Build step 01 / 05 ~3 min read

at the endA sliced plate on your printer: three parts, one plate, 1 h 28 min, 15.4 g.

Print it

One plate, 1 h 28 min, 15.4 g.

Send hardware/cad/head_plate.gcode.3mf to a Bambu X2D. PLA, 0.20 mm Standard, tree supports from the build plate, max z 36.4 mm. Other printers: head_plate.3mf is the unsliced plate, or the STLs below with supports on.

part	cm³	on the plate
yoke	8.89	wall face down
body	5.11	flange down, open top up
lid	1.23	on its side

STLs in hardware/cad/: yoke.stl, nicla_head_body.stl, nicla_head_lid.stl; the _oriented copies are already turned for the plate.

Orientation is measured, not guessed (fitcheck_print.py): any face over 30 mm² in the air, or a bridge past 10 mm, gets supports. The first plate went out without them and scrambled.

Every plate

The head is one plate; the body and the wrist cradle add four. The last card is the stock SO-101 follower for anyone starting without a kit. Every number comes from the sliced file itself, rebuilt on each docs build.

plate 01 head yoke, body, lid 1 h 28 min15.4 gsupports head_plate.gcode.3mf plate 02 pods wheel_pod x3 8 h 02 min130.2 gno supports pods.gcode.3mf plate 03 centre deck centre_deck 4 h 08 min106.3 gno supports centre_deck.gcode.3mf plate 04 upper deck upper_deck 2 h 34 min74.8 gno supports upper_deck.gcode.3mf plate 05 wrist nicla_wrist_body, nicla_wrist_lid 40 min7.2 gsupports wrist.gcode.3mf plate 06 follower, stock 12 SO-101 parts, skip it if your kit's follower is intact 1 d 5 h 39 min381.6 gsupports slice_follower.py

Printability, learned the hard way

The first plate shipped in whatever orientation the CAD exported and nobody measured it. It scrambled. fitcheck_print.py now measures every part in its plate orientation: overhang past 45°, flat ceiling hanging in air, longest bridge, and the area on the bed.

part	as first exported	now	how
yoke	1449 mm² overhang, 1403 mm² ceiling in air (the 26 mm flange as a mushroom), 46 mm bridge	279 mm² overhang, 7.5 mm bridge, 1289 mm² on the bed	wall face down, rotated −90° about X. Six candidates measured; flipping was no better than as-is because the servo pocket became the ceiling.
body	545 mm² overhang, worst 189 mm² at the cavity roof	342 mm², cavity opens up so the fit surfaces print clean	flipped 180° about X, flange down
lid	107 mm² overhang, 20.7 mm bridge	13.4 mm², 10.0 mm bridge, support-free	on its side

The rule that came out of it: any face over 30 mm² in the air or a bridge past 10 mm gets tree supports from the build plate, and the slicer is driven from the repo (slice_plate.py) so the supports are in the .3mf, not in someone's memory. The remaining 279 mm² on the yoke is the underside of the 24 mm boss, a horizontal cylinder; tree supports carry it.

A second lesson is in the numbers on The head: printed inner dimensions come back 0.1 to 0.2 mm undersize, so every fit in this design is 0.5 mm per side or wider.

After printing

• Board slides in through the lid face by hand, under the two lips.
• Servo 6 drops into the yoke, four tabs flat on the wall.
• Lid clicks on with thumb pressure.
• Screws at hand: 2 × M2.5 × 5, 4 × M2 × 4 self-tap, plus the four horn screws from the handle.

Regenerate from source: python hardware/cad/yoke.py && python hardware/cad/nicla_head.py && python hardware/cad/slice_plate.py. Then put it together.