LPBF From Prototype to Production: Print It, Check It, Then Machine It

An LPBF metal part can come off the build plate looking finished and still be the wrong part.

I've watched it happen more than once. The prototype looks great, the photos go around the office, and then somebody puts a micrometer on the bore that holds a bearing. Laser powder bed fusion (LPBF, the process EOS calls DMLS) is an amazing process, but for most production parts it's a near-net-shape process. The printer gets you close. Machining gets you to the print.

What most LPBF parts are printed on

A lot of the LPBF parts you'll buy from a service bureau come off an EOS M280 or its successor, the M290. It's the workhorse of the industry: a single fiber laser, a build volume of about 250 x 250 x 325 mm (roughly 9.8 x 9.8 x 12.8 inches) and layers in the 20 to 40 micron range. Bigger multi-laser machines exist for bigger parts, but this is one of the most widely installed metal printers out there.

It also runs a wide range of metals. The ones you'll see quoted most:

• AlSi10Mg aluminum: light, good thermal conductivity, the go-to for brackets, housings and heat exchangers.
• 17-4 and 15-5 stainless: strong, hardenable, corrosion resistant. Common for tooling, valves and general functional parts.
• 316L stainless: the corrosion-resistant choice for fluid, food and medical-adjacent parts.
• Maraging steel: a tool steel that ages to high hardness. The classic material for conformal cooling inserts and molds.
• Inconel 718 and 625: nickel superalloys for heat, pressure and corrosion, like turbine and exhaust parts.
• Titanium (Ti-6Al-4V): strength with low weight, for aerospace and medical parts.

As built, a well-run machine typically holds around ±0.005 in (0.13 mm) on many features, looser on big or thin parts. That's great for a lot of geometry. It's not a bearing fit, a sealing face or a thread class. Which brings us to the real point.

Here's how I'd take an LPBF part from prototype to production without surprises.

1. Get the prototype in your hands first

Before anyone talks about volume, print one and measure it. Not a render and not a simulation: a real part, built on the machine, in the material, with the orientation and supports you plan to use in production.

That first part tells you things the CAD model can't:

• Where the part shrank, warped or curled as it cooled
• Which surfaces came out rough from supports or down-facing angles
• Which features are in tolerance as built, and which aren't even close
• How much the part moves after stress relief and coming off the plate

Check it against the drawing, feature by feature. That inspection is the most valuable thing you'll pay for in the whole project.

2. Build the cut sheet from the real part

Once you know how the part actually prints, you can write the machining plan. I call it the cut sheet. It answers a few questions before a single chip is made:

• Which features get machined? Bearing bores, sealing faces, threads, mating surfaces and anything with a tight tolerance callout.
• How much stock do we leave? Add material in the CAD model on those features so the machinist has something to clean up. Too little and you machine air on one side. Too much and you're paying to cut metal you just paid to print.
• Where are the datums? A printed part has no flat, square edge to pick up from. Design in a datum face or a fixturing feature, and machine that first.
• How is it held? Thin printed walls can move under clamping. Plan the fixture before the parts show up, not after.

The cut sheet comes from measuring the prototype, not from guessing. That's why the prototype comes first.

3. Check the part, then machine it to tolerance

My rule is simple. Every part gets checked before it gets machined, and machined to tolerance after. Inspect the as-printed part, confirm there's enough stock where you need it, then machine the critical features to the drawing and inspect again.

Skip the first check and you find out about a short feature after you've already paid for the machining.

4. Lock down the system, the material, the setup and the traceability

A prototype proves the part can be made. Production proves it can be made the same way every time. That takes four things:

• The right system. Pick a machine and a shop that can hold the build size, the material and the volume you need, and keep running it on the same platform.
• The right material. Certified powder with a known chemistry, tracked by lot. Know how the shop handles virgin and reused powder.
• A locked setup. Same build parameters, orientation, support strategy and post-processing (stress relief, heat treat, HIP if you need it) for every build. If something changes, the part gets requalified.
• Traceability. You should be able to trace any part back to its build, its powder lot, its machine and its inspection report. In aerospace, medical and defense, that paper trail is part of the product.

When you don't need to machine it

Not every LPBF part needs a cut sheet. If it's a prototype for form and fit, a part with no tight tolerances, or a feature like an internal channel or a lattice that can't be machined anyway, print it and use it as built. Just decide that on purpose, feature by feature, instead of finding out at inspection.

The short version

• Print the prototype and measure it before you plan production.
• Write the cut sheet from what the real part tells you.
• Check every part, then machine it to tolerance.
• Lock the system, material, setup and traceability before you scale.
• Treat LPBF as near net shape unless the part truly doesn't need machining.

What's the feature that always needs machining on your LPBF parts? Tell me in the comments.

More from me

• 2026 in Additive Manufacturing: What Stood Out and What Needs to Change
• Which 3D Printing Process Fits Your Part? Take the 60-Second Quiz
• Your $2,000 Bambu Can Print Carbon Fiber Nylon. That Doesn't Make It an Industrial Printer

Want the whole playbook on picking a process, pricing parts and getting them into production? My book Print It. Price It. Sell It. is out now on Kindle and in paperback.

Tim Ruffner has spent almost 20 years in additive manufacturing and is an Account Manager at 3D Parts Unlimited, a service of MSC. Connect with him on LinkedIn.