Back to selected work
Project specifications
Schematic symbols81
PCB footprints106
Automated CI checks6 deterministic checks
Altium review4-part datasheet checklist
LicenseCERN-OHL-P-2.0

The problem

A large share of expensive PCB respins start long before routing: they begin in the component library. One swapped pin on a schematic symbol, a mirrored pin-1 mark, an undersized drill on a power header, or an unverified land pattern pulled from an anonymous forum can turn a two-week fabrication run into scrap hardware.

At the same time, native Altium Designer libraries (.SchLib and .PcbLib) are opaque binary compound files. Standard git diff cannot show whether a footprint pad shifted by 0.2 mm, whether a file was corrupted during a save, or whether temporary Altium history and lock files slipped into a commit.

Over several years of hardware work (2019–2026), I curated my reusable components into Personal-Altium-Library, published publicly under CERN-OHL-P-2.0. To make a binary EDA repository trustworthy for both personal reuse and external engineers, I paired the library files with a two-layer validation workflow modeled after software release discipline.

What the library covers

The repository tracks two checked-in Altium libraries whose exact sizes and SHA-256 digests are recorded in library-manifest.json:

  • AltiumLibraryDDT.SchLib (396,800 bytes) — 81 schematic symbols.
  • AltiumLibraryDDT.PcbLib (36,292,608 bytes) — 106 PCB footprints.

Across docs/INVENTORY.md, the catalog spans practical building blocks from real hardware builds rather than synthetic demo parts: SMD passives across 0402/0603/1206 packages, high-voltage diodes, barrier terminal blocks and shrouded headers, wireless modules (ESP32-DEVKITC-32D, ESP8266_ESP-12, HM-10), power modules (LM2596, MT3608, Mini-360), load-cell and current sensors (HX711, ACS758LCB-050B), relays (G5LE), and micro-SD holders.

Two-layer validation for binary EDA files

Because CI cannot replace an electrical engineer inspecting copper geometry in Altium, docs/VALIDATION.md separates what machines can guarantee from what requires human review in the EDA tool.

Layer 1: Automated repository integrity (CI)

Running python3 scripts/validate_repository.py locally and in GitHub Actions (quality.yml) enforces six deterministic checks on every push and pull request:

  1. Required project, licensing, and engineering-notice files exist.
  2. Both .SchLib and .PcbLib carry valid compound-file binary signatures (catching truncated or text-corrupted commits immediately).
  3. File byte sizes and SHA-256 checksums match library-manifest.json exactly.
  4. Declared symbol (81) and footprint (106) counts match the tracked inventory in docs/INVENTORY.md.
  5. Local Markdown documentation links resolve cleanly.
  6. Common generated artifacts, Altium backups, and autosave files are blocked from entering Git.

When a component is intentionally added or edited, scripts/update_manifest.py refreshes the manifest — and the policy explicitly forbids blindly accepting a regenerated hash without verifying that only intended library entries changed.

Layer 2: Manual Altium engineering gate

Before any modified symbol or footprint is accepted, docs/VALIDATION.md defines a four-part Altium review checklist:

  • Library integrity: Open and save both libraries without conversion or repair warnings and run built-in library checks.
  • Schematic review: Verify every pin number, name, electrical type, hidden power pin, designator, and footprint link directly against the current manufacturer datasheet.
  • Footprint review: Check pad geometry, numbering, drill sizes, thermal pads, solder mask, paste, silkscreen, courtyard, pin-1 indicator, and 3D body height against the manufacturer's recommended land pattern and fabricator tolerances.
  • Integration review: Place the symbol in a temporary schematic, push it to a temporary PCB, verify all pin-to-pad mappings, run ERC and DRC, and inspect both board sides in 2D and 3D.

Takeaway

Reliability in connected hardware starts with boring, repeatable habits: versioned manifests, datasheet-first verification, and honest boundaries about what automation can and cannot prove. Personal-Altium-Library is both a public engineering resource and a concrete look at how I structure PCB review workflows before releasing Gerbers to fabrication.

Sources

Explore consulting