- Import LFE5U-45F-8BG381I as a proper named/valued SamacSys symbol
(symbol, footprint, 3D model) instead of relying on the generic
ECP5U_45_CABGA381 placeholder, so it's findable by exact MPN in the
Symbol Chooser. Dedupe the vendor STEP file (byte-identical to the
one already added for the BG381 package) and repoint the footprint's
bare, unresolvable "LFE5U-45F-8BG381I.stp" model reference to
${KIPRJMOD}/3dmodels/fpga_cpld/LFE5U-45F-8BG381I.step.
- Also make the existing 9-unit ECP5U_45_CABGA381 symbol (split by
bank, easier to use on a schematic than the flat SamacSys one)
findable under the same MPN: Value/manf# set to LFE5U-45F-8BG381I
after verifying its 381 CABGA381 ball assignments match the SamacSys
symbol 1:1, and link its Footprint to the newly imported footprint.
- Import IS66WVE4M16EBLL-70BLI (SamacSys) symbol, footprint and 3D
model.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013gndCxJXQfouCQy4hP7NSX
Symbol ECP5U_45_CABGA381 and footprint LATTICE_BGA_CABGA381 already
cover this package generically; add the SamacSys-sourced STEP model
as a more accurate alternative body for manual per-instance
assignment, same treatment as the existing BG256 body.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SLWiTVT39zMvgWZfKNvYcP
- Add DMT6007LFG-7 (transistor) and TPS2121RUXR (dual-input power mux) with
footprints and TPS2121RUXR's 3D model.
- Import Lattice_ECP_FPGA symbol library (ECP5U-45 CABGA256/381/554,
CSFBGA285) from mattvenn/basic-ecp5-pcb's legacy .lib, converted via
kicad-cli sym upgrade; symbol-only, footprints resolved via the standard
Package_BGA KiCad library. Add a vendor-accurate BG256 3D body
(LFE5U-25F-8BG256I) for manual per-instance assignment.
- Fix check_library.py's 3D model reference regex to strip surrounding
quotes, which KiCad 10's re-save format adds around (model "...") paths
-- was causing a false "missing 3D model" error for SJ-3506-SMT-TR after
its footprint got re-saved in the new format.
- scripts/mikilab_lib.py / mikilab_cli.py: typed Python API + JSON CLI
for add/update/remove/query, for other apps to integrate without
shelling out to add_component.py
- lib_common.py/import_component.py/import_batch.py: regenerate
sym-lib-table.global/fp-lib-table.global on every mutating call, not
just the project-local tables
- scripts/find_missing_3d_models.py: lists components with no 3D model
and search links to fill the gap
- docs/GUIDA_USO.md: practical Italian usage guide (companion to README.md)
- checkup fixes: removed orphan duplicate footprint
footprints/other/CP_20_1_ADI.kicad_mod, and a dangling
${easyeda2kicad}/tmp 3D model reference in AXP2101's footprint that
pointed at a file that never existed on disk
The auto-generated FSC-BT1058 footprint used roundrect pads with a
slightly different body outline than the physical module. Since the
FSC-BT1035 footprint (pad geometry verified against the Feasycom
datasheet perimeter walk) matches the same physical package/pinout,
FSC-BT1058.kicad_mod now reuses that geometry directly.
Also recategorized FSC-BT1058 from "other" to "rf" (symbol, footprint,
3D model), matching FSC-BT1035 and the rest of the Bluetooth/RF parts.
Regenerated sym-lib-table/fp-lib-table (project and global) and synced
the real KiCad global library tables. Added "bt1058" to the rf
classification tokens in lib_common.py so a future re-import doesn't
land back in "other".
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
FSC-BT1058 had ended up with two footprint/symbol libraries on disk
(FSC-BT1058 and MIKILAB_FSC-BT1058) after a re-import baked the
"MIKILAB_" nickname prefix into the component name itself. Since that
prefix is normally added automatically when sym-lib-table/fp-lib-table
are generated, the double-prefixed copy (MIKILAB_MIKILAB_FSC_BT1058)
was invisible to collision checks and left the original as a stale,
outdated duplicate.
Consolidated onto the single correct library (symbols/other and
footprints/other without the redundant prefix, matching the rest of
the library's naming convention), fixed the symbol's Footprint
property and the footprint's 3D model reference, and regenerated both
lib-tables from disk.
