3D Printed Maze
Here's a surprisingly quick and easy-to-build little toy -- assuming you have a 3D printer and a little more. Note that this maze is too small to be safe for kids under age 3.
A few years ago, I made the big wooden maze (also shown here) as a technical demonstration of some fancy computer engineering technology my research group at the University of Kentucky developed. That technology allows GPUs (Graphic Processing Units) to efficiently execute programs written for cluster supercomputers. The four balls represent four independent programs that want to do different things (go in different directions) simultaneously, and the tilting of the maze to move them models how a GPU executes the same instruction at the same time for all programs. If you care, you can read about that at http://aggregate.org/MOG/ .
Anyway, I made these little mazes as handouts with the exact same design as the big wooden maze. They don't produce the impressively loud "clunk" sounds when the balls hit the walls, nor do they work anywhere near as smoothly as the big wooden one, but these tiny mazes do work. If you print them at about 1.5X the size shown here, and use correspondingly larger balls, they work much smoother. Either way, the balls are sealed behind a transparent plastic cover and even the tiny maze is fairly durable.
Parts & Tools
There are three components that you'll assemble to make the maze:
- The 3D-printed maze itself.
- The transparent plastic cover.
- Four chrome steel ball bearings, 1.5mm diameter.
- A spool of appropriate PLA filament for your 3D printer... you'll use very little for one maze.
- A sheet of overhead transparency material for a copier/laser printer will make at least 24 covers (under $0.04 each).
- Either buy the balls new (under $0.01 each) or salvage them from a discarded ball-bearing computer fan.
- A 3D printer (I used a MakerGear M2).
- Scissors, a programmable paper cutter (what I used), or a laser cutter.
- An ordinary household clothes iron.
The 3D Model
It was actually a bit of an unexpected pain to create the 3D model for this maze....
I still had an xfig drawing of all the pieces that I made for the large wooden maze. You'd think that it would be a simple matter of taking the 2D maze drawing, scaling it, and extruding it to a 3D model using something like freecad. However, the wall thickness choices were not right when scaled. Thus, I had to re-work everything and actually ended-up writing a script to generate (ugly looking) openscad code for the 3D design by translating and transforming the 2D xfig model.
I've posted the suitable-for-3D-printing SCAD and STL files at http://www.thingiverse.com/thing:65025 . You'll also find the 2D PS file I used for cutting the maze tops there.
I still had an xfig drawing of all the pieces that I made for the large wooden maze. You'd think that it would be a simple matter of taking the 2D maze drawing, scaling it, and extruding it to a 3D model using something like freecad. However, the wall thickness choices were not right when scaled. Thus, I had to re-work everything and actually ended-up writing a script to generate (ugly looking) openscad code for the 3D design by translating and transforming the 2D xfig model.
I've posted the suitable-for-3D-printing SCAD and STL files at http://www.thingiverse.com/thing:65025 . You'll also find the 2D PS file I used for cutting the maze tops there.
Print the Maze Body
Print the maze.
Using PLA filament, it took our MakerGear M2 only about 15 minutes to make this (not counting waiting for the bed to heat and cool). You can see a time-lapse animated GIF of the whole thing here.... There isn't much solid volume in this print, but the side walls of the maze are sufficiently stiff with anything over 10% fill, and even 25% fill barely changes the print time or quantity of PLA used.
Using PLA filament, it took our MakerGear M2 only about 15 minutes to make this (not counting waiting for the bed to heat and cool). You can see a time-lapse animated GIF of the whole thing here.... There isn't much solid volume in this print, but the side walls of the maze are sufficiently stiff with anything over 10% fill, and even 25% fill barely changes the print time or quantity of PLA used.
The Cover Up
The balls for the big wooden maze are from a croquet set. They stay in the maze because they and the maze have more than enough mass to ensure your tilting action will not beat gravity. However, with the tiny 3D printed maze, we need a transparent cover to keep the 1.5mm diameter ball bearings from bouncing out as you tilt the maze.
You really can't 3D print a transparent cover. PLA material is more translucent than transparent, and the "stringy" structure of the printed material causes more optical distortion. However, you can make a plastic cover out of overhead transparency film designed for use in copiers or laser printers.
