What Should a PCBA Supplier Quality Agreement Cover Before Production?

Overview

A PCBA supplier quality agreement defines the responsibilities shared by an OEM and its EMS supplier before manufacturing begins. It should clarify the applicable quality requirements, authority to handle nonconforming assemblies, customer approval boundaries, required records, and quality escalation procedures. The agreement needs to reflect the actual manufacturing scope without duplicating product specifications, test instructions, or purchase-order terms.

 

The Quality Agreement Has to Match the Work Being Purchased

An assembly-only order and a turnkey EMS project don't carry the same responsibilities.

If the OEM supplies the bare boards and components, the manufacturer may be responsible for receiving, handling, assembly, and the inspections included in its scope. When the EMS provider also sources the materials, additional supplier-management and incoming-quality obligations may apply.

Functional testing, programming, conformal coating, and box build raise similar questions. A supplier shouldn't be assumed responsible for a test or acceptance activity that was never included in the agreed work.

Equally, an operation shouldn't disappear from the quality requirements simply because someone else performs it.

For repeat PCB assembly programs, the agreement needs to identify the products, manufacturing locations, operations, and outsourced activities it covers.

When Part of the Work Leaves the Factory

Consider an outsourced conformal coating operation.

If coating is part of the EMS supplier's contracted deliverable, the applicable customer requirements still need to reach the subcontractor. The agreement should establish how that work is controlled, which subcontracting arrangements require customer notification or approval, and who is responsible for obtaining inspection records or resolving a nonconformance.

The OEM shouldn't have to chase an unfamiliar subcontractor for answers about a service purchased through its EMS provider.

Nor should the supplier have to guess which customer requirements apply to the outside operation.

The actual contractual scope determines the responsibilities, but the route for obtaining evidence and making decisions needs to be clear.

Keep Product Requirements in Controlled Documents

A supplier quality agreement shouldn't become a second engineering package.

The BOM, fabrication data, and assembly drawings define the product. Manufacturing instructions and test specifications establish the applicable technical requirements. The purchase order covers the authorized commercial work.

The quality agreement connects those requirements to the people responsible for carrying them out.

A phrase such as "built to IPC standards" is not precise enough on its own. Where IPC-A-610 or J-STD-001 applies, the agreed standard, revision, product classification, and relevant customer-specific requirements need to be identifiable.

Those standards provide a technical basis for evaluating the assembly or manufacturing process. They don't automatically authorize an unapproved repair, a design modification, or acceptance of a product that falls outside its specified requirements.

Document control matters for the same reason.

If a released BOM and an assembly drawing disagree about a component, a document-precedence clause may help identify the governing requirement. It doesn't necessarily resolve a genuine engineering discrepancy.

Someone with the appropriate engineering authority still needs to clarify the intended configuration before the affected work proceeds.

For an upcoming PCBA build with unresolved drawing revisions, acceptance criteria, or customer-specific quality instructions, OEM teams can share the relevant technical documents with Shenzhen STHL Technology Co., Ltd. (STHL) at info@pcba-china.com for an initial technical discussion.

It's a more useful starting point than a general request to meet all customer quality expectations.

A Hold Tag Does Not Authorize a Shipment

A supplier needs the authority to identify suspect material, prevent unintended use, and investigate a problem. That doesn't mean everyone involved in the investigation can decide whether the finished product is acceptable.

The distinction becomes clear when an assembly needs corrective work.

Suppose an SMT component has been installed with the wrong polarity.

An approved rework instruction may allow the manufacturer to remove and replace the component, inspect the affected area, and complete the necessary verification. If the assembly meets the original requirements afterward, it may be released under the supplier's established authority.

Now change one detail: a PCB pad lifts during component removal.

The work is no longer necessarily covered by the routine component-replacement instruction. The proposed repair may involve additional requirements for restoring the damaged feature and verifying the resulting assembly.

Whether the repair is permitted, and whether customer authorization is required, depends on the applicable product requirements and approved procedures.

A technically possible repair is not automatically an authorized one.

A third situation is different again. The board remains outside specification, but the supplier proposes shipping it under a concession.

That's a request to accept a defined nonconformance, not a rework operation.

The authority to approve it must be established under the customer requirements or an agreed delegation. A quality engineer's assessment that the deviation is unlikely to affect performance is not, by itself, permission to ship.

The Decisions Are Separate, Even When the Same Team Handles Them

An EMS supplier may be authorized to release conforming assemblies through its normal quality procedures while still needing OEM approval for a particular repair or concession.

