Metrology and quality control

In short

Metrology is the science of measurement — defining units, building standards, selecting instruments and estimating how much a measured value can be trusted. Quality control is the use of those measurements to decide whether a part, a batch or a process is acceptable, and to keep the process centred and stable. Together they decide whether interchangeable manufacture actually works: a shaft made in Pune must fit a housing made in Rajkot without filing. The topic carries steady marks in GATE (Production/Manufacturing section), ISRO, BARC and PSU exams, and is the most common "shop-floor" interview area for quality, inspection, NPD and tool-room roles at Maruti Suzuki, Bosch, Tata Motors, Bharat Forge, Sundram Fasteners, L&T, HAL, BHEL and every Tier-1 auto supplier.

Explained

Fundamentals: standards, accuracy and error

• Standards hierarchy: the metre is defined from the speed of light; CSIR-NPL, New Delhi keeps India's national standards, and traceability flows national → reference → working → shop-floor gauge. Calibration labs are accredited by NABL to IS/ISO/IEC 17025.
• Line standards carry graduations (steel rule, scales); end standards measure between two faces (slip gauges, end bars) and are far more accurate. Wringing makes slip gauges adhere by molecular attraction, not grease.
• Reference temperature is 20 °C (ISO 1 / IS 2102); all length standards and tolerances apply at that temperature.
• Accuracy = closeness to the true value; precision/repeatability = closeness of repeated readings to each other. A set of readings can be precise and still wrong (bias).
• Sensitivity = output change per unit input change; resolution/least count = smallest readable change; hysteresis = different reading going up versus coming down; reproducibility = agreement between different operators/instruments.
• Systematic errors (zero error, calibration error, temperature, parallax, cosine error from misalignment, elastic deformation of part and anvils, support/sag error) are correctable; random errors are only reducible by averaging.
• Abbe's principle: the measuring scale should lie along the line of the dimension being measured — micrometers obey it, vernier calipers do not, which is why a micrometer is more accurate.
• Uncertainty combines contributions in quadrature; a measuring instrument should be about 10 times more accurate than the tolerance it checks (the "10:1" or gauge-maker's rule; 4:1 is the practical minimum).

Limits, fits and tolerances (IS 919 / ISO 286)

• Basic (nominal) size is the common size from which limits derive. Tolerance = upper limit − lower limit. Deviation = limit − basic size; the fundamental deviation is the one nearer the basic size and fixes the position of the tolerance zone.
• Unilateral tolerance deviates one way only (25.00 +0.02/0.00) — preferred for gauging; bilateral deviates both ways (25.00 ± 0.01).
• Tolerance grade (IT) fixes the *size* of the zone: grades IT01, IT0 and IT1 upward (ISO 286 runs to IT18; editions differ in the count). The standard tolerance unit i = 0.45∛D + 0.001D microns, D in mm being the geometric mean of the diameter step, gives IT5 = 7i, IT6 = 10i, IT7 = 16i, IT8 = 25i, IT9 = 40i, IT10 = 64i, IT11 = 100i.
• Fundamental deviation letters: A–ZC for holes, a–zc for shafts (28 positions). H hole has lower deviation zero; h shaft has upper deviation zero; js/JS are symmetric. For the clearance shafts, FD = −5.5 D^0.41 for "f", −11 D^0.41 for "e", etc.
• Three classes of fit: clearance (shaft always smaller — H7/g6, H8/f7, H11/c11), transition (may clear or interfere — H7/k6, H7/js6), interference (shaft always bigger — H7/p6, H7/s6, H7/u6).
• Hole basis (H) is preferred in India (IS 2709) because holes are made by fixed-size tools — drills, reamers, broaches — and the shaft is easier to adjust; shaft basis suits drawn bar and long shafts carrying several components.
• Maximum material limit (MML): the condition with most material — largest shaft, smallest hole. Least material limit (LML) is the opposite. Allowance = intended difference = minimum clearance (positive) or maximum interference (negative).
• Selective assembly sorts parts into sub-groups when the required fit is tighter than economical tolerances (piston–pin, bearing grading).

