← All technical guides

Carbide part tolerances: what is actually achievable?

**The question we get most often after the quote: "Can you hold ±0.01 mm on this?" The short answer: yes — but only on the surfaces you're willing to pay to grind. Understanding the two-stage reality of carbide manufacturing will save you money and arguments.**

1. The problem

Most engineers order a carbide part the way they order a machined steel part. You put a tolerance block on the drawing, the supplier quotes it, and you assume the whole part ships at that precision. With carbide, that assumption is expensive. Sintering shrinks the part by roughly 20% linearly, and the natural spread after sintering eats a big part of your tolerance budget before any grinding starts.

2. The conclusion

There are two very different regimes, and the boundary between them is a cost step, not a gradient:

• **As-sintered (no grinding):** typically ±0.1 to ±0.25 mm on small-to-mid parts. Tighter than a casting, looser than machining. Free to give you, but geometry-dependent.
• **Ground (diamond grinding of functional surfaces):** ±0.01 mm is routine; fine lapping pushes key dimensions toward ±0.002–0.005 mm. Rods and simple cylinders are commonly supplied to h5/h6 classes.

Everything tighter than sintering needs diamond grinding — and grinding is where most of the cost and lead time live.

3. The basis

• **Why sintering spreads:** pressed powder compacts shrink about 20% linearly during sintering. Wall thickness, tooling wear, powder lot, and sintering position all shift the final dimension slightly. Repeatable within a batch, but not a tight-tolerance process on its own.
• **What the industry holds:** manufacturer handbooks and published practices place as-sintered tolerance around ±0.15 mm (some quote ±0.8% of dimension or ±0.005 in, whichever is larger, others ±0.2–0.4 mm on bigger parts). Ground tolerance is typically ±0.01 mm, and precision ground/lapped surfaces go to ±0.002–0.005 mm.
• **What makes a part hard to hold:** long length-to-diameter ratio, thin walls, internal holes (harder to grind and measure), and large size. Size alone is the biggest variable — what's ±0.01 mm on a 10 mm part is not the same job on a 100 mm part.
• **Position tolerances are a grinding story, not a sintering story:** concentricity, roundness and flatness are set by how the part is fixtured and ground, not by the sintered shape. If the drawing calls these out, say so explicitly — that is the dialogue that decides the process.

Standards anchor: general tolerances usually follow ISO 2768 (medium class is a common starting point); fits follow ISO 286 (h6/H7-class fits are routinely ground). Grade classification anchors: GB/T 18376 and ISO 513. Note these cover classification and general tolerances — they do not hand you a precision; the drawing and the application do.

4. The advice

• **Mark the fit, leave the rest alone.** Call out the two or three dimensions that actually mate or seal and grind only those. Leaving non-functional surfaces as-sintered typically cuts both cost and lead time meaningfully.
• **Give the part a stiffness reality check.** Long, thin carbide parts are both harder to sinter straight and harder to grind without flex. If the drawing allows, shorten or thicken — it buys tolerance for free.
• **Ask for the process in the quote.** "Ground on OD and both faces, as-sintered elsewhere" tells you more than a tolerance block you won't inspect.
• **Send the application, not just the print.** A seat that must seal at 300 bar and a wear pad that just guides a strip need completely different tolerance budgets. With the duty we'll tell you which dimensions matter — and quote accordingly.

Send us the drawing with the application. We'll confirm what is ground, what is as-sintered, and what that does to the price — within 1–2 business days.

---