Overview
The supplier says the SMT build is on hold. Before the OEM asks when production will restart, one thing needs clearing up: what exactly has been stopped?
An incorrect component reel found during kitting is not the same as a wrong part discovered after reflow. One may affect the next board to be assembled; the other could already be on boards waiting for test.
A PCBA supplier should stop affected work when it can no longer demonstrate that the material, process, or assemblies meet the approved requirements. It should seek OEM approval when the proposed way forward changes a customer-controlled requirement or involves accepting nonconforming product beyond its authorized disposition limits.
Neither situation automatically calls for shutting down unrelated manufacturing work.
The difficult part is establishing which boards are affected, what the supplier is authorized to do, and what evidence will be enough to start building again.
What Does "The Build Is on Hold" Actually Cover?
A stop-build notice should tell the OEM more than which SMT line has stopped.
It should identify the part number and revision, the affected work order or material, the operation being held, and whether any potentially affected assemblies have already moved downstream.
Those details change the decision.
If the specified component is unavailable, the immediate problem may be an incomplete material kit. If the wrong manufacturer part number has already been loaded into a feeder, the supplier needs to determine which boards could have received it.
And if the same material lot was used on another work order, holding only the job where the discrepancy was discovered may leave part of the problem untouched.
On a PCB assembly floor, one machine can process several products without those products sharing the same risk. The reverse is also true: separate work orders may share a suspect material batch, manufacturing setup, or inspection process.
A local hold is appropriate when the supplier can establish a reliable boundary around the affected work. A broader stop may be necessary when the risk extends across shared operations or cannot yet be contained within a smaller scope.
The aim is to avoid creating more suspect assemblies without unnecessarily interrupting work that remains demonstrably conforming.

Work Backward to the Last Check That Could Have Caught the Problem
The board that triggered an alarm may not be the first affected board.
Consider a feeder-loading discrepancy. A useful investigation asks when the reel was installed, what the earlier material checks actually verified, and which boards passed through the affected placement operation.
Simply finding the most recent PASS in the manufacturing records is not enough.
An AOI program may confirm component presence and orientation without being able to distinguish between two visually similar manufacturer part numbers. If that is the problem, earlier AOI passes do not establish that the correct part was fitted.
The relevant checkpoint might instead be a verified material-identification record or feeder setup check.
A printing-process issue calls for different evidence. The team may need to examine the printer setup, solder paste inspection results, equipment adjustments, and the last point where the process was confirmed to be operating within its approved limits.
And when a functional test fixture becomes unreliable, the question may reach back to its last valid verification, not merely the first board that produced an unexpected result.
The principle is straightforward: a previous check only helps define the affected population if it could have detected the condition now under investigation.
When the Records Cannot Narrow the Scope
Traceability depends on the product, the agreed controls, and the records actually maintained. Not every PCBA project has individual board-level identification at every manufacturing operation.
If the supplier cannot reliably exclude earlier boards, the initial hold may need to cover a wider population.
That does not mean every board in the hold is defective. It means the available evidence has not yet established which ones are unaffected.
The supplier can narrow the population later when suitable records or inspection results support that decision.
An OEM should be wary of a precise affected quantity when the supplier cannot explain how it was established.
The useful question is not "How many boards are bad?" before the investigation has answered it. It is "Which boards could be affected, and what evidence supports the boundary?"
The Hold Must Follow the Boards Through WIP and Inventory
Identifying suspect boards is only useful if the factory can keep them from moving forward unintentionally.
On the floor, that may involve clearly identified ESD-safe storage and physical segregation. In the manufacturing or inventory system, the corresponding work orders or product records may need a hold status or another effective restriction.
Both sides matter.
A batch can be placed on a designated hold rack while still appearing as available stock to production planning. Equally, an electronic hold record is of limited value if the boards cannot be distinguished from released material at the next workstation.
The exact method depends on the factory's systems. What matters is maintaining product identity and preventing unintended processing, allocation, or shipment.
There is a related point that often gets overlooked: the hold has to remain understandable after the people who discovered the problem leave the shift.
The next operator should be able to tell what is restricted, which operation is affected, and who is authorized to change that status.
A note saying "QC checking" is unlikely to be enough if it does not identify the product or explain what movement is prohibited.
When Does the OEM Need to Be Involved?
The answer depends on what the supplier is proposing, not just how disruptive the issue appears.
An EMS supplier should be able to take appropriate immediate measures to contain suspected nonconforming work. Waiting for a purchasing contact to respond while the same questionable setup continues producing boards defeats the purpose of containment.
