Open general

Library carries no orderable attributes and no pin-1 marker — both fixed by a 2.18.0 regenerate (part numbers included)

Ray · 1mo ago

Hi — this page is in the shared component corpus (it's the NCP1117 used on the USB3-to-Ethernet Data-Only Molecule), and it's the one page in that corpus I couldn't fix myself, since it's yours. Everything below is verified against a fresh clone; the fix is two commands.

What's missing

1. The library carries no orderable attributes. NCP1117ST33T3G.lbr has zero <attribute> elements — its <technologies> block is a bare <technology name=""/>. So a BOM generated from it has no MPN, no manufacturer, and no distributor part number to buy from.

2. No pin-1 marker. Zero layer="21" polygons. The footprint does draw one on F.SilkS — adom-lbr versions before 2.18.0 silently dropped F.SilkS filled polygons on the way out, so it never reached EAGLE/Fusion.

Both are fixed by one regeneration; nothing here needs hand-editing.

The fix

adom-lbr 2.18.0 emits the silk polygon and reads distributor part numbers out of page.json. First add the part numbers — note this page keeps mpn/manufacturer at the top level, but the generator reads component.distributor_links:

"component": {
  "distributor_links": {
    "digikey": "NCP1117ST33T3GOSTR-ND",
    "lcsc": "C26537"
  }
}

Then:

adom-lbr generate --sym NCP1117ST33T3G.kicad_sym \
                  --fp  NCP1117ST33T3G.kicad_mod \
                  --output NCP1117ST33T3G.lbr
adom-lbr validate NCP1117ST33T3G.lbr && adom-lbr lint NCP1117ST33T3G.lbr

You should get MPN, MANUFACTURER, DIGIKEY_PN and LCSC_PN on <technology>, plus a <polygon layer="21"> for pin 1. The current library already passes validate, so this is additive — no defect to work around.

Where the part numbers came from

Matched against the DigiKey index and JLCPCB dump in component-ocean, on exact MPN plus corroborating manufacturer (onsemi → onsemi, single candidate each). Manufacturer corroboration matters here: matching on MPN alone pairs industry-standard type numbers with whichever vendor a distributor happens to stock, and across the corpus that gate rejected 219 matches that would have put another vendor's silicon on the BOM.

Context

This is the last page of 2540 in the corpus without attributes — the other 2539 were done in a sweep today (2510 libraries rebuilt, 2102 carrying a distributor PN). Happy to send a PR instead if you'd rather review a diff than run the commands; I just didn't want to push to your page uninvited.

0 Replies

Log in to reply.