Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use X-Ray Inspection to Find Hidden Defects
Transcript
- Lucas: So we were talking last week about how startups use ultrasonic welding to seal enclosures, and it got me thinking about the step before that: what happens when you have a circuit board that looks perfect on the outside, but there's a hairline crack hiding under a component? That's where X-ray inspection comes in, and it's a topic we haven't touched yet. Luna: X-ray inspection. That sounds like something only the big guys do. Is that actually accessible to a small team? Lucas: It's more accessible than you'd think, and I've got a concrete story. There's a startup I know that makes smart thermostats. They started getting units back from the field with intermittent failures, maybe one in a thousand. They ran all the standard tests, functional, thermal, even some vibration, and everything passed. But the failures kept happening. Turns out there was a cracked solder joint under a shielding can on the microcontroller. You couldn't see it with a microscope, couldn't catch it with a functional test because it only failed when the board flexed just right. Luna: So how did they find it? Did they just RMA everything and hope? Lucas: They finally sent a few boards to a test lab that does 2D X-ray inspection. Within an hour, they saw the void and the crack. That's the moment they realized they needed X-ray in their process, not just as a post-mortem. And that's what we're going to unpack today: how startups can use X-ray inspection to catch these hidden defects before they become field failures. Luna: Before we dive deeper, Lucas, I've got to say, this kind of conversation is exactly why I love doing this show. It's so focused on real problems that hardware builders face every day. And you know, if you're listening and these episodes help you think through your own design or manufacturing challenges, we'd be grateful if you'd consider supporting the show. It's what keeps us ad-free and independent. You can buy us a coffee at buy me a coffee dot com slash fexingo. Lucas: Yeah, that's a nice way to put it. And we do appreciate that. So, back to X-ray. The first thing to understand is why visual inspection isn't enough. As boards get denser, components get smaller, and more connections are hidden under packages. Think ball grid arrays, or BGAs, where the solder balls are underneath the chip, completely invisible from above. Luna: Right, so unless you're willing to tear the board apart, you're blind. So what does X-ray actually see? Lucas: X-ray sees density differences. Solder is denser than the board and the components, so it shows up clearly on the image. You can see voids in the solder joint, which are basically air pockets that weaken the connection. You can see cracks, bridges where two adjacent joints are touching, and you can see if the solder has wetted properly to the pad. In 2D, you get a single projection, like a shadow. It's good for spotting gross defects, like a missing component or a major bridge. For more subtle issues, like a crack that's only visible at a certain angle, you might need 3D X-ray, which is essentially a CT scan. Luna: But 3D X-ray machines are probably way out of a startup's budget, right? Lucas: They're expensive, yes, but you don't have to buy one. This is where the test as a service model that we talked about in episode 154 comes back. You can send your boards to a contract manufacturer or a dedicated inspection lab and pay per batch. For the thermostat startup, the lab charged them about two hundred dollars per board for a 3D scan. That's a fraction of the cost of even one field failure when you factor in shipping, support time, and brand damage. Luna: So the economics make sense. But how do you decide when you actually need X-ray? I mean, not every board has a BGA. Lucas: Good question. The rule of thumb is: if you have any component with hidden solder joints, like BGAs or QFNs, or if you have RF shielding cans that cover multiple components, X-ray is worth it. Also, if you're dealing with high-reliability applications, like medical or automotive, or if your product experiences vibration and thermal cycling, that's when these defects become more likely. You don't need to X-ray every board, but you should sample from each production run, especially during early production ramp. Luna: I've heard that X-ray can also help with process control. Like, you can adjust your reflow oven profile based on what you see. Lucas: Exactly. That's the proactive use. Let's say you X-ray a few boards and you see consistent voids in the same location. That might indicate a stencil design issue or a thermal profile problem. You can tweak the profile, run another batch, and X-ray again to see if the voids shrink. It's like having X-ray vision into your manufacturing process. That's a huge advantage for a startup because you're making your process better, not just catching bad boards. Luna: So what are the limitations? I feel like there's always a catch. Lucas: The main limitation is that X-ray only shows you the internal structure, not the electrical performance. A joint might look perfect but still have poor contact resistance. So it complements, but doesn't replace, functional testing. Also, 2D X-ray can miss certain defects if they're oriented perpendicular to the X-ray beam. That's when you need to take multiple angles or go 3D. And there's a cost and time overhead, so you have to be selective. Luna: And you mentioned that the equipment can be expensive, but you can rent time on it. Is that common? Lucas: Very common. There are test houses that specialize in this. You can also ask your contract manufacturer if they have X-ray on-site. Many do, especially if they serve automotive or medical clients. You just have to negotiate the cost per panel or per board. Some will include it in the setup fees for a new program. It's all about knowing what to ask for. Luna: Alright, so you've got a startup, they're about to launch their product, and they've decided they need X-ray. What's the first step? Lucas: First, identify which components present the highest risk. Typically, it's the BGAs, the QFNs, the connectors with fine pitch. Second, decide on your sampling plan. Industry standards like ipc a-610 describe acceptance criteria, but you can also define your own based on your failure tolerance. Third, work with your CM or a test lab to set up the inspection criteria. Provide them with a drawing showing which joints to inspect, and the acceptable void percentage. You can't just say 'check it' and hope for the best. Luna: So you need to be pretty specific. What about interpreting the X-ray images? Is that something a startup can do in-house? Lucas: You can, but it takes training. The images are not intuitive at first. You're looking at shadows and grayscale. That's why many startups initially rely on the lab's expertise. As you get more experience, you can start to recognize common issues. There are also software tools that can analyze the images automatically and flag defects, but again, that's a cost. For a small team, it might be better to start with a service provider and then gradually bring it in-house if the volume justifies it. Luna: I remember reading a case study about a drone startup that found a problem with their motor controller. The X-ray showed that the solder was bridging two pins on a connector, and it was causing intermittent shorts. They thought it was a firmware bug. Lucas: That's a perfect example. The symptoms can be so random that you chase your tail in software. X-ray gives you the physical truth. For that drone startup, they fixed the stencil alignment and the issue went away. That's the kind of story that makes me a believer in this technology for even the smallest teams. Luna: So, what's the future? Are we going to see cheap X-ray machines that fit on a benchtop? Lucas: There are already benchtop systems that cost around fifty thousand dollars, which is still out of range for most startups. But the trend is towards more affordable systems, especially with the rise of ai based analysis. I've seen some startups that are building software to automate defect detection in X-ray images, which could bring the cost down further. But for now, the service model is the way to go. Luna: So, if a listener is thinking, 'this sounds interesting, but I don't know if I need it,' what would you tell them? Lucas: I'd say, if you're shipping products that could fail in the field and you're not doing any internal inspection of hidden joints, you're taking a risk. You don't have to do X-ray on every board, but at least do a sample on your first few production runs to make sure your process is sound. It's like insurance. You hope you never need it, but when you do, it pays for itself. And with the service model, the entry cost is low enough that you can try it without breaking the bank. Luna: Well, that's a solid pitch. I'm convinced. Lucas, where can our listeners learn more about this? Lucas: We'll put some links in the show notes, including IPC standards and a list of X-ray inspection service providers. We'll also have a checklist for evaluating your own inspection needs. So check that out. Luna: And if you have questions, you can always reach us on social media. We're @Fexingo on Twitter. But for now, let's leave it here. Thanks for listening, and we'll catch you next time.