X-Ray PCB Inspection: What You Get and What It Proves

X-ray PCB inspection shows you what’s under a BGA, inside a via, and between the copper layers — without cutting the board apart. You get 2D transmission images, oblique-angle views, and on multilayer work a set of per-layer images, plus a written report on solder joint condition, voiding, shorts, opens and internal routing. It’s the fastest way to answer “is that joint good?” or “where does that net go?” on a board you can’t afford to destroy.

This page is written for the engineer holding the board: someone chasing an intermittent BGA failure, verifying a first build, or trying to trace nets on a six-layer assembly with no documentation. It covers what happens in the lab, what the images can and can’t prove, and what we need from you to quote.

What X-ray PCB inspection actually shows you

X-rays pass through FR-4 and copper at very different rates. Solder — mostly tin, with lead in older assemblies — absorbs more than laminate, so joints show up as bright, dense shapes against a much fainter board background. Everything you learn from an X-ray image comes from that density difference.

In practice, here’s what an operator is looking at on screen:

• BGA and QFN solder joints. Ball shape, ball diameter consistency, collapse, head-in-pillow signatures, bridging between adjacent balls, and voiding inside the ball. Voids read as dark circles inside a bright sphere.
• Through-hole barrel fill. How far solder actually wicked up a plated through hole on a wave or selective-solder board — visible in an oblique view, not straight down.
• Hidden connectors and shields. Joints under RF cans and press-fit terminations you can’t reach with a scope.
• Internal copper and vias. Buried and blind vias, plane cutouts, thermal reliefs, and inner-layer trace runs when the outer copper isn’t too dense to see through.
• Component internals. Die size and bond wire pattern inside a plastic package, which is often enough to flag a remarked or substituted part.
• Cracks and delamination. Barrel cracks and separation show as thin dark lines when the geometry cooperates. They don’t always cooperate — more on that below.

2D, oblique, and 3D

Straight-through 2D imaging is the default and covers most questions. Oblique views — tilting the board or the detector 30 to 60 degrees — separate features that overlap in a top-down shot, which is how you see whether a via is filled or whether two stacked joints are actually touching. Computed tomography reconstructs a volume from hundreds of projections and gives you true cross-sections at any depth. CT is slow and expensive per board, so we reserve it for cases where a 2D answer isn’t good enough: a suspected crack inside a single ball, or a stackup question on an HDI board.

How X-ray fits into board copy and reverse engineering work

On a reverse engineering job, X-ray does two jobs that nothing else does cheaply. First, it maps buried and blind vias so the net list closes properly instead of ending at a pad that seems to connect to nothing. Second, it lets us confirm inner-layer geometry before we commit to destructive layer separation, so the grinding sequence is planned rather than exploratory.

It also helps identify what’s on the board. A package with a sanded-off marking still has a distinctive die and bond wire signature inside, which narrows the candidates enough to make an electrical guess testable. That feeds directly into building a bill of materials from a finished assembly, especially when the customer’s board is a decade old and half the markings are illegible.

The limitation to understand up front: X-ray images copper as density, not as a layer. On a four-layer board with sparse inner routing, a top-down image often reads clearly enough to trace. On a ten-layer board with two solid ground planes, the planes swamp everything behind them and you get almost nothing useful from transmission imaging alone. That’s when layer separation, not more X-ray time, is the answer. Our guide to using X-ray in reverse engineering goes deeper into where the crossover sits.

When to use X-ray and when something else is cheaper

What happens to your board in the lab

The process is non-destructive and, in the normal case, the board comes back to you in the condition it arrived.

1. Intake and photography. Both sides shot under diffuse light, serial or date codes recorded, any pre-existing damage noted so nothing gets attributed to us later.
2. Scoping. We agree on what you want answered — specific components, a whole-board sweep, or net tracing. This drives the imaging plan and the price.
3. Fixturing. The board is held flat in the stage. Warped boards and tall components limit how close the detector can get, which limits magnification, so we sometimes remove a heatsink or shield with your permission.
4. Parameter setup. Tube voltage and current are tuned per area. A thick heavy-copper power section needs far more penetration than a thin RF section, so a single setting rarely serves the whole board.
5. Imaging. Top-down pass first, then oblique views on anything ambiguous, then CT slices if the job calls for it. Each frame is logged with position and settings.
6. Review and measurement. Void area is measured as a percentage of ball area. Ball diameter and pitch are measured against the datasheet and against neighbors on the same package. Findings are marked on the images.
7. Report. Annotated images, measured values, a plain-language finding per item, and a recommendation.

What we grade against

Where an acceptance criterion is called for, we reference the standard the industry actually uses: IPC-A-610 for assembly workmanship and solder joint acceptability, IPC-A-600 for the bare board, and IPC-7095 for BGA design and inspection practice including void assessment. If your program has its own limits — a customer-specific void cap, for instance — send them and we’ll grade to yours instead of the default. We’ll say which document a call was made under so you can audit it.

What to send us, and what comes back

To get a real answer instead of a range, send this:

• Clear photos of both sides of the board on a dark background, whole board in frame, in focus, no flash glare.
• Board dimensions and thickness, plus layer count if you know it.
• The markings on the parts you care about — the BGA, the suspect connector, whichever device is at issue.
• What you need to know. One line is enough: “verify all four BGAs for voids and bridging” or “trace the buried vias under U7.”
• Quantity. One failure sample and a fifty-piece sampling plan are very different jobs.

Back from us: a feasibility statement (can X-ray answer your question, or do you need cross-section or layer separation), a lead time, a price, and a note on anything we’d want your approval for — removing a shield, for example. Deliverables are the annotated image set, the measured data, and the written report; if the work feeds a copy project, the imaging integrates with the rest of the file package we hand over on a board copy.

What drives the price

Number of components to image, magnification required (0.4 mm pitch needs more than 1.0 mm pitch), whether oblique or CT imaging is needed, board size relative to the stage, and sample count. A single-BGA check on a small board is quick. A full-coverage sweep of a 300 mm backplane with 2,000 joints is a different order of work. We quote per job rather than per square inch because the honest cost tracks the questions, not the area.

Honest limits — what X-ray can’t prove

This is the part most vendor pages skip. Knowing where the technique stops saves you from paying for an answer you won’t get.

Thin cracks can be invisible. A crack whose plane runs along the beam direction produces almost no density change. Oblique views help, CT helps more, but a negative X-ray result does not prove a joint is uncracked.

Cold joints often look fine. X-ray sees geometry, not metallurgy. A ball that wetted poorly but sits in the right shape can image as acceptable. Head-in-pillow is detectable when the interface is visible as a seam; it isn’t always.

Dense copper blocks the view. Solid planes, thick copper power sections, large heatsinks and metal shields all reduce contrast on whatever sits behind them. Sometimes the fix is removing the shield; sometimes there is no fix.

Overlap confuses double-sided assemblies. Components moun

Related reading

• Sep 2026Free PCB Sample Evaluation | Feasibility, Risk & Price
• Sep 2026PCB Copy Quote — Priced Answer in One Working Day
• Sep 2026Contact PCB Reverse Engineering Company | Engineers