Gauging and inspection of size, angle, form and texture

• Limit gauges inspect without measuring: GO gauge must enter, NO-GO must not. Types: plug (holes), ring and snap/gap (shafts), taper, thread, radius, feeler, screw-pitch.
• Taylor's principle: the GO gauge checks all related dimensions simultaneously at the MML (so it is full-form and full-length — a plug gauge of full cylinder length), while the NO-GO gauge checks one dimension at a time at the LML (so it is short, pin-type or two-point). Otherwise a bent or lobed hole passes.
• Gauge tolerance: IS 3477 practice puts the gauge tolerance inside the work tolerance zone (unilateral disposition) so no bad part is accepted; textbook rule is gauge tolerance = 10% and wear allowance = 5% of work tolerance, the wear allowance added to the GO gauge in the direction of wear.
• Comparators measure deviation from a slip-gauge setting, not absolute size: mechanical (dial indicator ~100:1, Johansson Mikrokator with twisted strip up to ~5,000:1, sigma comparator), mechanical-optical (Zeiss Ultra-optimeter, ~1,000–4,000:1), electrical/electronic (LVDT, inductive, capacitive — easy to amplify, record and feed to SPC), pneumatic (Solex back-pressure or flow type, up to ~10,000:1, non-contact, excellent for small bores and in-process gauging).
• Angles: bevel protractor (LC 5′), sine bar with slip gauges (h = L sin θ), angle gauge blocks (combined by addition and subtraction), spirit level/clinometer, autocollimator (arc-second resolution, used for straightness and squareness of machine tools).
• Flatness and surface plates by NPL flatness interferometer and optical flats: each dark fringe with monochromatic light represents a height change of λ/2 (≈0.294 µm for sodium light, λ = 0.5893 µm).
• Straightness by straight edge and feeler, wedge/autocollimator, or taut wire; roundness by V-block, bench centres or a dedicated roundness tester, with the circle fitted as LSC, MZC, MCC or MIC (least squares, minimum zone, min circumscribed, max inscribed).
• Surface texture (IS 3073, symbols per ISO 1302): Ra = arithmetic mean deviation of the profile (centre-line average, µm) — the parameter quoted in India; Rq rms; Rz mean of the largest peak-to-valley heights of five sampling lengths; Rmax; Rsk skewness. Measured by stylus profilometer (Talysurf), optical/laser profilometer, or compared against roughness specimen blocks. Typical Ra: turning 1.6–6.3, milling 0.8–3.2, grinding 0.1–0.8, lapping 0.012–0.1 µm.
• Screw threads: elements are major, minor and effective (pitch) diameter, pitch, angle, form. Pitch diameter by two- or three-wire method on a floating carriage micrometer, or by thread micrometer; pitch by pitch-measuring machine; form by profile projector; functional check by a GO/NO-GO thread gauge per IS 4218 / ISO 965.
• Gears: tooth thickness by gear tooth vernier (chordal) or constant-chord method, base tangent (span) micrometer, rolling test on Parkinson gear tester for composite error, runout, pitch error, profile and lead by gear-measuring machine.
• CMM (coordinate measuring machine): moving-bridge, cantilever, gantry or horizontal-arm, with touch-trigger or scanning probes; builds features from point clouds and evaluates GD&T directly. Advantages: one setup, any geometry, digital records; limitations: cost, temperature-controlled room, probe compensation and operator skill. Vision systems and laser scanners do non-contact and in-line inspection.
• GD&T (IS 8000 / ISO 1101, ASME Y14.5): 14 characteristics — form (straightness, flatness, circularity, cylindricity), profile (line, surface), orientation (parallelism, perpendicularity, angularity), location (position, concentricity, symmetry), runout (circular, total). Datums fix the inspection setup; MMC modifier permits bonus tolerance; true position tolerance is a circular zone of 57% more area than an equal ± coordinate box (zone diameter = √2 × the box side), so it is cheaper for the same function.

