Table of Contents
A human hair is about 70 micrometers. Some engineering parts need tolerances 10x tighter — down near 7 µm — which is the difference between a good part and one that fails in the field. That gap sounds small, but it decides whether bearings run quietly, whether a hip implant fits, or whether an engine crankshaft meets life targets. This guide walks through the notation you see on drawings, the three classic fit types, how surface finish changes effective tolerance, a short primer on GD&T, and how to think about measurement uncertainty. If you ever read a print and wondered why a dimension says “Ø20.00 ±0.02” or why an aerospace drawing calls out positional tolerance instead of +/- limits, this article explains the why and how with industry examples and practical checks.
1Tolerance notation: ±, bilateral and unilateral
Tolerance notation is the first language between design and shop. A plus-minus (±) tolerance is symmetric: Ø20.00 ±0.02 means the part diameter can be 19.98–20.02 mm. Bilateral tolerances allow different plus and minus values (for example +0.03/−0.01). Unilateral means tolerance only in one direction (for example 10.00 +0.00/−0.05). Reading limits instead of shorthand helps avoid errors. A drawing that lists limits (19.98–20.02) removes ambiguity about interpretation and rounding.
Plus-minus, bilateral and unilateral
Plus-minus is quick and common. Bilateral tolerances give more control when a process can add material both ways but one side is more critical. Unilateral is often used when only material removal is possible or when mating features must not exceed a single boundary, such as a bore that must not be larger than the mating pin.
Interpreting limits and decimal places
Decimal places imply measurement resolution. If a drawing shows 20.0 mm and a tolerance of ±0.1, the implied measurement resolution is lower than a drawing that lists 20.000 ±0.005. Always match inspection equipment resolution to the decimal and tolerance on the print.
2Fits: clearance, interference and transition
Fit selection determines how two parts behave when assembled. Clearance fits always leave space, interference fits require force or shrink-fit, and transition fits can be either depending on actual dimensions. The ISO system (ISO 286) gives tables of nominal-size-based tolerances and lettered holes/shafts to standardize selection.
ISO fit system and selecting a fit
ISO uses letter-number codes (for example H7/g6) to define tolerance zones relative to the nominal. H7/g6 for a shaft/hole combination might give a light running fit. Use tables to compare minimum and maximum clearances or interferences. For critical assemblies, simulate thermal expansion and stack tolerances before locking in a fit.
Practical examples: bearings, shafts and press fits
Automotive rotating assemblies often use clearance or transition fits for serviceability; bearings may require an interference fit on the shaft and a light press in the housing. A shrink-fit (heat the housing or cool the shaft) is common in aerospace for high-strength joints without welding.
3Surface finish and roughness
Surface finish isn't decoration — it changes the functional size and contact behavior. Ra (arithmetical average) and Rz (average max height) are common metrics. A rough surface can make a nominally tight tolerance act like a looser one because peaks and valleys change contact points and effective clearance.
Ra, Rz, lay and why they matter
Ra is simple to measure and widely used, but Rz or other parameters sometimes correlate better with sealing or bearing life. Lay indicates directionality of surface pattern and affects friction. For example, a sealing surface often needs a specific finish to avoid leaks even if dimensional tolerance is tight.
How finish affects tolerance stacking
When you stack many tolerances in an assembly, surface roughness and form error add to the effective variation. A nominally flat surface with ±0.01 mm flatness but roughness of 0.004 mm could reduce contact area or change preload calculations. Include surface finish in stack-ups where contact mechanics matter.
4GD&T basics (Geometric Dimensioning and Tolerancing)
GD&T moves from linear plus-minus to functional controls: perpendicularity, concentricity, position, profile and more. Instead of limiting size only, GD&T controls form, orientation and location relative to datums. That makes requirements clearer for manufacturing and inspection when used correctly.
Common symbols and datums
Symbols you’ll see often: position (⌀ with tolerance), flatness, perpendicularity (right angle symbol), concentricity, profile of a surface. Datums (A, B, C) establish reference surfaces. A positional tolerance referenced to three datums constrains feature location in a controlled way.
When to use GD&T vs traditional limits
Use GD&T when function depends on orientation, location or form rather than just size. Aerospace and medical drawings often prefer GD&T because it encodes how a feature must behave in assembly and inspection. However, GD&T needs a team that understands it; misapplied symbols can confuse shops rather than help them.
5Measurement uncertainty and real-world practice
Inspection results are measurements with uncertainty. Combine instrument error, calibration uncertainty, environmental factors and operator variability when you state a measurement. The combined standard uncertainty is the square root of the sum of squared individual standard uncertainties; expanded uncertainty multiplies that by a coverage factor (often k=2 for ~95% confidence).
Uncertainty math: combined and expanded
If you have independent standard uncertainties u1, u2, u3, the combined standard uncertainty uc is uc = sqrt(u1^2 + u2^2 + u3^2). The expanded uncertainty U = k * uc; k≈2 gives about 95% confidence. Use this when you compare inspection results to tolerances: a part at the edge of tolerance might be indistinguishable from out-of-tolerance if uncertainty is large.
Calibrations, gage R&R and common mistakes
Regular calibration against traceable standards reduces uncertainty. Gage R&R studies quantify repeatability and reproducibility. A famous mistake to remember: the Mars Climate Orbiter (1999) failed partly due to a units mismatch in software, a reminder that small measurement/interpretation errors can cause massive failures. In inspection, the #1 shop error is using a measuring device with worse resolution than the drawing requires.
Pro Tips
- 1Quick mental trick: 25.4 µm ≈ 0.001 in (1 thou). Divide µm by 25.4 to get thousandths of an inch.
- 2When a drawing shows 20.00 ±0.02 mm, plan inspection with resolution ≤0.005 mm to avoid ambiguity.
- 3For assemblies, run tolerance stack-ups with worst-case and statistical methods; include surface roughness in contact points.
- 4Use Gage R&R to check if your inspection system can tell parts apart near tolerance limits.
Tolerances tie design intent to manufacturing reality. Read a drawing not only for numbers, but for what those numbers mean in assembly: will parts slide, press, seal or locate? Consider fits, surface finish and GD&T together rather than as separate notes. Try the related converters to check quick transforms (micrometers to thousandths, mm to inches) and use uncertainty calculation templates when planning inspection. Small changes in tolerance or finish can save cost and prevent field failures — test stack-ups early and include inspection uncertainty in acceptance criteria.
