Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Thermal Imaging to Catch Defects
Transcript
- Lucas: So, Luna, you know how we've talked a lot about thermal cycling and environmental stress screening on this show? Well, there's a complementary tool that a lot of hardware startups are now adding to their test flow, and it's surprisingly accessible: thermal imaging. Luna: Thermal imaging? Like those FLIR cameras you see on construction sites? Lucas: Exactly the same technology, but in the last few years the cost of uncooled microbolometer sensors has plummeted. You can get a decent thermal camera module for under two hundred dollars now, and that's changing how startups do failure analysis. Luna: So instead of just baking a board and hoping it survives, you can actually watch it heat up in real time? Lucas: Right. And the key insight is that many defects show up as heat signatures before they show up as electrical failures. A slightly off resistor, a cold solder joint, a trace that's too thin—they all create localized hot spots under load. Luna: And you can catch those before they become a field failure that costs you thousands in warranty claims. Lucas: Exactly. Let me give you a concrete example. I was talking to a drone startup—they were seeing intermittent failures in their motor controllers, maybe one in fifty units would just stop mid-flight. They did all the standard tests: thermal cycling, vibration, even X-ray inspection on the boards. Luna: And nothing showed up? Lucas: Nothing. But when they ran a thermal imaging session under full load, they found a specific MOSFET that was running twenty degrees hotter than its neighbors on the same board. It wasn't failing during the test, but that heat signature told them the component was marginal. Luna: So they swapped the component and the failure rate dropped? Lucas: They swapped the supplier for that MOSFET, actually. The new parts ran within a couple degrees of each other. The fix cost them maybe a dollar per unit, but it saved them an estimated two hundred thousand dollars in warranty and return costs over the next year. Luna: That's the kind of return on investment that gets a CFO's attention. Lucas: And the beauty is, it's not just for drones. I've seen startups using thermal imaging to check everything from battery packs to LED drivers to medical devices. It's a general-purpose diagnostic tool. Luna: But isn't thermal imaging just a fancy way of looking at what you could measure with a thermocouple? Lucas: That's a fair point, but thermocouples give you a point measurement. Thermal imaging gives you a full spatial map. You can see not just that the board is hot, but where the hot spots are. And that's crucial for finding defects that are localized. Luna: So you can spot a single bad component among hundreds on a board, just by looking at the heat pattern? Lucas: Often, yes. And you can also catch assembly issues like a heatsink that isn't making proper contact, or a thermal pad that's misaligned. Those are almost impossible to catch with electrical testing alone. Luna: What about the accuracy? I've heard thermal cameras can be thrown off by things like emissivity and reflections. Lucas: That's a real limitation. The camera measures infrared radiation, and different materials emit that radiation differently. Polished metal, for example, has a low emissivity, so it looks cooler than it actually is. But with careful calibration and using a known reference, you can still get reliable relative measurements. Luna: So it's not about absolute temperature, but about spotting anomalies compared to a known-good board? Lucas: Exactly. You capture a golden unit, establish the thermal profile, and then compare every production unit against that. Any significant deviation is a red flag. Luna: That's a smart approach. But how do you scale that to high-volume production? I imagine taking a thermal image of every board would slow things down. Lucas: Good question. In high-volume, you don't necessarily image every unit. You might do it on a sampling basis, or you integrate it into an automated test fixture that runs a quick thermal sweep. Some startups are even using AI to analyze the thermal images and flag anomalies automatically. Luna: ai powered thermal inspection. That sounds like something we could talk about all day. But for now, what's the one thing our listeners should take away? Lucas: That thermal imaging is no longer a lab-only tool. It's affordable, it's fast, and it can catch defects that other tests miss. If you're a hardware startup not using it in your failure analysis, you're leaving money on the table. Luna: And that's a lesson that applies to more than just hardware. It's about being willing to look at your product from a different angle, literally and figuratively. Lucas: Yeah. Speaking of looking at things from a different angle, I want to mention something. If you're finding value in these conversations, and you want to keep them coming without ads, listener support is what makes that possible. It costs about the same as a coffee to help us out. Luna: And we genuinely appreciate it. We keep the show independent and focused on the topics that matter to you. Lucas: You can do that at buy me a coffee dot com slash fexingo. No pressure, just if it's useful to you. Now, back to thermal imaging—one more thing that's interesting is how it's being used in the field, not just in the lab. Luna: Oh, you mean like technicians using thermal cameras to diagnose failures in installed systems? Lucas: Right. For example, a startup making industrial IoT sensors found that their customers were using the thermal imaging feature on their phones to spot overheating components in their own equipment. That gave the startup an idea for a new service: they now offer remote thermal diagnostics as part of their monitoring platform. Luna: So a tool that started as an internal QA method turned into a revenue stream. That's a nice pivot. Lucas: It is. And it shows that thermal imaging isn't just about catching defects; it's about understanding your product's health over its entire lifetime. Which, for a hardware startup, is a huge differentiator. Luna: I hadn't thought about it that way. You're not just making a product; you're building a relationship with how it behaves in the real world. Lucas: Exactly. And with the cost of thermal sensors continuing to drop, I suspect we'll see even more creative uses. Maybe even in consumer devices. I'm thinking laptops that can self-diagnose cooling issues, or electric vehicles that can tell you a battery cell is degrading before it fails. Luna: That's the future. But for now, what's the quickest way a startup can start using thermal imaging? Lucas: Start simple. Buy a low-cost thermal camera module, set up your test bench, and just start looking at your boards under load. You'll be surprised at what you see. And if you're in the middle of a puzzling failure, it might just be the missing clue. Luna: And if you need help interpreting the images, there's a growing community of engineers sharing their experiences. A little humility and curiosity goes a long way. Lucas: Well said. That's all for today. We'll be back next week with another tool that hardware startups are using to build better products. Until then, keep your boards cool and your prototypes closer to the spec.