The Hard Lessons of PCB Auto-Routing: Part 2 of Seve's List

Seve's original post

Part 1 episode

This episode continues where the last one left off, diving into Part 2 of Seve's blog post “13 Things I Wish I Knew Before Building an Autorouter.” 🧠⚡ Seve (founder of tscircuit) and Matt (founder of atopile) explore the deep technical challenges of auto-routing printed circuit boards using code, compiler toolchains, and caching, rather than traditional schematics-based tools.

In this second half, they unpack:

* Why caching is crucial for speeding up PCB auto-routing

* Visualization as a debugging superpower

* The power of PyInstrument and visual profiling tools

* Why recursive functions and Monte Carlo methods often fail in optimization

* Game dev tricks (like greedy A*) adapted for PCB pathfinding

* The difference between grid-based vs intersection-based routing math

* Why keeping results grounded in physical space (like millimeters) makes debugging easier

* Using animation to catch stupid behavior before it goes live

* How a meta-router manages multiple strategies in parallel

* Insights from CNC machines and high-fidelity spatial modeling

This episode is packed with Seve and Matt's hands-on experiences, hard-won insights, and sharp advice for anyone building or using modern, code-first electronics design tools.

🛠 Whether you're a PCB engineer, systems architect, or startup founder in the hardware space, this conversation is pure gold.