Additive vs. Subtractive Manufacturing: A Quality Inspector’s FAQ from Star Micronics

Every week, someone on a kickoff call or quote review asks me the same things: Should we print or machine it? Why is this quote higher than the next one? Can one shop really handle a mixed-technology part? I’m the quality and compliance lead at Star Micronics, a custom fabrication company. For four years I’ve reviewed nearly every deliverable before it reaches a customer—roughly 200 unique parts a year. That position changes how you see trade-offs. I’ve delayed shipments because a supplier couldn’t show me a material certificate, and I’ve approved parts with tool marks because the drawing allowed them. Here are the questions I answer most, minus the marketing filter.

What’s the difference between additive and subtractive manufacturing?

Additive manufacturing (AM) builds parts by adding material layer by layer. FDM, resin printing, powder-bed fusion, and metal 3D printing all count. Subtractive manufacturing starts with a block, sheet, or bar and removes everything that doesn’t belong. CNC milling, turning, and laser cutting are subtractive processes.

The difference that matters isn’t the machine—it’s the trade-off. AM can create geometry you can’t reach with a cutter: curved internal channels, lattices, one-piece assemblies that replace ten welded parts. Subtractive gives you parent-metal properties, better surface finish, and tighter tolerance control. Holding ±0.05 mm on a CNC machine is routine. Holding that across the full build area of a large metal 3D printer is not.

ISO/ASTM 52900 defines these categories internationally, so teams can talk without confusion. My rule is simple: choose by function, not by process. A part isn’t good because it was printed or cut. It’s good if it conforms to a spec across the whole order.

When does a large format metal 3d printer make sense?

A large format metal 3d printer is not a universal answer. I tell customers that before they send me a model of a solid block and ask for a print. This equipment wins when geometry is complex and tooling is impractical. For example, a 500 mm frame currently produced by welding eight components can be printed as one piece. That removes weld distortion, source inspections, and a long list of potential failure points.

It loses when the part is mostly solid with a few holes. That shape belongs on a CNC mill or lathe, where material cost is lower, speed is higher, and mechanical properties are predictable. Many metal AM components still need finish machining on sealing surfaces and threads. If someone quotes you a pure print without mentioning secondary operations, ask why.

Full disclosure: I’m part of a shop that sells additive services. I’m also the person who tells clients the technology is wrong for maybe 60 percent of the metal AM requests we see.

What does a “durable model cutting tool” actually mean?

Some prototypes have to survive more than a desk photo. When a part will be dropped, mounted, or cycled hundreds of times, search terms like “durable model cutting tool” start showing up in our inbox. What people usually want isn’t a special tool. They want a cutting process that leaves a durable model with clean edges and no hidden damage.

A durable model starts with material and edge quality. Cutting 6 mm aluminum sheet on a fiber laser gives clean edges. Cutting the same sheet on an undersized router with a worn bit can leave burrs, melted polymer, or micro-cracks. Glass-filled nylon and carbon fiber are even less forgiving: they wear tooling fast, and once the tool dulls, dimensional accuracy drops without warning.

This isn’t a reason to fear hard materials. It’s a reason to ask what tooling will be used and how the shop checks edge quality. “Durable” is not a feeling. It needs a test plan or a number.

Which gives better quality: 3D printing or CNC machining?

I wince at that question because quality isn’t a property of the process. It’s conformance to a specification, and every process fails differently. 3D printed parts fail through porosity, layer adhesion, or build orientation effects, so I validate them differently. CNC parts fail through datum errors, tool deflection, or surface finish drift, so we check differently.

During our Q1 2024 quality audit, we traced the majority of rejected first articles to unclear specifications, not to machine capability. The drawing said “smooth finish” without a Ra value. Another said “align parts” without a positional tolerance. Good process selection doesn’t fix an ambiguous requirement.

What separates a quality shop is whether they test the part that matters. Ask how critical features will be measured—CMM, gauge pins, surface profilometer, or no? If the answer includes “we’ll just know,” be careful.

When you shop Star Micronics, what does multi-process actually get you?

If your search starts with “shop star micronics,” you’re probably hunting for one supplier who won’t play hot potato with tolerances. Your part might need a 3D printed internal bracket, a laser-cut and bent cover, and a few CNC machined mounting bosses.

Send that to three separate suppliers and each one can honestly say their piece is fine while the assembly still doesn’t fit. Tolerance responsibility disappears between purchase orders. When we run fiber and CO2 lasers, CNC machining centers, press brakes, and FDM/resin/metal printers in one facility, tolerance gaps become our problem—and that’s a good thing for you. We catch interference before it becomes scrap.

One supplier, one part number, one quality record. That’s the real value of multi-process one-stop fabrication, and it’s why engineers consolidate prototypes here in the first place.

Is the lowest quote really the best value?

Here’s where my quality background makes me unpopular with procurement: the lowest quote is often the most expensive one. I once approved cheaper end mills to trim $180 from our own shop costs. The tool chipped on the 14th part and scrapped 30 components. The rerun cost us far more than the savings. That mistake changed how I look at every bid that comes in “significantly under” everyone else.

I’m not saying ignore price. I’m saying compare what the amounts include. Does the quote list inspection? Tolerance verification? Material certificates? If a supplier is 30 percent lower and hasn’t asked about functional surfaces, they haven’t understood the part. A $200 saving per piece doesn’t help when 10 percent of pieces fail in your incoming inspection.

The part starts delivering value when it’s installed and working, not when the purchase order is signed. Cheap saves money once. Quality saves money for the life of the product.

What quality documentation should you demand before approving a batch?

If you’re not sure what to review, start with three items: a dimensional report with actual measured values, not just “pass”; material certificates or traceability documentation; and a defined method for checking critical features. This isn’t bureaucracy. In 2022, when we formalized our first-article review protocol, we caught an assembly interference that would have delayed a 50,000-unit annual order.

That one catch paid for the entire quality system for the year. If a manufacturer can’t show the data, it’s because they didn’t measure the process. And no process can guarantee what it doesn’t measure.