Quality & acceptance
AOI vs X-Ray vs ICT vs FCT: PCB Assembly Test Methods
Choose inspection and test methods by the failure you need to detect. Compare visible features, hidden joints, electrical networks and operating functions without treating one method as complete coverage.
Separate bare-board, assembly and product responsibilities
Bare-board fabrication creates the conductive and insulating structure. Assembly adds components and solder joints. The finished product then operates with firmware, connectors, an enclosure, and an intended environment. A test at one stage cannot automatically establish correctness at the others. Keep the stage visible in the test plan so a statement such as electrically tested is not mistaken for a powered product test.
For a bare board, continuity and isolation checks can compare the manufactured interconnect with the relevant reference data. They do not demonstrate that the schematic is functionally correct or that a populated product will meet its performance requirements. Other fabrication evaluations may address construction or material characteristics, but their scope and sampling need to be specified separately.
For a populated assembly, inspection can detect selected workmanship conditions and electrical tests can evaluate defined circuit properties. Product validation may need operating modes, loads, environmental conditions, and software behavior that are outside the assembly supplier's routine process. Identify who owns each part of that work. An unclear boundary can leave a critical requirement untested while both teams assume the other handled it.
Use a matrix with stage, risk, method, acceptance criterion, and retained evidence. Add a decision owner for failures and exceptions. This makes the plan reviewable before boards are built. It also gives purchasing a way to compare quotations for the same deliverable rather than treating every use of the word test as equivalent.
- Identify risksWhat could fail?
- Map coverageWhich method detects it?
- Provide accessPads + firmware + fixture
- Approve limitsResults + retest rules
What each method can and cannot tell you
For the printing stage, use SPI measurements and deposit limits. At the initial build gate, first-piece inspection and release connects the required evidence to a documented decision. These are process stages, not substitutes for the electrical or functional tests below.
Use the table to prepare a coverage discussion. Actual detection depends on the equipment, access, programming, component type and validated test method. The presence of a machine does not demonstrate coverage of every defect on a particular assembly.
| Method | Useful for | Important boundary |
|---|---|---|
| AOI / visual | Presence, placement, polarity where visible and inspectable solder features | Cannot directly see most hidden BGA joints or prove electrical operation |
| X-ray | Hidden-joint geometry, selected voiding and bridging conditions | An image alone does not establish complete electrical or product function |
| ICT | Accessible shorts/opens and selected component/value checks | Coverage depends on test access, circuit topology, guards and programs |
| Flying-probe electrical test | Accessible measurements with less dedicated fixture investment | Access, cycle time and circuit interactions still constrain coverage |
| FCT | Power, communications, outputs and behavior under a defined procedure | Only the tested operating states and limits are demonstrated |
| Bare-board electrical test | Connectivity/isolation of an unpopulated PCB against the reference | Does not test components, soldered assembly or firmware |
Use optical inspection for visible conditions
Automated or manual optical inspection evaluates features that the chosen view and method can observe. Depending on the setup, it may identify missing components, orientation issues, alignment, or visible solder conditions. It does not directly observe every hidden joint, internal connection, or electrical parameter. A component with the correct appearance may still have the wrong value or behavior.
Inspection needs a defined reference and acceptance criteria. The assembly revision, approved variant, and DNP list should match the inspection program or instruction. Otherwise a deliberately empty position may be reported as a defect, or a missing part on another variant may be accepted incorrectly. Program and drawing revision control are therefore part of the test system, not administrative details separate from quality.
Consider package visibility. A leadless package may provide limited visible solder evidence unless its construction supports a suitable side feature. TI's discussion of wettable flanks explains this distinction for relevant QFN packages; it does not mean all QFNs can be fully assessed optically. Use the exact package characteristics and the actual inspection capability when defining coverage.
Record how ambiguous findings are resolved. An image flagged by a machine is not necessarily a confirmed defect, and an operator override should have a reason. For important features, retain the evidence required by the agreed process. The objective is a consistent decision linked to the correct build, not simply a count of images or an unexplained pass result.
Reference for this review: .
Use X-ray for selected hidden structures
X-ray inspection can provide information about hidden solder and internal geometry that optical inspection cannot see. Its usefulness depends on the equipment, viewing method, package, and defect type. A two-dimensional projection can superimpose structures, while more advanced methods may resolve different details. State the requested evaluation rather than assuming that the word X-ray implies a particular technique or complete coverage.
Define what constitutes an acceptable result for the feature being examined. Voiding, bridging, insufficient solder, or alignment concerns need criteria appropriate to the component and application. Do not apply a universal void percentage to every package or every functional requirement. A thermal pad and a signal ball may have different consequences, and a geometric observation does not directly measure electrical or thermal performance.
Decide whether the inspection is full, sampled, or targeted at identified risks. Those are different scopes with different costs and evidence. If a failure is found in a sample, specify the escalation rule rather than improvising after the fact. The order should make clear whether the method is part of routine acceptance, first-article review, or a separate investigation.