To stop this class of bug from recurring:
- lib_common.check_component_name() rejects any --name that already
starts with "MIKILAB", used by import_component.py's import_symbol/
import_footprint (and therefore also by add_component.py and the
SnapEDA/UltraLibrarian importers, which call into the same core).
- check_library.py now flags any on-disk symbol/footprint whose name
already carries that prefix as an ERROR, independent of how it got
there.
- import_component.py gains --remove to cleanly delete a component
(symbol, footprint library, 3D model) and regenerate the lib-tables
in one step, for exactly this kind of cleanup.
Errors: 0 on check_library.py after the fix.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The importers previously refused to touch an already-imported component by
design, which meant re-importing an updated UltraLibrarian/SnapEDA zip for
the same part-number errored out. Add --update to import_component.py's
import_symbol/import_footprint (and thread it through import_snapeda.py and
import_ultralibrarian.py) so an existing symbol/footprint can be replaced
in place instead, with the manifest recording it as UPDATED rather than NEW.
Used it to refresh XTAC5212IRGER from a newer UltraLibrarian download
(single-unit symbol layout vs. the previous split power/signal units;
same 29 pins).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Add the footprint entries (missing in KiCad's real global fp-lib-table)
and the XTAC5212IRGER symbol entry for the components added in the
previous commit, so the tracked .global snapshot matches what's now
registered in KiCad.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Imports were done via import_snapeda.py for the SnapEDA parts and the new
scripts/import_ultralibrarian.py for XTAC5212IRGER (UltraLibrarian export),
mirroring import_snapeda.py but handling UL's timestamp-named symbol files
and its multiple pad-size footprint variants (nominal/-L/-M).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Model refs were missing the ${KISYS3DMOD} prefix used elsewhere for
KiCad's bundled 3D shape libraries, so check_library.py flagged them
as unresolved.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Symbol + footprint reconstructed from manvalan/DigiRadio's EasyEDA
project (same method as the FSC-BT1035 footprint), cross-checked
against the manufacturer datasheet found in that same project. The
datasheet's own circuit diagram shows two independent, mechanically
-ganged contact pairs (not a simple SPST), reproduced faithfully in
the symbol. Pin numbering preserved exactly as extracted, kept
internally consistent between symbol and footprint.
Extracted 52 pad records + silkscreen geometry from the raw JSON-Lines
document format inside manvalan/DigiRadio's fabricated EasyEDA Pro
project (epro2kicad couldn't handle this project's newer multi-doc
export format, so the relevant FOOTPRINT document was located and
parsed directly). Pad size/pitch independently match the official
FSC-BT1035 datasheet's mechanical spec exactly; pad numbering follows
the same perimeter walk as the datasheet's own pin table, so pad N =
pin N. The underlying component was originally a SnapEDA export for the
mechanically-compatible FSC-BT806 (same 13x26.9mm 52-pad family
package), relabeled BT1035 by the designer -- documented in README.
Validated with kicad-cli fp export + visual SVG comparison against the
datasheet's own footprint layout figure before import. No 3D model.
Realtek Gigabit Ethernet PHY, imported from a SnapEDA KiCad export zip
(symbol + QFN40P500X500X90-41N footprint + STEP 3D model, all bundled
together). This is the first real-world use of scripts/import_snapeda.py.
Si4684-A10-GM (Skyworks FM/DAB/DAB+ receiver): extracted from unmerged
kicad-symbols MR !4782 (cannot_be_merged, failing pipeline), then
independently validated with kicad-cli and cross-checked against two
unrelated sources (Skyworks public product summary block diagram, and
PE5PVB/SI4684-DAB-Receiver's built/fabricated symbol) -- all comparable
pin names/numbers agree. Symbol only, standard Package_DFN_QFN
footprint.
FSC-BT1035 (Feasycom BT 5.2 audio module): hand-authored from the
official Feasycom datasheet's Table 3-2 pin definition (all 52 pins),
validated with kicad-cli. Symbol only -- no reliable footprint/3D
source was found; see README for what was tried and how to add one
later (manvalan/DigiRadio has a fabricated footprint for this exact
part, but its geometry lives in EasyEDA's proprietary JSON format).
scripts/import_snapeda.py: unzips a SnapEDA "Download KiCad" export and
hands off to import_component.py's existing core logic (same
validation/collision/lib-table rules as every other import path).
