← 返回资讯
2026-11-14约 34 分钟阅读责任主体:芯芸达

What Does a CT Voxel Size Tell an Export Buyer?

Read CT voxel pitch with field of view,resolution and task-specific inspection evidence. Work through declared grid and storage calculations before agreeing the report package.

A supplier offers industrial computed tomography with a voxel size of 10 micrometers. Your drawing calls for control of a 40-micrometer internal feature. The arithmetic seems reassuring. Four voxel pitches fit across the feature. The purchase decision still needs evidence about the material, part geometry, imaging conditions, analysis procedure and acceptance rule.

A voxel size describes the sampling grid of a reconstructed three-dimensional image. Spatial resolution, defect detection and dimensional measurement each answer a further question. An overseas buyer can make a CT quotation much more useful by asking which inspection task has been demonstrated, under which conditions, and which records will accompany production parts.

Start with the decision the scan must support

Consider three requests for the same component. Engineering wants the distance between two internal surfaces. Quality wants to screen a region for pores above an agreed criterion. A failure-analysis team wants to understand the shape of an unexpected cavity.

Each request needs a different evaluation. A useful report should identify the feature, the measurement or classification to be made, and the decision that follows. A small voxel number alone leaves those requirements open.

The distinction also appears in current metrology research. NIST's June 22, 2026 literature-review abstract separates artifacts used to assess spatial resolution from artifacts designed for defect-detection studies. That is a practical reason to ask what a demonstration artifact was designed to establish. A resolution target and a representative internal flaw serve different purposes. NIST review of fabricated XCT artifacts

For a buyer, the first line of the inspection request could read “Locate and report the specified internal voids in region A under the agreed evaluation procedure.” A dimensional request would identify the actual dimensions, datum arrangement and applicable tolerances. Those statements give the supplier something more specific to demonstrate.

Read voxel size as a grid specification

A reconstructed CT volume contains small volume elements called voxels. In a simple cubic grid, the voxel edge length defines the distance represented by one cell along each axis. The total sampled extent depends on both that pitch and the number of cells.

ZEISS's technical note explains that voxel size depends on detector sampling and imaging geometry. It also describes additional contributors to image quality, including source blur, stability and imaging artifacts. The document comes from an instrument manufacturer, so its explanation can inform a technical question while each commercial performance claim remains attributable to that manufacturer. ZEISS technical note on resolution parameters

Here is an explicitly fictional calculation. Suppose a reconstructed width of 12 millimeters is divided into 1,200 cubic cells along that axis. The corresponding pitch is 0.01 millimeters, or 10 micrometers. A 40-micrometer feature spans four pitches in the idealized coordinate grid.

That calculation establishes a geometric sampling relationship. Accepting the inspection still requires a demonstration of how the feature appears and how the stated procedure identifies or measures it. Detection probability, dimensional uncertainty and classification error rates each require their own measured evidence.

Ask the supplier to identify whether a quoted value describes the proposed whole-part scan, a smaller region of interest, or a best-case instrument configuration. The relevant number is the one attached to the actual inspection plan.

Match the field of view to the production part

A component's envelope and mounting arrangement affect the usable imaging configuration. An attractive demonstration on a small specimen can leave the production part's accessible field of view unresolved.

NIST's AMB2022-01 benchmark page records a specific example in which specimen size and detector coverage constrained the achievable voxel edge length. Different legs and an extracted sliver were imaged at different pitches. The example shows why a sample description belongs beside the acquisition settings. Its numerical settings apply to that experiment. NIST AMB2022-01 benchmark measurements

A practical RFQ therefore identifies the production part's material, maximum external dimensions, relevant wall or penetration path, and the internal region under review. Include the drawing revision and a marked image locating that region. If a supplier proposes scanning a cut specimen, record the permitted preparation and the relationship between that specimen and the delivered component.

This also helps compare quotations. One provider may price a whole-part scan. Another may price a selected region with different preparation and handling. A comparison table should expose those differences before the buyer compares the headline voxel values.

Give detection and measurement their own evidence

Spatial resolution concerns an imaging system's ability to distinguish nearby features under stated conditions. Detection asks whether a specified feature is identified using a defined procedure. Dimensional measurement adds a numerical result and its measurement context.

The following questions translate those distinctions into a purchasing discussion.

Inspection objectiveEvidence to requestDecision the buyer can review
Distinguish neighboring structuresResolution method, target and acquisition conditionsWhether the demonstration addresses the required separation
Identify internal voidsRepresentative flaw types, sizes, locations and decision procedureWhether the demonstrated detection task matches the production request
Measure an internal dimensionFeature definition, calibration or reference route, analysis settings and uncertainty evaluationWhether the reported result supports the agreed dimensional decision
Investigate an unexpected featureImage data, region coordinates, rendering and interpretation notesWhether engineering can inspect the observation and its context

These are proposed procurement checks. The table assigns a review question to each task. Actual acceptance belongs in the engineering and quality agreement for the component.

