I Learned KiCad by Redrawing a Published Board That Had No Ground Pour

Same board, ground done right

The task was to re-create the drive board's schematic in KiCad from the author's published schematic, with one deliberate difference (a ground scheme that actually connected), and then lay out the PCB. The total time from never having opened KiCad to a finished schematic and a partial layout was about 4 to 6 solid hours (Estimated).

That number is small, which says something encouraging about modern KiCad and about learning under pressure. But these are also solid hours (focused, with a real board on the line). The schematic was never fabbed and never bench-tested, so I cannot call it verified, only finished. However, the distance from zero to a complete schematic of a real mixed-signal board is an evening or two, not a semester.

Re-drawing someone else's schematic turned out to be the fastest possible tutorial. Skimming a schematic feels like understanding, but re-creating it symbol by symbol, pin by pin, forces a line-by-line reading. By the end I understood the signal chain well enough to debug it later.

The hard part was knowing which line is which

The well-known beginner problems in PCB design are footprints and DRC errors, but I hit a wall with the connectors and terminals: positive, negative and ground, for every component and every wire. Keeping track of which line is which terminal across a whole schematic was the confusing part, and my workaround, color-coding each line by hand, was extremely tedious. At the time I was sure the software was missing a feature, but it was not.

What I know now is that the underlying skill is net thinking rather than visual tracking. Every wire belongs to a named electrical net, KiCad has net labels and net highlighting for this, and once the nets are named (+15V, −15V, GND, DA1_OUT) instead of traced by eye, the tracking problem mostly goes away (general KiCad knowledge, which I found later, the hard way). My hand-coloring phase was a manual version of what net labels do, and the tedium was a sign that I was missing the abstraction.

Deliverables

The redesign is saved in the project as dac_board_v2/: the schematic (dac_board_v2.kicad_sch), the partial layout (dac_board_v2.kicad_pcb), a placement-coordinates file, and REFERENCE_MAPPING_kicad_to_liao.md. That file is a table mapping my auto-assigned designators to the original board's, written so the two schematics could be cross-read.

The race ended before the finish line

When the original author finally replied with corrected design files (§M6 has the timeline), my schematic was done and my layout was partial. The decision was easy and a little anticlimactic. It was to fab from the author's corrected files, which carried his layout exactly, and to shelve my redesign unfabbed [project CLAUDE.md §2].

So the board I spent those hours designing was never manufactured, but it still paid off in two ways. As insurance, it gave the project a fallback path during the silent weeks instead of a dead stop. As a reference, my KiCad schematic became the net-level map of the circuit during bring-up and debugging, something I could search and highlight, which the original PDF schematic could never be. However, my designator mapping file became obsolete within a week, because the fabbed board used the original designators on its silkscreen.

Advice for beginners

Re-draw a real schematic you need, and name your nets on day one. A finished schematic with a partial layout gives you most of the value. And design work that gets overtaken by events is not wasted if it made the wait less risky and left you understanding the circuit.

Summary of the three tools

I learned three tools in three months, all for one bench. §M6 is where all three were tested for real, on the day the drive board's published files turned out to be broken and I had to deal with it as a beginner with a scope, an iron, and a half-learned CAD package.