Key takeaways
- Mark the critical characteristics on the drawing. An inspector cannot measure every dimension on every sampled part, so the drawing must say which ones decide the lot.
- Agree the sampling plan per characteristic: attribute sampling for go/no-go gauges, variables sampling where actual values are recorded, and often zero defects for critical fits.
- A gauge is only evidence if it is calibrated and can resolve the tolerance. Check the certificate, the due date and a zero or check-standard reading before measuring.
- Record actual values, not ticks. Readings show a drift towards a limit that a pass mark hides.
What a dimensional inspection covers
Engineering parts are bought on a drawing. A dimensional inspection answers one question: do the parts in this lot match that drawing, within its tolerances? It has three inputs:
- The drawing and its revision. The inspector first confirms the factory is working to the revision on your purchase order. Sub-contracted machining shops often hold an older print or a sketch with key sizes only.
- Critical characteristics. The fits, bores, threads, sealing faces and hole patterns that decide whether the part works. Mark them on the drawing, ideally ballooned and numbered.
- The general tolerance. Every untoleranced dimension takes the class in the title block, such as ISO 2768-mK for machined parts. Cast dimensions normally follow the grade named under ISO 8062-3. If the drawing cites no general tolerance, untoleranced sizes have no agreed limit.
The first parts off new tooling need a fuller check than a production lot. A first article inspection verifies every dimension and note on one representative part, usually at a pre-production inspection. Lot inspections then sample the characteristics that matter.
Sampling: attributes, variables and zero-defect characteristics
A lot of 5,000 flanges cannot all be measured. The sampling plan decides how many parts are checked and what result accepts the lot. Agree it per characteristic before the visit:
| Approach | How it works | Use it for |
|---|---|---|
| Attribute sampling (ISO 2859-1) | Each sampled part is conforming or not. The count of nonconforming parts is compared with the accept and reject numbers for the lot size, inspection level and AQL. | Go/no-go gauges: thread plugs and rings, pin gauges, snap gauges; visual checks of surfaces and marking |
| Variables sampling (ISO 3951-1) | Actual values are measured and the lot is judged from their mean and spread against the specification limit. It needs fewer parts than attribute sampling for the same protection, but assumes a stable process and roughly normal data for one characteristic. | A diameter, a length or a hardness reading recorded on every sampled part, on long runs from one process |
| Zero defects or 100% | A fixed number of parts with acceptance on zero nonconforming, or every part checked. | Safety and assembly-critical features, parts from a new tool, rework lots |
Two practical rules. First, draw the sample from every cavity, die, fixture or machine that made the lot, because each source has its own offset. Second, a sample that is large for visual checks may be impractical for a 40-feature dimensional layout; many buyers set a smaller fixed number of parts for full dimensions and the AQL sample for critical ones. For fasteners, ISO 3269 sets an acceptance inspection procedure for use where the purchaser and supplier have no prior agreement. See AQL sampling explained, the inspection level and the AQL calculator.
Gauges at the factory, and how they are checked
Measurements are taken with the factory's own calibrated instruments, by the inspector or by the factory's quality staff while the inspector watches and records. Clear To Load does not bring a coordinate measuring machine or metrology lab to the factory.
| Feature | Typical instrument | What the inspector checks |
|---|---|---|
| Outside diameters, lengths, steps | Vernier or digital caliper, outside micrometer | Resolution suits the tolerance; zero checked; reading taken at more than one position |
| Bores | Bore gauge, plain plug gauge, inside micrometer | Size and roundness at both ends of the bore |
| Threads | Thread plug and ring gauges, go and no-go | Gauge marked with the thread size and class on the drawing; go enters fully, no-go does not |
| Heights, hole positions, flatness | Height gauge and surface plate, dial indicator, feeler gauges | Datums taken as the drawing defines them |
| Hole patterns and fits on assemblies | Buyer's or factory's fixture or check gauge | Gauge approved by the buyer and in good condition |
| Surface finish | Roughness comparator plates or the factory's profilometer | Ra values on the drawing; a comparator is a screen only |
Before any reading, the inspector checks each instrument's calibration certificate, its due date, and that it traces to a recognised calibration lab. A common rule of thumb is that an instrument should resolve about a tenth of the tolerance band; a 0.02 mm caliper is not evidence for a ±0.01 mm bore. Parts are measured after they have cooled, because ISO 1 sets 20 °C as the reference temperature for dimensional specifications, and a freshly machined or heat-treated part is larger while warm.