The reverse extreme isn't useful either. Requiring a customer signature before every routine containment action can delay the controls intended to protect the product.

There is another detail worth settling early: concessions need limits.

Approval of a deviation on one production lot shouldn't quietly become permission to accept the same condition on future builds.

Likewise, placing customer-owned material on hold doesn't necessarily give the factory authority to scrap it.

These decisions may involve the same people, but they aren't interchangeable.

Consigned Components: The Receiving Record Matters

Customer-supplied material creates a handoff between two organizations. It doesn't settle who is responsible for every defect found afterward.

Suppose a tray of customer-supplied ICs arrives with damaged packaging.

The receiving team may record the condition, identify the affected material, and place it on hold. The next step depends on the agreed incoming-control requirements and material disposition authority.

That situation is different from finding a damaged component after SMT assembly.

By then, the investigation may need to consider the incoming condition, material identification, storage and handling, applicable moisture-sensitivity precautions, and the manufacturing operations the component passed through.

When the damage was discovered is useful information. It isn't proof of when or how the damage occurred.

Incoming inspection has limits, too.

A receiving check may identify damaged packaging, incorrect markings, or an apparent specification mismatch. It cannot be assumed to detect every latent electrical defect, particularly when inspection is sampling-based or electrical testing isn't part of the receiving plan.

STHL's official incoming-material inspection video shows factory-side material checks before PCBA production. It provides a practical view of the receiving process, although the authority to accept, reject, or replace customer-supplied material still depends on the individual project requirements.

Material Ownership Doesn't Establish the Cause of a Defect

If a component was defective before it reached the factory, the responsibility analysis may differ from a case where handling or an assembly operation caused the damage.

An OEM shouldn't automatically bear the loss simply because it purchased the component.

Equally, the EMS provider shouldn't automatically be held responsible for a latent defect that the agreed receiving controls were not designed to detect.

The evidence and contractual terms matter.

A rejected populated board may also involve several kinds of cost: the bare PCB, installed components, additional testing, rework labor, replacement material, or scrapped assemblies.

Those costs don't necessarily arise from the same cause.

A useful quality agreement provides a route from the technical findings to the applicable commercial responsibilities. It doesn't need to reproduce a complete pricing or liability schedule.

The important point is that material can be controlled and investigated without turning every incident into a new negotiation over basic responsibility.

Know Which Records Will Still Be There When You Need Them

A shipment may pass outgoing inspection and still require investigation months later.

The original acceptance result is useful. It may not answer the next question.

A lot-level inspection report can establish that the specified checks were completed with acceptable results. It may not identify which component batch went onto a particular board or whether that assembly underwent rework.

Manufacturing traceability serves another purpose: reconstructing what happened to the affected product as it moved through production.

A manufacturing execution system (MES) may connect material, process, inspection, and test information. How much detail can be reconstructed depends on the traceability controls established for the project.

Not every PCBA needs serial-number-level traceability. And having a serial number doesn't automatically mean the factory can trace every component lot or process condition associated with that unit.

The buyer needs to know what the available records can actually establish.

A factory can generate and retain inspection data without routinely sending every report to the OEM. That may be entirely appropriate for the project.

One customer may require a certificate of conformity and a defined inspection summary with each shipment. Another may need serial-number-linked functional test results, measured values, first-article documentation, or other manufacturing records.

The level of detail makes a difference when something fails later.

A functional test record marked PASS documents a passing result against the criteria used for that test run. By itself, it may not reveal the actual measurements, test program revision, or configuration needed to investigate a later failure.

A more detailed record can support a deeper investigation, provided that information was captured and linked to the affected product.

Neither format should be assumed without reviewing the requirements.

Having AOI, X-ray, first-article inspection, or functional test equipment also doesn't mean every inspection image or report must accompany every shipment.

The documentation package needs to match the agreed inspection, traceability, and reporting scope.

STHL's quality control and inspection processes provide context for the manufacturing checks and quality activities that may be relevant to a PCBA project. The specific records generated, retained, or delivered should be confirmed against the customer's production and quality requirements.

Audit access deserves similar care.

Reviewing records for an affected lot is different from conducting a supplier quality-system audit. Neither automatically grants access to unrelated production information or other customers' records.

Access rights, confidentiality, notice requirements, and applicable obligations need to be addressed in the agreed arrangement.

For OEM teams preparing a new production program or reviewing the requirements for a repeat build, the next step is to submit the PCBA project requirements, including available engineering files, expected quantities, inspection and testing needs, and any existing supplier quality requirements.

It's easier to agree on the documentation before production than to discover during a field investigation that the information was never required or linked to the affected assemblies.