Keep X-ray results connected to unit identity and process history. If a board is reworked, the original and subsequent observations may both matter. Do not describe a post-rework image as if it represented the untouched initial process. A useful report states what was inspected, which units were included, and what conclusions the method can and cannot support.
Distinguish in-circuit measurements from functional behavior
In-circuit testing can evaluate selected components and connections when the circuit and access support the measurement. Other components on the same net may influence a reading, so a simple resistance value is not always a direct measurement of one resistor. Test development must account for the circuit topology and the available isolation or guarding techniques. Access alone does not guarantee diagnostic coverage.
Functional testing powers the assembly and exercises defined behavior. It may verify an output, a communication exchange, programming, or a sequence of operations. It needs firmware identity, stimulus, loads, timing, and pass limits. A board that turns on is not necessarily a board that meets its functional requirements. State the actual behavior to be demonstrated and the conditions under which it is checked.
The methods can complement each other. A functional failure may identify a problem without locating it, while an in-circuit measurement may help isolate a component-level issue. Conversely, a set of component checks may not expose an interaction that appears only during operation. Choose the combination according to risk, volume, available access, and the required evidence rather than applying every method by default.
Separate development cost from execution cost. Fixtures, software, calibration, and reference units can be necessary before testing a production board. The Keysight test-plan overview illustrates that ICT is a structured program, not merely a fixture touching pads. The actual configuration and limits must be developed for the design. Request a quotation with those engineering inputs and exclusions visible.
Reference for this review: .
Worked planning example: a controller with a BGA and analog input
Illustrative planning example, not a customer case study: a controller has a BGA processor, a power stage, two analog inputs, a precision analog output and a communications port. Inspect placement and visible solder features; define X-ray scope for hidden joints, electrical checks for supply nets and accessible components, and powered checks for current, calibrated input/output response and communications. The coverage matrix below maps these different questions to records for the same hardware and firmware configuration.
| Risk | Proposed discussion | Evidence still needed |
|---|---|---|
| Incorrect firmware | Program a controlled image and read back identity | Approved binary, revision and verification method |
| Analog channel error | Apply known inputs and compare outputs to limits | Stimulus accuracy, loads, limits and calibration requirements |
| Intermittent connection | Review the defect mechanism and suitable test conditions | Product-specific stress/reliability plan if required |
| Hidden solder defect | Choose views, sampling and disposition rules | Agreed inspection criteria; correlation with other tests |
A coverage matrix for a mixed-signal controller
Use these fields to scope the report with your quotation. The table is a planning example, not a 吃瓜tv test result or a promise that every record is included by default.
| Record item | Information to include | Why it matters |
|---|---|---|
| Wrong or reversed visible part | AOI or appropriate visual review | Program and review depend on distinguishable visible features. |
| Hidden solder geometry | Appropriate X-ray views | Image interpretation needs a defined region and acceptance basis. |
| Accessible component/network value | ICT or suitable flying-probe measurement | Topology and access constrain isolation and interpretation. |
| Loaded output or communication | Functional test | Specify firmware, mode, stimulus, load and observed limits. |
Walk through an engineering example
Continue to place the selected methods into an assembly control plan.
The email link opens a draft for you to complete. Contact is a separate form; these pages do not send messages automatically.
Make test access part of the design release
Supply a schematic, netlist where applicable, test-point map, connector pinout, fixture interface and safe power-up sequence. Identify sensitive nets and allowable probing conditions. For FCT, include firmware, equipment requirements, stimuli, measurable limits and expected cycle time. A functioning demonstration board is useful, but it is not a complete repeatable test procedure.
For the functional part of the plan, use the functional test guide to prepare the procedure and fixture discussion.
Compare coverage before comparing inspection cost
A low inspection price is useful only when the scope answers the required question. List the failure modes or characteristics the plan needs to address, then identify the method, access condition and observable result for each. A fast optical inspection may cover many visible features, while a slower electrical measurement answers a different question. Do not count the same easy-to-inspect feature several times and mistake the total for broad coverage.
Separate setup work from recurring work. Programming, fixture design, adapters, reference samples and validation can matter greatly on a first build. Handling, measurement cycle, review of calls and authorized retesting affect recurring effort. A design change can repeat part of the setup. Ask suppliers to identify the assumptions behind the comparison and the work excluded from each option. The flying-probe cost comparison tool accepts your own allowances for an equivalent-scope comparison.
No coverage percentage is meaningful without a denominator. It might refer to components, unique nets, solder joints, tested requirements or a defined fault list. These categories cannot be added together as though they were the same unit. Keep inaccessible and unresolved items visible. Where a gap is accepted, identify the responsible decision maker and the reason rather than silently marking it as tested.