The next step is where authority matters.
Correcting the Process Back to the Released Requirements
An incorrect feeder setup may be corrected using the approved BOM and manufacturing instructions. A malfunctioning test fixture may be restored to its specified configuration. A soldering defect may be reworked using an authorized method.
Where these actions bring the process or product back into conformity, the supplier may have authority to carry them out internally, subject to the agreed procedures and any customer-specific restrictions.
That qualification matters. Particular products may restrict component replacement, rework methods, repeated thermal exposure, or other operations that would otherwise be routine.
Rework and repair should not be treated as interchangeable, either.
Rework is intended to return a nonconforming product to its specified requirements. Repair may make a product acceptable for its intended use without restoring every original characteristic, which can change the approval required.
The fact that an engineer knows how to perform a repair does not mean the supplier has authority to accept the resulting condition.

An Internal Fix Can Still Require Customer Notification
Suppose a suspect component batch has been contained and the factory is investigating whether it was used in other orders.
The supplier may be authorized to hold the material and carry out an approved correction. Yet the OEM may still need to be notified if delivery is affected or if assemblies already shipped could share the same exposure.
Notification and technical approval serve different purposes.
A purchasing representative might agree to a revised delivery date without having authority to approve an alternative component or a change to the product specification.
Conversely, the supplier may be required to notify the OEM about an incident even when no customer approval is needed for its immediate internal response.
The governing agreement and applicable product requirements determine those obligations.
When the Proposed Fix Is No Longer the Approved Build
This is the clearer approval boundary.
A substitute component can be pin-compatible and still be unsuitable for the OEM's application. Electrical ratings, temperature behavior, tolerance, reliability, qualification status, and other design requirements may need evaluation.
Before using an alternative, the supplier needs to establish whether the part is already approved and whether that approval covers the intended configuration.
The same concern arises if the proposed disposition changes a customer-controlled test limit, removes required inspection coverage, or asks the OEM to accept assemblies that do not fully meet specification.
These decisions should follow the applicable technical requirements and designated approval authority. For automotive programs subject to IATF 16949 and customer-specific rules, additional concession or deviation requirements may apply.
The broader rules for changing approved manufacturing arrangements are discussed in which PCBA manufacturing changes require buyer notification.
For a build already on hold, the immediate question is narrower: can the supplier restore the approved condition, or does the proposed disposition require the OEM to authorize a departure?
An Inspection Alarm Still Needs an Engineering Judgment
A rejected inspection result does not always establish a product defect. But dismissing a repeated alarm as an inspection nuisance can be just as risky.
Automated optical inspection (AOI) can flag missing components, placement offsets, polarity concerns, and visible soldering conditions within the capabilities of its configured program. The result may require review before the actual condition is established.
Solder paste inspection (SPI) provides information earlier in the process. A shift in paste volume, area, height, or position may indicate a printing issue before the boards reach reflow.
Neither checkpoint should be interpreted beyond what it can reliably measure.
Shenzhen STHL Technology Co., Ltd. (STHL) describes SPI, first article verification, AOI, and functional testing among its published quality control processes. The methods and acceptance criteria for an individual project depend on the agreed manufacturing and inspection plan.
What matters in a stop-build decision is whether the required controls remain effective for the risk being investigated.
More Inspection Is Not Always a Substitute for Process Control
If solder paste measurements begin moving outside an approved process window after a printer adjustment, continuing to feed more boards into placement and reflow can increase the amount of WIP requiring evaluation.
The supplier may need to correct the printing conditions and verify the result before continuing.
But what about boards printed before the adjustment was corrected? Their disposition remains open until the relevant evidence supports a decision.
Adding downstream AOI checks does not automatically compensate for an uncontrolled printing process. The proposed control must be suitable for the failure mode and authorized where necessary.
The reverse situation also exists.
An AOI program that produces excessive false calls may need adjustment without any change to the product's acceptance criteria. If the required inspection coverage is preserved and verified, that adjustment may remain within the supplier's internal process authority.
The number of inspection rejects alone does not determine whether an OEM deviation is required.
When the Test Setup Is the Suspect
Functional test failures can come from the board, the fixture, test software, wiring, contact resistance, or the test procedure.
If the test setup is unreliable, running more boards through it may generate more data without answering whether the assemblies meet their requirements.
The affected test and product-acceptance activities may need to stay on hold while the setup is checked.
