PlatformsFaststream SiliconFaststream RadioFaststream VisionConnected EdgeFaststream SecureMobility & Rail
ProductsSemiconductor IPWireless & RANEdge & GatewaysTracking & IdentificationSoftware & FrameworksConnected Systems
Technology5G protocol stackWireless and RF architectureBaseband and low PHYForward error correctionControl and data planeHigh-speed interfacesFirmware and bootSilicon root of trustSoftware-defined vehicleAutomotive OTAFunctional safety
AIAI Engineering ServicesEdge AI & Embedded MLComputer Vision EngineeringSensor Fusion & PerceptionAI Silicon & AccelerationMLOps for DevicesAI Visual InspectionPredictive MaintenanceDriver MonitoringVideo Analytics & Safety
SolutionsSemiconductorIndustrial AIConnected ProductsAsset TrackingBluetooth AoA RTLSWearable TrackingAutomotive & MobilitySmart InfrastructureSecure IdentityWireless & SatellitePrivate 5GSmart WashroomsFuel ManagementSmart BuildingsWorker SafetyEnergy MonitoringSmart AgricultureSmart CityAutonomous PlatformsAssembly AutomationLiDAR Rail SafetyHardware Wallet
IndustriesSemiconductorTelecommunicationsIndustrial & ManufacturingAutomotive & MobilityTransportation & RailAerospace & DefenceHealthcare & MedicalEnergy & UtilitiesOil & GasRetailConsumer ElectronicsMedia & EntertainmentSmart Infrastructure & IoT
ServicesSystem Integration overviewASIC & SoC DesignRTL to GDSIIVerification methodologyDFT and silicon testLow-power designMixed-signal integrationDesign enablementFPGA DesignFPGA-to-ASIC ConversionAnalog, Mixed-Signal & RFHardware & High-Speed PCBEmbedded SoftwareCloud, OTA & Device ManagementManufacturing TransitionHow we engage
CompanyAbout FaststreamEngineering ExcellenceLeadership & OrganisationHow We EngageQuality & ComplianceStandards & EcosystemPartners & EcosystemTrust CentreLocations & DeliveryNewsroom & MediaCareersCase StudiesKnowledge CenterWhite PapersGlossaryNewsletterResources & Support
ContactStart a projectHow we engage
Talk to us
ENGINEERING INSIGHT

The false-call problem, and why AOI is judged on trust

The failure mode of automated optical inspection is not the missed defect — it is crying wolf. A board has thousands of joints, most cosmetically imperfect and functionally fine. A system that flags every deviation buries the operator in false calls and, within a shift, gets ignored. The real metric is how few good parts it questions.

ShareLinkedInXEmail
An inspection is only as useful as it is trusted.
TUNED HOT (HIGH SENSITIVITY)TUNED ON COST (CALIBRATED)Cosmetic fillet variationFlagged as a defectLearned as normalOperator workloadFloods of false calls to clearOnly real questions raisedTrust in the systemMuted by shift twoKept, because it is rightA genuine bridgeCaught, among the noiseCaught, and believedA rare critical defectLost in the false callsWeighted by consequenceThe decision thresholdSet to catch everythingSet on cost of each errorDetecting defects is easy. A low false-call rate at an acceptable escape rate is the engineering.
THE LEVERS

What drives the false-call rate down.

Controlling false calls in AOI
LeverWhat it doesWhy it matters
Model of the good jointLearns the full range of acceptableStops cosmetic variation being flagged
Multi-angle imagingCaptures joint shape, not just top viewSeparates real 3D defects from appearance
Defect classificationNames bridge, insufficient, tombstoneBeats a generic anomaly flag
Cost-weighted thresholdBalances false calls vs escapesSet on consequence, not an accuracy score
Imbalance handlingLearns rare defect classesRare but critical faults not averaged away
Operator feedback loopFolds verdicts back into trainingFalse-call rate keeps falling
THE DISTINCTION

A missed defect is rare; a false call is constant.

On a populated board, most solder joints are cosmetically variable and perfectly acceptable. The rare genuine defect — a bridge, insufficient solder, a lifted lead — is the exception. So the everyday behaviour of an inspection is dominated not by catches but by how it treats the thousands of good, imperfect joints it sees. Tune it to catch everything and it flags a flood of them; an operator has to clear each false call by hand, and once the rate is high they stop trusting the machine and start rubber-stamping it. At that point the inspection has failed, regardless of how many real defects it could technically find.

The fix is to teach the system what a good joint really is, in all its variety. A good fillet varies enormously in shape and sheen, and a model that has only seen a narrow slice reads that legitimate variation as a fault. Training on the true distribution of acceptable joints is what lets the system question only genuine deviations. Multi-angle imaging helps, because some of the worst defects — a lifted lead, a hollow joint — are three-dimensional and look acceptable from directly above; capturing the joint's actual shape, not just its top-down appearance, is what tells a real defect from a cosmetic one.

Then the threshold is set on cost, not on an accuracy number. The right operating point balances the cost of a false call against the cost of an escape, weighted by the real consequence of each defect type rather than by how often it appears. A rare but critical defect is worth catching even at the price of a few more questions; a cosmetic quirk is not worth a single one. And every operator verdict is folded back into training, so the false-call rate keeps falling and the trust keeps building — which is the only thing that keeps the system switched on.

IN PRACTICE

Building an AOI people trust.

COMMON QUESTIONS

What engineers ask before they call.

01

Why is the false-call rate the key metric, not detection?

Because on a real board missed defects are rare and false calls are constant. Most joints are cosmetically variable but functionally fine, and a system tuned to catch every possible defect flags a flood of these good joints. Operators must clear each false call, and once the rate is high they stop trusting the machine and start passing its calls without looking. So the metric that actually decides whether the inspection is useful is how few good parts it questions, at an acceptable escape rate.

02

How do you tell a real defect from cosmetic variation?

By teaching the system the full range of what a good joint looks like, which is wide, so only genuine deviations stand out; by imaging the joint from multiple angles so its true three-dimensional shape drives the decision rather than a flat top-down view; and by classifying the specific defect type instead of raising a generic anomaly. The decision threshold is then set on the real cost of a false call versus an escape.

03

Why isn't a single top-down image enough?

Because some of the most serious defects are three-dimensional. A lifted lead, a hollow joint or insufficient solder can look perfectly acceptable from directly above and only reveal itself from an angle or under structured light. A single top-down image therefore misses them or, worse, forces the system to guess, which raises false calls. Multi-angle capture lets the system see joint shape, at the cost of a more careful imaging setup.

FOUND THIS USEFUL?

Pass it on.

Written for engineers. Share it with one.

ShareLinkedInXEmail
KEEP READING

Related work.

BUILD WITH FASTSTREAM

Bring us the difficult part.

Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.