Imports ESP32-C3/C5/C6/H2/S2/S3 MINI/WROOM/WROVER/SOLO modules and
ESP8684-WROOM-02C/U from the official espressif/kicad-libraries repo,
each with symbol + footprint + 3D model. Skipped bare SoC/die symbols
(no footprint of their own) and DevKit board symbols (not components
you'd place on your own board).
ESP32-S31-WROOM-3 is replaced: the previous easyeda2kicad.py export had
a fatal unquoted-URL syntax bug (fixed in an earlier commit, but the
official symbol/footprint/3D model is simply better -- real chip is
brand new, Espressif's own library already supports it).
All 19 footprints referenced their 3D model via ${KICAD8_3RD_PARTY} /
${KICAD9_3RD_PARTY} (KiCad Plugin & Content Manager install path, not
defined since this library isn't PCM-installed). Repointed to
${KIPRJMOD}/3dmodels/microcontrollers/ using the STEP files copied in
alongside each part, so every model resolves without PCM.
Adds ADAU1467WBCPZ300RL (the SigmaDSP chip previously missing), plus
PCM1808QPWRQ1, PCM9211PTR, Combo384, MW-1466CORE (ADAU1466 core module),
LME49720MA, CS8421-CZZ, PCM1861DBT, FDC608PZ, IMN10T108, AK5384.
Source repo used legacy KiCad .lib/.dcm format; converted via
kicad-cli sym upgrade. Parts already covered by the official mirrored
libraries (PCM5102A, INA194, AZ1117-3.3, ADAU1452) were intentionally
skipped to avoid duplication. PCM1681-Q1 skipped (KiCad 'extends'
variant, can't stand alone without duplicating PCM1681).
Fixed real defects found in the source footprints during import:
- PCM1808QPWRQ1 had an absolute 3D model path pointing at the upstream
maintainer's own machine (/Users/HILO/...)
- ADAU1467WBCPZ300RL and PCM9211PTR each had two dead 3D model
references (${KICAD_USER_TEMPLATE_DIR} and a bare filename)
All three now use a single ${KIPRJMOD}-relative reference to the STEP
files under 3dmodels/audio/.
LME49720MA imported symbol-only, Footprint property repointed at the
existing MIKILAB_ipc_soic:IPC_SOIC127P600X175-8N instead of duplicating
a generic SOIC-8 footprint.
The easyeda2kicad.py-generated file had an unquoted URL as the
(generator ...) value: '(generator https://github.com/uPesy/...)'.
Bare/unquoted s-expression tokens can't contain ':' or '/', which broke
KiCad's parser and made the whole library silently fail to load (the
symbol was present on disk and correctly registered in fp/sym-lib-table,
but invisible in KiCad). Wrapped the value in quotes.
Also added a check_library.py rule to catch this class of error (an
unquoted generator value containing ':' or '/') automatically going
forward.
- scripts/generate_global_tables.py generates sym-lib-table.global /
fp-lib-table.global using a custom env var (default MIKILAB) instead
of ${KIPRJMOD}, for use across every KiCad project
- README §1 now has concrete, tested install steps for this machine's
KiCad 10.0 setup (per-project vs. global), including a merge script
and a note about stale kicad-personal-library entries already
present in the user's global sym-lib-table
Self-contained KiCad library (symbols, footprints, 3D models, docs,
import/check tooling) with no dependency on kicad-personal-library.
- Fixed sym-lib-table/fp-lib-table: single (version 7) header, one entry
per library, MIKILAB_<name> nicknames, ${KIPRJMOD}-relative URIs
- Resolved the SOT95P280X145-5N footprint collision (TPS7A2012PDBVR vs
TPS7A2018PDBVR): confirmed byte-identical modulo KiCad's internal
tedit timestamp, unified into one shared footprint
- Resolved a case-insensitive filename collision between the official
Diode.kicad_sym and a custom diode.kicad_sym
- Wrapped standalone footprints (ESP32-S31-WROOM-3, IC_TPS63020DSJT,
SOT95P280X145-5N) into their own .pretty libraries so they're
actually registered in fp-lib-table
- Fixed broken symbol->footprint references (including one pointing at
a nonexistent easyeda2kicad library)
- Added scripts/check_library.py, import_component.py, add_component.py,
import_batch.py, and lib_common.py
micheleandClaude Sonnet 5