Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / Why Hardware Startups Use Accelerated Life Testing
Transcript
- Lucas: So there's this smart lock startup I was reading about, and they had a seal that was supposed to last years. But they ran an accelerated life test, and the seal cracked in about three weeks. That's the kind of story that makes you realize why hardware startups can't just ship and hope. Luna: Three weeks to simulate years—how does that even work? Are they just running it 24/7 or is there something smarter going on? Lucas: It's a mix of both, but mostly it's about cranking up the stresses that cause failure. Higher temperature, more humidity, heavier vibration, faster cycling. The idea is to force the same degradation that would happen over years into a short window. That's accelerated life testing in a nutshell. Luna: That sounds like exactly the kind of thing I'd want to know about if I were building a device. And if today's episode is useful for something you're building or running, the way we keep these conversations ad-free is through listener support. You can buy us a coffee at buy me a coffee dot com slash fexingo. It genuinely helps. Lucas: Yeah, and we appreciate that. But back to the test itself—the smart lock case is a perfect example because it wasn't a complex failure. It was just a rubber seal that dried out and cracked. But in the field, that would have taken maybe three years to show up. In a test chamber, they can push the temperature up and drop the humidity to near zero, and suddenly that same failure happens in days. Luna: So the environment is the enemy here. Humidity, heat, cold, vibration—pick your poison. But how do you decide which ones to apply? You can't test for everything. Lucas: Right, and that's the art. You start with a failure modes and effects analysis. You ask: what's likely to break? For a smart lock, it's the motor, the electronics, the seals. Then you pick the stresses that accelerate those specific failures. Temperature cycling is huge for solder joints. Vibration is big for connectors and mounts. Humidity attacks seals and PCB coatings. Luna: But you can't just crank the heat to a thousand degrees and call it a day. There's a reason we don't test everything at extreme levels, right? Lucas: Exactly. There's a sweet spot. If you go too extreme, you might create failures that would never happen in real life. Like, you melt the plastic housing, and that's not a realistic failure mode. So you use models to translate test stress into real-world years. The Arrhenius equation is the classic one for temperature. It says that for every ten degrees Celsius you raise the temperature, the reaction rate roughly doubles. Luna: So that's the math behind why a few weeks at high temperature can stand in for years at room temperature. But temperature isn't the only stress. What about thermal cycling? That's about expansion and contraction, right? Lucas: Right. Thermal cycling is for solder joints and mechanical attachments. The Coffin-Manson model relates the number of cycles to the strain range. So you might do hundreds of cycles from minus twenty to plus sixty degrees Celsius. That can reveal cracks that would take years of seasonal changes to appear. And vibration—that's another common one, especially for anything that's going to be mounted on a door or a wall. Luna: Do startups actually buy these expensive chambers? I've seen prices for a basic thermal chamber starting at like twenty grand. That's a lot for a seed-stage company. Lucas: A lot of them don't buy. They rent or they use third-party labs. There's a growing ecosystem of test as a service companies. You send them your prototype, they run the test, and you get a report. That's way cheaper than buying equipment. The smart lock startup I mentioned—they used a lab that specialized in accelerated aging for outdoor electronics. Cost them maybe five thousand dollars for a full test cycle. Luna: Five grand to catch a seal failure that would have caused a flood of warranty claims and bad reviews. That seems like a no-brainer. But I'm guessing there's a catch—you have to know what you're looking for, or you'll just get a list of failures that don't matter. Lucas: That's the real challenge. The test will always find something. The question is whether it's a real risk or just a test artifact. Experienced reliability engineers know how to interpret the results. They look at the failure mode and ask: would this happen in the field? For the smart lock, the seal crack was clearly real. But a startup might see a solder joint crack and think they have to redesign the board when it was just over-stressed in the test. That's why you want an engineer who's seen a few of these. Luna: So what's the actual return on investment? If you do this test, what do you get out of it besides knowing that your seal might fail? Lucas: You get a predicted lifespan. You can say with some confidence that the product will last, say, five years under normal use. That's huge for marketing, but also for design decisions. You can compare different materials, different seal designs, and see which one lasts longer. It's a way to make data-driven choices instead of guessing. Luna: That makes sense. But I've heard that some startups skip accelerated testing entirely because they're in a rush to launch. What happens then? Lucas: Sometimes it works out fine. But when it doesn't, it's brutal. There was a fitness tracker startup a few years back that skipped vibration testing. The first batch of devices started rattling apart after a few months because a connector wasn't robust enough. They had to do a recall, and that killed their momentum. A test would have cost them maybe ten grand. The recall cost them millions and their reputation. Luna: Ouch. So it's not just about engineering—it's about business survival. You're protecting your brand, not just your hardware. Lucas: Exactly. And the nice thing is that you don't have to do a full HALT test on every prototype. You can start with a simple thermal cycle or a basic humidity test. Even that can catch the low-hanging fruit. As you get closer to production, you can invest in more comprehensive tests. Luna: I guess the key is to start early. If you test too late, you might have to redesign the whole enclosure, and that's a huge delay. Testing a prototype that's still flexible is way cheaper than testing something that's already in tooling. Lucas: Yeah, you want to iterate on the test results before you commit to injection molding. That's when changes are cheap. Once you have a mold, every change is thousands of dollars and weeks of delay. So run the test, find the weak points, fix them, and then run it again to verify. Luna: And there's also the data angle. I've seen startups using AI to analyze test data and predict failures even before they happen. Is that a thing yet? Lucas: It's emerging. You can use machine learning to look at sensor data from the test chambers and find patterns that correlate with early failures. But you need a lot of data to train those models. Most startups don't have that volume yet. Still, even simple statistics can help. You can track how many cycles it takes for a sample to fail, and use that to estimate the distribution. Luna: So it's not about replacing the engineer with AI. It's about giving the engineer better tools. That makes sense. Lucas: Right. And one more thing: accelerated life testing isn't just for the final product. You can use it to qualify components. Say you're choosing between two batteries. Run a thermal aging test on both, see which one degrades more slowly. That's a quick, cheap way to make a better sourcing decision. Luna: That's a smart way to think about it. It's not just a pass-fail gate; it's a source of information for a bunch of decisions. Lucas: Exactly. So if you're a hardware startup and you haven't thought about accelerated life testing yet, I'd say start with the basics. Pick one likely failure mode, run a simple test, see what happens. You might be surprised what you find. Luna: And if you do, you'll save yourself a lot of pain down the road. Thanks for listening, everyone.