Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Potting Compounds for Vibration Protection
Transcript
- Lucas: You've designed a circuit board. It works perfectly on the bench. Then you put it inside a product that gets dropped, shaken, or splashed. And suddenly it stops working. Luna: This is the moment a lot of hardware founders discover potting compounds. Lucas: Exactly. Today we're talking about potting — which is basically encasing sensitive electronics in a solid resin. Epoxy, urethane, silicone. You pour it over the board, it cures, and now your components are locked in place, insulated from moisture and vibration. Luna: It's different from conformal coating, which we covered in episode 91. That's a thin layer — maybe 50 microns. Potting is much thicker. We're talking millimeters. Lucas: Right. Conformal coating is like a raincoat. Potting is like entombing the board in concrete. And that changes the design constraints completely. You have to think about thermal management because the resin traps heat. You have to think about rework because once it's potted, you're not swapping a capacitor. Luna: So when does it make sense for a hardware startup? I mean, it adds cost and complexity. Lucas: It makes sense when the product's operating environment is genuinely hostile. Think about a drone that might crash into a tree. Or a sensor that sits on a vibrating factory machine. Or a handheld device that gets dropped on concrete regularly. Those are the use cases. Luna: I was reading about a startup that makes compact lidar units for agricultural robots. They potted the entire optical assembly because dust and constant vibration were killing their field reliability. Lucas: That's a perfect example. And there's a specific decision they had to make: which resin? Epoxy is very hard, great for vibration damping, but it shrinks slightly as it cures, which can stress components. Urethane is more flexible, better for thermal cycling, but it doesn't adhere as aggressively. Silicone is the softest, best for extreme temperature ranges, but it's expensive and doesn't provide as much mechanical support. Luna: So you're trading off rigidity versus thermal performance versus cost. Lucas: Exactly. And the right choice depends on what's actually killing your boards. If it's high-frequency vibration — say a motor spinning at ten thousand RPM — you want a hard epoxy that prevents components from moving relative to each other. If the board goes through freeze-thaw cycles, you want urethane that can expand and contract without cracking. Luna: What about the process itself? Is it something a startup can do in-house or do you need to outsource? Lucas: You can absolutely do small runs in-house. There are manual dispensing syringes, two-part cartridges that mix as you dispense. The key is degassing — you have to vacuum out the air bubbles before the resin cures, otherwise voids create weak points. A lot of startups use a small vacuum chamber that costs maybe a thousand dollars. Luna: And then you have to factor in cure time. Some epoxies cure in 24 hours at room temperature. Others need heat to accelerate. Lucas: Right. And if you're doing a thousand units a month, you might invest in a heated cure oven to get cycle times down to an hour. That changes your cost per unit dramatically. At small scale, potting might add five to ten dollars per board in material and labor. At volume, it can drop below a dollar. Lucas: But there's a hidden cost: rework. If a potted board fails in the field, you can't just desolder a component. You have to either discard the whole assembly or use chemical strippers to dissolve the resin, which is messy and not always successful. Luna: So it's a commitment. You're basically saying: this board is going to be reliable enough that we don't expect to repair it. Lucas: Exactly. And that's why some startups design their boards so that only certain sections are potted. They use a dam — a silicone bead that contains the liquid resin to just the vulnerable area — and leave connectors or test points exposed. That way you can still access diagnostics. Luna: That's clever. So you get protection where you need it, but you don't sacrifice all repairability. Lucas: Right. And the dam itself is an additional step, but it's worth it if you think you'll ever need to diagnose a field failure. I've seen startups use 3D printed dams for prototyping, then switch to machined metal dams for production. Luna: Let's talk about a real example. Do you have a specific case where potting made or broke a product? Lucas: Yeah, there's a drone startup I've been following — they make a small agricultural spray drone. Early prototypes kept losing flight controllers after hard landings. The vibration sensors were reading spikes over 30 Gs. They tried conformal coating, but the components still came loose. So they potted the entire flight controller board in a urethane-based compound. Luna: And that solved it? Lucas: Completely. Field failure rate dropped from about eight percent to under half a percent. The trade-off: they had to add a small heatsink to the processor because the urethane trapped heat, and the processor temperature went up by about eight degrees Celsius. But it was within spec. Luna: Eight degrees is manageable. And the cost per board? Lucas: At their volume — about five hundred units a month — they were spending about three dollars per board on the potting material and labor. That's a small price for a tenfold reliability improvement. Luna: That's the kind of calculation that makes sense. And it's exactly the kind of thing we look at on this show. If these conversations are useful for what you're building or running, it's worth mentioning that we keep this show entirely ad-free by design. There is no sponsor, no mid-roll, no brand integration. Lucas: Yeah, and that's a deliberate choice. We want these episodes to be purely about what works in hardware, without any outside influence. If that matters to you, and you want to support that, there's a simple link: buy me a coffee dot com slash fexingo. Luna: It's a small gesture that keeps the whole library free and open. So back to potting — one thing we haven't touched on is testing. How do you verify that your potting process is actually working? Lucas: Great question. The standard is to do thermal cycling and vibration testing on a sample from each batch. You pot a few boards, then put them through say a hundred cycles of minus forty to plus eighty-five Celsius, while running a vibration profile that matches your product's expected environment. Luna: And look for cracks, delamination, or changes in electrical performance. Lucas: Exactly. One thing startups often miss is that the resin itself can absorb moisture over time, which changes its dielectric properties. So you want to do a moisture sensitivity test as well. Leave potted boards in a humidity chamber at 85 percent relative humidity for a week, then measure insulation resistance. Luna: That's a lot of testing for a small team. But it's cheaper than a recall. Lucas: Absolutely. And there are contract test labs that will do this for a few thousand dollars. For a startup that's about to ship a thousand units, that's cheap insurance. Luna: So if someone's listening and thinking about potting for their product, what's the first step? Lucas: Start with a prototype. Get a few bare boards, some cheap epoxy from a supplier like MG Chemicals, a vacuum chamber, and a mold made from silicone or 3D printed plastic. Pot one board and test it against an unpotted board in your actual use environment. Measure the difference in failure rate. That'll tell you if potting is worth the cost. Luna: And if it is, then you choose your resin carefully. Lucas: Right. Talk to suppliers. They have application engineers who will recommend a specific compound based on your temperature range, vibration frequency, and chemical exposure. Don't just guess. These materials are chemistry. Get it wrong and your boards crack or your components corrode. Luna: Good advice. I think that's a solid note to end on. Lucas: Potting is one of those techniques that seems old-school, but it's incredibly effective for the right problem. If you're building something that has to survive real-world abuse, it's worth a look.