An X-ray inspection system is a capital purchase and a long technical commitment. The specification that matters is rarely the one printed largest in the brochure, and the questions below are the ones that separate a system that fits the application from one that merely fits the budget.
The first five cover core performance and capability. The last five cover safety, reliability, and return on investment.
1. What resolution and power does the system actually deliver?
Tube voltage and current determine penetration and flux, but focal spot size is the specification that drives resolution. A smaller focal spot produces less geometric unsharpness, which is what allows fine features to stay sharp as magnification increases. A large spot blurs the edges of every feature regardless of what the detector is capable of.
The detector side is measured in microns per pixel. Neither number means much alone. Effective system resolution comes from the combination of focal spot, detector pixel pitch, and the magnification geometry the system can reach, and that combination is what determines the smallest defect that can be resolved.
Ask for demonstrated performance on a known resolution standard rather than a specification sheet figure.
2. What board types and materials can it handle?
Thickness, copper weight, and overall board size all affect whether a system can produce a usable image. Heavy copper and multilayer construction absorb aggressively, and a system that images a two-layer board cleanly may not penetrate a thick board with heavy internal planes.
There is a second question hiding inside this one. Fitting a board inside the cabinet is not the same as being able to inspect all of it. Usable inspection area is typically smaller than physical load area, and the difference concentrates at edges and corners. Verify that the system can reach every region that needs inspection, including at the oblique angles required for viewing under components near a board edge.
3. Does it support 3D or computed tomography inspection?
2D imaging handles surface voids and a large share of routine inspection. Buried vias, internal cracks, and layer misalignment often require volumetric data. High-density interconnect work in particular tends to push toward CT.
This question is worth answering honestly rather than aspirationally. CT costs acquisition time and throughput, and a system specified around it when the application does not need it is an expensive mismatch in the other direction.
4. What defects can it detect, and how accurate is it?
Ask for real sample images, and ask for them on parts resembling yours rather than on a demonstration board chosen to show the system at its best. Solder voids, bridges, QFN misalignment, via cracks, and head-in-pillow defects each stress the system differently.
Then ask for false reject and false accept rates, and ask under what conditions those rates were measured. A false accept rate ships defects. A false reject rate consumes operator time re-reviewing good parts, which is the cost that quietly erodes the throughput number in the next question.
5. What throughput and level of automation does it provide?
Boards per hour is only meaningful when quoted for an inspection recipe resembling the one you will actually run. Best-case throughput on a simple program tells you little about performance on a dense board with many regions of interest.
Automatic pattern recognition, programmed inspection routines, and inline integration all change real-world productivity substantially. Sample handling matters here too, since move and settle time frequently exceeds exposure time in production inspection, making cycle time a motion problem as much as an imaging one.
6. How does the software handle data and traceability?
Inspection data has value beyond the pass or fail decision. Determine whether the system generates annotated reports with images, serial numbers, and lot numbers, and whether that data can flow into an MES or QMS for yield tracking.
Also ask about export formats and interfaces. Data locked inside a proprietary viewer is difficult to use for process improvement, and traceability is what closes the loop between finding a defect and preventing the next one.
7. What are the maintenance and uptime expectations?
Ask about tube life and what happens at end of life, since a closed tube is replaced as a unit while an open tube can be serviced and rebuilt. That distinction changes both the cost and the downtime profile over the life of the system.
Cover detector calibration frequency, the preventive maintenance schedule, and mean time between failures. Spare parts availability and service coverage in your region matter as much as the MTBF figure, because a stable system with a long parts lead time still produces long outages.
8. What about safety and compliance?
Confirm built-in shielding, safety interlocks, and operator training. Cabinet X-ray systems sold in the United States are regulated under 21 CFR 1020.40, which sets the emission limit alongside requirements for interlocks, key-actuated controls, and indicators showing when X-rays are being generated. CE marking applies for European deployment.
Many US states also require registration of radiation-producing equipment separately from the federal standard, so confirm what your jurisdiction expects before the system arrives. There is no flexibility available on any of this.
9. What is the total cost of ownership?
Purchase price is the beginning of the calculation. Power consumption, consumables, software licensing, training, service contracts, and facility requirements all accumulate over a system lifetime measured in years.
The return side deserves equal specificity. Ask what yield improvement or scrap reduction is realistic for an application like yours, and account for inspection currently being outsourced, defects currently reaching customers, and engineering time currently spent on destructive analysis.
10. What is the vendor’s track record and upgrade path?
Ask which industries the vendor serves. Aerospace, medical, and automotive customers impose qualification and documentation requirements that leave a mark on how a company builds and supports its equipment.
Then ask what the system can become. Whether it can be upgraded to computed tomography, extended with automation, or updated with new inspection algorithms determines whether this purchase serves you in five years or has to be replaced. A vendor whose product line has a path forward is a different proposition from one selling a fixed configuration.
The point of asking
These questions are not about extracting concessions from a vendor. They are about arriving at a specification that matches the inspection problem, which is what produces consistent results after installation rather than a system that images impressively during a demo and disappoints in production.