What should I name this puzzle?

It pays to share with non-puzzlers.  This puzzle was inspired by a non-puzzler's comments on a study model, and it reached completion in just a few days.  And it's remarkably ideal for building on my SD300 because it can mass-produce them quickly, easily, and economically.

 It began when I shared this geometric model with my coworkers--several people enjoyed playing with it.  It's not really a puzzle, but when they took it apart they had trouble putting it back together.  One coworker attributed it to the circular shape, the most distinguishable feature when the pieces are apart, because it encourages the user to instinctively follow the wrong contours.

His comments let me to design this puzzle with two congruent pieces, whose circular shape and curved surfaces are intended to mislead the user.  This is a breakthrough puzzle for me because it's my first that uses congruent pieces that exhibit symmetry relative to each other, yet it has only one unique solution out of the four permutations in which the pieces could be oriented.

 The side walls give a small hint at how it works.  Not much, though.

I'd originally planned it as a colorful opaque puzzle, with an appearance like a bonbon.  But it looks stunning in the SD300's transparent material, which shows off the delicate internal contours when the puzzle is assembled.

I'm also experimenting with other applications of these principles.  This trillion-shaped puzzle has contours that imply 3 axes of rotation, running through each of the 3 corners.  Naturally that's not really possible, but the illusion is sufficiently distracting that my testers had a really hard time with it.

Not-so-wonderful Waterful

Inspired by a wasp trap, a friend suggested a water-filled dexterity puzzle that challenges the user to guide a floating ball into an inverted funnel.  It's a cool idea, but why not use two balls: one that floats and another that sinks.  I designed a two-piece box that had transparent windows and two funnel-shaped goals.

I put two little balls inside and sealed it full of water.  One ball is polypropylene, which floats, while the other is made of PTFE, which sinks.  The challenge is to get the floating ball into the funnel at the top and the sinking ball into the funnel at the bottom--at the same time!  My first test model seemed to work perfectly.

Problem!  The PVC windows gradually fogged up as the PVC absorbed moisure.  After a couple of days it turned completely opaque.  This effect is reversible: it turns transparent again if it dries out, but the puzzle needs to be filled with water so this design just wasn't practical.

So I tried a different approach: I incorporated the funnels into two caps that could be bonded to a piece of acrylic tubing.  Acrylic doesn't absorb moisture so it would stay crystal clear, right?  I assembled a test model and filled it through a fine needle and sealed it up.

This worked much better: the side walls were perfectly clear and the cylindrical shape worked like a lens.  Here's how it looks when solved, with the floating ball in the funnel at top (barely visible) and the sinking ball in the bottom funnel.  At first it seemed perfect: no fogging, no bubbles!

But a small bubble appeared a few days later, and gradually grew.  After two months it had grown so large that it interfered with the balls, and it's obviously going to keep growing.  So this won't work, either.

Although the containers seem to be watertight--no leaks--I'm guessing the PVC isn't vaportight.  So the water vapor could be escaping by diffusion through the PVC end caps.  But my coworkers liked this puzzle while it worked, so it's worth another try.  Back to the drawing board!

SD300 Issues from emails

Through email I've learned a lot of tricks and solutions to SD300 issues.  Occasionally I've relayed a solution from one user to another, who then devises an improvement of his own.

A user in Romania found his Anti-Glue cartridge had dried out while the machine wasn't being used, so here's a trick I learned from a British user: add water to the cartridge by injecting it into the breather hole with a syringe or needle-tipped squeeze bottle.  (49 gauge fits)  One user drilled a larger hole and refilled it through a funnel.

This technique could also be used to extend the life of an Anti-Glue cartridge in case you've spoiled one.

Occasionally users have reported an odd phenomenon where unused support material doesn't tear away cleanly because the layers are stuck together and the unused plastic has a gooey, melted texture.  I'm not sure it's the same, but I recently experienced something like it while cleaning the model in the picture.  Most of this material is unused support material.  It should all be shiny like the area at lower left.  The problem area is at top-right, which didn't peel away cleanly.  Each layer of unused material was littered with tiny spots that were tightly bonded to the layers underneath--and it appears that the shiny Anti-Glue film was interrupted by little spots where the layers were unintentionally glued together.

