Positioning mistakes substance.
When a misaligned 0201 resistor, reversed diode, missing out on QFN, or wrong-value capacitor enters reflow, a correctable placement event can become a firm problem that consumes professional time, pads, flux, production capability, and shipment margin.
Why pay to solder an error you could reject seconds earlier?
That is the actual disagreement for Automated Optical Evaluation after pick and place. It is not just an additional video camera inside an already pricey SMT line. Correctly executed, it is the initial objective checkpoint in between component mounting and thermal processing– a point where mistakes remain relatively inexpensive to correct.
My hard-line view is this: suppliers ought to quit treating AOI as a ceremonial pass/fail box. AOI is a process-control instrument. If its information does not influence feeder configuration, nozzle upkeep, placement offsets, element libraries, printer settings, or line choices, the manufacturing facility is spending for images as opposed to knowledge.
What AOI After Choose and Location Really Means
AOI after choice and area is typically called pre-reflow AOI, post-placement assessment, or component placement examination. The equipment examines the populated PCB after installing however before the board enters the reflow stove.
A common process sequence is:
Solder paste printing → SPI → choice and area → pre-reflow AOI → reflow → post-reflow AOI
Not every line has all three examination points. Prototype operations might rely on SPI plus post-reflow AOI. High-volume automobile, telecoms, clinical, or commercial lines may validate pre-reflow examination because the expense of a duplicated placement error can escalate swiftly throughout countless boards.
Producers planning an incorporated line should assess AOI as part of the complete complete SMT line configuration, not as a separated acquisition added after conveyors, printers, mounters, and stoves have actually already been picked.
That difference issues. Conveyor size, SMEMA communication, board clearance, cycle time, CAD-data availability, line balancing, repair service routing, barcode traceability, and factory-system combination all influence whether the AOI station assists manufacturing or comes to be a traffic jam.
What Post-Placement AOI Need To Discover
An experienced SMT optical examination program should find more than “element present” or “element missing.” At minimum, it must assess:
- Missing and additional parts
- Incorrect part identity
- Polarity and alignment mistakes
- Rotation past the configured tolerance
- X-axis and Y-axis placement offset
- Tombstoned or standing components where visible
- Changed fine-pitch packages
- Raised leads or package-body abnormalities visible before reflow
- Harmed, contaminated, or improperly significant parts
- International material within the inspection area
This is the heart of pick and area quality control. The AOI outcome ought to be mapped to a likely process reason: feeder loading, centroid information, nozzle selection, vision acknowledgment, pickup elevation, board support, fiducial placement, part packaging, or machine calibration.
A generic “NG” alarm system is weak process engineering. A valuable alarm says what moved, by just how much, on which designator, where feeder, under which program revision.
Manufacturing facilities comparing equipment can begin with the readily available SMT inspection system classifications and then match them versus the speed, part variety, and data user interfaces of their existing pick-and-place device platform.
The Pre-Reflow False-Call Issue Nobody Must Disregard
Pre-reflow AOI has a built-in problem: solder paste is still soft, elements have actually not yet resolved with molten-solder surface stress, and a placement that appears low prior to reflow may end up being acceptable afterward.
That is not a theoretical edge case.
A peer-reviewed Binghamton University study provided at FAIM 2024 evaluated 30,000 production-line information factors entailing SPI and pre-reflow AOI. The scientists defined specific pre-reflow placement-offset alarm systems as false calls when the parts ultimately got to appropriate positions via self-alignment throughout reflow. Their machine-learning classifier achieved 89.3% accuracy in identifying those cases.
That searching for subjects the awkward trade-off:
Set placement resistances also tight, and drivers spend their change assessing boards that reflow would certainly have fixed.
Establish them also loose, and real placement issues enter the oven.
The answer is not to disable alarms. It is to segment tolerances by package kind, pad geometry, solder-paste deposit, element mass, lead style, board design, historical reflow habits, and product risk.
A 0402 resistor, a polarized electrolytic capacitor, a 0.4 mm-pitch QFN, and a large secured inductor should not share one generic displacement threshold. Yet several badly commissioned AOI programs behave as though they should.
AOI Should Make Use Of Upstream Information
A modern-day AOI device sees the board. A contemporary high quality system understands the procedure that generated it.
In 2023, researchers examining PCB manufacturing made use of data representing 6 million pins, 2 million parts, and 15,387 PCBs to model defects at pin, component, and board levels. Their job integrated SPI-derived attributes with AOI results rather than treating each inspection device as an isolated island. The complete data-centric production study sustains a more comprehensive lesson: evaluation quality improves when printing, placement, and final confirmation data can be linked through element and PCB identifiers.
This is where numerous manufacturing facilities underperform.
They have SPI. They own AOI. They own qualified mounters. Yet the records being in different databases, utilize inconsistent identifiers, and vanish right into month-to-month spread sheets nobody depends on.
The far better design web links:
Paste down payment → feeder and nozzle → placement coordinates → pre-reflow result → reflow result → driver verification → repair personality
That chain transforms PCB assembly inspection right into root-cause analysis.
