Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Flexible Printed Circuits
Transcript
- Lucas: If you've ever opened a smartphone or a laptop, you've seen a flexible printed circuit — that thin, orange-brown ribbon that connects the screen to the main board. Hardware startups often treat them as a speciality item, something for consumer electronics giants. But for a growing number of early-stage companies, especially in medical wearables and compact robotics, flexible circuits are becoming the default choice, not the exception. Luna: I've definitely seen them inside devices, but I don't think I've ever heard a startup founder say they started their design with flex in mind. Usually it's a last-minute fix when they run out of space. Lucas: Exactly. And that's what I want to dig into today — why more hardware startups should be considering flexible printed circuits from the beginning, not as a fallback. I've got a concrete case that makes the numbers really clear. A startup I've been following, they're building a continuous glucose monitor — the kind that sticks on your arm for two weeks. Originally, they designed it with a standard rigid PCB stacked in two layers. Luna: That's a pretty common approach for a small device. Stack boards, connect them with a header or a ribbon cable. Lucas: Right. But the problem was thickness. They needed the device to be under 6 millimetres to be comfortable on the arm. With two rigid boards and a connector, they were at 8.2. So they looked at moving to a single flexible circuit that could bend into a U-shape inside the enclosure. That single change dropped the total thickness to 5.1 millimetres. They also eliminated three separate connector points — the header between the two rigid boards and two wire to board connectors for the sensor and the antenna. Luna: So fewer failure points and a thinner device. That's a pretty compelling double win. What about cost? Lucas: The flex circuit itself was more expensive — about $2.80 per unit versus $1.10 for the two rigid boards combined. But when they factored in the elimination of the connectors — which were about 40 cents each — and the reduction in assembly labour because they didn't have to hand-solder those connectors, the total board-level cost came in at $3.10 for the flex versus $3.50 for the rigid stack. So actually cheaper at volume. Luna: That's a great example. It's not just about space — it's about system cost. I think a lot of founders don't run that full cost comparison. Lucas: They don't. And frankly, the flex pricing has come down a lot in the last five years. The material itself — polyimide film, often Kapton — is more expensive than FR4, the fibreglass used in rigid boards. But the fabrication process has matured. You can now get prototype quantities of two-layer flex boards for maybe $200 to $500 from domestic quick-turn shops, and production pricing in the $0.50 to $2.00 range for simple designs at tens of thousands of units. It's not exotic anymore. Luna: That's actually less than I expected. I remember hearing horror stories about flex circuits being ten times the cost of rigid. Lucas: That was true maybe fifteen years ago, when you had to use specialist fabricators with limited capacity. Now the market is much more competitive. Companies like Sierra Circuits, Advanced Circuits, and even some Chinese fabs have standard flex processes. The key is to design with their design rules in mind from day one. One startup I spoke with had to re-spin their flex board twice because they put a 90-degree corner trace right at the bend area. The copper cracked after a few hundred flex cycles. Luna: Right — so the design rules for flex are different. You can't just take a rigid PCB layout and make it flexible. Lucas: Exactly. The biggest pitfall is the bend radius. If you fold a flex circuit too tightly, you stress the copper. A good rule of thumb is the bend radius should be at least ten times the thickness of the circuit. So for a two-layer flex that's 0.2 millimetres thick, you want a bend radius of at least 2 millimetres. And if it's a dynamic flex — something that will move during use, like a hinge in a foldable device — you need even more margin. You'll also want to use rolled annealed copper instead of electrodeposited copper; it's more ductile and handles repeated bending much better. Luna: And you mentioned trace routing — no sharp corners at the bend, right? Lucas: Correct. Traces should run perpendicular to the bend axis, not parallel. And you want to avoid placing plated through-holes in the bend area entirely — they're stress concentrators and will fail. Also, you should stagger traces across layers so they don't stack on top of each other, which creates a stiff ridge. Most flex fabricators provide a design guide that covers all of this, and I'd recommend any startup download those before they start their layout. Luna: If today was actually useful to you — and I'm already thinking about how this applies to a drone project I've been tinkering with — the way these episodes stay ad-free is listener support. It's buy me a coffee dot com slash fexingo. Just a simple way to say this kind of deep-dive engineering content matters. Lucas: Yeah, we don't run sponsors for a reason. We want to talk about what's actually relevant to builders, not what a sponsor wants you to hear. So if you find value in that, the coffee link is a direct way to keep it going. And hey, if you're that drone project — flexible circuits are perfect for arms that fold. I'd love to hear how it goes. Luna: Alright, back to the bench. So we've talked