Statistical quality control and process capability

• QC detects defects in the product (inspection, gauging, sorting); QA builds the system that prevents them (procedures, audits, training, documentation). Inspection may be 100%, sampling, first-piece, patrol, in-process or final; destructive tests force sampling.
• Chance (common) causes are inherent and random; assignable (special) causes — tool wear, a new batch of material, a changed setting, a different operator — shift the process and are what control charts are designed to catch.
• Control charts: variables — X̄–R (n = 2–9), X̄–s, individual-moving range; attributes — p (fraction defective), np (number defective), c (defects per unit), u (defects per unit area). Limits are ±3σ of the plotted statistic, drawn from the process itself; specification limits never go on a control chart.
• Out-of-control rules: a point outside the limits, seven or more consecutive points on one side of the centre line, a run of seven rising or falling, cycles, or hugging the limits.
• Process capability compares the natural spread (6σ) with the tolerance: Cp ignores centring, Cpk penalises it. Cp requires a stable chart first; a capable but uncontrolled process is a fluke. Indian automotive practice (IATF 16949 PPAP) typically asks Cpk ≥ 1.33 ongoing and 1.67 for short-term/special characteristics.
• Six Sigma targets a tolerance half-width of 6σ with 1.5σ drift allowed, i.e. 3.4 defects per million opportunities, through the DMAIC cycle (Define, Measure, Analyse, Improve, Control).
• Measurement System Analysis / Gauge R&R: repeatability (equipment variation) plus reproducibility (appraiser variation); %GRR under 10% acceptable, 10–30% conditional, above 30% unacceptable, with at least 5 distinct categories. Also bias, linearity, stability; attribute gauges are studied by kappa/cross-tab.
• Acceptance sampling (IS 2500 Parts 1–3, aligned to ISO 2859/3951): single, double, multiple and sequential plans defined by lot size, sample size n and acceptance number c, selected by AQL and inspection level. The OC curve shows probability of acceptance versus incoming quality; producer's risk α is rejecting good lots, consumer's risk β is accepting bad lots; LTPD is quality at β = 0.10; AOQL is the worst average outgoing quality under rectifying inspection.

Quality management, tools and Indian regulation

• Seven basic QC tools: check sheet, histogram, Pareto chart (80/20 — attack the few big causes), cause-and-effect/Ishikawa fishbone (6M: man, machine, material, method, measurement, mother-nature), scatter diagram, control chart, stratification or flow chart.
• Cost of quality: prevention + appraisal (control costs) versus internal failure (scrap, rework) + external failure (warranty, recall, goodwill). The 1-10-100 rule — a defect caught in-process, at final inspection, or at the customer.
• Improvement systems: PDCA (Deming cycle), Kaizen, 5S, poka-yoke (mistake-proofing fixtures and sensors), SMED, TPM, QFD (house of quality), 8D problem solving, root-cause with 5-Why.
• FMEA: Severity × Occurrence × Detection = RPN in the classic form; the current AIAG-VDA handbook replaces RPN ranking with an Action Priority (High/Medium/Low) table, so quote whichever your syllabus uses.
• Standards and law in India: ISO 9001:2026 QMS (risk-based thinking, process approach, PDCA; succeeded IS/ISO 9001:2015 on 16 September 2026, three-year transition) with IATF 16949 for automotive, ISO 14001 and ISO 45001 alongside; BIS Act 2016 governs the ISI mark, hallmarking and Quality Control Orders/CRS; Legal Metrology Act 2009 with the Packaged Commodities Rules 2011 governs weights, measures and declarations in trade, enforced by state Legal Metrology departments (periodic verification and stamping of weighbridges, fuel dispensers, shop balances). Penalties and fee schedules are revised from time to time — check the current rules rather than quoting a figure.

Key points

• Metrology - science of measurement, its standards, instruments and uncertainty; quality control - accepting or rejecting and keeping the process centred.
• Interchangeability - parts made independently assemble without fitting; it exists only because tolerances and gauges exist.
• Tolerance - UL − LL, always positive; deviation - signed distance of a limit from basic size.
• IT grade - size of the tolerance zone (i = 0.45∛D + 0.001D µm); letter - position of the zone.
• H7/g6 reading - capital = hole, small = shaft; H means hole lower deviation = 0.
• Hole basis - preferred, because hole size is set by standard tools.
• MML/LML - most/least material condition; GO gauges work at MML, NO-GO at LML.
• Taylor's principle - GO checks all dimensions together and is full form; NO-GO checks one dimension at a time.
• Wear allowance - 5% of work tolerance added to the GO gauge only; gauge tolerance (10%) lies inside the work zone.
• Comparator - amplifies a deviation from a slip-gauge datum; pneumatic gives the highest magnification without contact.
• Sine bar - h = L sin θ; accurate below about 45°, poor above it because sensitivity falls.
• Ra - mean absolute deviation of the roughness profile in µm, India's standard texture parameter.
• Two-wire method - E = M − 3d + 0.866p for a 60° metric thread; best wire d = 0.57735p.
• CMM - probes points and computes features and GD&T in one setup; needs 20 °C control.
• True position - a circular/cylindrical tolerance zone, 57% more area than the equivalent ± box.
• Cp vs Cpk - spread only vs spread and centring; Cpk ≤ Cp always, equal only when perfectly centred.
• Control limits - from the process (±3σ); specification limits - from the designer. Never mix them.
• Gauge R&R - %GRR < 10% accept, 10–30% conditional, > 30% reject the measurement system.
• AQL, α, β, LTPD - sampling-plan risk language; the OC curve is the plan's fingerprint.

