Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / Hardware Startups Should Use Selective Laser Melting Early
Transcript
- Lucas: So you're a hardware startup, you've got a design for a metal part — maybe a drone motor mount, maybe a custom bracket — and you need, say, fifty units to ship to beta testers. The traditional move is CNC machining, right? But there's another option I think more founders should know about early: selective laser melting. Luna: That's the metal 3D printing process — powder bed fusion with a laser. I've seen it mostly used in aerospace and medical implants. Is it practical for a startup with a tight budget? Lucas: That's the exact question. And the answer is: it depends on how many parts you need and how much you're willing to pay per part. But the trade-offs are shifting. Let me give you a concrete example. I was talking to a founder in Austin who makes autonomous drone frames. His first version was CNC'd from 6061 aluminum — twelve weeks lead time, and he had to pay for the fixture up front. That's a few thousand dollars before he even sees a chip. Luna: Ouch. And if the design changes after the first prototype, you're paying for a new fixture. Lucas: Exactly. So for his second revision, he tried selective laser melting — SLM. He sent a STEP file to a bureau in Austin that runs an EOS M 290. Five days later, he had twenty parts in 316L stainless steel. Total cost: about forty dollars per part. That's more per unit than CNC at volume, but the lead time was a fraction, and he had zero tooling cost. He could make a tweak to the file, resubmit, and get new parts in the same five-day window. Luna: But stainless steel is heavier than aluminum. For a drone frame, weight is everything. Did that hurt performance? Lucas: Great point. Actually, he redesigned the part for SLM. Instead of machining a solid block, he used a lattice structure inside the arms — like a truss. The final part was actually lighter than the original CNC'd aluminum version because he could remove material where it wasn't needed. That's the design freedom you get. With CNC, you're limited by the tool's access. With SLM, you can build organic shapes, internal channels, lattices. You can consolidate multiple parts into one. Luna: So the real benefit isn't just speed — it's enabling a different geometry that can outperform the original. Lucas: Right. And that's the argument for using SLM early in product development. Before you commit to injection molding or die casting, you can iterate on the actual metal part, not a plastic prototype. Plastic doesn't behave like metal under load, so you're guessing on stiffness and thermal performance. With SLM, you're testing in the final material — or close to it. Luna: What's the catch? There's always a catch with additive. Lucas: Several. First, surface finish. As-built SLM parts have a rough surface — about 6 to 10 micrometers Ra. If you need a smooth mating surface for a seal or a bearing, you'll need post-processing: machining or maybe vibratory finishing. That adds time and cost. Second, support structures. Overhangs need supports, and those have to be removed manually. That can add a day or two and some labor cost. Luna: And the per-part cost doesn't scale down the way CNC does. With CNC, the more you make, the cheaper each part gets because the setup is amortized. With SLM, you're paying for machine time and powder — and the machine time is roughly the same for one part or ten, because you fill the build plate. Lucas: Exactly. So the economics flip. For small batches — under about five hundred units — SLM can be cheaper than CNC if you factor in tooling. But above that, CNC wins on per-unit cost. There's a nice paper from MIT that modeled this for aerospace brackets. The break-even point was around four hundred parts. For startups, that sweet spot is usually in the low hundreds. Luna: So the advice would be: if you need fewer than five hundred metal parts and you're still iterating on the design, SLM is worth a serious look. But if you're scaling to thousands, get your design locked and then invest in CNC or casting. Lucas: Exactly. And there's another dimension: material choice. The most common SLM materials are 316L stainless, Ti-6Al-4V, and AlSi10Mg — an aluminum alloy. 316L is great for corrosion resistance and strength, but it's heavy. The titanium alloy is expensive but offers the best strength to weight. And the aluminum alloy is lighter but not as strong as 6061-T6. You have to match the material to the application. Luna: And the aluminum alloy for SLM isn't the same as wrought 6061. The powder is a cast-type alloy with silicon and magnesium. So your mechanical properties might be different from what you'd expect with standard aluminum. Lucas: Right. So you really need to understand the datasheet. But here's the thing — if you're a startup, you probably don't have a materials engineer on staff. So you rely on the service bureau. And good bureaus will help you with design for additive manufacturing — DFAM. They'll tell you, 'Hey, this wall is too thin for powder removal,' or 'You need a bigger radius here to reduce stress concentration.' Luna: Have you seen any startups that really messed this up? Lucas: Yeah, one story that sticks. A friend of mine was designing a custom heat sink for a high-power sensor. He sent an STL file to a bureau, and they printed it in 316L. But he didn't account for the fact that SLM parts have internal stresses from the rapid heating and cooling. When he bolted it down, it warped. The solution was to design a stress-relief cutout and to specify a stress-relief anneal after printing. That added two days and a couple hundred dollars, but it solved the problem. Luna: So you need to think about the entire process, not just the printing. The post-processing is half the story. Lucas: Exactly. And that's why I think every hardware founder should at least get a quote from an SLM bureau before committing to CNC tooling. The quote is usually free, and you'll learn a lot about what's possible. Even if you don't use SLM for production, you might use it for prototypes, jigs, fixtures, or tooling inserts. I've seen startups print custom end of arm tooling for their assembly line — parts that would take weeks to machine, printed overnight. Luna: That's a good point. The same technology that makes your product can also make your factory more efficient. Lucas: And honestly, if today's episode was useful for something you're building or running, there's a link on the show page where you can buy me a coffee — it's buy me a coffee dot com slash fexingo. Small gesture, but it genuinely helps keep this show ad-free and focused on practical content. Luna: Yeah, it's a nice way to say this was worth your time. And we appreciate it. Lucas: Alright, back to the tech. I want to mention one more thing: the cost of SLM machines themselves has come down. Ten years ago, a mid-range system was half a million dollars. Today, you can get a desktop metal printer for under a hundred thousand. Some startups are even buying their own machines for in-house prototyping. The trade-off is that you then need to learn powder handling, safety protocols — it's not trivial. Luna: What about lead times from bureaus? Are they consistent? Lucas: Generally, yes. Most bureaus quote one to two weeks for standard parts. Rush orders can be three to five days for a premium. The bottleneck is usually the build plate — they batch parts to fill it. So if you design a small part that fits in a gap, you might get a discount because they can nest it with other jobs. Luna: So design your parts to be compact and stackable. That's a DFAM principle. Lucas: Exactly. And one final thought: if you're considering SLM for a structural part, make sure you understand the fatigue properties. Layer-by-layer build creates anisotropy — the part is weaker in the Z direction. For a bracket that experiences cyclic loading, you want the layers oriented perpendicular to the load. That's something you discuss with the bureau upfront. Luna: So it's not a magic bullet. But for the right use case — small batches, complex geometry, fast iteration — it can be a game changer. Lucas: Yeah. And the key is to try it early. Get a test coupon printed in your material, measure it, test it. Don't wait until you're in production to discover that your part warps or your surface finish is too rough. Prototype in the real material, with the real process. Luna: Good advice. I think a lot of listeners will be looking at that quote from a bureau this week. Lucas: I hope so. It's one of those tools that's still underused in the startup world, but the barriers are lower than most people think.