Molding and Casting #

Intro #

In this page, I will document the process of making a keycap using molding and casting.

Designing #

I started with a standard keycap model like this:

A mold was made by subtracting this model from a cube. Immediately I realized the walls of the model has angle which made it impossible to mill the mold.

After re-evaluating the shape of the model, I decided to flip the model around so that the cross-shaped peg is now facing upwards, thus providing the chance of making it a two-part mold.

For validating the design, I made a 3D printed mold to test it out.

Turns out the test was a complete failure, but I learned some valuable things:

• The shape for the upper part of the mold is too complicated. The central part broke inside the mold, probably because it’s too thin.
• Print quality should be higher, in order to get a smoother finish, which will be much easier to release.

With that in mind, I changed the model design again. This time I changed from hollow to a solid infill design. Also all of the vertical faces are tilted with a small angle to avoid friction.

Milling #

Reference video:

Calculating feedrate:

Feedrate = Number of flutes * Chip load * Spinal speed

For this machine and tools used, nf = 2, ss = 14000, so

Feedrate = 2 * cl * 14000 = 28000 * cc

Foam #

Flattening #

Vcarve

• Measure the dimensions of the material
• Input Vcarve
• Create square, covers all the area
• Pocket toolpath, cut depth 10mm (or more)

• Clear pocket –> Pocket allowance –> -11mm

• Save toolpath

Roland Modela MDX-40

• VPanel
• Make sure is on G54 coordinate system (in XYZ and Set origin point section)
• Install tool
• Move material under the tool, lowering the tool
• Place the sensor
• Make sure sensor wire are connected and is under the tool
• Set Z origin using sensor –> Apply
• Move tool to start point
• Set XY origin here –> Apply
• Cutting speed –> 30%
• Cut –> Delete all –> Add .nc file –> Output
• Gradually add cutting speed

Roughing #

Vcarve

• File –> Import 3D model
• Check orientation, size, central
• Z plane should be lower the lowest point of the pocket
• Hit OK
• Toolpaths –> Material Setup –> change if needed
• Toolpath –> roughing

fr = 28000 * cc = 28000 * 0.07 = 1960mm/min

• Machining limit boundary –> Model boundary
• Machining allowance –> 0.1mm
• Roughing strategy –> depends on your model

Finishing #

Reflections after the session:

• Foam milling should not only serve as the test cut for verifying the shape of the model, but also should be used as a reference in terms of time usage as the ratio between the feedrates used for cutting foam and wax is known.
• Roughing strategy for most “flat” design model could be z level, but for models that contains tiling vertical faces like mine, maybe 3D raster is the better choice.

Wax #

Roughing #

We use the chip load for steel here when milling wax to make sure that the machine can handle it.

fr = 28000 * cc = 28000 * 0.038 = 1960mm/min

Finishing #

For saving time I used the same 6mm tool for finishing because I want to get back home before sunrise😇. Lesson learned: reserve at least 4h for milling both the foam and the wax.

Milling results #

Overall it looks good but there’s definitely stair-shaped edges on the vertical faces. Several reasons:

• 6mm tool rather than 3mm
• 0.5 stepover is too much
• z level rather than 3D raster

Will look into those factors and see which is the deciding one next time milling.

Casting #

Darren had a super detailed introduction for all the materials used in the casting. Shout out to him for making such a helpful documentation.

Casting silicon #

For the silicon mold I used the Moldstar 15 slow:

After mixing we did a long degassing session, probably 20-30 minutes which is significantly longer than what I expected.

Casting plastics #

For the actual model, I used the Smooth cast 305.

Final results #

Casting process was successful. Removal of the 3D printed part was easier than I think, for I was worrying about the cross-shaped peg would be stuck inside and broke off if I try to pull it out.

Sanding could potentially remove the stair-shaped edges, but I think the current appearance is acceptable, as the strips now appear as if they are intentionally designed with that feature🤣.