X-ray Inspection for PCB Assembly: What It Finds

Optical inspection sees the surface of a solder joint. X-ray inspection sees through it, which is the only way to judge a connection under a BGA or ball grid array, a land grid array or a package with a thermal pad. It does not replace visual inspection, and it does not answer every question, but without it a large part of a modern assembly is simply unverified.

What X-ray Inspection Adds

The technique produces an image based on how much radiation passes through the board. Denser material attenuates more, so solder, copper and the voids inside them appear at different intensities and the internal structure of a joint becomes visible as a contrast pattern.

That is what allows an operator, or a machine, to detect defects that no camera can reach: a missing ball under a package, a bridge between two adjacent joints, an incomplete fillet on a through hole barrel or a large void in a thermal pad.

How the Image Is Formed

A source emits radiation through the assembly and a detector on the other side records the transmitted intensity. The board is usually positioned at an angle so the beam is not obstructed by the dense plane layers, and the result is a two dimensional projection of a three dimensional object.

Because the image is a projection, features on different layers overlap. The skill in interpretation is separating the solder of interest from the copper planes, the vias and the components above and below it.

Transmission and Computed Tomography

The standard system produces a single projection, sometimes from two angles, which is fast and adequate for most inspection. It shows the presence, shape and approximate size of a joint but cannot isolate a slice of it.

Computed tomography reconstructs many projections into a three dimensional volume, which allows individual slices to be examined. It is slower and more expensive, so it is used for failure analysis, for qualifying a new package or process, and for the occasional problem that transmission imaging cannot resolve.

What It Detects Reliably

Missing or displaced balls, bridges, large voids, insufficient solder and open joints are all visible. Head in pillow defects, where two balls meet without merging, are harder but can often be identified from the shape of the interface when the imaging is good.

Outer lead joints on quad flat packages are also inspected this way when the lead is under the package, and the technique is the only practical check on the integrity of a through hole barrel that has been filled and capped.

Void Measurement and Acceptance

Voids are measured as a fraction of the joint area, and the criteria come from a standard or from a customer specification. The measurement is influenced by the imaging parameters, so two systems can produce different numbers for the same joint unless the settings are agreed.

A single large void and an equivalent area of small ones are not equivalent in service. The solder void that matters is the one that reduces the cross section carrying current or heat, and the acceptance criterion should reflect that rather than a simple percentage.

Limits of the Technique

Resolution falls as the sample becomes thicker or denser. A board with heavy copper planes attenuates much of the beam, and the contrast available for the joints above and below is reduced accordingly.

The image is also a projection, so a defect can be masked by a feature on another layer. Good practice includes imaging at more than one angle, which separates structures that would otherwise overlap in a single view.

Coverage and Sampling

Full inspection of every joint on every board is slow and rarely necessary. Most programmes combine optical inspection of the accessible joints with X-ray of the hidden ones, and some use a sampling plan where the process is stable.

Sampling has to be designed rather than assumed. It is most useful where the defect rate is low and the process is controlled, and least useful on a new product where the failure modes are not yet known, which is why the assembly programme should be reviewed as the product matures.

Programming and Image Quality

Automated X-ray systems compare images against a programmed reference, so the programme determines what is found. Poorly chosen thresholds produce false calls on good boards and, worse, missed defects on bad ones.

Image quality depends on the settings as much as the hardware: voltage, current, exposure time and the number of averaged frames all affect the contrast. Those settings should be recorded with the programme so that results remain comparable between runs.

Following Up on a Finding

An X-ray image shows a symptom rather than a cause. A void rate that rises points to the printing, the profile or the paste, and the investigation follows the same logic used for any reflow defect.

Where the defect is confirmed, cross sectioning a sample provides the definitive answer, since it shows the actual interface and the grain structure. The image tells you where to cut; the section tells you what happened.

Equipment Types

Benchtop systems use a sealed source and a flat panel detector, and they are adequate for occasional inspection and for small assemblies. Production systems add motorised axes, programmable inspection routines and automatic defect classification.

The choice follows the volume and the requirement. A line that inspects every board needs throughput and repeatability, while a laboratory that investigates a few samples needs resolution and the ability to rotate the sample freely.

Fixturing and Sample Handling

The board has to be held in a defined position so that repeated inspections are comparable. A fixture that allows the board to move slightly changes the projection geometry and therefore the apparent size and position of every feature.

Bare boards are easy to hold; populated assemblies are not. Components and connectors create obstacles and add material in the beam path, and the fixture should present the region of interest at a consistent angle rather than the whole board at whatever angle fits.

Correlating X-ray With Other Data

X-ray results are most useful when combined with the process data from the same period. A void rate that rises together with a change in paste lot, a profile adjustment or an increase in ambient humidity points to a cause rather than to a coincidence.

The same applies to defect classification. A pattern of findings concentrated on one package identifies a stencil or placement issue, while a spread across the board suggests something more general, and the placement records usually settle which.

Cost and Throughput

X-ray inspection is a cost per board rather than per order, and the time is dominated by the number of images and the resolution required. Reducing the region of interest to the joints that need checking lowers the cost without reducing the value.

Programming and validation are one-off costs that are amortised over the production run. On a short run they can dominate the total, which is one reason a lower cost optical strategy is sometimes chosen where the risk of a hidden defect is acceptable.

FAQ

Can X-ray see every solder joint? Most of them, but resolution falls as the board becomes thicker and denser, and a projection can hide one feature behind another. Imaging at several angles helps.

Does X-ray damage the assembly? At the energies and exposures used for inspection, no. The technique is non destructive and is used on production boards.

Is X-ray always needed? Not on a board with only accessible joints. It becomes necessary as soon as the design includes area array packages, thermal pads or any joint that cannot be seen.