Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Acoustic Leak Detection for Sealed Products
Transcript
- Lucas: So Luna, you've had one of those moments where you drop a gadget in water and you just freeze, waiting to see if it survives. Luna: Oh, definitely. I once dropped my phone in a puddle — it was fine, but the panic was real. And now I expect every new device to survive that, honestly. Lucas: Right — and that expectation is why hardware startups are obsessed with leak testing. If your product claims to be waterproof, you better be sure. But here's the thing: traditional leak testing — like submerging a device in water and looking for bubbles — is slow, messy, and can actually damage the product. Luna: Yeah, I've heard of 'dunk testing' — they fill a tank, pressurize it, and watch. But you're saying there's a better way? Lucas: There is. It's called acoustic leak detection. And it's been quietly adopted by everyone from Tesla to the people making the Apple Watch. The idea is elegant: you pressurize a sealed product with air, then listen for the sound of a leak using an ultrasonic microphone. Luna: Wait — you literally hear the leak? Lucas: Not with your ear. The escaping air creates a high-frequency sound above our hearing range — think 40 kilohertz and up. A sensor picks that up, and software maps it to a leak rate. You can detect a hole as small as half a micron — that's 500 nanometers. Luna: Half a micron. So a bacterium could swim through it, but water vapor might not. That's incredibly sensitive. Lucas: Exactly. And the test takes about three to five seconds per unit. Compare that to dunk testing, which can take a minute per unit, plus drying time. For a startup trying to do high-volume production, that speed difference is huge. Luna: So who's actually using this? I imagine it's not just big companies. Lucas: It's spreading fast in the startup world. I talked to a founder last month — she runs a small company making waterproof drones for agricultural surveying. Their first batch had a 12 percent return rate because of water ingress. After switching to acoustic leak detection in production, returns dropped to under 2 percent. Luna: Twelve percent returns — that would kill most hardware startups. The cost of replacements, shipping, customer trust... Lucas: Exactly. And the fix wasn't expensive. They bought a desktop-sized acoustic chamber for about eight thousand dollars. It fits on a workbench. They test every unit before it goes in the box. Luna: Eight thousand dollars — that's within reach for a seed-stage startup. What about the air supply? Do you need a clean dry air system? Lucas: You need a compressed air source with a filter, yes. Most factories already have that. The real cost is the fixture — the custom jig that holds your product and seals against it. That can be another few thousand, depending on complexity. But overall, you're looking at under twenty thousand for a basic setup. Luna: That's not nothing, but it's in the same range as a good 3D printer. And it saves you from a recall. Lucas: Right. And there's an even cheaper option for early prototyping: you can send units to a test lab that offers acoustic leak testing. A typical test costs about fifty to a hundred and fifty dollars per unit. You send five prototypes, you get detailed leak rate data. Luna: That's smart. Validate the design before you invest in your own chamber. Lucas: Exactly. And the data tells you more than just pass-fail. It gives you the actual leak rate in cubic centimeters per minute. So you can set a threshold — say, anything under 0.1 cc per minute passes, anything above fails — and track trends. If you see the leak rate creeping up over a production run, you know something is wearing out in your sealing process. Luna: That's a huge quality advantage. So what kinds of products need this? Obviously anything with an IP67 or IP68 rating. Lucas: Yeah, consumer electronics are the big category: smartwatches, fitness trackers, Bluetooth earbuds, action cameras. But also medical devices that need to be sterilized — if you seal a device, you need to know it stays sealed after autoclaving. And automotive components like sensors and ECUs. Luna: I've also seen it in outdoor IoT devices — like solar-powered sensors for agriculture. They sit in rain and dust for years. Lucas: Exactly. And one thing that surprises founders: the seal doesn't have to be perfect on day one. A lot of gaskets and O-rings seat better after a few thermal cycles. So you might test, then run the device through some temperature changes, then test again. Luna: That makes sense. The rubber relaxes into place. Lucas: Right. And that's why you need to correlate your leak test with real-world performance. No test is perfect — acoustic detection is very