X-Ray Inspection — When It's Needed

Section Overview

Everything you have used to verify a repair so far has a blind spot. A light microscope sees only exposed surfaces (10.3); a meter and even boundary scan confirm a net is connected but never show you the joint itself (10.4); and the visual criteria apply only to joints you can actually see (10.2). That leaves the hardest joints unseen — the ones hidden under a ball-grid array*, the fill deep inside a plated barrel, the* void buried in solder — and this final section is the tool that reaches them. X-ray inspection passes X-rays through the whole assembly, and because dense solder absorbs far more than the plastic and laminate around it, the solder images darker — so a hidden joint shows right through the package and the board. This section is about the decision and the reading, not operating a machine. You reach for X-ray when a joint is invisible to light and unreachable by a probe AND the stakes justify it (10.1). It reveals the defects nothing else can: voids as lighter spots inside the darker solder, bridges as solder joining balls that should be separate, missing or open balls, head-in-pillow (a ball that melted but never fused to its paste), ball alignment and collapse, and barrel fill. But it has a real limit: a flat, straight-down 2D image is a density projection, so features stacked in the beam can hide behind one another — which is why angled views tilt the beam to separate them, and why computed tomography takes many views and reconstructs a full three-dimensional model when 2D is not enough. There is manual X-ray for inspection and rework, and automated X-ray inspection on production lines. And there is a practical reality: X-ray systems are expensive and demand radiation-safety compliance, so most bench repairers do not own one — you send the board to an X-ray service, and the real skill is knowing when it is needed and how to read what it shows. One hazard sets this tool apart from everything before it: X-rays are ionizing radiation, contained by shielding and interlocks you never defeat. When the joint you must trust is one you cannot see, X-ray is how you finally see it.

Why This Matters

Some of the most important joints on a modern board are the ones you cannot see, and a repair is only as trustworthy as your ability to verify its worst-hidden connection. This matters because area-array packages hide their joints by design: a ball-grid array carries its connections as a grid of balls underneath the package, where no light and no probe can reach them, so without X-ray a BGA rework is verified by hope (10.3; 10.4). This matters because the hidden defects are real and common: a void*, a bridge between balls, an open, a* head-in-pillow — each can fail a board that looks and even tests fine, and each is visible only through the assembly. It matters because interpretation has traps: a flat 2D image overlaps everything in the beam, so a bridge can hide behind a ball and a reader who does not know that limit can pass a defective joint — knowing when to tilt the view or call for computed tomography is part of reading X-ray correctly. It matters because the step is expensive and situational: X-ray costs time, money, and radiation-safety overhead, so you spend it where the joint is invisible and the acceptance class demands certainty, not on every board (10.1). It matters because the safety is non-negotiable: X-rays are ionizing radiation, and the entire reason a cabinet system is safe is its shielding and interlocks — understanding that is what keeps this a safe tool. And it matters because it completes the inspector's reach: between visible inspection, electrical test, and X-ray, there is no joint on the board you cannot verify by some means. Learn when the hidden joint must be seen, and you can stand behind even the repairs you cannot look at directly.

Required Prerequisites

• Electrical Verification After Rework — Section 10.4 verified function and reached un-probeable nets with boundary scan; X-ray is how you see the hidden joint itself. You should know the visual joint criteria (10.2), that a light microscope sees only exposed surfaces (10.3), and what a ball-grid array is and how it is reworked (Chapter 8). This is an awareness and decision section — you will not operate an X-ray machine here.

Recommended Consumables

• Access to an X-ray inspection service, shared lab, or contract inspector — the realistic way most repairers get an X-ray when they need one (few own a system)
• A set of annotated example X-ray images — good and defective BGAs — to learn to read voids, bridges, opens, and head-in-pillow
• A reference guide to X-ray defect signatures — to match what you see to a named defect
• An inspection log — to record what X-ray showed and the decision it drove

Recommended Practice Hardware

• Sample X-ray images or a defect-signature atlas (good/void/bridge/open/head-in-pillow) — to practice interpretation without a machine
• A boneyard BGA board you could send for X-ray — to compare your visual and electrical verdict against what X-ray reveals
• If you have supervised access to a trained facility — observe a manual X-ray and a CT scan of a real board (with an authorized operator)
• No X-ray equipment of your own is expected — this section is decision and interpretation, not operation