A rendered image can help explain the result. Retaining the underlying region coordinates and analysis settings helps the next reviewer find the same feature. Where the contract requires reproducibility, define the data format, available image volume, software dependencies and permitted access before ordering the work.

Review a detection claim at the stated operating conditions

NIST researchers studied acquisition parameters, image quality and probability of detection for additive-manufacturing defects in a 2019 paper. The authors explicitly describe their subsequent probability-of-detection analysis as expository. They explain that a more rigorous study would require independently established defect sizes with appropriate metrological traceability and characterized uncertainties. A buyer should preserve that qualification when using the paper. NIST-hosted acquisition-parameter and detection study

For a supplier demonstration, ask how the reference features were established. Record the inspected material and geometry, feature type, location, scan configuration and analysis procedure. A claim about pores in one material needs a defined basis before it is used to accept another material or a different defect shape.

If a quotation includes a detection probability, request its associated feature-size definition, population, decision threshold, study method and uncertainty or confidence statement. A percentage separated from those conditions gives the buyer little basis for applying it to a new part.

The same discipline applies to an automatic classification feature. Identify what the software classifies, which configuration was used, and what happens to ambiguous results. Arrange the demonstration around an authorized sample set and a declared evaluation rule.

A finer grid also changes the data volume

Another fictional calculation helps procurement teams plan deliverables. Assume a cubic reconstruction with a single unsigned 16-bit value per voxel, stored densely without compression or file overhead.

ReconstructionVoxel countRaw value storage
1,024 × 1,024 × 1,0241,073,741,8242,147,483,648 bytes, or 2 GiB
2,048 × 2,048 × 2,0488,589,934,59217,179,869,184 bytes, or 16 GiB

Doubling the cells along all three axes multiplies the cell count by eight. For a fixed physical extent, that corresponds to halving the grid pitch in this idealized example. The calculation describes raw storage under the stated assumptions. Actual delivered files depend on cropping, compression, metadata and additional data products.

This is a reason to agree on the handover package. An overseas engineering team may need a report and selected images. Another may require the reconstructed volume and a compatible reader. Put those deliverables, transfer arrangements and retention period into the quotation so the file can be used after the inspection is complete.

Write a capability page that can answer a buyer's question

An English supplier page can connect the headline CT capability to a short, controlled inspection description. Identify the component families and materials within the published scope. Explain the information needed for feasibility review. Give a route for submitting the relevant drawing and inspection objective through the company's approved inquiry process.

A useful public example can include a clearly labeled demonstration part, the inspected region, the acquisition conditions relevant to the claim and a sample report excerpt that the company is authorized to publish. Keep the engineering approval and publication permission attached to that material internally.

For manufacturers working with Xindar, the practical content task is to turn approved technical information into readable English pages and a clear inquiry path. The manufacturing service page provides the commercial starting point. Component acceptance and published technical limits require the manufacturer's approved evidence. Xindar manufacturing and export website services

Use a short acceptance exercise before buying recurring inspections

  1. Identify one real part revision and one inspection decision. Mark the relevant region and define the feature or dimension to be evaluated.
  2. Ask the supplier to propose the acquisition and analysis procedure for that part. Retain the proposed sampling pitch with its field of view and conditions.
  3. Agree on the reference method or representative artifacts used in the demonstration. Record which conclusions that evidence supports.
  4. Review the resulting report and data package against the agreed decision rule. Resolve ambiguous results through the declared engineering process.
  5. Confirm the recurring production scope, change-control process and delivery records. Revisit the demonstration when a material, geometry or method change affects its applicability.

This sequence produces a reviewable connection between the requested task, the demonstration and the production record. It also gives the English website team precise information to publish when describing the capability.

Questions overseas buyers ask

Can voxel size be used to estimate how many grid pitches span a feature?

Yes. Divide the feature's stated length by the pitch using consistent units. Interpret the result as a sampling calculation, then assess the relevant detection or measurement evidence separately.

Which voxel value should appear in the quotation?

Ask for the proposed value for the specified part, inspected region and acquisition configuration. Include whether the work covers the whole part or a selected region.

What should accompany a claimed detection probability?

Request the feature-size definition, representative population, operating conditions, decision procedure and study support. Those details determine whether the claim can be applied to the intended inspection.

Which CT records help an overseas engineering review?

A useful package identifies the part revision, region, acquisition and analysis conditions, feature results, applicable decision rule and agreed data files. The exact package follows the contract and engineering purpose.

Sources and calculation scope

Official sources were reviewed on November 12, 2026. The NIST 2026 review is cited at abstract level, the 2019 study within its stated methodological limits, and the ZEISS document as manufacturer technical guidance. The grid and storage examples are fictional calculations. They establish the arithmetic under the stated assumptions. Source access details and calculation checks are retained in the editorial record.