My first theory was glue might have dripped after ironing a new layer to the top of the model block, as pictured in this previous post.  So I watched the machine carefully during the next build.  No droplets.  But when the machine applied Anti-Glue there were spots in the film--holes through which glue might unintentionally bond the base layer to the model and support material.

I watched more carefully as the machine ironed the next layer of material, and noticed it already had smears of glue across the top surface--right out of the feeder!  How was that even possible?

Aha!  There were scraps of plastic clinging to the underside of the iron bridge, which contains the jets that spread glue on the top of the model.  My current theory is that those scraps hung down enough to touch the PVC as it was being rewound into the feeder, drawing glue into the feeder by simple capillary action.  Right or wrong, the issue hasn't occurred since I cleaned around the glue jets underneath the iron bridge.

It's not obvious how to get access to inspect and clean the underside of the iron bridge so I made a video.

Puzzle Overview: Moiré Maze

I've previously posted about the development of Moiré Maze so here's an overview of how it works as a puzzle.  There are a few clues about its solution but no spoilers.

At left is Oskar's Boston Subway puzzle from 2006.  It's a sandwich of acrylic sheets containing laser-cut channels; the bottom layer has 10 parallel channels running north-south and the top layer contains 10 parallel channels running east-west.  It's a criss-cross grid.  There's a layer in between that selectively connects the top and bottom grids at some places, thereby turning it into a puzzling maze.  George Miller constructed the puzzles and put inside a small magnet the user can guide through the maze using a magnetized wand.

It occurred to me to build a similar layered maze using circles instead of straight lines.  Ultimately I developed two layers of concentric ring-shaped channels, one over the other as depicted at left.  The layer at right is sandwiched between the other two layers to selectively allow the user to transfer between the two sets of rings like a maze.

The layers are stacked like this with a mirror on the bottom and a transparent cover on top.  I carefully selected a combination of layer thicknesses and a much thinner magnet to prevent the magnet from becoming turning sideways and becoming wedged between layers, a nuisance Boston Subway sometimes exhibited.

Laser cut acrylic comes with a protective film over the surface.  To avoid getting dust and fingerprints on the layers inside the puzzle, I only removed the protective film after each layer had been assembled into the puzzle.

But there's more to this puzzle than a mere shape variation on Boston Subway--it introduces a visual subterfuge that misdirects the user away from the solution.  The critical secret is plainly visible, even in the photo, but it's tough for a viewer to choose to see it once their mind has resolved the image.  Some testers couldn't even force themselves to see it when the solution was shown to them.

In Boston Subway it was sometimes difficult to locate the little silvery magnet because it reflected the same colors as the neutral mirror and irridescent sheets.  I chose to use a bronze-colored mirror with Moiré Maze to enhance the contrast, so the silver magnet is easier to distinguish against the bronze-tinted background.

As a final touch I gave Moiré Maze a built-in stand that secures the layers and holds the puzzle at a convenient slope for playing.  Initially I'd planned to make the stand with laser-cut acrylic like the rest of the puzzle, but after some experimentation I substituted dark-blue pieces 3D-printed from my SD300 because the PVC material is so resilient that it can squeeze the acrylic sheets tightly together.  If I had made those piece of acrylic, then the stand's outer pieces would be strained in tension and would thus be vulnerable to breakage--acrylic tends to fracture under stress.

A Shower Runs Through it

This is one of those "Gee whiz!" stories about 3D printing...

A coworker recently renovated the guest shower in his home, resulting in damage to a 1970's era plastic drain fitting.  He couldn't find a suitable replacement at any of the hardware stores so I offered to make a replacement.

It took only a minute to model a replacement in CAD.  The fitting just needed a flat grid for water to drain through and a collar (not shown) that extends into the opening in the shower floor.

