How to Set Realistic Stamping Tolerances: Grades vs Cost

Every dimension on a stamped part carries a tolerance, whether the drawing says so or not. The question is who pays for the precision. A general class such as ISO 2768-m costs almost nothing. A fine-class callout on every feature can raise tooling cost by 30 to 60 percent and slow production for parts that never needed it. This guide explains what stamping can actually hold, how tolerance grades shift cost, and how to set callouts that survive contact with the press.

1. Why Tolerances Decide Project Cost

Tolerance is the most expensive three lines in a drawing. A tighter value does not just change a number; it changes the tool, the press speed, the inspection plan, and the scrap rate. The same part drawn with a general tolerance class and drawn with precision callouts on every feature can differ in tooling cost by tens of percent before the first part is made.

That is why tolerance setting belongs at the start of a project, not at the end. Buyers who specify first and ask questions later discover the gap between the drawing and reality when the first article fails inspection. Buyers who understand capability before they write callouts get quotes that match the part they actually need.

Consider the same bracket quoted two ways. One drawing carries a general tolerance class with two critical callouts. The other applies a fine class to every dimension. The second drawing can arrive with a tooling price 30 to 60 percent higher, a slower production rate, and a longer first article inspection, yet the parts assemble identically. Tolerance is not a drawing convention; it is a cost line, and it deserves the same scrutiny as material grade and quantity.

2. How a Stamp Gets Its Dimensions

A stamped dimension is produced by three systems working together: the die, the press, and the material. The die defines the nominal geometry, the press delivers the force, and the material decides how faithfully the shape survives the stroke. Any variation in one system moves the final dimension.

Tool wear drifts hole diameters over a production run. Die clearance controls burr height and edge quality. Springback bends the material away from the tool angle. Material thickness varies from coil to coil within its own specification. Feeding accuracy in a progressive die shifts hole patterns. None of these effects can be eliminated; they can only be controlled, which is what a tolerance budget does.

The measurement system is part of the same story. A dimension held to ±0.05 mm is only meaningful if the gauge can measure it, which is why shops use CMMs, vision systems, and calibrated micrometers for precision features and why measurement system analysis matters. Capability is not the nominal position of a dimension; it is the spread of that dimension across a run, and the tolerance defines how much spread the design can accept.

3. ISO 2768: The Default Tolerance Language

Most stamped part drawings need a general tolerance block, and ISO 2768-1 is the common language for it. The standard defines four classes for linear dimensions, fine, medium, coarse, and very coarse. Each class gives permissible deviations that grow with the nominal size, because a long dimension accumulates more variation than a short one.

Values are in millimeters and show the permissible deviation for each class. Medium, noted as ISO 2768-m, is the most common default for stamped parts, and it is a sensible starting point for most dimensions. Fine, ISO 2768-f, is where cost starts climbing noticeably. For stamped sheet metal, some suppliers prefer DIN 6930-2 general classes, which work the same way: state the standard and class in the drawing block so nobody has to guess.

The general class covers what is not otherwise marked, so it does the heavy lifting for the 85 to 95 percent of features that only need to fit together. ISO 2768-1 also includes angular tolerance classes with their own values, which is convenient for bend angles, and ISO 2768-2 extends the system to geometric tolerances such as flatness and straightness. If your drawing block says ISO 2768-m and nothing else, a shop knows exactly what that means; if it says nothing, the shop must choose a default, and the two of you may disagree about which one applies.

4. What Tolerances Stamping Can Realistically Hold

Before you write a tolerance, know what the process can do in production, not just in a tryout. Conventional stamping commonly holds about ±0.10 to ±0.25 mm on general dimensions. Precision stamping, with dedicated tooling and controlled processes, reaches roughly ±0.05 mm on critical features. A skilled progressive die shop treats ±0.05 mm on critical dimensions as a standard capability and can hold about ±0.02 mm on a single controlled feature when the volume justifies the tooling.

Feature types differ. Hole-to-hole position typically lands in the ±0.10 to ±0.20 mm range with good pilot design and strip stability. Blanked hole diameters hold about ±0.05 mm in normal production and ±0.025 mm in precision work. Bend angles usually sit between ±0.5 and ±1.0 degree, because springback and forming control limit them. Flatness over a 50 mm span runs about ±0.10 mm conventionally and ±0.05 mm with precision tooling. These are working numbers, and a supplier should confirm the ones that apply to your material and geometry.

