Most machined part drawings use plus-minus dimensioning alone. For simple parts that is adequate. For anything where features must relate to each other precisely, it produces a specific and expensive problem: the drawing does not actually say what the part needs to do.<\/p>\n\n\n\n
Geometric dimensioning and tolerancing solves that by describing relationships rather than only sizes. It is also frequently applied badly, with symbols scattered across drawings by engineers who learned the notation without learning what each callout costs to achieve.<\/p>\n\n\n\n
This guide covers the callouts that earn their place on a machined part drawing, what each one asks the machinist to do, and where GD&T saves money rather than adding it.<\/p>\n\n\n\n
The Problem GD&T Solves<\/strong><\/p>\n\n\n\n Consider a plate with four mounting holes, dimensioned with plus-minus tolerances from an edge.<\/p>\n\n\n\n Under that scheme, each hole’s position is controlled independently, in X and Y, from a datum edge. The tolerance zone for each hole is a square. But a bolt passing through the hole does not care about a square zone, it cares about radial clearance, so the square zone is a poor description of what actually matters.<\/p>\n\n\n\n Worse, the tolerances stack. If hole one is toleranced from the edge and hole two from hole one, their relationship carries both tolerances. On a part with several chained dimensions, accumulated tolerance can exceed what any individual feature allows.<\/p>\n\n\n\n GD&T addresses both. Position tolerance defines a round zone matching the functional requirement, and every feature relates to defined datums rather than chaining from each other.<\/p>\n\n\n\n Datums Come First<\/strong><\/p>\n\n\n\n Everything in GD&T references datums. Get them wrong and the rest of the scheme, however carefully applied, describes the wrong thing.<\/p>\n\n\n\n A datum is a reference derived from a physical feature of the part, and datums are usually established in a sequence of three: primary, secondary and tertiary. Together they constrain the part fully, so any measurement is repeatable.<\/p>\n\n\n\n Choose datums by function.<\/strong> The primary datum should be the surface the part actually sits on or locates from in its assembly. If the part bolts to a plate, the mating face is the primary datum. If it locates on a shaft, the bore is.<\/p>\n\n\n\n The most common GD&T error is choosing datums for convenience of drawing rather than function. When the datum scheme does not match how the part is used, the drawing controls the wrong relationships and parts that pass inspection can still fail to assemble.<\/p>\n\n\n\n State the datum sequence explicitly.<\/strong> A|B|C means something different from B|A|C, because the order determines which surface constrains the part first.<\/p>\n\n\n\n The Callouts Worth Knowing<\/strong><\/p>\n\n\n\n Fourteen geometric characteristics exist. A handful cover most machined part requirements.<\/p>\n\n\n\n \uc704\uce58<\/strong><\/p>\n\n\n\n The most useful and most used callout on machined parts.<\/p>\n\n\n\n Position controls where a feature is relative to datums, with a round tolerance zone rather than a square one. For holes and bosses this matches function directly, since clearance is radial.<\/p>\n\n\n\n Position also allows maximum material condition<\/strong>, indicated by a circled M. This grants additional positional tolerance as the feature departs from its maximum material condition. In practice, a hole at the larger end of its size tolerance has more clearance available, so it can be further out of position and still assemble.<\/p>\n\n\n\n MMC is genuinely free tolerance where the requirement is assembly rather than precise location. Applying it where appropriate reduces cost without reducing function, and it is one of the clearest cases where GD&T saves money rather than adding it.<\/p>\n\n\n\n \ud3c9\ud0c4\ub3c4<\/strong><\/p>\n\n\n\n Controls how flat a surface is, independently of any datum. Useful for sealing faces, mounting surfaces and anywhere a surface must sit evenly.<\/p>\n\n\n\n Tight flatness over a large area is genuinely difficult and can require grinding or lapping rather than milling. Specify it where function requires and leave it off where it does not.<\/p>\n\n\n\n \uc218\uc9c1\uc131<\/strong><\/p>\n\n\n\n Controls how square a feature is to a datum. Common on bores relative to a mounting face, and on walls relative to a base.<\/p>\n\n\n\n Parallelism<\/strong><\/p>\n\n\n\n Controls how parallel a feature is to a datum. Common on opposing faces where a consistent gap or thickness matters.<\/p>\n\n\n\n Concentricity, Runout and Total Runout<\/strong><\/p>\n\n\n\n For rotating parts, runout is usually more useful and considerably cheaper to verify than concentricity.<\/p>\n\n\n\n Concentricity is defined in terms of derived median points and is difficult to measure. Runout is measured by rotating the part and reading indicator movement, which is straightforward.<\/p>\n\n\n\n For most rotating applications, runout describes what matters and should be preferred. Concentricity is often specified where runout was intended, adding inspection cost for no functional benefit.