Category: Software

We buy olive oil in large bottles, then fill smaller bottles for easier handling. The caps on those bottles were never meant to last as long as we keep them and the thin, deeply drawn aluminum tends to crack after a while.

So I conjured a cap cover from the vasty digital deep:

Which looks exactly like you’d expect when printed in black PETG:

You can see the raggedy edge of the original cap just inside the cover’s rim. A snippet of double-sided tape holds the cover in place, after de-oiling the cap with alcohol.

Having gotten one to fit, I made enough for All The Bottles:

Only two of those see regular service: one in use and another filled when the first is nearly empty. The remaining pair huddle in the back of the shelf against future need.

The OpenSCAD source code produces those fancy knurls with BOSL2’s textured cyl() :

// Shower soap dish
// Ed Nisley - KE4ZNU
// 2026-01-17

include <BOSL2/std.scad>

/* [Hidden] */

HoleWindage = 0.2;
Protrusion = 0.1;
NumSides = 5*3*4;

$fn=NumSides;

ID = 0;
OD = 1;
LENGTH = 2;

WallThick = 1.0;
BaseThick = 2.0;

CapOD = 36.0;

CoverOA = [CapOD,CapOD + 2*WallThick,20.0 + BaseThick];

//----------
// Build it

  render()
    difference() {
      cyl(BaseThick,d=CoverOA[OD],chamfer1=1.0,anchor=BOTTOM) position(TOP)
        cyl(CoverOA[LENGTH] - BaseThick,d=CoverOA[OD],
            texture="trunc_pyramids",tex_size=[2,6], style="convex",
            anchor=BOTTOM);
      up(BaseThick)
        cyl(CoverOA[LENGTH],d=CoverOA[ID],anchor=BOTTOM);
    }

A SquidWrench meeting discussion about printing transparent objects prompted me to conjure a soap dish from the vasty digital deep:

They’re all done in “natural” PETG with sufficient variations in speed, temperature, extrusion multiplier, and fill pattern to stock the shower & tub:

The single-thread sidewalls came out reasonably translucent in all the variations, but the baseplate remained stubbornly white-ish, even at 20 mm/s and 250 °C with 100% infill. The seams where the extruder retracts and lifts to the next layer remain conspicuous, with a scarf joint forming the white slab in the left-rear dish.

Quite a while ago, I’d considered making soap dishes with shattered-glass bottoms, but came to my senses. These have some key advantages:

• Exactly the right size for narrow shower shelves
• Light enough to not damage anything when it inevitably falls off
• Reasonably unbreakable when that happens
• Easily replaced

They’re also test pieces for the whole transparency thing, so it’s all good.

The OpenSCAD source code as a GitHub Gist:

The mixed flock attending the bird feeder in the back yard scatters enough seeds to attract the deer, so I added a tray underneath to catch the overspray:

Well, two trays, because it took a couple of iterations to make the solid model match reality:

The n-1 iteration was Close Enough™ and two trays are obviously better than one.

The “trays” are stray lids from the six gallon buckets we use for many purposes, including root-cellaring the vegetable garden harvest. The lid’s solid model was straightforward:

Removing the lid from a solid block produces the most complex part of the mount:

An aluminum plate on the outside (the gray slab in the overall view above) distributes the strain from the two M6 screws across the block.

A smaller block on the inside of the lid has a pair of square nuts:

All three parts build from their flattest side:

The downward facing clamp arch in the main block didn’t need support, but the square nut sockets in the segment definitely came out better with little support blocks inside; PrusaSlicer does a good job with most support structures.

The n-1 iteration used M6 rivnuts that were slightly too long after making the lid model match reality, so I switched to square nuts. The OpenSCAD code calculates the segment block length to match the actual screws, but 75 mm M6 screws and square nuts are barely long enough.

I clamped the outer block to the lid as a drill guide for the first hole, then pinned the block with a screw to ensure it didn’t slip while drilling the second hole:

Those were freehanded in the drill press at low speed with serious concentration; some things you just gotta do that way.

The mixed flock overwhelmingly approves the trays, to the extent a dozen birds clamor to use them: definitely a crowd-pleaser!

I’m certain you can buy pole-mounted trays, but what’s the fun in that?

The OpenSCAD source code as a GitHub Gist: