
Sheet metal formingDesignerDrawing 47 of 47
Metal
Keycaps
Keycaps pressed out of flat aluminium sheet, one stroke of a punch into a 3D-printed die. Before printing any tooling, I simulated the press to see whether the corners would tear, wrinkle or come out at all.
- StatusSimulated, dies drawn, 2026
- Material0.5 mm soft aluminium, 3003-O or 5052-O
- Cap1u low-profile, Choc stem, 18 mm across
- Toolingforming die and punch, plus trim dies for 1 and 4 caps
- SimulationCalculiX and OpenCourant, up to 464,000 elements
Sheet 02 of 08/Why
Caps from flat sheet
Each cap starts as a flat laser-cut square of aluminium. A punch presses it into a die in one stroke and folds the skirt down, then the cap gets a Choc stem glued underneath and gets anodized.



Sheet 03 of 08/Tooling
The dies
Fusion 360 scripts build the tooling straight from my own keycap body: a female die, a male punch and the flat blank, with two 5 mm steel pins to line them up. The dies are 3D printed. A second pair trims the scrap ring off the edge, and a four-up version trims four caps at once.






Sheet 04 of 08/Simulation
Pressing it in a computer first
The first runs in CalculiX showed how much the die's edge matters: with a 0.5 mm rounded edge the press got to 79 percent, and with a 0.2 mm edge it stopped at 57. The full runs moved to OpenCourant on my PC, covering the press, the cut and the punch lifting off.

Sheet 05 of 08/Notch
The missing corners
The first blank had a square notch cut out of each corner, so the metal wouldn't bunch up there. The simulation showed the notch reached past the cut line into the cap itself, so the finished cap came out with no corners.


Sheet 06 of 08/Result
Full corners
With no notch at all, the skirt formed all the way round, the cut went cleanly round the edge, and no wrinkles showed at the corners. Heating the metal to 250 °C didn't change the shape.



Sheet 07 of 08/Detail
A finer mesh
The last run used a 0.055 mm mesh with nine layers through the 0.5 mm sheet, 464,000 elements for a quarter of one cap. It took about six hours on my PC.
Sheet 08 of 08/Limits
What the simulation can't show
The solver doesn't really remove metal at the cut. It squeezes it down to a thin film, and the videos hide that layer to show the separation. How clean the cut edge comes out can only be checked on a real press, and that's the next step: I haven't pressed a cap yet.