Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Friction Stir Welding for Aluminum Enclosures
Transcript
- Lucas: Friction stir welding — FSW — is one of those processes that sounds almost too niche to matter for a small hardware startup. But I've now talked to three founders in the past six months who've put it at the center of their assembly strategy, and the results are pretty striking. Luna: I've heard the term, but I'll admit I don't have a clear picture of how it actually works. Isn't it basically just rubbing a spinning pin along a seam until the metal softens? Lucas: That's actually pretty close. A rotating tool — with a shoulder and a smaller pin — plunges into the joint line and literally stirs the material together in a plasticized state. But here's the key: it never melts the base metal. The peak temperature stays below the melting point, usually around 80 to 90 percent of that. So you avoid all the problems that come with fusion welding — porosity, hot cracking, distortion from thermal contraction. Luna: Right, so it's a solid-state join. That makes sense for aluminum, which is tricky to weld conventionally without a lot of pre- and post-weld treatment. Lucas: Exactly. And for startups, the biggest win is that you can join thin-walled extrusions — say 1.5 millimeter thick 6061 aluminum — into a strong, hermetic enclosure with almost zero filler material. No epoxy to cure, no fasteners to torque, no gaskets to leak. One drone startup I know switched from epoxy bonding to FSW on their main battery housing. They cut assembly time by 40 percent and got an IP67 seal on the first try. Luna: Forty percent is huge when you're doing runs of a few hundred units. What was the catalyst for the switch? Was it just speed, or were there failures with the epoxy? Lucas: Both, actually. The epoxy had a 24-hour cure cycle, which created a bottleneck. And they had field failures where the bond line failed after thermal cycling — the drone would go from a hot landing pad to a cold altitude, and the epoxy would get brittle and crack. FSW gave them a metallurgical bond that's essentially the same strength as the base metal. No cure time, no consumables, no adhesive shelf-life issues. Luna: So the cost equation shifts. Instead of buying epoxy in bulk and managing dispensing robots, you buy a tool and a spindle. But FSW equipment isn't exactly cheap, right? Lucas: It's not a hundred-dollar gadget. A dedicated FSW machine can run $100,000 to $200,000. But what I'm seeing more of is startups retrofitting their existing three-axis CNC mill with an FSW toolholder and a custom clamping fixture. That retrofit kit — toolholder, controller, and a few polycrystalline cubic boron nitride tools — runs around $50,000. If they already have a Haas or a Tormach, they're in business. The learning curve is maybe two weeks to get consistent welds. Luna: And that fits the startup pattern: leverage existing capital equipment rather than buying a dedicated machine. But what about the geometry limitations? FSW is typically linear or 2D curved seams, right? Can you do corners? Lucas: You can do corners, but it's trickier. The tool has to dwell or change direction, and that can leave a keyhole at the end of the weld if you're not careful. Most startups design around it — they use FSW for long straight seams on a box-like enclosure and then seal the corners with a small fillet weld or a corner bracket. The trade-off is still worth it because the main seam is the one that takes the most stress and the most leak risk. Luna: I want to come back to the thermal side. You mentioned no distortion — that's a huge deal for precision enclosures like optical housings. How does FSW manage that? Lucas: Because the heat-affected zone is much narrower. In fusion welding, the molten pool spreads heat widely, and as it solidifies and cools, the shrinkage pulls the workpiece out of shape. With FSW, the heat is generated by friction and plastic deformation right at the tool interface, and it's conducted away quickly through the clamp and the backing plate. The result is typically less than 0.1 millimeter of distortion per meter of weld. That's fine for a drone battery box, but for a camera housing where lens mounts need to stay within 50 microns, you still need post-weld machining. Luna: So it's not a cure-all, but it eliminates the need for stress relief and straightening that you'd need with MIG or TIG. That saves days of lead time. Lucas: Exactly. And there's another angle that I think is underappreciated: the cosmetic finish. With the right tool and feed rate, FSW leaves a smooth, almost brushed-metal surface that looks great out of the fixture. A startup making LED architectural lighting told me they stopped painting their enclosures after switching to FSW — they just anodize the welded assembly, and the weld line is nearly invisible. That's a whole step removed from the BOM. Luna: That's a strong argument. But let's talk about the practical gotchas. What are the failure modes that a startup that's new to FSW should watch out for? Lucas: The most common one is a wormhole defect — a void that forms inside the weld if the tool feed rate is too high or the plunge depth is too shallow. You can't see it from the outside, but it destroys the strength and the hermeticity. The fix is to do a process qualification: weld a few test coupons, cross-section them, and look for voids. Another issue is surface galling if the tool shoulder gets too hot and starts sticking to the aluminum. That's usually a sign you need to adjust the tool offset or use a different tool material. Luna: And you mentioned spiral marks — I've seen those on some consumer products. Is that purely cosmetic or a structural concern? Lucas: