Statistical Process Control PCB Guide

Statistical process control is the practice of measuring a process while it runs and reacting to the result before the product is out of specification. It is not a report, and it is not a one off study, both of which are common misunderstandings. The method works because it separates the ordinary variation of a process from the variation that signals a change, and it tells the operator which one is present.

What the Method Is

The method plots a measurement against time on a chart with a centre line and control limits calculated from the process itself.

The limits are not the specification limits. They describe what the process normally does, so a point outside them means the process has changed rather than the product has failed.

The rule that follows is simple: an out of control point is investigated even if the part is inside the tolerance, because the next part may not be.

The method is therefore preventive, while inspection is detective. Both are needed, and a shop that relies only on inspection is paying for the defects it finds.

Common and Special Causes

Common cause variation is the ordinary noise of a stable process. It comes from many small sources and it can only be reduced by changing the process itself.

Special cause variation comes from an identifiable event, such as a new drill, a change of bath chemistry or an operator who set a machine differently. It shows as a pattern on the chart.

Reacting to common cause variation by adjusting the process makes it worse, because each adjustment adds variation. That behaviour is called tampering and it is the most common way a control programme fails.

The distinction is the whole value of the method, since the response to the two kinds of variation is opposite.

<img src="https://www.gopcba.com/wp-content/uploads/2024/09/202007291740128037.jpg" alt="Control chart displayed on a factory screen” />

Choosing the Chart

A chart of a continuous measurement, such as a hole diameter or a plating thickness, plots the individual values and the moving range.

A chart of a count, such as the number of defects on a panel, uses a different calculation and a different set of limits. The two are not interchangeable.

The sample size and the frequency are chosen so that a change is visible within the time that it takes to produce a batch. A chart that updates once a shift is useless for a process that drifts within an hour.

The measurement system has to be good enough for the chart to be meaningful, since a gauge that is noisy produces a chart of its own variation rather than of the process.

Limits and Reaction Rules

The control limits are calculated from the data, usually as the mean plus and minus three standard deviations of the plotted statistic.

A run of points on one side of the centre line, a trend and a repeating pattern are all signals even when every point is inside the limits. Those rules are written into the standard and they should be applied literally.

The reaction should be defined in advance, so that the operator knows what to do at a signal rather than deciding in the moment. A rule that is not written is a rule that is not followed on the night shift.

The reaction is usually to stop, check the last known good point and look for the cause, and the record of the investigation is as important as the chart.

Sampling Plans

The sampling plan defines how many parts are measured, how often and from where. It has to cover the variation between machines, shifts and batches rather than a single convenient position.

A rational subgroup is a set of samples taken close together in time and under the same conditions, so that the variation within the subgroup reflects only common causes.

A panel is a natural subgroup in a PCB shop, and a coupon from a panel is the natural sample. The plan should say which coupon and at what point in the process it is measured.

The plan should also define what happens to the material produced between the last good sample and the signal, since that is the material that may be affected.

Using It in a PCB Shop

The dimensions that suit the method are those that are measured routinely and that have a tolerance: hole diameter, plating thickness, copper weight, board thickness and outline.

The plating thickness is the classic case, because the bath drifts and the measurement is already taken for the process. The chart turns a periodic measurement into a control.

The etching process is a second case, since the trace width drifts with the bath and the chart shows the drift before the impedance leaves its tolerance.

The plating specification and the quality criteria give the limits that the chart is compared against.

Practical Rules

Confirm that the process is stable before setting limits, and never adjust a process in response to common cause variation.

Write the reaction rules down, including the action to take and the person to tell, so that the response does not depend on experience.

Choose a sampling plan that covers machines and shifts, and take the sample from a defined position on the panel.

Keep the chart and the investigation record together, and record the process data with the fabrication records for the batch.

Process Control and Verification

On a design of this kind, control chart is the item that decides how the rest of the board is arranged. The process window is set by the narrowest step in the flow, so an improvement anywhere else shows up as margin rather than as yield until that step is addressed. A short note on the drawing about handling, storage or packaging is often worth more than an extra decimal place on a tolerance.

Where a value sits close to a process limit, the drawing should say so, since the shop can then open the process window rather than working to a nominal figure that carries no tolerance. A stack-up that is drawn rather than described removes most of the ambiguity from a quotation, and it lets the fabricator price the board against the dielectric and copper weights that will actually be used.

Process Control and Verification

On a design of this kind, control chart is the item that decides how the rest of the board is arranged. The process window is set by the narrowest step in the flow, so an improvement anywhere else shows up as margin rather than as yield until that step is addressed. A short note on the drawing about handling, storage or packaging is often worth more than an extra decimal place on a tolerance.

FAQ

Are control limits the same as specification limits? No. Control limits describe what the process normally does, while specification limits describe what the product needs. A process can be in control and still be incapable.

What is tampering? Adjusting a stable process in response to ordinary variation. Each adjustment adds variation instead of removing it, and the process gets worse.

How often should samples be taken? Often enough that a drift is visible before the affected material is shipped. The frequency should come from the speed of the process rather than from habit.