Plan reliability and environmental work as a separate question
A manufacturing acceptance test asks whether the delivered unit meets a defined condition at that stage. Reliability or environmental evaluation asks how the design and construction behave under specified stresses and operating conditions. Passing the first does not automatically answer the second. A short functional test cannot establish a product lifetime, and a stress test without a relevant model cannot honestly predict field performance.
Choose environmental conditions from the product's intended use and applicable requirements. Temperature, humidity, vibration, contamination, and electrical loading may matter, but not every product needs the same set of tests. Specialized testing may require qualified facilities and procedures. Identify who defines the conditions, who performs the work, and how the results affect release decisions.
Keep exploratory testing distinct from qualification. Driving a prototype until it fails can reveal a weak point, but the result is not automatically a pass or fail against an agreed product specification. Record the purpose, stress sequence, and monitoring so the evidence can be interpreted. Avoid changing several conditions at once unless the test plan deliberately calls for that combination.
If a failure occurs, preserve the sample and its history before repair or destructive analysis. Compare with an appropriate control and choose the next investigation based on the observed symptom. A broad label such as reliability failure does not identify whether the cause lies in design, materials, assembly, handling, or the test setup. The test plan should lead into a disciplined investigation rather than an unsupported diagnosis.
Read the combined result without losing its limits
A board can pass the communication exchange and still fail the loaded output requirement. It can pass a production screen and still require separate environmental or life testing for design qualification. Preserve this distinction in the shipment summary. A clear report names the completed checks and exceptions, allowing engineering and purchasing to decide on the actual evidence rather than an unexplained 鈥渇ully tested鈥� statement.
For a quotation, share the current assembly data, firmware requirements, intended modes, critical measurements and any existing test procedure. If some limits are undecided, identify them as open questions. Continue to assembly layout and test access when access or observability must be designed into the next revision, or use Contact to discuss the implementation scope.
Write a test scope that leads to an actionable inquiry
For every requested method, state the unit population, stage, acceptance basis, and record required. Identify whether the work is routine acceptance, first-article review, or development. Include firmware, fixtures, loads, and safety constraints where applicable. If the limits are not yet known, ask for help defining the scope rather than describing an undefined test as mandatory and complete.
Define failure handling before production. Who can authorize rework, what must be recorded, and which tests must be repeated afterward? Does a design-related problem pause the lot? How are questionable results escalated? These decisions affect both schedule and confidence in the delivered units. A test with no disposition process can generate data without resolving whether the board is acceptable.
Keep reporting proportional to the decision. Some projects need unit-level results; others may need a summarized acceptance record plus traceability. Request what your engineering and quality process actually uses. Avoid equating a thick report with good coverage, or a short report with poor quality. The key is whether the evidence supports the agreed conclusion and can be connected to the correct revision.
The downloadable worksheet organizes stage, risk, method, limit, and evidence. Use it when discussing assembly scope or preparing a prototype validation plan. It does not certify a design or imply that every method listed is provided in-house. A productive inquiry identifies the faults or functions you care about and the decisions the evidence must support. That gives the reviewer a clear basis for proposing an appropriate, bounded test plan.
Define the identifiers and record relationships with the PCBA traceability checklist.
Prepare a test-plan discussion
This optional tool prepares a discussion using your inputs. It does not approve the design, place an order or upload project files. You can contact engineering directly without completing it.
Assembly Test Plan Builder
Working document
Prepare your handoff checklist
Open the editable text worksheet, save a copy and record the requirement, owner and evidence for each item. Keep unanswered points visible when requesting review.
- Product failure modes and critical functions
- PCB/PCBA revision and schematic
- AOI/X-ray scope and acceptance criteria
- Electrical access and test-point map
- FCT stimuli, firmware, loads and pass/fail limits
- Fixture/program owner and validation plan
- Sampling, serial or lot linkage and record format
- Rework, retest and unresolved coverage gaps
The Contact page is a separate enquiry step. Attach your completed worksheet and relevant files there using its available submission options; this page does not automatically transfer your answers.
Questions before you release the files
Can AOI replace ICT?
No. Visible-feature inspection and electrical measurements detect different defect classes. Choose the combination from the risks and available access rather than treating one as a universal replacement.
Is FCT the same as reliability testing?
No. Production FCT normally checks defined functions under specified conditions. Environmental, life or reliability qualification is a separate plan when required.
Should prototypes use the full production fixture?
Not always. Balance the validation need, fixture development effort, access and expected future volume. Agree a prototype test scope and document any gaps before using the results to release production.
Continue with the next engineering task
Technical references
Explains hidden-joint and electrical-access considerations. It is not a coverage guarantee for an individual board.
Background on developing a PCBA test system; the product-specific procedure and limits must still be defined.
Project worksheet and supporting references
Download the project review worksheet (CSV). Record evidence and open questions before submitting files. This download does not send information to 吃瓜tv.
Technical references: ; . References support the stated scope, not blanket manufacturing approval.