Inspection and test

This is the job: read the requirement, plan the check, pull the sample, measure, grade the finding, identify the material, and dispose of it under authority. GD&T items are Analyze-level. Slow down on those.

III.A Drawings, GD&T, MBD (model-based definition) Apply / Analyze

A print is a legal contract in most shops. Title block (name, number, revision, material, default tolerances, scale, units). Revision / change block. Notes (heat treat, finish, NDT (nondestructive testing), “break edges”). Bill of materials. Views: orthographic, section, detail, auxiliary, isometric. Model-based definition (MBD) puts the same information on a 3D model instead of (or in addition to) a 2D sheet. You still need datums, tolerances, and a way to inspect them.

Default tolerances and title-block traps

If a dimension has no tolerance, the title-block block tolerance for that decimal place applies. A dimension written 1.00 is not the same as 1.000. Count the places.

ASME Y14.5 (the U.S. dimensioning and tolerancing standard) ideas you must own

Y14.5 is the U.S. dimensioning and tolerancing standard the BoK (Body of Knowledge) names. Core ideas:

• Feature. A physical portion of a part (hole, slot, face, pin).
• Feature of size. A feature with an opposed size dimension (hole, tab, slot width, pin diameter).
• Datum. A theoretically exact plane, axis, or point. The datum feature is the real surface you set up on.
• Datum reference frame. Three mutually perpendicular planes derived from the datum features in the feature control frame (typically primary, secondary, tertiary — 3-2-1 contact idea).
• Feature control frame. The rectangular box: symbol | tolerance | datum references.
• MMC (maximum material condition) / LMC (least material condition) / RFS (regardless of feature size). Maximum material condition (smallest hole, largest pin). Least material condition (largest hole, smallest pin). Regardless of feature size (no bonus).
• Bonus tolerance. Extra geometric tolerance available when a feature of size departs from MMC (or LMC, if specified that way). Bonus = how far the actual size is from the MMC size, for that feature.

Position of a pattern, perpendicularity of a pin, flatness of a face, profile of a surface, circular runout of a shaft — know the symbol and what setup it implies. See the GD&T field card in this site.

Part alignment: primary datum establishes the first plane (typically three-point contact on a surface), secondary constrains rotation (two-point), tertiary stops the last motion (one-point). You do not “level it until it looks good” if the drawing named A, B, and C.

III.B Sampling Apply

100% inspection is expensive and still misses things when people fatigue. Sampling inspects a subset and makes a lot decision. It accepts the risk of being wrong in both directions.

• AQL (Acceptance Quality Limit). The poorest quality level (as a percent nonconforming or defects per hundred units) considered acceptable as a process average. It is not a “guarantee that the lot is better than AQL (acceptance quality limit).” It is the quality level that will be accepted most of the time by the plan.
• Producer’s risk (α). Chance of rejecting a good lot.
• Consumer’s risk (β). Chance of accepting a bad lot.
• LTPD (lot tolerance percent defective) / RQL (rejectable quality level). A poor quality level that should be rejected most of the time.
• Random sampling. Every piece in the lot has an equal chance. Do not grab the top layer.
• Lot size N, sample size n, acceptance number Ac (or c), rejection number Re (rejection number). Find n defective. If defectives ≤ Ac (acceptance number), accept. If ≥ Re, reject. On single sampling Re is usually Ac + 1.

ANSI/ASQ Z1.4 (the civilian cousin of MIL-STD-105E) is the classic attribute plan family: inspection levels (I, II, III), code letters from lot size, switching rules among normal / tightened / reduced. C=0 plans accept the lot only if the sample has zero defectives — popular with some customers, steeper OC curve at low n. The exam wants you to know the vocabulary and how a plan is used, not to memorize every table cell. Bring the tables if your handbook has them.

III.C Inspection planning and processes Apply / Analyze

Types

Inspection errors

Bias, fatigue, flinching (calling a borderline good because you do not want the fight), distraction, rushing, bad setup, cosine error on an indicator, parallax on a dial, using the wrong end of a go/no-go, measuring a dirty part, measuring a hot part. Name the error and the source. That pairing is in the 2025 BoK.

Traceability

Lot number, heat number, serial number, date code, barcode, RFID (radio-frequency identification), FIFO (first in, first out), shelf life / age control. If a gage or a material later fails, you need to know which product it touched.

Nonconforming product

Identify it immediately: tag, label, quarantine location. Do not leave a rejected shaft on the accept pallet. Severity:

• Critical — likely to cause unsafe conditions or prevent performance of a major function.
• Major — likely to cause failure or material reduction of usability.
• Minor — not likely to reduce usability much; departure from standard that is a workmanship issue.

Disposition: rework (to print), repair (not fully to print, needs approval), reprocess, reinspect, scrap, return to supplier, use-as-is / customer waiver. MRB (material review board) decides. You record.

III.D Testing methods

Nondestructive (NDT) Understand

The part survives. Visual, optical, dye / liquid penetrant (surface cracks on non-porous metal), magnetic particle (surface and near-surface on ferromagnetic parts), ultrasonic (internal), eddy current (conductivity, cracks, thickness, sorting), radiography / X-ray (internal volume), profile. Know which method finds which class of defect. Penetrant will not find a subsurface void. Mag particle will not work on aluminum.

Destructive Apply

Tensile (strength, yield, elongation), force-to-failure, drop test. You sacrifice a sample to learn about the lot.

Functional Apply

Does it do the job: leak, torque-to-turn, compression, tension under service load.

Hardness Apply

Do not convert scales with a casual chart and call it data unless a standard conversion is specified. Support the part. Mind minimum thickness.

III.E Software for test equipment Remember

Treat test software as part of the measuring system. Safeguard it (access control, version). Run functional checks. Compare results against a known standard or a second method. Know which parameters the program is actually evaluating. Validation: does this version, on this machine, still measure what you think it measures after a change.