Board camera evaluation checklist for inspection systems

A board camera can produce a recognizable image and still fail an inspection requirement. The missing detail may come from insufficient sampling, an unsuitable lens, a cropped output mode or a connector entering the optical path. Receiving controls can also matter before the camera reaches the bench.

Use this checklist to qualify a complete camera module, not just its sensor specification. It separates documented requirements, measured image performance and receiving evidence so you can make an acceptance decision before committing the module to a machine.

1. Define acceptance criteria before requesting samples

Start with the inspection decision: what must the system detect, measure or distinguish? A general requirement such as “high image quality” cannot support a repeatable acceptance test.

Record the following in a qualification sheet:

• [ ] Smallest relevant feature: specify its size, contrast and permitted variation. A dark scratch and a low-contrast surface mark need different tests.
• [ ] Required field coverage: include part position variation, not just the nominal part outline.
• [ ] Operating conditions: state working distance, part motion, exposure constraints and lighting arrangement.
• [ ] Image delivery: define output dimensions, pixel format, acquisition rate and permitted dropped frames.
• [ ] Mechanical configuration: identify the lens, mounting hardware, mating connectors and installed cable routing.
• [ ] Receiving controls: define packaging, identification and documentation requirements separately from image performance.

Give each criterion a pass condition and an evidence field. “Connector clearance acceptable” is weaker than “no obstruction or shading across the required image area with the specified mating cable installed.”

For the architecture decision that precedes qualification, use the board-level camera selection guide. This checklist begins once you have a candidate configuration to evaluate.

2. Reconcile active, total and delivered pixel counts

Different pixel counts do not automatically indicate an error. Sensor documentation can distinguish total array dimensions, active imaging dimensions and border or reference regions. A camera may deliver a smaller region, or software may resize the image after acquisition.

• [ ] Identify each published count: label it as total sensor array, active imaging area, configured region or delivered image dimensions.
• [ ] Record the acquisition mode: capture the region of interest, binning, subsampling and scaling settings where supported.
• [ ] Inspect the decoded image: confirm its width, height, pixel format and row stride rather than relying on a preview window.
• [ ] Resolve unexplained differences: request documentation identifying which pixels contribute scene information.

For a hypothetical module, a total array of 1640 × 1240 pixels might contain an active region of 1600 × 1200 pixels. The difference is 40 columns and 40 rows overall. That arithmetic does not establish how those pixels are distributed or what they do; only the documented array map can establish that.

An image enlarged from 800 × 600 to 1600 × 1200 contains more output samples, but it does not recover the scene detail absent from the original capture.

If the software representation is unclear, follow image bytes versus pixel data before interpreting buffer size as resolution.

3. Measure field coverage and visible detail together

A pixel count becomes useful only when related to the field of view. Measure coverage with a target at the actual inspection plane, using the intended lens, focus position and image mode.

• [ ] Measure both image axes: record field width and height at the required working distance.
• [ ] Include positioning margin: place the target at the permitted extremes of part location.
• [ ] Test representative features: use actual defects or a justified target with comparable contrast and orientation.
• [ ] Check center and edges: passing at the image center does not establish performance across the field.

For an approximately planar scene viewed close to perpendicular, average object-space sampling can be estimated as follows:

Suppose your hypothetical output is 1600 × 1200 pixels and the measured field is 80 × 60 mm. Horizontal sampling is:

Assume your acceptance requirement calls for at least 6 pixels across that feature. This is a project requirement, not a universal detection threshold. The required image width becomes:

If positioning margin increases the required field width to 88 mm, the requirement rises again:

This calculation screens candidates; it does not prove feature visibility. Lens contrast, focus, motion blur, noise and image processing still affect the decision. The small-pixel camera selection guide explains why additional samples alone may not produce usable detail.

For measurement tasks, also distinguish sampling from measurement accuracy using camera resolution for dimensional inspection. Verify local scale where perspective or distortion makes the average sampling estimate inadequate.

4. Test connector clearance in the assembled configuration

A product illustration cannot establish whether the installed assembly clears the optical path. The mating plug, cable boot and mounting hardware may occupy more space than the board connector itself.

• [ ] Fit the actual mating connectors: include cable boots, adapters and strain relief in the test assembly.
• [ ] Check the complete image: inspect all corners and edges, not just the central inspection region.
• [ ] Exercise permitted cable positions: look for intermittent obstruction as the cable moves within its intended routing limits.
• [ ] Check focus access: confirm you can adjust and secure the lens without loading nearby components.
• [ ] Check mechanical loading: provide cable support so connector forces do not shift the camera or lens.