Formulas and examples

1. Fit calculation, Ø50 H8/f7. Diameter step 30–50 mm → D = √(30 × 50) = 38.73 mm.

• i = 0.45 × 38.73^1/3 + 0.001 × 38.73 = 0.45 × 3.384 + 0.0387 = 1.56 µm.
• IT8 = 25i = 39 µm; IT7 = 16i = 25 µm (matches the IS 919 table).
• FD for shaft f = −5.5 × 38.73^0.41 = −5.5 × 4.479 = −24.6 ≈ −25 µm.
• Hole 50 H8 = 50.000 / 50.039 mm; shaft 50 f7 = 49.975 / 49.950 mm.
• Max clearance = 50.039 − 49.950 = 0.089 mm; min clearance (allowance) = 50.000 − 49.975 = 0.025 mm → clearance fit, typical of a running journal.

2. Plug-gauge design for a Ø25 H7 hole (25.000 / 25.021). Work tolerance = 21 µm.

• Gauge tolerance = 10% = 2.1 µm; wear allowance = 5% = 1.05 µm.
• GO gauge sits at the low limit (MML) + wear allowance = 25.00105, zone inside the work tolerance → 25.0011 / 25.0032 mm, full cylindrical form, length ≥ hole depth.
• NO-GO gauge sits at the high limit (LML), zone inside → 25.0189 / 25.0210 mm, short pin-type, checked on one diameter at a time (Taylor).

3. Sine bar set to 7° 30′, 200 mm centres.

• h = 200 × sin 7.5° = 200 × 0.130526 = 26.105 mm.
• Slip-gauge stack from an M87 set: 20 + 3 + 1 + 1.1 + 1.005 = 26.105 mm (five blocks, largest first, all wrung).
• Sensitivity: a 0.01 mm stack error gives dθ = 0.01/(200 × cos 7.5°) = 5.04 × 10⁻⁵ rad = 10.4 arc-seconds — so the same error at 60° would give about 20″, which is why sine bars are not used above ~45°.

4. Two-wire check of an M20 × 2.5 external thread.

• Best wire d = 0.57735 × 2.5 = 1.443 → use 1.45 mm wires; micrometer over wires reads M = 20.505 mm.
• E = M − 3d + 0.866p = 20.505 − 4.350 + 2.165 = 18.320 mm.
• Basic pitch diameter = 20 − 0.6495 × 2.5 = 18.376 mm, so the thread is 56 µm below basic — inside the 6g band (about 18.33 / 18.16 mm; read exact limits from IS 4218 / ISO 965 tables). Accepted.

5. SPC and capability on a turned shaft, spec 25.00 ± 0.05 mm. 25 subgroups of n = 5 give x̄̄ = 25.012 mm, R̄ = 0.040 mm.

• σ̂ = R̄/d₂ = 0.040/2.326 = 0.0172 mm.
• X̄ chart: 25.012 ± 0.577 × 0.040 → 24.989 / 25.035 mm. R chart: UCL = 2.114 × 0.040 = 0.0846 mm, LCL = 0.
• Cp = 0.100/(6 × 0.0172) = 0.97; Cpk = min[(25.05 − 25.012)/0.0516, (25.012 − 24.95)/0.0516] = min[0.74, 1.20] = 0.74.
• Expected rejection: z_upper = 0.038/0.0172 = 2.21 → 1.36%; z_lower = 3.60 → 0.02%; total ≈ 1.4% (≈14,000 ppm), almost all oversize.
• Action: re-centre the tool offset by −0.012 mm to make Cpk = Cp = 0.97, then reduce σ (rigidity, tool wear control) to reach the Cpk ≥ 1.33 the customer wants.

Where it is used

• Automotive and Tier-1 suppliers: in-process gauging and SPC on engine bores, crankshafts and gears (Maruti Suzuki, Tata Motors, Bosch, Bharat Forge, Sundram Fasteners), with PPAP, Gauge R&R and Cpk ≥ 1.33 as supply conditions under IATF 16949.
• Tool rooms and standards rooms: slip gauges, sine bars, profile projectors and CMMs for die, jig and fixture acceptance at CIPET, MSME tool rooms (Indo-German Tool Room, CTTC) and HAL/ISRO vendor shops.
• Aerospace and defence: HAL, BrahMos, DRDO labs — GD&T-driven CMM inspection, first-article inspection reports, full traceability to NPL.
• Calibration and testing laboratories: NABL-accredited labs issuing ISO/IEC 17025 certificates for industry instruments; roles such as calibration engineer and metrologist.
• Legal metrology and consumer protection: verification and stamping of weighbridges, LPG and fuel dispensers and retail balances by state Legal Metrology inspectors under the Act of 2009.
• Job roles: quality inspector, QA/QC engineer, CMM programmer, SQA/supplier quality engineer, Six Sigma Green/Black Belt, reliability and NPD engineer.