about cost and design rules. What about reliability testing for flex circuits? Is it any different from rigid? Lucas: It is. And this is an area where startups sometimes get caught out. The standard IPC test for flex is IPC-6013, which classifies flex boards into three types. Type 1 is simple single-sided flex, Type 2 is double-sided with plated through-holes, Type 3 is multilayer flex and rigid-flex combinations. For most medical and wearable startups, you want Type 3 certification at the very least. You'll also want to run a dynamic flex test — essentially bending the circuit repeatedly to a specified radius for tens of thousands of cycles. For a continuous glucose monitor that's worn for 14 days and flexes with arm movement, you might need 50,000 cycles at a 5-millimetre bend radius. Luna: That's a lot of cycles. Is that something startups can do in-house, or do they need to send it out? Lucas: You can build a simple test fixture with a servo motor and an Arduino for a few hundred dollars. The key is to instrument the circuit — measure continuity through the traces during the test. If the resistance spikes, you know a crack is forming. One startup I know did exactly that with a $400 setup and discovered their initial design failed at 12,000 cycles. They adjusted the bend radius from 3 millimetres to 5, switched to rolled annealed copper, and passed 60,000 cycles on the next spin. That kind of testing is cheap compared to a field failure. Luna: And it's something a hardware startup can absolutely do on a shoestring budget. So let's talk about when you shouldn't use flex. I assume it's not always the right answer. Lucas: No, definitely not. If your device doesn't have tight space constraints, if it's a stationary product like a desktop sensor, rigid boards are almost always cheaper and easier to design. Also, if you need very high component density or very fine trace widths — say, below 75 microns — flex becomes more challenging and expensive. And if your product is going into a high-temperature environment — like inside an engine bay — the polyimide can handle it, but the adhesives used in some flex laminates may not. You need to specify adhesive-less construction for extreme temps. Luna: So the decision really comes down to: do you need to save space, reduce weight, or eliminate connectors? Those three criteria. Lucas: Exactly. And I'd add a fourth: do you need to improve reliability by removing mechanical connectors? Connectors are often the most failure-prone part of an assembly. A flex circuit that replaces two connectors and a cable can dramatically improve mean time between failures. That's especially important for medical devices or industrial sensors where downtime is expensive. I've seen startups in the IIoT space — industrial Internet of Things — switch to flex for vibration-prone applications. They had connectors shaking loose on rigid boards. Flex eliminated that problem entirely. Luna: Right, because the traces are embedded in the flexible substrate, so there's nothing to rattle loose. That makes a lot of sense for drones, robotics arms, anything with movement. Lucas: Exactly. And there's a middle ground too — rigid-flex boards. That's where you have rigid sections for components and flexible sections connecting them. It gives you the best of both worlds: you can mount large processors and connectors on the rigid parts, and route signals between them through the flex. The downside is cost — rigid-flex is more expensive than either pure rigid or pure flex because the fabrication is more complex. But for a startup building a compact robot with multiple PCBs, it can save space and assembly time. Luna: I remember seeing a rigid-flex design in a handheld barcode scanner. It folded like origami inside the handle. The assembly time dropped because they didn't have to install any cables. Lucas: Exactly. And that's the kind of win that shows up in unit economics. Less labour, fewer parts to source, simpler assembly. For a startup scaling from prototype to thousands of units, those factors can be decisive. One more point: if you're considering flex, engage your fabricator early. Send them your stack-up and bend requirements before you finalise the layout. They can tell you if your bend radius is too tight, if your material choice is wrong, or if you need stiffeners in certain areas. The worst time to discover a design rule violation is after you've ordered a thousand boards. Luna: Yeah, and that's true for any PCB, but especially for flex because the design rules are less intuitive. So if I'm a startup listening, what's the one concrete takeaway? Lucas: Run the full system cost comparison, not just the board cost. Include connectors, assembly labour, and reliability. You might find flex is cheaper overall. And download the design guide from your fabricator before you start routing. That alone will save you a re-spin. For the continuous glucose monitor startup I mentioned, they went from a failed design to production in three spins. The first was rigid, the second was flex but with bad bend radius, the third was production-ready. They launched six months ahead of their original schedule because flex let them solve the space problem without a custom battery or a thicker device. Luna: That's a great outcome. And it's a reminder that hardware is full of these trade-offs that most people only learn through painful experience. Thanks for walking us through it. Lucas: Happy to. And if you're building something with flex circuits, we'd love to hear about it. Drop us a note — we might feature it in a future episode.