Limn

Limn: the electronics in KiCad

A KiCad 9 project, limn.kicad_pro. The root sheet is the overview, with one sheet per role or board:

How to read it:

• Sheets meet through global labels (USB_*, I2C1_*, MOT_*, ENDSTOP_*, PROBE_*, BED_*, ...) and named power rails (+12V, +7V5_ZK, +9V, +5V_LED, +5V_TOOL, +5V_TOOL_IN, +5V_HUB, +5V_MELZI, +3V3_PI, +3V3_AUX, +3V3_PB, +3V3_PH). All GND is one net.
• Solid boxes: one board or assembly. Dashed boxes: planned, or going; their parts are DNP.
• Boards are blocks with the pins Limn uses, named as the code names them. The Pro Micro is drawn with its full header in physical order. A USB link is one pin (the cable).
• References per sheet: power 1xx, host 2xx, tool dock 3xx, Melzi 4xx, toolhead 5xx, Aux 6xx, the dock's pogo 7xx (J701), calibration bed 8xx (the beam bed 85x), plot bed board 95x, hub board 10xx.
• Set aside (2026-10-23, not drawn): the bed on three load-cell feet (the old Plot bed sheet, refs 90x), the RTP node, a toolhead MCU, the QRE1113 on the toolhead (U954, R961, C957, GP28). Their drawing code is in kicad_gen/aside.py, to bring one back as it was.
• Wires turn at 45°.

Gathered from the code and the docs on 2026-10-16: notebooks/pinout.md, micropython/nodes/*.json, klipper/*.cfg, projects/limn/content/electronics.md, docs/block-diagrams.md, docs/load-cell.md, TODO.md.

Regenerating

python3 docs/kicad/gen.py --force --check

kicad_gen/ holds the drawing code: core.py (the writer, wires, labels, enclosures), boards.py (the blocks), sheets_a.py / sheets_b.py (one function per sheet), aside.py (what is set aside). gen.py keeps KiCad's settings in limn.kicad_pro and only rewrites its sheet list. --check runs ERC, audit.py (a net joining two signal pins of one part: the shorts ERC can't see between passive pins) and writes one SVG per sheet into svg/. Today (2026-10-23): ERC 0 errors 0 warnings, audit ok. R958 fitted (as traced on the PCB), TOUCH_GP on GP2 only.

Footprints: Plot bed board has them (perfboard: R and C in 0805 since 2026-10-23, the fuse in 1812; the modules in limn.pretty, written by kicad_gen/footprints.py, listed in fp-lib-table), and so do the three boards of their own below. The other sheets don't yet.

One PCB per board. limn.kicad_pcb is the plot bed board's (Update PCB from Schematic, F8). Its perfboard's copper is drawn on Dwgs.User between the hole dots (perfboard.strips, 2026-10-23): the top and bottom rows one strip each, the 15 rows between joined down each column in 5 groups of 3. The hub board, the Aux PCB and the beam bed are boards of their own; their parts are "not on board" in the schematic (sheets_b.own_board), so F8 never puts them on limn.kicad_pcb. Their boards are made from the netlist by tools/mkboard.py hub|aux|beam|head: the sheet's parts with a footprint, their nets, a first placement (not routed). It never writes over an existing board (--force does); open them in the PCB editor on their own (no project of their own, so no F8: a schematic change goes in by hand or a sync like tools/sync.py).

The RP2040-Zero and RP2040-Tiny (same 23 castellated holes, rows 15.24 mm apart) use Waveshare's pin numbers (GP0-15 = 1-16, GP26-29 = 17-20, 3V3 21, GND 22, 5V 23; 24 = the USB), with Waveshare's STEP models in limn.3dshapes/ (gzipped, .stpZ). The library footprints point at KiCad's 3D models (${KICAD9_3DMODEL_DIR}): install kicad-packages3d to see them.

Once the schematic is edited in KiCad, regenerating overwrites the edits: don't, or move the change into the generator first.

To fill in

• The FE1.1s breakout (which board) and its pad map: limn:FE1.1s_USB_hub_breakout is a placeholder (U1002).
• The Adafruit USB-C breakout's pin order (U1001: VBUS, D+, D-, GND in a row for now).
• The beam bed's pogo: its pitch and pin order (J851: a 1x05 2.54 mm row for now); where the LEDs/PTs sit against the slots; the board's thickness.
• The toolhead PCB: the K flex pads' pitch and place (3.0 mm past the motor's end, a placeholder), the key switch's tabs (from the model), the screws' size and places, the outline; the breakout's STEP (the user, 2026-10-30).
• The FFC: 1 -> 1 (the cell reads right through pins 2-5, 2026-11-01; reversed they'd land on 16-19). Meter 13-15 before power anyway.
• The bed pogo connector's pin order (power, GND, DETECT, TX, RX).
• The cameras on the Pi: which is on which port, and which devices are on the hub vs straight on the Pi today.
• The buzzer circuit: gate resistor, supply, flyback diode.
• The Pico's VSYS (USB only, or the Aux PCB's LM2596 through a diode?), and where the MPU9250 gets its VCC.
• The MCP23017 breakout's RST pin.
• The ADS1220 module's header order: from the photo (2026-10-18), DIP numbering, top view; footprint limn:ADS1220_module.
• The Pololu Mini MOSFET Slide Switch's pad map (U1006 on the hub board).
• Where the Pro Micro's D+/D- pads are underneath (footprint pads 30/31 are placeholders by the USB-C).
• klipper/leds.cfg for the RP2040-Zero (lamp gpio16 → 14, indockator gpio17 → 13) and the merged Dock/plot bed firmware (dock pins GP29, GP0/1, GP4 = BED_EN, GP5 = SW_5V_EN, high = on; CS2 on GP21, LED_R/LED_G gone).
• The FSR board's series resistors on the rows (values).
• The Melzi's endstop headers (2 or 3 pins), and whether the BLTouch and the OR gate are already out.