Without it, a persisting skewed-component alarm may be condemned on the AOI program when the actual reason is paste-volume inequality, feeder indexing, worn nozzles, negative centroid information, insufficient board support, or bundle variant.
Pre-Reflow AOI Versus Post-Reflow AOI
| Examination point | Primary purpose | Strongest defect protection | Main limitation | Best functional usage |
|---|---|---|---|---|
| SPI after publishing | Measure solder-paste deposition | Quantity, area, height, countered, linking and insufficient paste | Can not confirm whether the proper part will certainly be mounted | Avoid print issues from infecting later phases |
| Pre-reflow AOI | Verify pick-and-place implementation | Missing out on, wrong, turned, reversed and displaced components | Can flag components that would self-align throughout reflow | Stop systematic positioning errors prior to heat |
| Post-reflow AOI | Inspect the visible firm setting up | Last component placement, polarity, noticeable joints, bridges and gravestones | Concealed joints remain outside straight optical sight | Final noticeable craftsmanship and process verification |
| AXI or X-ray | Take a look at internal structures | BGA, QFN and other hidden solder interfaces | Higher price, slower programming and potentially longer cycle time | High-risk bundles and hidden-joint confirmation |
The right answer is hardly ever “pick one maker.” The appropriate solution is to develop a coverage version.
Pre-reflow AOI is strongest where the procedure is still relatively easy to fix. Post-reflow AOI is toughest where the last soldered problem has to be evaluated. X-ray addresses evidence that optical line-of-sight can not straight record.
The 2024 YOLO-pdd research study standard shows how quickly vision designs are advancing: the reported version reached 95.8% precision, 99.6% recall, a 97.4% F1 score, and 92 frames per second on its PCB dataset. Those numbers are encouraging, however they are research-benchmark results– not consent to presume comparable manufacturing facility efficiency on reflective solder, combined part surfaces, warped boards, brand-new bundles, or insufficient training information. Evaluation the YOLO-pdd research study before repeating the numbers in a capital-equipment reason.
2D AOI or 3D AOI?
Two-dimensional AOI continues to be useful for presence, markings, polarity, rotation, shade, synopsis, and lots of positional checks. It can be fast, relatively simple to program, and adequate for less facility products.
Yet 2D systems presume elevation from appearance. Shadows, reflective terminations, dark package bodies, silk-screen interference, board color, and illumination variation can misshape that inference.
Three-dimensional AOI adds measured elevation and surface geometry. That makes it much better suited to coplanarity evaluation, lifted leads, package-height variant, element tilt, and intricate solder-joint forms.
Still, 3D is not magic. Poor collections produce inadequate cause 3 measurements also.
A buyer contrasting systems such as Saki BF-series systems, MIRTEC MV-series machines, Koh Young Zenith models, Pemtron Athena or Zeus systems, and Jutze AOI equipment must run a controlled evaluation utilizing real boards, real issues, real reflective coatings, and real takt-time demands. The readily available Saki AOI and SPI item range is one useful beginning factor for evaluating configurations.
How to Compensation AOI Without Developing an Alarm Factory
The very first manufacturing dish must not be approved due to the fact that one gold board passed.
It ought to be challenged.
Run appropriate variant. Run recognized defects. Turn polarized components. Remove components. Change plans in regulated increments. Introduce alternative approved vendors with different markings. Include board color variation, solder-mask variant, component-body variation, and realistic paste deposits.
After that gauge four separate outcomes:
True favorable: The system appropriately flags an actual problem.
True adverse: The system appropriately passes an acceptable attribute.
False positive: The system flags an acceptable function.
Incorrect adverse: The system misses out on an actual problem.
Accuracy alone can hide an unsafe assessment program. A line processing 99.9% appropriate components could appear “precise” by passing everything while stopping working to discover minority defects that matter.
Track recall by flaw class. Track false calls per board. Track driver testimonial time. Track repeated alarms by designator, feeder, nozzle, plan, vendor, change, and equipment.
And lock program modifications. An operator must not be able to broaden a polarity limit at 2:00 a.m. without leaving a traceable record.
Your acceptance criteria should likewise be linked to a concurred workmanship standard. IPC determines IPC-A-610J, launched April 4, 2024, as the current Alteration J acceptance standard, while IPC’s modification history records the original IPC-9716 AOI process-control criterion in December 2024 and Modification A in April 2026.
A vendor’s default library is not your quality plan.
The Metrics That Subject a Negative AOI Program
I would certainly decline any type of AOI proposal that discusses electronic camera megapixels yet can not define how these production metrics will be measured:
| Metric | Calculation | What it discloses |
|---|---|---|
| Issue recall | True flaws spotted ÷ all validated true flaws | Whether purposeful problems are leaving |
| False-call price | Duds ÷ all AOI alarms | Whether operators are losing testimonial time |
| False phone calls per board | Duds ÷ boards checked | The useful work troubled production |
| First-pass return | Boards passing without fixing ÷ overall boards | Overall process stability |
| Testimonial time | Overall verification time ÷ alarms examined | Labor and bottleneck influence |
| Retreat price | Defects located downstream ÷ boards evaluated | Failings missed out on by the evaluation plan |
| Repair confirmation price | Confirmed repair work ÷ repair service efforts | Whether rework is in fact regulated |
| Recipe-change regularity | Program adjustments per item and period | Whether programming is secure or responsive |
Here is a basic instance.