Two distinctions matter when reading these numbers. First, capability in a tryout is not capability in mass production; a die can hold tight values for a hundred parts and drift across ten thousand as tools wear and material changes. Second, capability applies to individual features, not to every feature at once; a part with five features each held to ±0.05 mm is harder than a part with one. Part size, material thickness, and feature depth all move the practical numbers, which is why the supplier's confirmation belongs in the engineering review, not the delivery dock.

5. Tolerance Grades vs Cost: The Price of Precision

The cost curve between tolerance classes is not gentle. Industry estimates put the jump from ISO 2768-m to ISO 2768 fine at a 30 to 60 percent increase in tooling cost, with more frequent inspection and higher scrap rates on top. The fine class buys real capability, but it should be earned by function, not applied by habit.

Why so steep? A tighter tolerance demands higher-precision tool construction, more stations or special processes, slower press speeds to keep the process centered, and more measurement at every inspection stage. The die builder must hold tighter relationships between punches and pilots, and the production team must monitor drift closely enough to catch a part leaving the band before a batch runs. Every one of those activities costs money, and the cost repeats on every order, not just the first one.

Volume changes the math but not the principle. A fine-class tolerance spread over a million parts adds less per part than over ten thousand, but it still adds tooling complexity, inspection time, and scrap risk that a medium class avoids. The right question is never whether the shop can hold the tolerance; it is whether the function needs it. When the answer is no, the cheapest improvement to the drawing is deleting the callout.

6. Where Tight Tolerances Hide Their Cost

The visible cost of a tight tolerance is tooling, and that is only the beginning. The hidden costs show up in four places. Tooling must be built with tighter relationships between punches, pilots, and die sections, which adds machining precision and often more stations or special processes. Production must run slower, because high press speeds widen the spread of every dimension. Inspection must sample more frequently, measuring more features on more parts. And scrap rises, because the process has a narrower band before a part falls out of tolerance.

Small form tolerances can force restriking, where parts pass through a second calibration operation to hold flatness or angles. Restriking adds a process step, a second set of checks, and a new source of variation. Multiply these effects across an assembly, and tolerance stack-up compounds the cost: every tight part makes the next one tighter. The cheapest tolerance is the one a feature does not actually need.

The hidden costs also surface in the quote structure itself. Two suppliers can quote the same drawing with different assumptions about which tolerances are serious, and the cheaper quote may simply be the one that plans to ignore a callout until the first article. Comparing quotes is only meaningful when the tolerance block is explicit enough that both shops price the same requirement. A clear general class plus named critical features removes most of that ambiguity.

7. Feature-by-Feature Tolerance Guidance

• General dimensions. Default to ISO 2768-m or the equivalent DIN class. Most features on a stamped part work fine at medium, and the drawing stays readable.

• Pierced hole diameters. Plan for about +-0.05 mm in normal production and +-0.025 mm when precision tooling is justified. Smaller holes in thicker or harder material shift the practical limits.

• Hole-to-hole positions. Expect +-0.10 to +-0.20 mm with well-designed pilots and stable strip feeding. Tighter positions require tighter pilots and more controlled material.

• Blank and cut profiles. About +-0.05 mm is a realistic working number for precision progressive tooling, with standard production landing wider.

• Bend angles. Plan for +-0.5 to +-1.0 degree. Tighter angles mean springback compensation work, secondary operations, or both.

• Flatness. About +-0.10 mm over a 50 mm span conventionally, +-0.05 mm with precision tooling and coining. Long, thin parts need larger allowances.

• Burr height. A small, even burr under about 0.05 mm is normal in production. If the edge must be burr-free, specify the side and add deburring to the plan.

These numbers describe common capability, not a promise. Material, part size, feature depth, and tooling quality all move them. Use the list to draft the drawing, then ask the supplier to confirm the values that apply to your geometry before the tool is committed.

Formed features deserve their own callouts. Tapped holes need enough material thickness for thread engagement, and the tap adds a tolerance chain on top of the stamping. Countersinks should leave at least half the base material to avoid tearing through, and their depth tolerance interacts with the sheet thickness. Slots, tabs, and cutouts follow the same edge-spacing rules as holes. When a feature is produced by a secondary operation, welding, tapping, or riveting, the drawing should make clear which dimensions are controlled by the stamping and which by the operation that follows, so inspection targets the right stage.