Real-World Applications

X-ray is where the last, invisible joints are finally verified, and it routinely catches what every other method missed. A technician who reballed and reflowed a BGA cannot see a single one of its joints by eye, so the rework is confirmed with an X-ray that shows every ball formed, none bridged, none open (10.3). An inspector chasing an intermittent fault that passed a functional test cold X-rays the suspect BGA and finds a head-in-pillow — a ball that touched its paste but never fused — which no surface look or continuity check had revealed. A repairer worried about a power joint X-rays it and sees excessive voiding inside the solder, a void fraction beyond what the acceptance class allows, and reworks it (10.1). Someone unsure whether a shadow on a 2D image is a real bridge has the board tilted for an angled view — or sent for computed tomography — to separate the stacked balls and confirm it. And a shop that does not own an X-ray sends its critical BGA reworks to a service, because knowing the joint must be verified matters more than owning the machine. The failures X-ray prevents: a hidden open or bridge under a BGA passed as good, a head-in-pillow that fails in the field, an over-voided power joint, and a misread 2D shadow — all caught by seeing through the assembly when nothing else can.

Common Challenges

• Reading a 2D image as if it were 3D. A flat X-ray overlaps everything in the beam — a bridge can hide behind a ball, so tilt the view or use computed tomography when features stack (10.4).
• Reaching for X-ray too soon, or too late. It is expensive and radiation-controlled — use it when the joint is genuinely invisible and the acceptance class justifies it, not as a first look and not never (10.1).
• Assuming you need to own the machine. Most repairers do not — the skill is knowing when a hidden joint needs X-ray and sending it to a service.

Safety Notes

Risk Level: Medium. Handling and interpretation are low-risk, but X-ray systems produce ionizing radiation — a genuine hazard that is controlled entirely by the machine's engineered shielding and interlocks.

Professional Tips Before Starting

• Know what you are looking for before you look. Decide what defect you suspect — a void*, a bridge, an open, a* head-in-pillow — so you can read the image against a specific question rather than hoping something jumps out (10.2).
• Respect the 2D limit. A single flat view overlaps stacked features — when a call is ambiguous, tilt for an angled view or ask for computed tomography rather than guessing.
• Spend it where it counts. X-ray is costly and radiation-controlled — reserve it for the invisible joints on boards whose acceptance class justifies the certainty (10.1).

Seeing the Joints Nothing Else Can

What X-Ray Inspection Is

The tool exists to solve one specific problem: seeing a solder joint you cannot look at, because a package or the board itself is in the way. X-ray inspection works by passing a beam of X-rays through the entire assembly to a detector on the far side. X-rays penetrate matter, but not equally — denser, higher-atomic-number materials absorb more of the beam than light ones — and solder, being dense metal, absorbs far more than the plastic packages, the laminate, and the copper around it. So on the resulting image the solder stands out, imaged darker (more absorbing) against the lighter surroundings, and crucially it shows through the parts and board that hide it from the eye. That is the whole reason X-ray can do what a microscope cannot: a light microscope sees only the surfaces light can reach (10.3), while X-ray sees the solder itself wherever it is — under a package, inside a barrel, in the middle of a joint. A basic system produces a two-dimensional transmission image, a shadowgraph of the assembly's density, in real time as you move the board. What you are reading, then, is a density map: bright where little metal is in the path, dark where solder is thick. (That dark-for-solder look is the usual display convention, and the one this section uses; some systems invert it and show dense solder lighter, so what matters is the relative contrast — a void is always less absorbing than the solder around it — not the absolute shade.) Learning to inspect by X-ray is largely learning to read that density map — to recognize what a good hidden joint looks like and what each defect does to the image. At its core the idea is simple and powerful: make the solder visible through everything in front of it, so the joints you could never see become joints you can inspect.