Capture a uniformly illuminated matte target with and without the suspect cable or bracket present, where removal is permitted by the documented procedure. A repeatable image change helps distinguish mechanical intrusion from lens shading or illumination variation.

Do not assume a large-looking lens has a particular angle of view. Coverage depends on focal length, active sensor dimensions and imaging geometry. Repeat the clearance test whenever the lens or output region changes.

For lens fitting, use the C-mount and CS-mount compatibility guide. Carrier pinout and electrical compatibility require a separate review, covered in CSI camera carrier board compatibility when that interface applies.

5. Select samples without assuming qualification

Use board-level cameras and the camera finder to shortlist documented interfaces, shutter types and lens mounts. A suitable catalog filter is a starting point, not an acceptance result.

The LS-BCM-0056 has a GigE interface, a C-mount lens mount and rolling shutter. Its qualification should include the intended motion and exposure conditions.

The LS-BCM-0047 has a USB3 interface, a C-mount lens mount and global shutter. Evaluate host compatibility and image delivery in the actual acquisition configuration.

For an optical candidate, the LS-FFL-1582 is a C-mount lens. Confirm sensor coverage, focus range and installed clearance before including it in the accepted assembly.

Compare other machine vision lenses through the lens finder. None of these listings establishes your required packaging method, serial-label arrangement or application-level inspection performance; confirm those separately.

6. Define ESD packaging and receiving traceability

Packaging acceptance belongs in the purchase requirements, not in an informal assumption made after delivery. A protective-looking package is not evidence that its ESD properties meet your receiving process.

• [ ] Separate protection from sealing: an ESD-protective package and a tamper-evident seal address different requirements.
• [ ] Specify identifier behavior: state whether you need individual serial numbers, lot identification or both.
• [ ] Define machine-readable content: if a code is required, agree on its data format and confirm the receiving scanner reads it.
• [ ] Inspect before opening: record visible damage, incorrect identification or missing closure evidence.
• [ ] Handle in an appropriate ESD-controlled area: follow your organization's documented handling process.

Do not infer packaging performance from color or appearance alone. If receiving evidence is missing, quarantine the item under your quality procedure and request clarification. An item that powers up has not thereby demonstrated compliance with packaging requirements.

7. Repeat the test and preserve the evidence

Qualification should produce an evidence package that another person can reproduce. Save original captures alongside the configuration that generated them, rather than retaining only screenshots or annotated examples.

• [ ] Record identity: associate each image set with the unit identifier and hardware revision where available.
• [ ] Record settings: include exposure, gain, pixel format, image dimensions, lens aperture and working distance.
• [ ] Record the physical arrangement: include target position, lighting arrangement and installed cable routing.
• [ ] Repeat acquisition transitions: test startup, acquisition stop and restart through documented procedures.
• [ ] Retest the operating extremes: check required part positions, lighting conditions and motion states.

An illustrative pilot might use 3 units, 10 acquisition start-stop cycles per unit and 100 saved frames at each selected condition. Choose quantities from your risk assessment. Such a pilot can expose configuration and repeatability problems, but it does not establish a population reliability figure.

Classify failures by cause: documentation mismatch, sampling shortfall, optical limitation, mechanical intrusion, acquisition failure or receiving nonconformance. This prevents a lens problem from being treated as proof that the camera electronics are unsuitable.

8. Make the acceptance decision explicit

Use separate decisions for imaging performance, integration and receiving compliance. Passing one does not compensate for failing another.

The usual answer—selecting more pixels—is wrong when detail is lost through defocus, motion blur, low contrast or obstruction. Likewise, a satisfactory image does not resolve missing traceability or undocumented packaging.

Before approval, complete this final gate:

• [ ] Required field coverage and feature visibility pass together.
• [ ] Delivered pixel counts and acquisition modes are documented.
• [ ] Lens, connectors and cables pass in their installed positions.
• [ ] Acquisition repeats without unexplained failures under the agreed test.
• [ ] Packaging and identification meet the written receiving requirements.
• [ ] Evidence, deviations and configuration are attached to the acceptance record.

Approve the tested configuration, not an unrestricted family of possible substitutions. A lens change, different cable assembly or revised image mode may require partial requalification.

Start a candidate configuration with the camera configurator. For unresolved packaging, identification or sample-selection requirements, send the acceptance checklist through technical support before committing to the configuration.