One more habit pays off: dimension the part the way it is measured. A bend angle is checked off the formed surfaces, a hole position is measured from the datum edge, and a flatness callout references the surface itself. When the drawing and the inspection plan agree on where measurement starts, first article results are repeatable and the two sides argue about facts instead of about who measured wrong. That alignment is a small detail that costs nothing and prevents the most common kind of rework argument.

8. Springback: The Tolerance Killer

Springback is the reason bend angles carry their own tolerance logic. When the press releases, the material relaxes elastically and the bend opens slightly. Mild steel typically springs back 1 to 3 degrees on a 90-degree bend. Advanced high-strength steel can spring back 5 to 10 degrees, which is why AHSS parts demand so much more from the tool.

The die compensates by forming beyond the target angle. A 90-degree requirement may be formed at 85 to 88 degrees so the part relaxes into spec. In rotary bending practice, compensation runs about 3 degrees for mild steel and about 10 degrees for AHSS. The compensation value changes with material batch, bend radius, and geometry, so a fixed die angle holds a range rather than an exact angle.

Modern die design simulates springback before steel is cut. One documented case adjusted die angles by 1 to 2 degrees from finite element analysis and held deviations under 0.5 degree; a fender project corrected 1.8 degrees of springback to within 0.2 degree and cut rework by a fifth. That is the practical answer to tight bend tolerances: let the supplier model the material, then set the angle range your assembly can live with.

Springback also varies with the ratio of bend radius to material thickness and with the temper of the coil. A generous radius relaxes less than a tight one, and a harder temper springs back more. That means a bend angle tolerance tighter than about ±0.5 degree usually requires either simulation-guided tooling, a secondary calibration operation, or a design that accepts the range. When in doubt, ask the supplier what angular range their process can hold for your exact material before writing the callout.

9. Materials Change What You Can Hold

The material is the third partner in every tolerance, and it is the least cooperative one. Coil stock arrives with its own thickness tolerance, surface condition, and mechanical properties, and those vary between suppliers and batches. A stamping process centered on one coil can drift when the next coil arrives with different properties, which is why material specification belongs on the drawing and material verification belongs in the inspection plan.

Grades behave differently. Stainless steel work-hardens during forming, so dimensions near bends shift more than carbon steel. High-strength steel springs back harder and wears tools faster. Aluminum tempers range from soft and formable to stiff and crack-prone. Galvanized coatings change friction and can flake at tight bends. Grain direction matters for fatigue and for bend consistency. All of these effects mean the same tolerance request can be routine on SPCC and demanding on AHSS, and the drawing should reflect the actual grade, thickness, and finish.

Documentation closes the loop. Material certificates with heat number traceability, chemical composition reports, and mechanical property reports let the shop confirm the coil matches the drawing before it runs. Incoming inspection verifies grade, thickness, and surface condition, because a coil that is off-spec moves every dimension downstream. If your tolerances matter, the material spec matters as much as the numbers on the drawing, and both should be fixed before tooling starts.

10. Drawing Callouts That Work

A tolerance drawing is a budget, and a good one starts with the general block. Write the standard and class once, for example ISO 2768-m, and let it cover everything that is not otherwise marked. Then call out the small number of features that truly need tighter control. Industry experience puts that critical fraction at roughly 5 to 15 percent of the dimensions on a typical part, which means the other 85 to 95 percent should ride the general class.

For the critical features, use the language that communicates intent. Geometric tolerancing such as position and flatness, applied per ASME Y14.5, tells the shop what matters and how to measure it. An explicit callout overrides the general block, so the drawing stays clean while the important relationships get their own numbers. Avoid the habit of dimensioning every feature individually with tight values; that is how CNC thinking creeps into a stamping drawing and turns a low-cost process into an expensive one.

Datums are the hidden half of a good tolerance drawing. A position callout is meaningless until the drawing says which edges or holes establish the reference frame, and the supplier must measure from the same datums you designed from. Redundant dimensions, where the same feature is located twice, create contradictions that inspection cannot resolve. A clean drawing names the standard, sets the general class, marks the critical features, and defines the datums, and that is enough for a shop to build and measure with confidence.