Interview questions

• Q: What is the difference between tolerance and allowance? A: Tolerance is the permissible variation on one dimension (UL − LL); allowance is the intended difference between mating parts — minimum clearance in a clearance fit, maximum interference in an interference fit.
• Q: Why is hole-basis system preferred? A: Hole sizes are fixed by standard tools such as drills, reamers and broaches, so it is cheaper to keep the hole standard and vary the shaft, which can be turned or ground to any size.
• Q: State Taylor's principle and its practical consequence. A: The GO gauge must check all related dimensions at once at the maximum material limit, so it is full-form and full-length; the NO-GO gauge checks one dimension at a time at the least material limit, so it is short and two-point. Otherwise a lobed or bent hole would pass a GO plug.
• Q: Why is wear allowance given only to the GO gauge? A: The GO gauge rubs the whole surface of every accepted part and wears towards the part, so extra metal is provided; the NO-GO gauge should not enter and hardly wears.
• Q: Accuracy versus precision — give an example. A: A micrometer with a 0.03 mm zero error gives tightly grouped but wrong readings — precise, not accurate. Accuracy is closeness to truth; precision is closeness of repeats.
• Q: A process has Cp = 1.6 but Cpk = 0.8. What is wrong and what would you do? A: The spread is fine but the mean is badly off-centre; shift the process mean (tool offset, machine setting) towards the target — no variance reduction is needed.
• Q: Can specification limits be plotted on a control chart? A: No. Control limits come from the process variation of the plotted statistic; specification limits apply to individual parts. Mixing them invites wrong reactions to normal variation.
• Q: Why does a micrometer read more reliably than a vernier caliper? A: It obeys Abbe's principle — the measuring screw is in line with the dimension — and has a constant-force ratchet, so there is less wear, play and operator force error.
• Q: What does a %GRR of 35% tell you? A: The measurement system itself consumes 35% of the variation, so the data cannot be trusted for judging the process; fix the gauge, fixture or method before drawing any conclusion about capability.
• Q: How do you measure the effective diameter of a screw thread? A: By the two- or three-wire method on a floating carriage micrometer: measure over the best-size wires and compute E = M − 3d + 0.866p for a 60° metric thread.
• Q: Why is true-position tolerancing better than ± coordinate tolerancing? A: It gives a round zone matching the function of a bolt in a hole, is 57% larger in area than the equivalent ± coordinate box, and keeps the datum reference explicit, so fewer good parts are rejected.
• Q: Which Indian law covers the weighbridge at a plant gate? A: The Legal Metrology Act 2009 and its rules — the weighbridge must be verified and stamped by the state Legal Metrology department and re-verified periodically; separately, calibration for internal quality use is done by a NABL-accredited lab.

Common mistakes

• Calling the hole "H7/g6" the shaft and vice versa — capital letters are always holes, small letters always shafts.
• Using the nominal diameter instead of the geometric mean of the diameter step in i = 0.45∛D + 0.001D, and then reporting a tolerance that does not match the IS 919 table.
• Computing maximum clearance from max hole − max shaft; it is max hole − min shaft.
• Placing the gauge tolerance outside the work tolerance zone, which lets out-of-spec parts be accepted.
• Making a short GO plug gauge or a full-form NO-GO gauge — a direct violation of Taylor's principle.
• Confusing Ra with Rz, or quoting Ra without the sampling length and cut-off.
• Reporting Cp or Cpk from an unstable process, or from data taken with an unvalidated gauge; capability is meaningless before the chart is in control and the Gauge R&R is acceptable.
• Ignoring temperature: measuring a 500 mm steel part at 30 °C and not applying ΔL = 500 × 11.7 × 10⁻⁶ × 10 = 0.059 mm to reach the 20 °C reference value.

Related topics

Jigs and fixtures · Machining and machine tools · Grinding and finishing processes · CNC machining and CAM · Engineering drawing and GD&T · Material testing and non-destructive testing