When It Is Needed

X-ray is powerful but costly and radiation-controlled, so the skill is knowing exactly when a joint earns it — and the answer is: when the joint is invisible to every other method and the stakes justify the look. The clearest case is the area-array package. A ball-grid array carries its connections as a grid of solder balls underneath the body, and a QFN or other bottom-terminated part hides its pads beneath it — none of these joints can be seen by eye or microscope, and only some can be reached electrically (10.3; 10.4). If you have reworked one and must be sure it is good, X-ray is the primary way to see it. Beyond hidden position, there are hidden internal defects: a void trapped inside a joint or a plated barrel, the fill of a through-hole barrel (hole fill), a bridge or an open beneath a package, a head-in-pillow on a BGA ball, and the alignment or collapse of a whole ball field. Each of these is either fully buried or under a part, and each can fail a board that looks and even tests acceptable. But invisible alone is not enough to justify X-ray — the stakes must too. You weigh it against the acceptance class (10.1): a hidden joint on a Class 3 medical or aerospace board earns the certainty X-ray gives, while the same package on a throwaway consumer item may not warrant the cost. So the decision is a two-part test: is the joint genuinely unseeable by cheaper means, and does the product's criticality justify the expense and effort? When both are yes — a hidden joint that must be trusted — that is when X-ray is needed, and reaching for it any sooner or any later is a judgment error.

What It Reveals and Its Limits

Once you are looking at an X-ray image, the value is in reading it correctly — knowing both the defect signatures it shows and the one big limitation that can fool you. The signatures are learnable. A void appears as a lighter spot within the darker solder of a joint, because the gas pocket passes X-rays more freely than the metal around it — and a large or clustered void fraction is a defect. A bridge shows as solder joining two balls or pads that should be separate. An open or missing ball shows as an absent or malformed joint in an otherwise regular grid. A head-in-pillow shows as a ball that is separated from its paste by a visible gap or seam — the ball sitting on the paste rather than merged with it. Ball collapse, alignment, and the uniformity of a whole BGA field read directly from the pattern. And the fill of a plated barrel — how far solder rose up the hole — is visible in a way no surface view gives (hole fill). But here is the limit that must be respected: a plain, straight-down 2D X-ray is a density projection — it sums up everything in the beam's path into one flat image — so features stacked on top of one another overlap. A bridge on the far side of a ball can hide behind it; a defect on the bottom of a joint can be masked by the solder above; two things at different depths blur into one. A reader who forgets this can look straight at a defect and not see it. That single limitation — flat images overlap depth — is the reason the next tools exist, and knowing it is the difference between reading X-ray and misreading it.

2D, Angled, and CT

Because a flat image overlaps depth, X-ray offers ways to see around that limit, escalating from a simple tilt to a full three-dimensional reconstruction. The everyday mode is straight 2D transmission: the beam goes straight down through the board to the detector, giving a fast, real-time shadowgraph that handles most inspection — reading a BGA field, spotting an obvious bridge or void*, checking barrel fill.* When features stack and a 2D view is ambiguous, the first step up is an angled, or oblique, view: the system tilts the beam or the board so you look through the joint at an angle, which separates features that were on top of one another and lets you see the sides of a ball, the space between stacked balls, or a defect the straight view hid. Many manual systems can angle the view interactively, and it resolves a great many uncertain 2D calls. When even angled views are not enough — when you need to know exactly where inside a joint a defect sits, or to measure a void in three dimensions — computed tomography is the answer. A CT scan takes many X-ray images from many angles all around the assembly and reconstructs them into a full three-dimensional model you can slice and rotate, resolving stacked internal features that no single projection can separate. CT is the most powerful X-ray mode and also the slowest and most expensive, so it is reserved for the hardest questions and the most critical parts. The progression is the point: 2D for most work, an angled view to defeat overlap, and computed tomography when you truly need the three-dimensional truth — you escalate only as far as the joint and its stakes require.

Manual X-Ray Versus Automated AXI

X-ray shows up in two quite different roles, and it helps to know which one you are talking about. The first is manual X-ray — an operator-driven system, where a trained person loads a board, moves and tilts it, adjusts the view, and interprets the live image with judgment. This is the mode of the repair bench and the inspection lab: it is flexible, good for investigating a specific suspect joint, confirming a rework, or answering an ambiguous question with an angled view or a computed tomography scan. The reading is done by a human who knows what a good and bad joint look like. The second role is automated X-ray inspection, abbreviated AXI, on a production line: the system X-rays every board automatically as it passes through, compares each joint against programmed pass/fail criteria, and flags the failures without a person reading each image. AXI is about throughput and consistency — inspecting one hundred percent of high-density boards, especially their BGAs, at production speed — rather than the flexible investigation a manual system gives. For the repair and rework world of this volume, manual X-ray is the relevant mode: it is how a hidden rework gets verified and how a puzzling fault gets seen. Knowing AXI exists rounds out the picture — it is how the factories that built the board inspected it in the first place — but the tool you would actually use, or send a board out to, is a manual X-ray or CT system operated by someone trained to run it.