An approved backup fixture may provide a way forward. So might an alternative method that has been verified to cover the required functions and authorized under the project arrangements.
A simpler test is not necessarily an equivalent one.
Upstream assembly may sometimes continue while testing is unavailable, provided that work remains authorized and the resulting WIP is clearly identified and controlled. Those boards cannot be treated as accepted finished goods until the required acceptance activities are complete.
What Should Be in the Supplier's Request for Approval?
A vague message saying production is stopped and the customer must approve urgently leaves the OEM with too much to reconstruct.
The supplier should identify the released requirement, what was actually found, the affected product identifiers, and the proposed disposition.
An OEM may also need to know whether the request concerns future production, boards already built, or both.
Those are not necessarily the same approval.
The technical basis could involve component identification, comparison with the approved BOM, inspection photographs, process records, test results, or an engineering assessment. Which evidence matters depends on the discrepancy.
A schedule impact belongs in the communication, but it cannot replace technical justification.
And the approval must come from the function authorized to make that decision. A purchasing acknowledgment is not necessarily engineering sign-off.
Where relevant, the authorization should state the applicable part number, revision, work order, material lot, quantity or serial range, together with any verification or time limits.
A temporary approval for one build should not quietly become the default configuration for the next purchase order.
For an existing project with STHL, technical information and affected-product details can be discussed through the established project contact or at info@pcba-china.com. Confidential project files should follow the agreed exchange and approval process.
Approval Conditions Are Easier to Establish Before the First Build
For a separate new manufacturing project, the OEM and EMS supplier can address hold points, deviation authority, inspection requirements, and escalation contacts before production starts.
That preparation is especially useful when a project involves customer-controlled components, restricted manufacturing processes, or additional quality requirements.
These conditions can be included with the BOM, Gerber files, assembly drawings, expected quantities, and testing requirements when an OEM is ready to request a PCB assembly quotation.
This is a separate project-planning step, not a substitute for resolving an existing production hold or a supplier's contractual corrective-action obligations.

The Line Can Restart While Earlier Boards Remain on Hold
This is the distinction that matters most when a production delay is being resolved.
Suppose the supplier identifies an incorrect feeder setup, loads the approved material, verifies the setup, and confirms that subsequent placement can proceed under the released configuration.
The affected operation may now be ready to restart.
Earlier boards produced while the incorrect component could have been installed still need their own evaluation and disposition.
The same applies after a printing-process correction or test-fixture repair. Restoring the process does not retrospectively prove that assemblies made under the questionable condition were conforming.
Nor does an OEM's limited concession authorize every board associated with the event.
The approved disposition must be applied to the products and conditions it actually covers.
How Much Verification Is Enough to Restart?
There is no universal restart package for every PCBA production hold.
A feeder correction may call for material identification, setup verification, and appropriate checks of the initial output.
A printing-process adjustment may require evidence that the process is back within its approved conditions.
A repaired test fixture needs verification that it can again perform the required tests.
Additional first article inspection, AOI, X-ray, or functional testing may be appropriate depending on the risk and the agreed plan. Requiring every method for every interruption adds paperwork without necessarily improving the decision.
The evidence should establish that the specific condition that caused the hold is now adequately controlled.
An investigation may remain open after manufacturing resumes if effective containment, appropriate verification, and the governing authorizations support continued work.
That is not a shortcut around root-cause analysis. Where the remaining uncertainty cannot be reliably contained or detected, restarting on the strength of a few passing boards may be unjustified.
Some work may also continue while other operations remain restricted. A missing through-hole connector, for example, might delay final assembly without preventing authorized upstream SMT work, provided the incomplete boards can be properly identified and handled.
The point is to keep each decision tied to its actual technical dependencies.
The Next Shift Should Not Have to Guess
A production hold is only fully useful when the decision survives the shift change.
The next team should know which work orders or boards remain restricted, what was corrected, who authorized the disposition, and which checks supported the restart.
The OEM should be able to understand the same record without reconstructing it from scattered emails.
Some interruptions will be resolved through routine controls. Others will require customer authorization, further investigation, or longer-term corrective action. Not every stop-build event calls for a full 8D report.
But an affected board should never become accepted product simply because the line has started running again.
A capable EMS supplier will occasionally have to stop work. What matters is how confidently it can explain the reason, identify the affected boards, keep them under control, and make the next manufacturing decision on evidence rather than assumptions.
That is the difference between a useful production hold and a problem that has merely been postponed.