Here's the replacement fitting next to the broken one.  My replacement part is PVC, the SD300's default material, which is ideal for plumbing.

The replacement fitting is thin but nice and strong because the layers are aligned with the natural laminations of the PVC material.  Since it's going to be used in a wet environment I treated it with PVC pipe primer to seal the edges.

It's not a very exciting model, but it was fun and satisfying because I produced it just a few hours after inspecting the original part.  And I posted it to Thingiverse.

Moiré Maze for laser cutter

I took last month's injection-molded prototype for Moiré Maze and adapted it back for laser cutting by projecting it into 2-dimensional layers and tracing their outlines.  I don't own a laser cutter, but plenty of businesses offer laser cutting services including mail-order Ponoko.  But why wait for the mail?

The SD300 can make good simulations of laser-cut parts, given adequate model data.  I created accurate simulations of laser-cut models by first modeling a sheet about the anticipated thickness of the part, about 3.0 mm to simulate the typical "eighth inch acrylic" (which usually runs a tad thin).  To simulate a laser's cutting behavior I cut the outlines using a "thin wall cut" with a 2 degree draft angle.

Models were easy to clean because the sheet-like design meant there weren't any overhangs, undercuts, or loose ends to deal with.  It was easy to stab leftover support material and peel it away from both sides of each finished piece.

I also designed a 2-piece base, shown upside-down here, on which the maze layers are stacked.  The base aligns the maze pieces and holds them together.

The magnetic wand can be stored in a compartment at the top edge of the maze, and it's held in place because the magnets in the wand are attracted to steel beads embedded inside the model.

Above is a brief video tour of Moiré Maze.

Moiré Maze was inspired by Boston Subway, Oskar van Deventer's design for IPP26 shown at left in the picture.  Boston Subway has layers of criss-crossing channels built from laser-cut acrylic, and the goal was to guide a little magnet from "work" to "home" using a magnetic wand.  Moiré Maze substitutes ring-shaped channels for the straight ones, and thereby creates a visual hiding place for its solution even though it's entirely transparent!

Evaluating Protomold and Moiré Maze

A friend from IPP suggested I investigate a company called Protomold for their expertise in rapid injection molding and wide selection of materials.  Although he was not a puzzle builder himself, he had apparently sponsored production of another designer's puzzle via Protomold and was pleased with their work.

As an exercise I chose to adapt Moiré Maze, a design I briefly tested in 2009.  The original test model contained 5 layers of transparent laser-cut acrylic containing ring-shaped channels.  The goal is use a magnetic wand to guide a small magnet through the channels from start to finish.  The lowest layer is a mirror, which I hoped would create an illusion that would disguise one secret feature so its solution could literally hide in plain sight.

I gave the prototype (pictured above) to several testers, and their toils demonstrated the secret feature was a success.  But the maze as a whole was much too elaborate, the nine pairs of overlapping rings merged into a dizzying kaleidoscope of curves.  The key feature worked, but testers agreed the overall maze was just too complicated.

So I consolidated the inner layers of Moiré Maze into a single 3D model, simplified the maze to just six rings, and added details for a built-in stand that could be injection-molded in a standard 2-part mold.  I requested a quote from Protomold, and in the meantime I built a test prototype on the SD300.

As it turns out, the SD300 is great at building a model that's designed for injection molding because there aren't any overhangs to get in the way while cleaning unused material off the model.  I did need to use a fine probe to clear the smaller channels, but it was easy because I had unobstructed access from both sides.

Both sides of the model needed some sort of lip to hold the cover pieces, so I designed these little tabs.  This tab is an example of a sliding shutoff.  It looks like an overhang but it really isn't--the other half of the mold would project through the hole in order to form the underside of the tab.

It's always good to build a prototype: during inspection I discovered an error in my maze.

Upon correcting the error I built another test model, this time using the SD300's transparent material because the user is supposed to see through the mazes.  My PVC material isn't nearly as transparent as acrylic, and there are blemishes from the SD300's construction process.  But it's an excellent test model!