Access and the Practical Reality

The honest, practical truth of this section is that you almost certainly will not own an X-ray system — and that is fine, because the valuable skill is the judgment, not the machine. X-ray systems are expensive — far beyond the cost of the irons, hot-air tools, and microscopes of the earlier chapters — and they carry an ongoing burden a bench tool does not: radiation-safety compliance, operator training and authorization, dosimetry, licensing, and regulatory obligations. For the great majority of repair shops and individual technicians, owning one is neither justified nor practical. So the real-world path is to use someone else's: an X-ray inspection service, a shared lab or makerspace with a machine and a trained operator, a contract inspector, or the X-ray capability of a nearby manufacturer. You send the board out — or take it in — when a hidden joint genuinely must be verified, and you get back an image or a report. This is why the whole section is framed around the decision and the reading rather than operating a machine: the skill that actually serves you is knowing when a joint is invisible enough and critical enough to warrant X-ray (10.1), what it can and cannot show, and how to interpret the result someone hands you. A technician who can look at a BGA rework and say "this one must be X-rayed, here is what I expect it to show, and here is how I will read a void or a head-in-pillow*" is doing the professional part of the job* — the beam itself can be rented. Know when to reach for it and how to read it, and you have the part of X-ray that matters most, machine or no machine.

Common Mistakes

• Trusting a BGA rework you never verified. Its joints are invisible by eye and probe — a critical BGA rework needs X-ray, not a hopeful power-up (10.3; 10.4).
• Passing a joint on a single flat 2D view. Stacked features overlap in projection — tilt for an angled view or use computed tomography when a call is ambiguous.
• X-raying everything, or nothing. It is costly and radiation-controlled — reserve it for invisible joints whose acceptance class justifies it (10.1).
• Defeating a machine's shielding or interlock. Those controls are the entire reason it is safe — never bypass them, and leave operation to trained personnel.
• Thinking X-ray fixes the joint. It only shows the defect — the rework is still done by the earlier methods (Chapter 8).

Troubleshooting Guidance

X-ray questions trace to what you suspect, what the image shows, and its depth limit. If you cannot see a BGA's joints at all: that is exactly what X-ray is for — no other tool images them (10.3). If a lighter spot sits inside a joint's solder: that is a void — judge its size and fraction against the class (10.1). If two balls appear joined: a bridge — confirm with an angled view in case it is an overlap artifact. If a ball looks separated from its paste by a seam: a head-in-pillow — an open or intermittent joint. If a 2D call is ambiguous: tilt the board for an angled view, or escalate to computed tomography for a 3D answer. If you need the exact internal location or a measured void volume: that is a CT job. If you do not have a machine: send the board to an X-ray service — owning one is not required. If the board is not critical enough to justify the cost: the honest answer may be that X-ray is not warranted — weigh it against the acceptance class (10.1). The throughline: decide what you suspect, read the density map, respect the 2D depth limit, and escalate to angled or CT only as the joint and its stakes require.

Verification & Testing Methods

Use this as a decision-and-interpretation check, not an operating procedure:

• [ ] I can explain that X-ray inspection images hidden joints because dense solder absorbs more of the beam and shows through the package and board (10.3).
• [ ] I reach for X-ray when a joint is invisible to light and unreachable by probe — under a ball-grid array, inside a barrel — and the acceptance class justifies it (10.1).
• [ ] I can read the common signatures — a void as a lighter spot, a bridge, an open, a head-in-pillow, ball alignment, and barrel fill.
• [ ] I respect the 2D limit — a density projection overlaps stacked features — and use an angled view or computed tomography when a call is ambiguous.
• [ ] I know manual X-ray from automated AXI, and that most repairers use an X-ray service rather than owning a system.
• [ ] I treat the radiation safety as absolute — shielded, interlocked systems, never defeated, operated only by trained and authorized people.

Then try the practice exercises below — interpretation and decision practice; scenarios differ from the quiz.