Feel free to cut an appropriate piece using a pair of scissors. However, if you happen to have one of those cheap programmable paper cutters, that will work even better. Driven by a postscript file (mazetops.ps from http://www.thingiverse.com/thing:65025) fed into graphtecprint, it took only a few minutes for my sub-$200 Silhouette paper cutter to extract a 4x6 array of these covers from a single sheet of transparency material. The best thing is that they are all the same exact size with nicely rounded corners and no roughness at the edges. Be warned that scissors sometimes leave an edge rough enough to cause some corner lifting after thermal bonding....
If you wish, you can even copy/print/draw a design onto the cover, preferably before cutting it out. If you do that, print the design backward so it will look right when the side it is printed on is sealed against the maze body.
You really can't 3D print a transparent cover. PLA material is more translucent than transparent, and the "stringy" structure of the printed material causes more optical distortion. However, you can make a plastic cover out of overhead transparency film designed for use in copiers or laser printers.
Feel free to cut an appropriate piece using a pair of scissors. However, if you happen to have one of those cheap programmable paper cutters, that will work even better. Driven by a postscript file (mazetops.ps from http://www.thingiverse.com/thing:65025) fed into graphtecprint, it took only a few minutes for my sub-$200 Silhouette paper cutter to extract a 4x6 array of these covers from a single sheet of transparency material. The best thing is that they are all the same exact size with nicely rounded corners and no roughness at the edges. Be warned that scissors sometimes leave an edge rough enough to cause some corner lifting after thermal bonding....
If you wish, you can even copy/print/draw a design onto the cover, preferably before cutting it out. If you do that, print the design backward so it will look right when the side it is printed on is sealed against the maze body.
Going Undercover
Now we're ready to attach the cover to the maze.
- Sit the printed maze face up on a table and carefully place four of the tiny ball bearings in it.
- Set the clothes iron to around 2-3 on the temperature scale and let it heat to that set point.
- Lay the cover material over the maze and, while holding it in place at one corner, gently apply the hot iron to tack down the cover. The non-stick surface of the iron will not bond to the plastic if the heat level was set correctly.
- Once one corner is tacked, turn the part around and hold it by that corner as you work from there to tack the remainder of the cover. Make sure the cover is sitting flat on top of the maze as it is tacked.
- Make several passes over the surface of the cover until you see that the tops of the internal walls of the maze have slightly melted against the cover. The edges of the cover should also sink-into the outer walls of the maze enough to be smoothly inset. When that happens, it's bonded.
- Let it cool slowly. Cooling too fast can cause the different expansion rates of the cover and PLA material to make the cover partially lift -- if that happens, simply repeat from #4 above.
Built! Now Go Solve the Maze....
Congratulations! You've now built the maze.
See the start and end positions for each of the balls in the rendering? The bad news is the little balls are not in the start position, are they? With the big wooden maze, we simply reach in and move them to the start position. With the tiny balls sealed under a sheet of plastic, you'll find getting the balls to the start position is more of a challenge. Getting the balls back and forth will provide hours of frustration fun!
Incidentally, the chrome steel balls can be colored using dye or a permanent marker. Thus, you can make them match the colors of the balls in the big wooden maze -- also shown here. However, that makes it even harder to get each one to its start position, so we don't recommend it.
Part of the frustration fun is that the balls don't roll too smoothly unless your 3D print is really good. For a smoother-operating maze, print the model scaled-up 150% and use larger balls. We've printed at both 150% and 200%, but there really wasn't much additional improvement going past 150%; it just depends on the smoothness of the maze floor in your 3D print.
See the start and end positions for each of the balls in the rendering? The bad news is the little balls are not in the start position, are they? With the big wooden maze, we simply reach in and move them to the start position. With the tiny balls sealed under a sheet of plastic, you'll find getting the balls to the start position is more of a challenge. Getting the balls back and forth will provide hours of frustration fun!
Incidentally, the chrome steel balls can be colored using dye or a permanent marker. Thus, you can make them match the colors of the balls in the big wooden maze -- also shown here. However, that makes it even harder to get each one to its start position, so we don't recommend it.
Part of the frustration fun is that the balls don't roll too smoothly unless your 3D print is really good. For a smoother-operating maze, print the model scaled-up 150% and use larger balls. We've printed at both 150% and 200%, but there really wasn't much additional improvement going past 150%; it just depends on the smoothness of the maze floor in your 3D print.