From breadboard to OEM run

A breadboard that works is a wonderful thing and a misleading one. It proves the idea is sound. It says almost nothing about whether you can make ten thousand of them, identically, for a price someone will pay. The distance between that first working prototype and a production run is where most hardware projects quietly stall.

Prove the idea, then design for the factory

Early on, speed matters more than elegance. We breadboard with dev kits and jumper wire, accept the mess, and answer one question: does the core concept work at all? Once it does, the mindset has to flip completely. From the first custom PCB onward, every decision is made for the factory rather than the bench — not 'can I build one,' but 'can a contract manufacturer build thousands without me in the room.'

That shift shows up in unglamorous choices. We design for manufacturability and for test: parts a pick-and-place machine can place, footprints with sane tolerances, test points and a bed-of-nails fixture so every board is verified in seconds on the line rather than debugged by hand. We pick components with second sources and healthy lifecycles, because the perfect part that goes end-of-life or onto a twenty-week lead time will halt a production run cold.

EVT, DVT, PVT, and the long tail

Hardware ramps in stages, and skipping one always costs more than it saves. Engineering validation proves the design works, design validation proves it against real-world stress and the certifications it needs, and production validation proves the factory can hold quality and yield at volume. Each gate exists to catch a class of failure while changing a board still costs a few hundred euros — not a scrapped reel of finished units.

Anyone can build one. The engineering is in building the ten-thousandth one exactly like the first.

— Protocore · Hardware engineering

Then there is everything that outlives the launch. Firmware needs a signed, resumable update path before the first unit ships, because recalling devices to reflash them is not a plan. The supply chain needs second sources and a bill of materials that survives a part going obsolete. Field returns need a way to be diagnosed. A product is not the unit on your desk; it is the unit, plus the line that makes it, plus the years it has to keep working after you have moved on.

We have taken access-control readers and NFC wristbands down this whole path — from a prototype on the bench to certified hardware coming off an OEM line at volume. The early prototype is the easy, exciting part. The discipline that turns it into a product you can ship by the pallet is the actual work, and it is the part worth getting right.