 

When a Customer Reports a Failure, Don't Wait for the Final 8D

A returned assembly may be the first indication of a wider manufacturing problem.

Or it may be an isolated failure with no connection to the EMS supplier's process.

The initial investigation needs to keep both possibilities open.

If other units share a component batch, manufacturing lot, or relevant process condition with the failed assembly, the supplier may need to identify and assess that population.

That doesn't mean every related unit is defective. It means the potential exposure needs to be understood while the cause is investigated.

Containment Comes Before a Completed Root-Cause Report

Containment deals with the immediate exposure: suspect work in progress, inventory awaiting shipment, or other potentially affected products.

Root-cause analysis addresses why the failure occurred.

Corrective action addresses the established cause. Effectiveness verification then checks whether the action achieved its intended result.

These are different stages of the response.

A customer shouldn't have to wait for a completed investigation before learning what containment measures have been taken.

At the same time, a supplier shouldn't present an early hypothesis as a confirmed root cause simply to complete a report.

An 8D process or another formal corrective-action method can be useful when the circumstances require it. The completed document, however, isn't automatically proof that the action was effective.

Depending on the defect, closure may require subsequent production results, verification of revised process controls, or another agreed form of evidence.

A repeated field failure, a potentially safety-related problem, and an isolated in-process defect won't necessarily require the same escalation or reporting arrangements.

The response expectations should reflect the product, the risk, and the obligations accepted by the parties.

The OEM May Have Evidence the Factory Cannot See

The EMS supplier may hold manufacturing records, material history, and inspection results.

The OEM may have the failed assembly, system-level symptoms, firmware details, operating conditions, or field history.

That information can be essential.

An intermittent electrical fault that appears only after the PCBA is installed in the finished equipment may be difficult to investigate using production records alone.

A quality agreement should identify how the parties exchange relevant evidence, who coordinates the investigation, and what is required before a significant issue can be considered closed.

A fast acknowledgment is useful. A defensible investigation and an effective corrective action require more.

The OEM Also Has Decisions to Make

A supplier quality agreement won't work if it assigns every obligation to the manufacturer.

The OEM needs to provide controlled product requirements and make decisions that it has reserved for itself.

That sounds obvious until a production lot is on hold and the only available customer contact is the buyer who issued the purchase order.

That buyer may be authorized to discuss price and delivery without having authority to approve a repair or a concession affecting the product specification.

The supplier needs to know who can make the technical decision and how to escalate the matter if that person is unavailable.

Being copied on an email doesn't establish approval authority.

Neither does silence.

Where affirmative customer approval is required, a delayed response shouldn't be treated as permission to proceed.

Some agreements expressly allow defined notification-only changes to take effect after a specified period without objection. That is different from assuming permission to ship nonconforming product or bypass an approval-controlled change.

The governing requirements determine the distinction.

Manufacturing change control also needs a clear approval interface. The quality agreement can identify who must be notified, who may approve the change, and which controlled procedure applies.

It doesn't need to reproduce every equipment replacement, process adjustment, component substitution, or factory-transfer scenario.

Likewise, it shouldn't replace the OEM's engineering change process.

Its role is to establish the authority and responsibilities behind those controls, not create a competing set of technical instructions.

The Final Test Is Whether the Agreement Works on an Ordinary Production Day

Not every project needs a lengthy standalone supplier quality agreement.

A one-off prototype may be adequately covered by a shorter, clearly documented set of requirements, depending on the product, its risks, and the applicable obligations.

A long-life industrial program, a regulated application, or a repeat-production relationship with detailed customer-specific requirements may need more formal controls.

Some OEMs already establish the necessary responsibilities through a master supply agreement and controlled quality attachments.

The document title matters less than whether the obligations are clear and fit the work being purchased.

Before production begins, it's worth reading the agreement against the decisions a manufacturing team might actually face.

An inspection flags a suspect board. Can the team identify who may hold it and what must happen before release?

A repair is proposed. Is the work covered by an approved procedure, or is customer authorization required?

A buyer requests the history of a shipped assembly. Will the available records identify the product and provide the information the buyer is entitled to receive?

A failure is reported from the field. Who coordinates containment, investigation, and corrective-action closure?

These aren't unusual demands to place on a quality agreement. They're ordinary decisions that become harder when the answers are left open.

A sound supplier quality arrangement gives the factory room to perform authorized work while preserving the customer's approval rights over matters outside that authority.

When a problem occurs, the teams should be discussing the evidence and the next technical action-not trying to work out who was supposed to make the decision in the first place.