Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Pick and Place for Rapid Prototyping
Transcript
- Lucas: So a question I get a lot from hardware founders is: at what point do you stop hand-soldering prototypes and start using a pick and place machine? And the answer might surprise you — because the threshold has dropped a lot in the last few years. Luna: I remember visiting a startup last year that was placing components with tweezers under a microscope. And they told me they thought automation was only for factories doing a million units. Lucas: Right, and that used to be true. A full-size pick and place line was a seven-figure investment. But today you can get a desktop machine like a Yamaha YS100 for under fifty thousand dollars, and it can place over a hundred thousand components per hour. Luna: A hundred thousand per hour — that's wild. So what's the catch? There's always a catch. Lucas: The catch is that the machine is only as good as your setup. You need feeders for each component reel, you need to program the placement coordinates, and you need to calibrate the vision system. That setup time can eat up any speed advantage if you're only placing a handful of boards. Luna: So what's the break-even? Like, how many boards justify the setup? Lucas: It depends on component count. But a good rule of thumb I've seen is that once you're placing more than about a hundred components per board, and you're doing at least ten boards, a pick and place machine starts to beat hand placement both in speed and consistency. Luna: And consistency is huge for prototyping, right? Because if you hand-solder a prototype and it fails, you don't know if it's the design or the soldering. Lucas: Exactly. That's the hidden cost of manual assembly — introducing variability. With a pick and place machine, every component is placed at the same orientation, with the same pressure, every time. So when your prototype doesn't work, you can trust it's a design issue, not a manufacturing defect. Luna: I want to dig into the machine a bit. What are the key features a startup should look for? Lucas: First, feeder capacity. A desktop machine might only hold eight or twelve eight-millimeter tape feeders. If your board uses fifteen different resistor values, you're swapping feeders mid-run. That kills throughput. Luna: So you want as many feeder slots as you can get. What else? Lucas: Nozzle changer. Some machines have a single nozzle and you have to manually swap it for different component sizes. Others have an auto-nozzle changer that holds four or six nozzles. That matters if your board has both small zero-four-oh-two resistors and larger QFP packages. Luna: And the vision system — I've heard component centering can be a pain. Lucas: Yes. Most mid-range machines use downward-looking vision to find fiducials on the board, and some have upward-looking vision to check the component's leads before placement. That's critical for fine-pitch parts. Without it, you'll get tombstoning where the component stands up on one end because of uneven solder paste. Luna: Tombstoning is such a frustrating defect. It looks like a tiny gravestone on your board. Lucas: And it's almost always a placement issue — the component wasn't pressed down evenly, or the nozzle picked it up off-center. A good machine with vision alignment can reduce tombstoning to near zero. Luna: So let's talk about a real-world example. You mentioned a medical device startup earlier? Lucas: Yeah. I spoke with a team in Austin that was building a wearable glucose monitor. Their prototype boards had about two hundred components, and they were hand-placing them. A single board took four hours. They bought a used Yamaha YS100 — I think they paid thirty-eight thousand dollars — and after a weekend of setup, they could place a board in under three minutes. Luna: Three minutes instead of four hours. That's not just faster — that changes how many design iterations you can do in a week. Lucas: Exactly. They went from one prototype per week to four per day. Their lead time from design to test dropped from three weeks to two days. That's the kind of velocity that can make or break a hardware startup. Luna: But I imagine not every startup should buy a machine. When does it make sense to use a contract manufacturer instead? Lucas: If you're doing fewer than ten prototype runs a year, or if your components are mostly through-hole, then a contract manufacturer's pick and place line probably isn't cost-effective. But if you're iterating weekly and your design is surface-mount, bringing assembly in-house can save you weeks of waiting. Luna: And there's also the learning curve. You have to learn to program the machine, maintain it, source feeders… Lucas: That's real. The first few runs will be slow. But the skills transfer. Once your team knows how to operate a pick and place machine, they understand DFM — design for manufacturing — much better. They start designing boards that are easier to assemble, which pays off even if you later move to a high-volume factory. Luna: That's a great point. The knowledge is as valuable as the machine. Lucas: And speaking of knowledge — if you're finding these conversations useful for what you're building or running, listener support is what keeps this show ad-free and focused on practical insights like this. You can help at buy me a coffee dot com slash fexingo. Luna: Yeah, it's a small way to keep us going. We really appreciate it. Lucas: Anyway, back to pick and place. One more thing — startups often overlook the importance of the solder paste printer. The pick and place machine is only as good as the paste deposit underneath. If the stencil isn't aligned or the paste is too thick, even the best placement won't save you. Luna: So it's really a system — stencil printer, pick and place, reflow oven. Each step matters. Lucas: Exactly. And that's why I think the next big opportunity for hardware startups is not just buying a pick and place machine, but building a mini SMT line that integrates all three steps. We're seeing companies like Voltera and Essemtec offer all-in-one solutions for under a hundred thousand dollars. Luna: That's still a big investment for a pre-revenue startup. But if you're post-series A and shipping hundreds of units a month, it starts to make sense. Lucas: For sure. And the resale market is healthy — you can often sell a well-maintained machine for seventy percent of what you paid. So it's not a sunk cost. Luna: Good to know. So if I'm a founder listening, what's the first step? Just buy a used machine? Lucas: No — first, take a board you've already designed and get a quote from a contract manufacturer. Ask them for a per-placement cost and a setup fee. Then calculate what your in-house cost would be, including the machine amortization, labor, and consumables. If your volume justifies it, then start looking at machines. Luna: And if you're not there yet, you can still design for pick and place — use standard footprints, avoid odd-sized components, and add fiducials. That way, when you do move to automation, your boards are ready. Lucas: Right. And that's the mindset shift I think is most valuable — treating prototyping not as a one-off craft, but as the first step in a scalable process. pick and place is a tool that forces you to think that way. Luna: I like that. It's not just about speed — it's about discipline. Lucas: Exactly. And that discipline is what separates hardware startups that iterate quickly from ones that get stuck on prototype number three.