3D AOI Versus 2D AOI: What Each Method Measures

A 3D AOI system projects structured light or a laser line onto the board and reconstructs a height map of every joint, while a 2D AOI system records an image and judges contrast, colour and shape. The distinction sounds technical but it decides which defects each machine can find and which ones pass straight through. Choosing between them is therefore a question about the defect spectrum, not about image resolution.

Both methods align the board to fiducials, compare against a reference and produce a pass, review or reject result. What differs is the physics of the measurement: one is a geometric reconstruction, the other is a photometric comparison. Every strength and every weakness of the two techniques follows from that single difference.

How 2D Inspection Works

A 2D machine illuminates the board with coloured or angled lighting and captures a single image per field of view. The algorithm then compares regions against the reference image and against a set of extracted features such as the fillet outline, the centroid of the pad and the grey level of the joint surface. It is fast, simple and inexpensive. For that reason a 2D system remains the default choice on many lines that were specified when board densities were lower.

Because a single image has no depth, the result depends entirely on how the surface reflects light. A change of solder mask colour or of board finish can therefore shift the whole result without any change to the process. A joint that is well formed but has an unusual surface texture can look like a defect, and a joint with too little solder can look acceptable if the lighting happens to flatter it. That is the fundamental limit of the method, and it is the reason two identical boards can be judged differently by two different machines.

How 3D Inspection Works

A 3D machine projects a known pattern and calculates the height of each point on the surface from the deformation of that pattern. The output is a dense height map from which volume, area, height and slope can all be derived. Calibration against a reference standard is what keeps those derived values traceable from one machine to another. The measurement is geometric, so it does not depend on how shiny the surface happens to be, and it does not depend on solder mask colour either.

The cost is complexity. The projection and camera geometry must be calibrated, the board must be flat enough for the working range of the sensor, and the reconstruction needs more computation than a 2D comparison. Setup and maintenance are correspondingly more demanding. Vibration and conveyor motion both affect the reconstruction, so the mechanical installation of the head matters as much as the optics.

What Height Data Adds

Height and volume are the two quantities that most directly describe a solder joint. Solder joint height can be compared against a target band, and volume can be compared against the paste volume that was printed. Those comparisons turn the inspection from a shape check into a process measurement.

That is why print inspection data and 3D AOI data can be correlated to close the print-to-reflow loop. A joint that is short of solder at AOI can often be traced back to an aperture or a print setting rather than to the reflow profile.

Shadowing and Occlusion

Both methods suffer from occlusion, but in different ways. In 2D, a tall component can shade a neighbouring joint from the light source, changing its appearance without changing its geometry. In 3D, the same component can hide the joint from the projection or the camera, leaving a hole in the height map.

The practical answer is the same for both: inspect the board from more than one direction, or accept a defined blind area and cover it by another means. Lighting angle can be changed on a 2D system to reveal a joint that was previously in shade, though each new angle costs programming time. The blind area should be documented, because it is a known escape path rather than an unknown one.

Coverage of Different Joint Types

2D inspection remains competitive for joints whose quality is expressed mainly by outline and contrast, such as chip component fillets on a clean board. It is also fast enough to be used in line on high-volume products where the defect spectrum is narrow and well understood. Inspection coverage should therefore be judged against the defect list rather than against the fraction of board area the camera can capture.

3D inspection is the stronger choice where height or volume is the controlling parameter. That includes fine-pitch devices, large thermal pads, connectors with coplanarity requirements and any product where the process window is narrow enough that a small volume error matters. On those products the measurement is the inspection, and a shape comparison adds very little. The extra cost is then repaid by a reduction in escapes rather than by a change in the false-call rate.

Throughput and Programming Effort

A 3D system is slower per unit area, because the pattern must be projected and reconstructed, and the projection sequence adds time that a 2D system does not have. Modern machines close much of that gap, but the difference never disappears entirely.

Programming effort is the other practical difference. A 3D program is built around measurement windows with limits, which is more work up front but less sensitive to cosmetic variation. A 2D program depends on lighting and colour thresholds that must be re-tuned whenever the board finish changes. That re-tuning effort is recurring, and it belongs in the cost comparison rather than in the list of one-off tasks.

Choosing Between the Two

The decision should start from a defect list with expected escape rates, not from a product brochure. If the controlling defects are missing parts, polarity and gross bridging, a well-programmed 2D system can be adequate. If they are volume, height or coplanarity, the geometry has to be measured directly. The machine choice then follows from the measurement rather than from the marketing description.

Many lines end up with both, using 2D for speed on simple products and 3D where the assembly demands it. That combination is reasonable provided the two systems share a common reference and acceptance philosophy, which is covered further in this AOI guide.

Verification and Correlation

Whichever method is used, the system should be verified against a defect set and against a physical measurement. For 2D that means a set of known-good and known-bad samples, and for 3D it should also mean a height or volume check against a reference standard.

Correlation between the AOI result and X-ray void data is a useful cross-check, because the two methods measure different physical quantities on the same joints. Divergence between them usually points to a programming or calibration problem rather than to a real process change. Periodic re-verification against the reference sample keeps both systems anchored to the same standard.

FAQ

Is 3D AOI always better than 2D? No. It measures more, but it costs more and is slower, so it pays only where height or volume actually controls the defect spectrum.

Can 2D AOI measure solder volume? Only indirectly, through fillet area and contrast. It cannot produce a true volume figure without a height measurement.

What causes a blind area in AOI? Tall components occluding neighbouring joints or the projection path, which should be mapped and covered by another inspection method.