Validate a PCB Sorting Screen Against the Defect It Must Detect

A temporary sorting operation is useful only if its screen can distinguish the defect of concern. Checking every board with an insensitive method creates complete-looking records without effective detection. Validate the screen’s discrimination before using a passed sort as evidence for release.

Minitab’s attribute agreement explanation distinguishes agreement among evaluators from agreement with a known reference. Inspectors can consistently reach the same wrong decision, so a screen needs a justified reference condition as well as repeatability.

Define the detectable condition

Describe the defect and its observable signature under the proposed inspection. If the issue is intermittent, hidden, or dependent on operating conditions, state how the screen exposes it. Do not assume that a convenient visual check can detect a failure mechanism that produces no visible indication.

Select representative challenge samples with known condition where possible. Include difficult boundary cases and acceptable examples that resemble the defect. Record how each reference condition was established so that the validation is not circular. Keep challenge specimens identifiable and monitor their condition, since repeated handling or testing can change the feature that originally made them useful.

Measure both kinds of mistake

Keep individual results rather than reporting only overall accuracy. A sample set dominated by clearly acceptable boards can produce a high accuracy percentage while the screen misses the rare defective examples that matter most.

Validate the production implementation

1. Freeze the screen instructions and settings.
2. Run representative challenges through the actual process.
3. Review misses, false rejects, and repeatability.
4. Resolve weaknesses or state the remaining limitation.
5. Monitor the screen while sorting the suspect population.

Include ordinary station conditions such as lighting, access, fixture loading, and operator handoff when they affect the method. A laboratory demonstration performed by the method’s author may not represent a sustained production sort.

Keep sorted material and results traceable. Record the screen revision, inspector or station, date, and population identity. If the method changes after sorting begins, determine whether earlier passed material remains covered by the new evidence.

A validated screen still has a defined detection scope. It does not prove the absence of every possible defect on the board. The release decision should state which condition was screened, the evidence supporting that method, and any additional requirements that remain outside it.