by adom
0805 MLCC · 10 µF ±10% X5R · 16 V
All 8 CL21 MLCC pages: centered-origin STEP renders the cap half-sunk (one export-run fix)
All 8 CL21 (Samsung) MLCC pages ship a centered-origin STEP: the cap renders half-sunk into the board
Found in a Z-origin audit of every hardware page's STEP model (945 pages measured). The eight CL21-family caps all carry the same defect, so it is one export-run fix, not eight:
| page | vertex Z range |
|---|---|
| aravk/cl21a106koqnnng | -0.625 .. +0.625 |
| aravk/cl21a225kafnnng | -0.625 .. +0.625 |
| aravk/cl21a226mpqnnne | -0.625 .. +0.625 |
| aravk/cl21a475kaqnnng | -0.625 .. +0.625 |
| aravk/cl21a476mqynnne | -0.625 .. +0.625 |
| aravk/cl21b104kbcnfnc | -0.625 .. +0.625 |
| aravk/cl21b105kbfnnne | -0.625 .. +0.625 |
| aravk/cl21b474kbfnnng | -0.625 .. +0.625 |
The 0805 body is 1.25mm thick and the model's origin sits at MID-BODY, so half the cap is below Z0. KiCad (and any consumer that places the model at the footprint origin, including the wiki's "Open in Footprint Editor" flow, which wires the page's own .step as the 3D model) expects the origin on the board surface with the body extending up, so these render half-sunk into the board.
Fix: re-export the family with the origin at the bottom face (Z 0 .. +1.25). The rest of the aravk basic parts measured clean, so whatever produced the CL21 batch is the outlier, likely one source/export run.
Method note: measured from VERTEX_POINT-referenced geometry only (construction points excluded), so the numbers reflect rendered geometry.
0 Replies
Log in to reply.