For TMD-1 the nine different tiles were created with the magnets embedded into the tile itself. This was accomplished by defining magnet sized voids inside the tiles and pausing the 3D print at the appropriate time to insert the small magnets. This task was simplified by the fact that the voids were all at the same depth within the tile, plus the polarity of the magnet was not relevant, so a big batch of different tiles could be created at the same time. 

There were no magnets in TMD-2 so printing them was pretty straight forward. 

For TMD-3 the sixteen different tiles will have the magnet voids at different heights depending on the tile, so smaller batches would be required when printing.

I thought I might make the job a bit easier by assembling tiles from pieces. That way I can print the parts in larger batches. Here is what I am thinking.

We start on the left with a tile base. There will be four different types of these with the magnet pedestals at different heights (1 mm to 4 mm).  The magnet is placed on the pedestal and held in place with a printed cap. Finally the tile face is snapped in place. All of these pieces can be printed in large batches and assembled later. 

Once I had the "system" in place I decided to make a complete set of tiles to test with. I printed the bases in one batch, the tops in another (with a pause to change filament), and the caps in a third print run.

The caps an tops fit pretty snugly, but still needed a drop of glue to set them in place. I used Gorilla Super Glue with the "Micro Precise" applicator which really helped administer the very small amounts of glue necessary.

My next step is to figure out the best way to determine the appropriate measurement settings necessary to recognize the differences between tiles. Will I need to have a "calibration" set when setting up the state transition table? I guess I'll find out.