X-ray Inspection of Hidden Solder Joints
Why Some Joints Cannot Be Seen
As packages have moved to area array terminations, the solder joint has moved out of sight. A ball grid array, a chip scale package and a land grid array all connect on the underside of the body, so optical inspection reaches the perimeter and nothing else. The joints that fail first, the ones at the centre of a large array where the thermal mass is greatest and the reflow is least uniform, are exactly the ones that cannot be seen from above. X-ray inspection exists because the alternative is to rely on the process without knowing what it produced.
How X-ray Imaging Works
An X-ray image is a shadow. The beam passes through the assembly and the detector records how much of it is absorbed, and because solder absorbs far more than the laminate or the mould compound, the joints appear as dark shapes against a lighter background. The contrast and the geometry both matter: a denser feature attenuates more, and the viewing angle determines what is hidden behind what. Two dimensional systems look straight down and can be tilted, which is enough to spot bridges and missing balls, while computed tomography rotates the sample or the source and reconstructs a slice, which is needed to judge a void or a crack inside a joint.
What the Image Reveals
The common defects have recognisable appearances. A missing ball is a gap in the regular pattern. A bridged pair is a joined dark shape where two should be separate. A void is a light region inside a dark ball, and its size and position are judged against a specification rather than by eye. An open joint, where the ball has not wetted the pad, often shows as a ball that is rounder and more uniform than its neighbours, because it never collapsed. Head in pillow defects, where the ball and the paste have not coalesced, are among the hardest to find and are usually detected by a change in shape at a particular angle or by a density difference at the interface.
2D, Oblique and CT
A transmission image is fast and covers the whole board, which is why it is used for screening: a bridge or a missing ball is usually obvious at a glance and does not need a three dimensional view. Tilting the source, sometimes called oblique or angled view, separates the joints of one row from another and reveals a bridge that a straight down view would hide. Computed tomography is slower and more expensive, and it is reserved for the questions that need a slice: the internal structure of a single critical joint, the extent of a void, or the failure analysis of a returned assembly. Choosing the technique is a matter of matching the cost to the question.
Reading Voids and Their Limits
A void is a pocket of gas trapped in the joint, formed from flux volatiles or from the surface finish, and it is normal to some degree. The engineering question is not whether a void exists but how much of the joint’s area it occupies and where it sits, since a void at the interface or at the perimeter matters more than one in the bulk. Standards give acceptance limits, and the measurement depends on the technique and the threshold used, so two inspectors can report different numbers on the same joint. The limits should be agreed with the customer and the method fixed, otherwise the report becomes an argument.
Limits of the Technique
X-ray is powerful and not universal. It is poor at judging the surface of a joint, so it does not replace optical or automated optical inspection for the visible perimeter. It depends on the density difference, so a joint under a heavy copper plane or a metal stiffener may be hidden, and a very thin joint may show little contrast. Its resolution is finite, so a small crack may be invisible until it becomes a break. And the equipment consumes time per board, which means the technique is normally applied to samples, to a first article or to a suspect lot rather than to every unit of a high volume run.
Making Inspection Worthwhile
Inspection finds defects; it does not prevent them. The value of an X-ray programme is proportional to what is done with the result: a bridge that is traced to a paste volume, a void rate that is traced to a profile or a finish, a head in pillow that is traced to a thermal imbalance across the package. Sampling should be designed to see the process, so that the boards are taken across the run and across the panel rather than from a single convenient position. Where the process is stable and the data is good, the sampling rate can be reduced; where a change has been made, it should go up again.
FAQ
What is X-ray inspection for? Inspecting solder joints that optical methods cannot see, such as those under a BGA or a chip scale package.
What is the difference between 2D and CT? A 2D image is a fast shadow suitable for screening; CT reconstructs slices and is used for a single critical joint or a failure analysis.
Are voids always a defect? No. They are normal within limits, and the acceptance depends on their size, position and the agreed specification.
Can X-ray find every defect? No. It is weak on surface conditions, may be blocked by heavy copper and has finite resolution.
Should every board be X-rayed? Usually not. Sampling across the run is more informative than inspecting every unit of a stable process.
Conclusion
X-ray is the tool that makes a hidden joint visible, and it is most valuable when the result changes the process rather than only the record. Match the technique to the question and sample to see variation. Inspection strategy belongs to quality management, the joints themselves are produced by SMT PCB assembly, and the testing that follows is part of PCBA testing. Inspection planning for a new assembly begins during prototype PCB assembly in 2026.