Intend a line evaluates 300,000 boards each month. AOI produces alarms on 1% of them, 70% of those alarms are incorrect, and each review takes 45 secs.
That creates:
300,000 × 1% × 70% × 45 secs = 94,500 seconds
That equates to 26.25 operator-hours monthly spent assessing acceptable boards. And that computation leaves out line stops, board handling, manager escalation, and lost concentration.
Incorrect telephone calls are not totally free.
The company’s wider procedure quality structure should as a result cover program recognition, measurement-system evaluation, issue seeding, driver disposition, maintenance, calibration, information retention, and corrective-action ownership.
Picking the very best AOI Device for SMT Assembly
The best AOI maker for SMT setting up is not automatically the fastest, newest, or most costly system. It is the equipment that fulfills the validated defect-coverage and takt-time requirements of the item combine with an acceptable false-call concern.
Ask suppliers to show:
- Discovery of your seeded flaws
- False-call performance throughout normal item variation
- Cycle time on your densest board
- Shows time for a brand-new product
- Changeover effort for high-mix production
- CAD, Gerber, BOM, centroid and IPC-2581 information import
- Barcode and MES traceability
- Closed-loop communication with SPI or positioning devices
- Offline shows and review capacity
- Regional parts, calibration, software application and design assistance
For a high-volume line, one second of cycle-time mismatch can identify whether the AOI terminal constricts the entire manufacturing facility. For a high-mix line, a four-hour programs concern can matter greater than academic examination speed.
The examination board makes a decision. Not the pamphlet.
Regularly Asked Inquiries
What is AOI after choice and area?
AOI after pick and area is a pre-reflow optical inspection step that contrasts each placed element’s existence, identity, polarity, rotation, X-Y balanced out, and noticeable problem versus set CAD, centroid, bill-of-materials, and image-library recommendations prior to the PCB gets in the reflow stove.
Its main financial worth is capturing positioning errors while elements can still be eliminated, rearranged, or replaced without desoldering.
Just how does AOI work after pick and location?
Post-placement AOI works by illuminating the inhabited PCB from numerous directions, catching adjusted 2D photos or 3D height information, situating parts through fiducials and CAD collaborates, gauging variances, and identifying each feature against tolerance policies prior to transmitting the board as pass, review, repair service, or turn down.
Advanced systems may additionally use analytical designs or trained classifiers, yet image top quality, collection precision, board assistance, and recipe recognition remain definitive.
Should AOI be set up prior to or after reflow?
Pre-reflow AOI is the placement-control checkpoint, while post-reflow AOI is the soldered-assembly checkpoint; the initial catches missing out on, incorrect, reversed, revolved, or displaced components prior to warmth, and the 2nd assesses the final noticeable outcome after solder paste has thawed, moistened, and possibly self-aligned elements.
Risky lines might validate both. Lower-volume lines need to choose inspection points utilizing issue background, part mix, getaway danger, fixing cost, and readily available cycle time.
Can AOI inspect BGA solder joints?
AOI is a line-of-sight surface area assessment approach, so it can validate a BGA plan’s presence, orientation, body setting, markings, and some coplanarity hints, but it can not straight see the hidden solder rounds below the bundle; X-ray or another complementary approach is needed for inner joint evidence.
This limitation additionally affects various other bottom-terminated parts. A solid inspection plan integrates methods instead of claiming one machine offers universal protection.
What is the very best AOI machine for SMT setting up?
The best AOI machine for SMT setting up is the system that fulfills the line’s validated defect-coverage, false-call, cycle-time, programs, traceability, solution, and product-mix needs on the customer’s own boards, rather than the model with the greatest camera-resolution case or the most sleek sales presentation.
Need a board-level acceptance trial with seeded flaws and typical manufacturing variant prior to accepting the acquisition.
Exactly how is AOI roi calculated?
AOI return on investment is the yearly value of avoided runs away, reduced rework, shorter medical diagnosis, recouped driver time, and safeguarded throughput minus ownership costs such as devices, programming, maintenance, calibration, training, flooring space, and line combination, divided by the complete spent price.
Usage confirmed factory information rather than generic cases. One stopped area failing might justify the system in a safety-related item, while a consumer line may depend extra heavily on throughput and labor savings.
Quit Soldering Unknowns
AOI after pick and place is most useful when it avoids a repeatable positioning trouble from ending up being a repeatable soldering issue.
Specify the defect classes. Seed the defects. Measure the incorrect phone calls. Attach SPI, placement, AOI, repair, and traceability data. Then buy the maker that verifies it can regulate your process– not the equipment whose presentation board was prepared to make it look foolproof.
For line preparation, board tests, design contrast, and inspection-station assimilation, speak to the SMT engineering group with your board measurements, minimum component dimension, package listing, target cycle time, production quantity, and existing problem background.
