After playing around a bit with my TMD-3 I'm ready to say it's finished. Tiles are recognized with great accuracy which was my first goal. The end result turned out to quite compact which was something else I had hoped to achieve. Here are my three Turing Machine Demonstrators for a comparison. 

One thing that I noticed is that the tiles themselves tend to lump together when loose due to the magnets that I used (6 mm). They are fine when placed into the state panels. I did not notice this with TMD-1 because the magnets that I used were smaller (3 mm) and the tiles much bigger and heavier. For TMD-2 the tiles didn't have any magnets at all. I did a quick test with 4 mm magnets and the effect was much less. I could also print the small tiles with 100% infill so they would be heavier. For now given the number of tiles I have printed and the time invested in calibration I'm not going to make the switch. In fact I plan to take advantage of the stronger magnets by creating a tile organizer backed with a steel sheet to hold the tiles in place.