good, but it's not infallible. You still need a sample batch to go through actual submersion testing for validation. Luna: So you use acoustic testing for 100 percent production screening, and dunk testing for engineering validation. Lucas: Exactly. And if you're a startup, that's a smart split. You don't want to be dunking every unit; it's slow and it can get water into connectors. But you want every unit tested. Luna: We talk a lot on this show about methods that save startups from costly mistakes. And this one seems like a classic: spend a little upfront on test equipment, avoid a nightmare later. Lucas: It really is. And you know, if you find these conversations useful for what you're building or running, listener support is what keeps the show ad-free and independent. We don't run sponsors, we don't sell your data. It's just a small group of curious operators and builders who chip in at buy me a coffee dot com slash fexingo. No pressure, no perks — just helps us keep the lights on. Luna: Yeah, it's pretty straightforward. And we appreciate everyone who's already contributed — it literally pays for the editing and hosting. Lucas: Alright, back to leak testing. One topic that comes up a lot is how to design your enclosure to make acoustic testing work well. You can't just seal any product in a chamber and get a clean reading. Luna: What's the gotcha? Lucas: Volume. The test works by pressurizing the product to a certain level — typically one to two PSI above ambient. Then you stop the flow and measure the pressure decay or listen for leaks. But if your product has a large internal volume, the pressure drop from a small leak is very slow. You need longer test times or higher sensitivity. Luna: So a big waterproof camera housing might be harder to test than a tiny earbud. Lucas: Exactly. You might need to use a different method — like mass flow or tracer gas — for very large volumes. But for most handheld or wearable devices, acoustic is perfect. Luna: And what about the test chamber itself? Do you need to isolate it from factory noise? Lucas: Good question. The acoustic sensor is inside the chamber, which is sealed. So it blocks ambient noise. But if the chamber itself vibrates — say from a nearby stamping press — that can create false readings. So you want the chamber on a stable surface, not bolted to a machine that shakes. Luna: That seems manageable. What about the type of seal on the product? Does it matter if it's a gasket versus an adhesive bond? Lucas: It does. Acoustic detection works best when the leak is a discrete hole — like a crack or a pinhole. For a gasket that's uniformly leaking along a whole edge, the sound might be distributed and harder to detect. In that case, you might see a pressure decay curve that tells you there's a leak, but the acoustic method may not pinpoint it. Luna: So you'd use pressure decay as a go-no go, and maybe a visual or dye test to find the exact location. Lucas: Right. And that's fine — the goal is to catch bad units, not necessarily to diagnose every failure mode on the line. You can do root cause analysis offline. Luna: One thing I'm curious about: calibration. How often do you need to calibrate the chamber? Lucas: Most manufacturers recommend every six months or after a thousand tests, whichever comes first. You use a calibrated leak — basically a tiny orifice with a known leak rate — to verify. Some systems have an automatic self-check. Luna: That's not too onerous. For a startup running a few hundred units a week, you'd calibrate maybe once a year. Lucas: Exactly. And the nice thing is that the data is digital — you can log every test result, build a database, and use it for continuous improvement. If you see a batch where leak rates are higher, you can trace it back to a specific shift, a specific operator, or a specific material lot. Luna: That's the kind of traceability that impresses investors, too. They love seeing a quality system that's data-driven. Lucas: Absolutely. And it's not just about avoiding returns — it's about building a brand that people trust. When you say your device is waterproof, you have to mean it. Acoustic leak testing gives you the confidence to make that claim. Luna: Alright, so to wrap: if you're a hardware startup making a sealed product, consider acoustic leak detection. Start with a test lab to validate your design, then buy a chamber when you hit production volumes. Lucas: And don't forget to design your enclosure with testing in mind — include a test port if possible, and minimize internal volume. That'll make your life a lot easier. Luna: Great advice. Thanks, Lucas. Lucas: Thanks, Luna. Next time, we'll look at another unsung hero of hardware reliability — something that keeps your device alive when it's sitting on a shelf for years.