11. Reading Supplier Capability Data

A supplier who can show capability data is giving you the answer before you ask. Process capability indices such as Cpk tell you whether the process can hold a tolerance over time, not just on the first article; values above 1.33 or 1.67 are common acceptance targets, depending on the industry. Statistical process control charts show drift before parts fall out of spec, and measurement system analysis, such as GR and R studies, confirms the gauges themselves can be trusted.

The first article inspection is the concrete proof. The supplier measures overall dimensions, hole positions, bend angles, material, surface finish, and functional features against the drawing, and production does not start until that part is approved. For automotive chains, PPAP documentation and traceability records are standard. Ask for the reports, first article inspection, CMM results, coating thickness data, and you will learn more about the supplier than any website tour.

Inspection is a cost too, and the drawing should set the level of verification, not leave it to chance. State which dimensions require full measurement on the first article, which need in-process sampling, and which simply ride the general class. That keeps the inspection plan proportional to the risk. If a supplier offers reports you did not ask for, treat it as a signal; shops that measure are shops that know where their process drifts.

12. Common Tolerance Mistakes Buyers Make

• Applying the fine class to everything. A 30 to 60 percent tooling premium for precision nobody uses is the most expensive drawing habit in stamping.

• Copying machining tolerances. Values that make sense on a CNC part are often impossible or ruinously expensive on a stamped one.

• Ignoring springback. A bend angle tolerance tighter than the material's natural relaxation range buys restrikes, not accuracy.

• Forgetting material variation. Coil-to-coil differences move dimensions; the drawing should specify grade and thickness so the process can be set correctly.

• Leaving no general tolerance block. Every dimension individually toleranced is slower to read, easier to contradict, and more expensive to hold.

• Specifying without volume context. A tight tolerance that is cheap at a million parts can be absurd on a thousand-part run, and the quote hides the difference.

• Accepting silent quotes. A supplier that promises an impossible tolerance without comment is signing you up for rework, not delivering a bargain.

• Skipping capability questions. Not asking for Cpk, FAI data, or inspection reports leaves tolerance risk entirely on your side.

• Ignoring burr direction. The burr side of a pierced edge is a dimensional fact, and discovering it at assembly is a rework event.

• Leaving measurement undefined. Two shops can measure the same hole from different datums and get different answers; the drawing should say where measurement starts.

13. From Drawing to Approved First Article

A tolerance that survives production travels a known path. The project starts with inquiry and drawing review, moves through engineering evaluation and quotation, then into tool design and manufacturing. Samples are produced and inspected, the first article is measured completely, and only your approval releases mass production. Surface finishing, final inspection, packaging, and shipment close the loop.

Along the way, the tolerance story is told by measurement. In-process checks monitor hole diameter, burr height, flatness, and tool wear during the run. Final inspection verifies dimensions, appearance, and function before packaging. Outgoing checks confirm quantity, identification, and documentation. Every stage exists to keep the process inside the band the drawing defines, which is why the band itself should be set with care.

Sample approval is the buyer's checkpoint. Most custom projects allow several rounds of samples before tooling is finalized, and each round is a chance to measure, assemble, and adjust the tolerance plan while changes are still cheap. For tight-tolerance parts, a prototype run before the production tool confirms the geometry and the measurement method. Once the tool is cut, changing a tolerance means changing steel, so the time to be realistic is before the first order, not after the first rejection.

Delivery closes the tolerance loop with paperwork. Depending on the project, buyers receive first article inspection reports, material certificates, dimensional inspection reports, CMM reports, coating thickness reports, and salt spray test results. Those documents are the audit trail for every tolerance on the drawing, and agreeing which ones ship with each order belongs in the RFQ, not in the final week of the project.

Setting realistic stamping tolerances comes down to three decisions. Choose a general class that matches the process, protect only the features that matter to function, and verify the result with capability data rather than hope. Parts drawn that way cost less, arrive on schedule, and pass the first article the first time. Tolerances are not a formality; they are the contract between your design and the press, and both sides deserve an honest one.

The final step is asking the supplier to show its capability matrix before you commit. A shop that publishes its tolerance ranges by feature type, shares Cpk data, and reviews your drawing before quoting is telling you how the project will go. A shop that avoids the subject is telling you something too. Realistic tolerances are a negotiation, not a form filling exercise, and the best time to negotiate is while the drawing is still open.