SPC and Process Capability in PCB Fabrication

A PCB line runs dozens of chemical and mechanical processes, and each one drifts slowly. Plating thickness creeps with bath age, etch rate changes with chemistry, and drill diameters wear with tool life. SPC, or statistical process control, is the discipline that makes those drifts visible before they produce scrap. This article explains the charts, the capability indices, and the practical reactions that follow from them.

What SPC Actually Measures

Statistical process control separates variation into two categories. Common cause variation is the natural spread of a stable process, and it can only be reduced by changing the process itself. Special cause variation comes from an identifiable event: a worn bit, a mis-set temperature, a new lot of chemistry. A control chart is designed to distinguish the two, so that an operator reacts to a real change instead of adjusting a process that is behaving normally.

The distinction matters because unnecessary adjustment makes a process worse. Chasing every point on a chart increases variation, a phenomenon familiar to anyone who has seen a line tuned continuously by a well-meaning operator. SPC gives the line a defined band within which no action is required, and a set of rules that identify when the process has genuinely moved.

Control Charts and Their Signals

An X-bar and R chart plots the mean and the range of small samples taken at intervals. The centre line and the limits come from the process itself, calculated from the data rather than from the drawing tolerance. A point outside the control limits, a run of seven points on one side of the centre line, or a steadily rising trend all indicate a special cause, and each triggers a defined investigation.

Individual and moving range charts suit processes where a sample is expensive or destructive, such as a microsection or a peel test. Attribute charts track defect rates where the measurement is pass or fail, which is common for visual inspection. The choice of chart follows the measurement, and the limits should be recalculated whenever the process is deliberately improved, or the chart will keep signalling a shift that is actually the new normal.

Process Capability and What Cpk Means

Capability indices compare the natural spread of a stable process with the tolerance it must meet. Cp relates the tolerance width to six standard deviations, while Cpk accounts for how well the process is centred within the tolerance. A process with a Cpk of 1.0 produces about 0.3 percent out of tolerance, which is far too high for a controlled impedance or a plating thickness that must be held within a narrow band.

The target for most PCB processes is a Cpk above 1.33, which corresponds to roughly 63 parts per million outside the limits, and safety critical work may ask for 1.67 or better. A low Cpk has two possible causes: a wide spread, which is a process problem, or an off-centre mean, which is often a calibration or setpoint problem. The two are corrected in entirely different ways, which is why the index alone is not enough and the chart must be examined alongside it.

Applying SPC to Plating and Etching

Plating thickness is a good candidate for SPC because it is measured routinely and its variation has a direct effect on impedance and current capacity. Thickness measurements are taken at defined points on a coupon, plotted on a chart, and compared with the tolerance band. A gradual upward drift usually means the bath is ageing and the current density has been increased to compensate, while a sudden step suggests a rectifier or a contact problem.

Etch rate behaves similarly. Line width measurements on a coupon plotted over time reveal a drift that would otherwise be invisible until a lot failed. Because etch compensation is applied in CAM, a drifting etch rate also changes the relationship between the drawn width and the finished width, so the compensation table should be reviewed whenever the chart shows a shift rather than at a fixed calendar interval.

SPC in Assembly

Assembly processes lend themselves to SPC as well. Solder paste deposit volume from an inspection system produces a continuous measurement that can be charted by aperture and by board position. Placement offset is measured by the placement machine and can be charted by nozzle. Reflow peak temperature is logged for every board, and its variation across a panel is a process characteristic that responds to conveyor speed and airflow changes.

The value of charting these variables is that most assembly defects are multivariate. A tombstoning outbreak may require a paste volume that is slightly high and a placement offset that is slightly large at the same time. A chart for each variable shows whether either has moved, which turns a difficult root cause investigation into a comparison of two plots against their historical behaviour, as described in quality improvement practice.

Setting Limits and Reacting to Signals

Control limits must be calculated, not chosen. They come from twenty to thirty subgroups of data collected while the process is stable, and they are reviewed whenever a change is made. Specification limits come from the customer or the design, and they belong on the chart as reference lines so that the distance between the two is visible. A process whose control limits sit outside the specification limits is not capable, no matter how stable it is.

The reaction plan is what makes a chart useful. Each signal should have a documented first response: who is notified, what is checked first, and how the affected material is handled. Where a process is stopped, the material produced since the last good measurement must be quarantined and evaluated rather than shipped. Writing that plan before the chart is introduced prevents a debate during a production emergency.

Records, Traceability, and Audits

SPC records are evidence. A customer audit will ask not only for the current chart but for the reaction that followed a signal, and for proof that the corrective action was verified. Keeping the chart, the investigation note, and the verification together makes an audit straightforward and turns the data into something more useful than a compliance artefact.

The same records support continuous improvement. When a process improves, the chart shows the new distribution and the capability index is recalculated against the original baseline. Over a year, that history identifies which process changes actually reduced variation, and it gives the next improvement project a place to start.

Process Control and Verification

Reviewing the design before the data is released is far cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end. 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.

FAQ

Is SPC the same as inspection? No. Inspection finds defects in the material already produced, while SPC acts on the process that produced them. The two are complementary, and a line that relies only on inspection will always be reacting late.

What Cpk should a PCB process achieve? A common target is 1.33 for a controlled characteristic and 1.67 or better for safety critical dimensions. Capability should be evaluated on a stable process, otherwise the index reflects the drift rather than the capability.

How often should control limits be recalculated? After any deliberate process change, and periodically to confirm that the original baseline still describes the process. Recalculating after every out-of-control signal hides real shifts instead of correcting them.