Mostly cosmetic. A spiral mark is the imprint of the tool shoulder's scroll pattern — it's designed to help move material, but if the feed rate is too slow relative to the spindle speed, it leaves a visible ring pattern. On a consumer product, that's a finish issue. You can minimize it by optimizing the feed to speed ratio, or you can just polish it off in a secondary operation. But for a startup, the lesson is: don't assume your first weld parameters are good. Spend a day dialing in speed, feed, and plunge depth. Luna: You know, this conversation actually reminds me why I appreciate shows like this — it's the kind of practical, real-world detail that you don't get from a vendor datasheet or a YouTube tutorial. And it's ad-free, which honestly makes it feel more trustworthy. Lucas: That's exactly the feedback we hear most. The show stays independent because listeners who find it valuable occasionally throw a few bucks in the tip jar. If today's episode saved you a week of assembly headaches, you can find us at buy me a coffee dot com slash fexingo. No subscription, no patreon tiers — just a coffee's worth of appreciation. Luna: Yeah. And it keeps us free to talk about whatever actually moves the needle for builders and operators. So, back to FSW — what about dissimilar metals? Can you weld aluminum to steel or copper with this process? Lucas: That's a frontier area. FSW can join some dissimilar combinations — aluminum to magnesium is fairly well established, and aluminum to copper is possible with careful parameter control. But aluminum to steel is tough because the intermetallic compounds that form at the interface are brittle. Some researchers have done it with a zinc interlayer, but it's not production-ready for a startup. For now, if you need an aluminum enclosure with a steel insert, you're better off designing a mechanical interlock or using a threaded insert. Luna: Got it. So FSW is really for when the whole assembly is aluminum, or maybe aluminum to its own alloys. But that covers a huge range: drone frames, robot arms, lighting, sensor housings, EV battery enclosures... Lucas: Exactly. And one more application I've seen recently is thermal management. A startup making LED grow lights uses FSW to join an aluminum heat sink base to a fin stack. Because the weld is solid-state, the thermal conductivity across the joint is nearly as good as a monolithic piece. That's a big deal compared to an epoxy bond, which can add 5 to 10 degrees Celsius of thermal resistance. Luna: So it's not just structural — it's functional. The same weld that seals the enclosure also conducts heat. That's the kind of multi-functionality that hardware startups love. Lucas: And it ties into the broader trend of bringing production in-house. When you control the welding process yourself — rather than outsourcing to a job shop that uses MIG or TIG — you get faster iteration cycles and you can tweak the design without waiting for a supplier's schedule. One founder told me he can go from a revised CAD file to a welded prototype in the same afternoon. Luna: That speed is the startup advantage. So if a team is evaluating FSW, what's the minimum viable setup they should look at? Lucas: Honestly, start with a manual mill or a small CNC with a rigid frame — something like a Haas Mini Mill or a Tormach 1100. You'll need a custom fixture that clamps the parts tightly to a backing plate. The FSW tool itself — a shoulder diameter of about 12 millimeters for 2-millimeter aluminum — costs maybe $200 for a basic H13 steel tool, though polycrystalline cubic boron nitride tools that last longer are more like $500 each. Then you need a toolholder that fits your spindle taper. Total non-recurring investment: under $5,000 if you already have the mill. The $50,000 number I mentioned earlier is for a fully integrated retrofit with force control and data logging. Luna: That's a much lower barrier. And the learning curve — you said two weeks to get consistent welds? Lucas: For simple butt joints, yes. You spend the first few days just finding the right spindle speed and feed rate. A good starting point for 6061-T6 at 2 millimeters thick is 1500 RPM and 100 millimeters per minute feed. Then you adjust based on the surface finish and the sound — a good weld sounds like a steady hum, not a chatter. Once you can make a sound weld, you move to more complex geometries. But I'd recommend any startup considering FSW to first run a design of experiments with a few key parameters: spindle speed, feed rate, plunge depth, and tool tilt angle. Luna: That DoE approach is exactly what we covered in episode 56, but for tolerances. So there's a nice cross-reference. Lucas, is there any product category where you think FSW is a bad fit? Lucas: Yeah, a few. If your enclosure has complex 3D curves — like a handheld device with organic contours — FSW is hard to program and the tool geometry limits you. For those, you're better off with die casting or injection molding. Also, if your volumes are under 100 units per year, the setup time might not be worth it compared to just using epoxy or mechanical fasteners. And if you need to disassemble the enclosure for servicing, FSW is a permanent joint — you can't un-stir it. So think about your repairability strategy. Luna: Good points. So it's a tool in the toolbox, not a silver bullet. But for the right product — sealed, structural, aluminum — it's a process that can give startups a real edge in quality and speed. Lucas: Exactly. And I think we'll see more of it as the equipment costs come down and the knowledge spreads. The fact that you can do it on a standard CNC mill means it's accessible to any startup that's already invested in machining. That's a pretty compelling on-ramp.