<\/p>\n\n\n\n Profile of a Surface<\/strong><\/p>\n\n\n\n A versatile callout controlling the form, orientation and location of a surface within a tolerance zone. Useful for contoured surfaces where plus-minus dimensioning becomes unwieldy, and increasingly used as a general-purpose control.<\/p>\n\n\n\n Cylindricity and Circularity<\/strong><\/p>\n\n\n\n Control the roundness of cylindrical features. Relevant for bearing bores, sealing surfaces and precision fits. Both are demanding to hold tightly and to verify, so reserve them for features that genuinely need them.<\/p>\n\n\n\n What Each Callout Costs<\/strong><\/p>\n\n\n\n This is the part missing from most GD&T training, and it is what determines whether your drawing is economical.<\/p>\n\n\n\n Tight position tolerances<\/strong> require accurate setups and good fixturing. Where features on different faces must relate tightly, they should ideally be machined in one setup, which may push the part toward multi-axis machining<\/a> that would otherwise be unnecessary.<\/p>\n\n\n\n Tight flatness over a large area<\/strong> may require grinding or lapping after machining.<\/p>\n\n\n\n Tight cylindricity<\/strong> may require honing or grinding rather than boring.<\/p>\n\n\n\n Tight perpendicularity between features on different faces<\/strong> is one of the most expensive things to ask for, because it demands either a single setup or very precise refixturing.<\/p>\n\n\n\n Anything requiring specialised inspection<\/strong> carries measurement cost. A coordinate measuring machine programme takes time to write and run, and complex GD&T schemes take longer to verify than simple ones.<\/p>\n\n\n\n The practical principle: every geometric callout should trace to a functional requirement. If you cannot say what fails when the callout is not met, it should not be on the drawing.<\/p>\n\n\n\n Where GD&T Saves Money<\/strong><\/p>\n\n\n\n It is worth stating clearly, because GD&T is often seen as adding cost.<\/p>\n\n\n\n Round tolerance zones give more area than square ones<\/strong> for the same functional clearance. Converting a plus-minus scheme to position tolerance often grants the machinist more room while controlling the part better.<\/p>\n\n\n\n MMC grants bonus tolerance<\/strong> where assembly is the requirement.<\/p>\n\n\n\n Eliminating tolerance stacking.<\/strong> Referencing features to datums rather than chaining them removes accumulated error, which often means individual tolerances can be looser than a chained scheme required.<\/p>\n\n\n\n Communicating intent removes guesswork.<\/strong> A machinist who understands what matters can choose setups and fixturing accordingly. A drawing that tolerances everything equally forces conservative assumptions everywhere.<\/p>\n\n\n\n Well-applied GD&T frequently produces a cheaper part than plus-minus dimensioning of the same requirement. Badly applied GD&T produces a more expensive one.<\/p>\n\n\n\n Practical Points for Overseas Manufacturing<\/strong><\/p>\n\n\n\n State the standard.<\/strong> ASME Y14.5 and ISO GPS differ in specific respects, and the drawing should say which applies. Assuming is a source of genuine disagreement.<\/p>\n\n\n\n Provide a fully dimensioned 2D drawing alongside the 3D model.<\/strong> Model-based definition works where both parties are set up for it. Where they are not, the 2D drawing is the contractual document.<\/p>\n\n\n\n Identify critical features explicitly.<\/strong> A note marking which characteristics are critical helps the supplier plan setups and inspection around what matters. Many suppliers will ask; providing it unprompted saves a cycle.<\/p>\n\n\n\n Specify inspection requirements.<\/strong> Which features need reporting, at what sample rate, and in what format. Agree this at quotation stage rather than after delivery.<\/p>\n\n\n\n Expect questions on ambiguous schemes.<\/strong> A supplier who queries your datum scheme is engaging with the drawing. One who quotes without questions on a complex GD&T drawing may not have read it closely.<\/p>\n\n\n\n Common Mistakes<\/strong><\/p>\n\n\n\n Datums chosen for drafting convenience<\/strong> rather than function. The most consequential error, because everything else follows from it.<\/p>\n\n\n\n Over-tolerancing.<\/strong> Applying tight geometric controls to features that do not need them, which raises cost and inspection burden without improving the part.<\/p>\n\n\n\n Concentricity where runout was meant.<\/strong> Adds substantial inspection cost for no functional gain in most applications.<\/p>\n\n\n\n Incomplete datum reference frames.<\/strong> A callout referencing only one datum where three are needed leaves the part under-constrained and the measurement ambiguous.<\/p>\n\n\n\n Mixing GD&T and plus-minus inconsistently<\/strong>, so it is unclear which controls a given feature.<\/p>\n\n\n\n Not applying MMC where assembly is the requirement.<\/strong> Free tolerance left on the table.<\/p>\n\n\n\n Deciding What Your Drawing Needs<\/strong><\/p>\n\n\n\n Work through the part functionally.<\/p>\n\n\n\n Anything that survives that review belongs on the drawing. Anything that does not is adding cost.<\/p>\n\n\n\n Our guide to CNC machining cost drivers<\/a> covers how tolerance decisions translate into quoted price.<\/p>\n\n\n\n\n