Practice Exercises

1. Read the image (6 minutes, interpretation). Given described X-ray images of BGA joints — a lighter spot in the solder, two joined balls, a ball with a seam above its paste — name each defect (void, bridge, head-in-pillow) and say what it means (10.2).
2. Decide whether to X-ray (5 minutes, reasoning). For several hidden-joint situations across different product classes, decide which warrant X-ray and which do not, justifying each from invisibility and the acceptance class (10.1).
3. 2D versus CT (5 minutes, reasoning). For a case where a suspected bridge may be hiding behind a ball in a flat image, explain why a 2D view is insufficient and whether an angled view or computed tomography is the right next step.
4. Plan the outsource (4 minutes, applied). Describe how you would use an X-ray service for a critical BGA rework: what you would ask them to look for and how you would interpret their report.

These core ideas — what X-ray inspection is, when a hidden joint warrants it, the defect signatures and the 2D limit, angled views and computed tomography, and the radiation-safety rules — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

• X-ray inspection images hidden joints by passing X-rays through the assembly — dense solder absorbs more and shows darker, right through the package and board — so it sees what a light microscope (exposed surfaces only) and a probe cannot (10.3; 10.4).
• Reach for X-ray when a joint is genuinely invisible and unreachable — under a ball-grid array or inside a barrel — AND the acceptance class justifies the cost; it reveals voids, bridges, opens, head-in-pillow*, ball alignment, and barrel fill* (10.1).
• A flat 2D X-ray is a density projection, so stacked features overlap and can hide — use an angled/oblique view to separate them, and computed tomography for a full 3D reconstruction when 2D is not enough.
• Manual X-ray is the operator-driven inspection-and-rework mode; automated AXI inspects every board on a line — and most bench repairers do not own a system, they use an X-ray service, so the skill is the decision and the reading.
• X-rays are ionizing radiation, made safe only by a cabinet system's shielding and interlocks — never defeat them, follow radiation-safety rules and dosimetry, and leave operation to trained, authorized people (Chapter 2).

Skills Learned

• You can now explain how X-ray images a joint hidden under a package.
• You can now decide when a hidden-joint situation warrants X-ray for its class.
• You can now interpret common X-ray defect signatures and 2D limits.
• You can now explain when angled or CT imaging is needed over flat 2D.
• You can now state the radiation-safety rules for operating an X-ray system.

Glossary Additions

• X-ray inspection — a radiographic inspection method that passes X-rays through an electronic assembly to image features that light cannot reach; because dense material such as solder absorbs more of the beam than the surrounding plastic and laminate, the solder images darker and shows through the package and board, making visible the hidden joints under area-array parts (such as a ball-grid array), the fill inside a plated barrel, and voids buried within a joint. A basic system gives a two-dimensional density projection (a shadowgraph); angled views and computed tomography extend it to resolve features stacked in the beam. It is the primary way to inspect joints a light microscope, which sees only exposed surfaces, cannot.
• head-in-pillow — a hidden ball-grid-array solder defect in which the package ball and the solder paste beneath it each melt during reflow but fail to coalesce into one joint, leaving the ball resting on top of the paste "pillow" separated by a thin gap or oxide seam rather than fused to it. The result is an open or intermittent connection that typically looks fine from the outside and can pass a casual check, so it is detectable in practice only by X-ray (where the ball appears separated from its paste) or by cross-sectioning the joint.
• computed tomography — X-ray computed tomography (CT), an advanced X-ray inspection technique that captures many X-ray images of an assembly from many different angles and computationally reconstructs them into a full three-dimensional model that can be sliced and rotated; this resolves internal features stacked on top of one another in the beam that a single flat two-dimensional X-ray, being a density projection, cannot separate. CT is the most capable and detailed X-ray mode and also the slowest and most costly, so it is reserved for the hardest inspection questions — locating a defect precisely inside a joint, or measuring a void in three dimensions — on the most critical assemblies.

Suggested Next Sections

Must read next:

• What Is a PCB? — this completes the soldering, rework, and inspection arc of Volume 3; the next volume turns to the board itself — how a printed circuit board is built up from copper and laminate, how its traces, vias, and planes carry signals and power, and how to repair the board rather than the parts on it.

Recommended:

• Electrical Verification After Rework — the electrical half of verification that proves function and reaches un-probeable nets, which X-ray complements by imaging the hidden joint itself.
• Visual Inspection Criteria — the joint-quality criteria X-ray extends to the joints an eye can never see.