Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Fused Deposition Modeling for Functional Prototypes
Transcript
- Lucas: So you're building a hardware prototype — probably something with an enclosure, maybe a few moving parts. You could go straight to CNC machining or injection molding, but that's expensive and slow for early iterations. That's where Fused Deposition Modeling, or FDM, comes in. It's not just for keychain toys anymore. Luna: Right — I think a lot of people still associate FDM with those hobbyist machines that print in basic PLA. But the material options now are pretty wild. Lucas: Exactly. We're talking carbon fiber filled nylon, polycarbonate, even PEEK for high-temperature applications. The key shift is that startups are now using FDM for functional prototypes — parts you can actually drop, torque, or expose to heat — not just visual mockups. Luna: Can you give me a concrete example? Who's doing this well? Lucas: There's a startup I spoke with — they're building a portable water quality sensor for field use. The device had to be waterproof, drop-resistant from a meter, and hold a battery pack plus a microcontroller. They started with SLA for the first visual models, but switched to FDM using carbon-fiber nylon for functional testing. Luna: And what did they learn from those FDM prototypes that they wouldn't have caught otherwise? Lucas: A lot. They went through 17 iterations before committing to injection molding. One issue was the snap-fit lid — it kept cracking along the layer lines. With FDM they could tweak the orientation and infill patterns overnight, test it the next morning, and iterate. They figured out that printing the lid flat with the snap-fit features oriented vertically gave the strongest engagement. If they'd gone straight to a hard tool, they'd have had to re-cut the mold — that's easily fifty to seventy thousand dollars and three weeks per change. Luna: Seventeen iterations is a lot — but at what cost? FDM is cheap per part, but the labor adds up. Lucas: They told me each prototype cost about $12 in material and maybe 4 hours of print time. Total material cost for all iterations was under $250. Their tooling was around $45,000 for the final mold. Without FDM, they estimated four extra mold revisions at roughly $17,000 each. So they saved at least $68,000 and cut four months off their timeline. Luna: That's a huge number for a startup burning through seed funding. So the financial case is clear — but what about the engineering limitations? FDM parts aren't isotropic. Lucas: Right — that's the biggest gotcha. The layer adhesion is the weak link. In standard FDM, the Z-axis strength can be only 40 to 60 percent of the XY strength. If you design a part that's loaded perpendicular to the layers, you'll get failures in the field. Smart startups design around this — they orient their prints so that the primary load paths run parallel to the build plate. Or they use annealing processes to improve layer bonding. Luna: I've seen some companies use polycarbonate for that reason — it's tougher than nylon and has better heat resistance. But it's also harder to print, right? Requires an enclosure and a hot end that can reach 300 degrees Celsius. Lucas: Yeah, polycarbonate is not a desktop machine material. You need an industrial-grade FDM printer with a heated chamber. But for a startup that's prototyping a device that will sit in a hot car or near a motor, it's worth the investment. One medical device startup I know uses FDM with ULTEM — which is a high-performance polyetherimide — to print surgical guide jigs. Those jigs have to be sterilizable and biocompatible for a single use. They're not making production parts, but they're making the tools that make the production parts. Luna: Interesting — so they're using FDM for jigs and fixtures, not just the prototype itself. That's a whole other use case. Lucas: Exactly. A lot of hardware startups overlook that. Low-volume assembly runs might be 50 or 100 units. Instead of machining a metal fixture for every step, you can print custom FDM jigs that hold parts in place for gluing, soldering, or fastening. They're cheap to replace if they wear out, and you can modify them in an afternoon. One startup making smart locks printed all their assembly fixtures on a single FDM machine — total cost under $500, versus maybe $5,000 for machined aluminum. Luna: That makes me think about the trade-off with other 3D printing technologies. When would you choose SLA or SLS over FDM for a functional prototype? Lucas: Good question. SLA gives you much better surface finish and finer detail, which matters for optics or fluid channels. But SLA resins are typically brittle and degrade under UV. SLS — selective laser sintering — produces nylon parts that are more isotropic and durable than FDM, but the machines are pricier and the post-processing is messier. FDM