PCB Process Audit Checklist for Board Shops

A process audit checks whether the shop is doing what it says it does, at the moment the work is happening, rather than whether the paperwork exists. That distinction is what makes the exercise useful. A single page audit checklist applied on the floor every month will find more real problems than a thick procedure manual that is reviewed once a year, because it sees the setup sheet that was signed before the parameters were set and the bath that was logged from memory at the end of a shift.

What the Audit Is For

The audit exists to confirm that manufacturing process control is real rather than documented. Every shop has a specification for plating thickness, a rule for bake times and a standard for stencil cleaning, and the question the audit answers is whether those standards are being followed on this shift, on this machine, with this operator. The answer is useful even when it is uncomfortable, and it is far cheaper to obtain than the defect that would otherwise reveal the same fact.

The second purpose is trend. A single finding is an incident, while the same finding in three consecutive audits is a system that does not work as written. Recording findings in a consistent form is what allows that pattern to appear, and it is the reason a short standard checklist beats a long bespoke one, since the same questions must be asked in the same way every time to be comparable.

Building the Checklist

An audit checklist should be short enough to complete in one shift and specific enough that two auditors would answer the same way. Items should be written as observations rather than as questions about compliance, so that the auditor records what is present and not what the operator believes. Between ten and twenty items is usually enough to cover the critical steps of a board shop without turning the audit into a project.

Each item needs a defined evidence source. Plating thickness is checked against a coupon measurement, bake time against a log with a timestamp, drill parameters against the machine program, and stencil cleaning against the schedule and the cleaning record. Where the evidence does not exist, that absence is itself a finding, and it is usually the most valuable one because it points at a control that is assumed rather than performed.

Auditing the Front End

The front end of the flow is where most audit value sits, because errors there multiply downstream. The auditor should confirm that the data was verified against the drawing, that the stackup matches the order, that the drill program corresponds to the current revision, and that the material issued matches the specification. A missing revision control on artwork is a small finding that can turn into a whole batch of boards built to an obsolete design.

Tooling control belongs in the same section. The auditor checks that the tools are identified, stored and retrievable, and that the tool used for the current job is the one the job requires. Shops that audit this rarely find a wrong tool in use, but they frequently find that tools have no index and no owner, which is the condition that produces the wrong tool eventually.

Auditing the Process Steps

The process steps are audited by watching the work rather than by reading the record afterwards. The auditor observes the setup, compares the parameters with the specification, and checks the first article measurement before the lot proceeds. A parameter that is set from memory, a first article that was measured on a previous job, or a bath that was analysed on a different shift are all findings, and none of them would appear in a review of the paperwork.

Measurement practice deserves its own attention, because it is where good processes are often misreported. The auditor should confirm that the gauge is the one specified, that it has a current calibration status, and that the measurement is taken at the position the specification requires. A plating thickness measured on a different part of the coupon than the specification intends produces a number that looks correct and means nothing.

Recording Audit Findings

Audit findings should be recorded as an observation, a requirement and a location, in that order, without a conclusion attached. Writing the finding as a judgement invites an argument about the judgement, while writing it as an observation invites a discussion about the process. The record should also carry the date, the step, the machine and the person who was present, since all four are needed if the finding is later linked to a defect or a customer complaint.

Findings should be graded so that the response can be proportionate. A finding that creates an immediate safety or product risk should stop the work, while a documentation gap can be corrected within a week. A shop that treats every finding with the same urgency will exhaust its own attention within a month and the audit will quietly stop being performed, which is the commonest way an internal audit programme dies.

Corrective Action and Closure

Corrective action has to address the cause rather than the instance. Replacing a missing log sheet fixes the finding and leaves the cause untouched, while moving the log to the point of use and requiring a signature before the batch moves addresses the cause. The difference between the two responses is what separates an audit programme that improves a shop from one that produces a growing file of repeated findings.

Closure needs verification, and verification needs a date. The corrective action should be checked at the next audit, and the check should be recorded against the original finding so that the history is visible. Findings that were closed and then reappeared are the most informative records in the file, because they show where the shop’s process control is weakest and where a permanent change is still required.

Practical Rules

Keep the checklist short, audit on the floor while the work is running, and record findings as observations with evidence. Grade them, assign an owner with a date, and verify closure at the next audit against the quality records and the fabrication notes for the batches involved.

Audit the steps that carry the most risk rather than every step equally, and rotate the focus so that the whole flow is covered over a few months. Review the process flow once a year against what actually happens on the floor, because a process that has quietly changed shape is a set of controls that no longer matches the work they were written for. A control that has drifted away from the work is worse than no control at all, since it is still followed and it still consumes attention while protecting nothing.

FAQ

How long should a process audit take? One shift is usually enough when the checklist is short and focused on the highest risk steps. A longer audit is rarely more thorough, only slower.

Who should perform the audit? Someone who knows the process but does not own the step being audited. Independence matters more than seniority.

What is the most common finding? A control that exists on paper but is not performed in practice, such as a log completed from memory or a first article measured on an earlier job.