My First PCB Prototype with AI: The Connector Was Harder Than the Design

language

eng

date

Aug 31, 2026

slug

first-pcb-prototype-with-ai

author

status

Public

category

🤖 Computer Science

updatedAt

Aug 31, 2026 03:15 AM

With AI's help, I designed a simple auxiliary board for use at work and placed my first order with JLCPCB. As a two-layer board, fabrication itself was straightforward. The bottleneck was elsewhere. Sourcing connectors took longer than the design, and once I had the physical board in hand, I had to revisit its size and connector placement.

Because it was a simple two-layer board with few components, I ordered it as an assembled PCBA. My remaining job was just to source and solder the unavailable parts and connectors. That became the hardest part of the project. Perhaps because Korea's hardware market is small, too many connectors are not readily available. This board was an add-on for a Chinese-made board, so I needed to find several of the exact connectors used on that board. That proved difficult. I had no choice but to spend time searching Mouser and Digi-Key, comparing candidates one by one. I eventually found the connectors, though I still have not made a cable that fits them exactly. Building a BOM and sourcing these parts did not happen automatically, and AI could not handle it perfectly either.

The second difficulty was lack of experience. I was building something I had never made before, so it was hard to know whether each design decision was right. It helps that we now have AI to ask without feeling self-conscious, but AI can be wrong, and I cannot trust an immediate answer completely. In the end, it either tells me to try it or asks for measurements that are almost impossible to make in order to gain confidence. I had to accept that there would be trial and error. As soon as I received the first PCB, I realized it was too large and that I had placed the connectors poorly. Still, the prototype let me verify the core behavior, and it worked as intended.

Next, I plan to reduce the board size by removing unnecessary ICs and revise the connector placement for a final version. This experience made one thing clear: AI is lowering the barrier to hardware development. It has made circuit design and production files much faster to prepare, but it does not replace parts sourcing or physical validation. Even in the AI era, the important work is deciding what to build, checking real constraints, and repeating that cycle until the lessons become experience.

My First PCB Prototype with AI: The Connector Was Harder Than the Design

Related posts

ThunderScope: The Open-Source Oscilloscope That Moves the Display and Processing to a PC

Cerebras WSE-3: Why Turn an Entire Wafer into a Single AI Chip?

What does the CPU think? A great introductory article on building a 16-bit CPU directly in C.