wins on material variety, cost per part, and speed of iteration if you don't need pristine surfaces. Luna: And the software side — are there good tools now for optimizing FDM prints for strength? I know some slicers let you vary infill patterns for different load areas. Lucas: Yeah, the slicers have gotten smarter. You can assign variable infill — dense near screw bosses, lighter in the core. Some even generate lattice structures that mimic bone for weight savings. And there are simulation tools that predict failure points before you hit print. But honestly, most of the startups I talk to still rely on practical testing — print it, break it, adjust the design, repeat. That iteration loop is the real advantage. Luna: It's a mindset shift too. Instead of trying to get the design perfect before prototyping, you embrace rapid iteration. That's hard for engineers who spent years in traditional manufacturing. Lucas: Totally. I've seen PhDs struggle with it — they want to simulate everything perfectly first. But the fastest path to a good design is often 20 quick FDM prints that fail early, rather than 2 injection mold iterations that fail expensively. The key is to know when to stop iterating on FDM and commit to the final process. Luna: So when do you draw that line? What's the signal that FDM has done its job? Lucas: For that water sensor startup, the signal was when they'd passed all their functional tests — drop test, water ingress test at one meter, battery life test — and the only remaining changes were about surface finish and cosmetic appearance. At that point you lock the design and go to injection molding. FDM can't replicate the smoothness or the material properties of molded plastic, but it can validate the geometry and the mechanics. Luna: One concern I hear from founders is that investors might look at an FDM prototype and think it's unprofessional. Does that ever hold them back? Lucas: It can, but it depends on the investor. Savvy hardware investors know that a rough-looking prototype that works is better than a glossy one that doesn't. Some startups keep a 'hero' SLA print for show and tell, but do all their testing with FDM. The real red flag is when a startup shows an FDM prototype that clearly hasn't been tested — you can see the layer lines aren't sanded, the fit is sloppy. That's a sign they're not iterating rigorously. Luna: So the FDM part should look like it's been used — maybe a bit scuffed, with tape holding something together. That's actually a good sign. Lucas: Exactly. I'd rather see a prototype with zip ties and hot glue than a pristine SLA model that's never been dropped. The FDM approach forces you to confront real-world constraints early. And honestly, if these conversations are useful for what you're building or running, that's exactly why we do this show — we want to help founders avoid the expensive mistakes we've seen others make. Luna: And a couple of dollars a month is genuinely what keeps these going. If you've gotten something out of it, buy me a coffee dot com slash fexingo makes a real difference — keeps us ad-free and focused on the practical stuff. Lucas: Yeah, it's that simple. No tiers, no perks, just knowing people find it useful. And it lets us keep digging into topics like FDM — which, by the way, has one more twist I want to mention. Luna: What's that? Lucas: There's a trend now toward using FDM to print the actual production tool — like the injection mold itself. Some startups are printing mold inserts out of high-temperature resin or metal-filled filament, then using them for short runs of 50 to 200 parts. It's a bridge between prototyping and full production. The parts won't have the same cycle time or surface quality, but for a pilot batch or clinical trial, it can cut tooling costs by 80 percent. Luna: That's fascinating — so FDM becomes part of the production chain, not just the design phase. That really changes the ROI calculation. Lucas: It does. But you have to be careful about the thermal and mechanical limits of the printed mold. It's not for high-volume or tight-tolerance parts. But for a startup that needs 100 units for a beta test, it can be the difference between a $50,000 mold and a $5,000 printed insert. The technology is evolving fast — I think we'll see more hybrid workflows where FDM is used for the prototype, the jig, and even the mold. Luna: And then the final product goes to traditional manufacturing. So FDM is really the glue that connects the idea to the factory floor. Lucas: That's exactly the right way to put it. If you're a hardware founder today and you're not using FDM for functional testing, you're leaving time and money on the table. The materials are there, the machines are accessible, and the iteration loop is faster than ever. The question is just whether you're willing to break a few prototypes before you find the one that works.