a difference in practice that no parameter change would explain.

Training is more effective when it is built around the defects that actually occur in the shop rather than around a general syllabus. Showing an operator a magnified image of the defect their station produces, and the measurement that detects it, teaches more than a procedure that lists parameters without explaining what they control. The aim is that the person running the process notices a change before the quality department does.

FAQ

Where should a yield investigation start? With the defect itself, examined under magnification, and with the position of the affected feature on the panel. That evidence usually narrows the possible process steps to two or three, which is far faster than reviewing every parameter in the shop.

How do I know whether a problem is drift or a step change? Plot the relevant measurement over time. A gradual slope indicates drift that process control should have caught earlier, while a sudden jump indicates a change in material, equipment or operator practice. The two require different responses.

Should every defect trigger a full investigation? No. Isolated defects are recorded and monitored, while recurring ones, or those affecting a customer critical feature, justify the full treatment. The skill is in recognising the difference, and the scrap rate trend is usually the best guide.