Small CoreXY printer

What changes from the original

1. Frame — The donor printer's V-slot extrusions, cut to the Aether's three lengths, with the design's V-slot brackets.
2. Printhead — Its Creality Sprite (direct drive, all-metal, 300 °C) whole, on a new printed plate that keeps the back half of the original's printhead plate — the part that bolts to the Z carriage — and replaces the front, with its own fans. Beside it, an eddy-current probe (Beacon) scans the bed in seconds and sets the height by touching the nozzle itself to the bed.
3. Bed — Our own 3 mm aluminium plate on the design's four adjustable corners, a stock 100 mm 24 V silicone heater pad between the corner screws, a thermal fuse taped to the plate, and the donor's PEI spring-steel sheet on a magnet, cut to size.
4. Electronics — Its BTT SKR Mini E3 V3 board in a new printed box under the frame, in place of the Manta M5P, with a separate Raspberry Pi running Klipper and the donor's power supply on the bench.
5. Motion — Its 1.8° stepper motors on X and Y (weaker than the design's 0.9° motors, so it starts at 2,500 mm/s² instead of 20,000), five bought MGN9H rails, and a bought 4 mm-lead screw motor for Z.
6. Gantry — The design's printed gantry. Its aluminium version is the upgrade if a test print shows the gantry is what limits the print.

Build log

• 2026-10-06Every bought part is now a standard catalogue part — nothing made to order. The bed heater became a stock 100 mm square pad, the cutoff a standard thermal fuse, and the probe's two screws ordinary button heads; the probe slid 5.5 mm along its arm so those heads clear the bed's cable clamp at every position. It will run one nozzle only: a 0.4 mm diamond-tipped one, for abrasive filaments.
• 2026-10-05The front feet. The original's own drawing stands the front of the printer on its electronics box and power supply; ours keeps the power supply on the bench, so the design's other front foot is drawn in: a 200 mm bar cut from the donor's 20 × 20 extrusion, with the original's bracket and two rubber feet. Three feet, level, the controller box 9 mm above the table, and the machine's centre of mass at least 26 mm inside the feet wherever the bed is.
• 2026-10-05The model now prints. Press Print it and the machine makes a 1¼-inch hose coupling in one continuous spiral, 0.25 mm layers at 80 mm/s shown ten times faster: the nozzle never moves sideways, the bed and its beam carry the part around under it, and the printhead climbs the lead screw a layer at a time. Every part moves exactly as the CAD says it does.
• 2026-10-05The probe, decided the long way. The donor's pin probe was dropped for nozzle-contact sensing like today's better commercial printers. Some feel the nozzle tap through sensors under the bed, which works because their bed only moves up and down; ours moves fast in both directions, so the sensor went on the head instead — a load cell between the arm and the extruder. Modelled and checked, it made the nozzle three times softer sideways, and even with a steel-leaf guide it stayed softer than the plain head. So the head went back to the rigid one-piece plate, and the probe became an eddy-current sensor 2.6 mm above the nozzle: it scans the bed's shape in seconds and finds the height by tapping the nozzle on the bed. No springs, no moving pin, no stiffness lost.
• 2026-10-05The printhead plate is modelled. The Sprite comes from published CAD — a community model that matches the maker's own dimensions to about a millimetre — and mounts by four screws on its side; the plate keeps the original's fixings to the Z arm and replaces the rest. A first try lowered the nozzle 12 mm and cost that much build height, so it was redrawn: the head now sits at the original height, 15 mm to one side, which trims the width the nozzle reaches from 130 to 120 mm. The probe moved beside it, along the long axis. One printed part, four screws. Checked against the original's CAD at 19 bed positions and at full height; the last check is a test print of the plate fitted to the real extruder.
• 2026-10-05First two parts modelled, in blue-green above. The controller box holds the SKR board on the original's four mounting points under the frame, with its fan on a sloped end wall so it prints without supports. The bed plate is a plain 108 × 159 mm rectangle of 3 mm aluminium, cut and filed by hand and drilled through a printed jig, on the original's adjustable corners; a heater pad up to 108 × 133 mm fits under it between the corner screws. Both were checked against the original's CAD: no collisions, and the bed reaches exactly as far as the original's before it meets the Z motor.
• 2026-10-05Design decided. The numbers: the bed is what accelerates (about 800 g with ours), the whole Sprite on Z weighs about the same as the original head, and the donor's motors hold the starting speeds with half their torque to spare. An adversarial review then changed three things before anything was bought: a heater pad with mounting holes in it can't be made this small, so the pad sits between the corner screws and the design's adjustable corners stay; the Z screw is a 4 mm lead, which holds the head when the motor is off; and the gantry is printed first. Next: measure the Sprite, then model its plate.
• 2026-10-05Planning round started. The Aether's own CAD was read into 172 parts in assembled position, the faces our new parts must meet were measured from it, and the specs of the donor parts we are re-using were looked up at their sources.