What changes for castings, forgings and machined parts
| Product | Dimensional points that cause most rejections |
|---|---|
| Iron castings | Wall thickness and core shift, machining allowance left on faces to be machined, cast dimensions outside the ISO 8062-3 grade, porosity that opens up on machined faces |
| Steel forgings | Die mismatch across the parting line, flash remnants after trimming, draft angles, machining stock |
| Brass precision parts | Thread form and class, burrs in cross holes, plating thickness affecting thread fit |
| Flanges and pipe fittings | Bolt circle and hole positions, facing type and finish, bore matched to the pipe schedule |
| Fasteners | Thread fit with go/no-go gauges, head height and across-flats, length |
| Engine and transmission components | Ground diameters drifting as the wheel wears, run-out, spline fit, results read against the PPAP dimensional record |
| Industrial valves | End-to-end length, flange drilling, stem and seat dimensions |
| Hand tools | Spanner openings and socket sizes on go/no-go gauges |
| Structural steel fabrication | Overall length, hole positions for bolted connections, squareness |
Indian engineering clusters include Rajkot (castings, forgings, pumps), Kolhapur (iron foundries), Jamnagar (brass parts), Ludhiana (bicycle parts and hand tools), Coimbatore (pumps and motors) and Pune (auto and engineering parts).
CMM reports, material and NDT: what is witnessed
Some checks need equipment or qualifications an inspector does not carry. They are witnessed or reviewed, and the report says so.
- CMM and profile tolerances. Where the factory has a coordinate measuring machine, the inspector selects the parts and watches the program run, then checks the printout against the drawing. Otherwise the parts can be sealed and sent to an accredited metrology lab.
- Material grade. The inspector reads the EN 10204 3.1 certificate against the order and matches heat numbers to marks on sampled parts. Where you order PMI, the inspector witnesses the factory's analyser on parts the inspector picks. See mill test certificates and EN 10204.
- NDT. Ultrasonic, magnetic-particle or dye-penetrant testing is done by the factory's qualified technician and witnessed; the procedure, calibration block and acceptance level are checked against the order. See NDT.
- Hardness and mechanical tests. Witnessed at the factory's test bench, or specimens sealed for an ISO/IEC 17025-accredited lab through lab testing coordination.
Clear To Load does not issue or co-sign EN 10204 3.2 certificates, and is not yet accredited: ISO/IEC 17020 accreditation is in progress.
What to put in your purchase order
- Drawing number and revision, with critical characteristics marked and numbered.
- The general tolerance standard and class for machined and cast features.
- Sampling plan per characteristic: AQL and inspection level, variables plan, or a fixed number with zero defects.
- Sampling across every cavity, die or machine, identified on the parts or the tags.
- The gauges to be used, including any buyer-supplied check fixtures, and that they must be in calibration.
- Whether actual values must be recorded, and in what format.
- Certificate type (3.1 or 3.2), and the witnessed tests you require.
Clear To Load runs these checks at during-production and pre-shipment inspections anywhere in India, priced by the man-day. The report arrives within 24 hours and ends in one verdict: Cleared to load, Hold or Not cleared. Start with castings, forgings and fasteners inspection or book an inspection.
Sources
- ISO 2859-1, Sampling procedures for inspection by attributes, Part 1 (iso.org)
- DIN ISO 3951-1:2024-08, adopting ISO 3951-1:2022, sampling procedures for inspection by variables (DIN Media)
- ISO 3269:2019, Fasteners: acceptance inspection (Standard Norge listing)
- ISO 2768-1:1989, General tolerances, Part 1 (iso.org)
- ISO 8062-3:2023, tolerances for castings (iso.org)
- ISO 1:2022, GPS standard reference temperature (BSI listing)