Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Vibration Testing for Reliability
Transcript
- Lucas: Luna, I was looking at return data for a smart-home hardware startup, and the number one cause of field failures wasn't a software bug or a bad component — it was a solder joint that cracked because the device got rattled around in shipping. Luna: Shipping vibration — the thing nobody thinks about until the first wave of returns comes back. Lucas: Exactly. And that's why vibration testing is one of those hidden disciplines that separates hardware startups that survive their first production run from the ones that don't. So today, let's talk about what vibration testing actually looks like for a small team with a limited budget. Luna: I'm already thinking about the cost. A proper vibration table — aren't those like fifty grand and up? Lucas: They can be. A new electrodynamic shaker with a basic controller starts around thirty to forty thousand dollars. But you don't have to buy one upfront. Most startups rent time at a test lab. A full day of random vibration testing on a commercial shaker might run you eight hundred to fifteen hundred bucks. That's cheap compared to a recall. Luna: So what are you actually doing in that day? Strapping the product down and shaking it? Lucas: More or less. But there's real methodology behind it. You typically run two types: sine sweep and random vibration. Sine sweep is where the shaker oscillates at a single frequency, sweeping from low to high — like twenty hertz up to two thousand hertz. That finds resonant frequencies, the natural frequencies where the product amplifies vibration and can self-destruct. Luna: Resonant frequencies — that's where something like a loose screw or a poorly supported PCB starts vibrating like a tuning fork and cracks everything around it. Lucas: Right. And random vibration is more realistic — it simulates the broad-spectrum jostling of truck transport or air cargo. The test profile is often based on mil std-810, which is the military standard for environmental testing. But there are also commercial standards like ASTM D4169 for shipping containers. Luna: I've heard of HALT and HASS as well. Where do those fit in? Lucas: HALT stands for Highly Accelerated Life Testing. It's a step beyond basic vibration — you combine vibration with thermal cycling to find failure modes fast. The idea is to push the product until it breaks, then fix those weaknesses. HASS is the production-line version — Highly Accelerated Stress Screening — where you test every unit briefly to catch infant mortality. But HALT is more for the design phase, and it's usually done at a specialized lab. Luna: So for a startup with a prototype, what's the practical first step? Do they need to build a vibration test jig? Lucas: You need a mounting fixture — a plate that attaches to the shaker and holds your product securely. Often it's just an aluminum plate with threaded inserts. The key is that the fixture itself must be stiff enough that its own resonant frequencies don't interfere with the test. If the fixture rings like a bell at two hundred hertz, you won't know if the failure came from the product or the fixture. Luna: So you're basically designing a fixture that is as boring mechanically as possible. Lucas: Exactly. And you verify that by accelerometers — small sensors glued to the product and the fixture. You measure the input at the fixture and the response at the product. Then you know exactly what the product experienced. Luna: How many accelerometers do you typically need? Lucas: At least three — one on the fixture to measure input, one on the product's PCB, and one on the enclosure. But more is better. I've seen startups use six or eight for a complex assembly. Each accelerometer costs about a hundred to two hundred dollars for a decent ICP type. The data acquisition system to read them is another few thousand. But again, test labs provide all that. Luna: Let's talk about a concrete example. I remember a smart lock startup that had a prototype that kept failing after about ten thousand cycles in the lab. They sent it for vibration testing and discovered a resonant frequency at around a hundred and ten hertz that was shaking the circuit board's mounting screws loose. Lucas: That's a classic find. A hundred and ten hertz is right in the range of heavy truck suspension vibration. So that product would have failed in shipping for sure. The fix might be as simple as adding a dab of thread-locker or changing the screw material to one with a higher coefficient of friction. Luna: They ended up adding a small rubber grommet between the PCB and the enclosure. It shifted the resonant frequency above two hundred hertz, which is much less common in transport. Lucas: That's a great example of why you test early. If they had caught that after tooling for mass production, changing the enclosure would have cost tens of thousands in mold modifications. At the prototype stage, it was a few hours in SolidWorks and a new 3D print. Luna: And speaking of early-stage — if these conversations are useful for what you're building or running, a couple of dollars a month is genuinely what keeps these going. Buy me a coffee dot com slash fexingo, if you've gotten something out of them. Lucas: Yeah, it really does make a difference. No ads, no sponsors — just listener support. So if you can, that's where to go. Luna: Back to vibration testing — what about the difference between random and sine? When would you choose one over the other? Lucas: Sine sweep is best for identifying resonant frequencies. You sweep slowly — say one octave per minute — and watch the response spike. Random vibration is better for simulating real-world transport, because it excites all frequencies simultaneously. Most startups do both: a sine sweep first to find resonances, then a random test for a set duration, like thirty minutes per axis. Luna: Per axis — so you test in X, Y, and Z? Lucas: Yes, and sometimes rotational axes too, though that's less common for small products. Each axis requires reorienting the product on the shaker. So a full test suite might take a full day: setup, sine sweep in three axes, random in three axes, and then data analysis. Luna: What about the duration? How long should a random vibration test last to be meaningful? Lucas: There's a rule of thumb from mil std-810: one hour of random vibration at the standard profile simulates about a thousand miles of truck transport. So a typical test is thirty to sixty minutes per axis. But if you're shipping internationally, you might go longer. Some startups do a 'torture test' of two hours per axis to add margin. Luna: And what do you look for after the test? Visual inspection? Functional test? Lucas: Both. First you do a functional check — does the product still power on, does it communicate, does it meet specs? Then you disassemble and look for cracked solder joints, loose fasteners, broken plastic clips. Sometimes you'll find hairline cracks in the PCB that only show up under a microscope. Luna: That's why some startups invest in a digital microscope — they're like two hundred bucks and can save a whole production run. Lucas: Absolutely. And one more thing: vibration testing is not just for shipping. If your product has a motor or a pump, it will generate its own vibration. I worked with a drone startup that had a camera gimbal that kept drifting. Turns out the motor harmonics were exciting a resonance in the gimbal arm. A simple vibration test during development would have caught that before they ever flew. Luna: So the takeaway is: vibration testing is cheap relative to the cost of failure, and it should happen early, ideally at the prototype stage. Lucas: Exactly. And if you're a startup founder listening, don't skip this step just because it seems like a big-company thing. You can rent a shaker for a day. The data you get will directly inform your mechanical design and save you from a recall that could kill the company. Luna: And maybe avoid that awkward moment when the first batch of units arrives at customers already broken. Lucas: That is a very expensive awkward moment. So test early, test often, and let the vibration tell you what you need to fix.