Bubble packing puzzle

I designed a new puzzle this morning, it's a relatively simple packing puzzle where you have to fit 21 circle-ish pieces into a case.

It ended up easier to solve than I intended, because the bubble parts were printed in the "solved" configuration, which means the 3d printing lines all point in the same direction, so it's easy to know which way they should be oriented.

If you want to print it yourself, you can download the print files from prusaprinters.org.

I haven't used prusaprinters.org before, but so far it seems vastly superior to Thingiverse: you can upload designs on the same day you sign up instead of having to wait 24 hours to be "approved", and the pages load like they're not served over dial-up.

In comparison to normal jigsaw puzzle geometry, in principle, this puzzle should be harder, because there's no clue as to the piece orientation (jigsaw pieces can only fit at 4 angles, circles can fit at any angle), and there's no clue as to which pieces form the outer border.

Design

I started by drawing a "master sketch" in FreeCAD, with a bunch of overlapping circles.

This sketch is used to model both the pieces and the casing that they fit inside.

For the pieces, I made a flat plate and then raised up the pieces with small gaps in between, in a single part:

Ideally each piece would be a separate part, but I couldn't work out an efficient way to do that in FreeCAD.

I think a taxonomy of the pieces would identify 4 distinct properties, based on the overlapping circle shapes:

1. pieces that are a complete circle
2. pieces with all sides concave
3. pieces with a sharp concave corner
4. pieces that form the outer border of the shape

I made sure to include at least one of each type.

I exported the single part with all the pieces as a single STL file and then used the "Z Cut" feature in PrusaSlicer to chop the bottom off, so that they would all be loose once printed:

To make the case, I just made a solid block and pocketed in the outline of the master sketch, with a gap around the outside:

I left 0.5 mm clearance between the pieces, but this turned out to be too much and they are too loose in the final puzzle.

Improvements

For a second iteration, the following changes should be made:

• the pieces are too loose: the clearance should be reduced
• the pieces should be separated from each other and printed in a random orientation
• the pieces are too small and fiddly: the entire puzzle should be scaled up by a factor of 2 or 3

It might be nice to make a larger version in wood using the CNC machine.

Another idea would be a 3-dimensional version, where instead of a bunch of overlapping circles with bits cut out, it would be a bunch of overlapping spheres with bits cut out.