Also using the SD300, I built a pair of transparent covers.  Above, I installed the back panel into the test model.  The back panel ought to be cut from acrylic mirror material, but a simple transparent panel would suffice for testing.

I dropped a small magnet inside and then installed the top cover.

The challenge is to navigate the maze from start to finish by using a magnetic wand to guide the little magnet through the channels between the two covers.

Protomold replied within 1 business day, giving me an interactive quote that enabled me to gauge the cost of various options.  These folks obviously know their business!  Their versatile quoting system permits the customer to view and evaluate their quote in four distinct ways, at the user's discretion:

• As a straightforward web page (pictured above) with tabs and interactive elements.
• As an enhanced web page with a powerful 3D plug-in.
• As a printable PDF document with all the information comprehensively laid out.
• As an enhanced PDF document that exploits Adobe Reader's ability to view 3D models.

The report called attention to six "required changes" that would need to be addressed for the model to be injection molded.  Five of these were the result of a single error in my model, which caused an undesirable indentation in the walls.  Pictured above, Protoquote's interactive plug-in highlighted the error and explained exactly what action was required and why.

Additionally, the report explained the ramifications of certain attributes that might have cosmetic or functional ramifications on the finished part even though they wouldn't prevent the model from being built as-is.  For instance, the blue marking show where "sink marks" could appear in the finished model because injection molded plastic shrinks as it cools.

After I resolved the "required changes" Protomold supplied an updated report.  Now that the model was completely build-able the report added an informational section calling out foreseeable (but harmless) side effects of the tool-making process.  Pictured above is a sample of the "Printable PDF" version of the report, denoting where sharp corners would come out very slightly rounded as a machining artifact.

It was a valuable experience, and this particular model was probably an ideal choice for learning how to make an injection-molded design.  But, in my final analysis I concluded injection-molding isn't suitable for this particular project in the quantities I needed because:

• This particular design still required laser-cut parts for the mirror and top cover.  Injection molding merely reduced the volume of laser-cut parts per model (from 5 down to 2), it didn't eliminate laser cutting entirely.
• To attain the required transparency the molds would require laborious polishing to a mirror finish.  That polishing would cost more than all other expenses combined, which rendered the project hopelessly uneconomical for a small production run.
• Critical geometry includes a knife-edge "gore" where two arcs converge toward an intersection, a feature not suited for injection molding.  Hence injection molding would diminish the quality of that particular detail.
• The previous test model had already demonstrated the design was suitable for being laser cut.

Still, it would be nice to devise a design that would be optimal (and economical) for injection molding--preferably something I couldn't otherwise fabricate.  Protomold offers a wide selection of engineering materials, including elastomers.  Hmm.  I have an idea that could use Santoprene™...

3d Printer Brings New Light to Antique Japanese Lantern

Ben Sherwood experimented with the use of modern technologies to help repair this 150 year old Japanese lantern.  It was missing a curved wooden rib like the one in the picture, so we 3D printed a replacement on the SD300.

The 3D printed rib fit the size and curvature of the lantern perfectly, thanks to careful measurements that Ben took prior to building it.  But naturally the modern plastic looked a bit funky among all the wood and paper components of the lantern.

Ben Purdy, a maker colleague, took my CAD data and adapted it to a vector drawing so he could make a replacement rib in acrylic using a laser cutter.

It looked even funkier than the 3D printed rib, but the acrylic rib fit the lamp so neatly that it could have been used to make a perfect repair.

Ben turned up the intensity of the laser and cut another replacement rib from solid wood.

The laser charred the sides of the wooden rib, which makes it a nice match for the original wood in the Japanese lantern.

The laser didn't touch the top edge of the rib, so it's still pale.  But Ben can choose to apply stain and lacquer to make it a perfect match if he chooses.  Although the 3D printer didn't manufacture the final replacement rib, it was a helpful catalyst for completing the project.

It was certainly a fun, satisfying project.  Ben Sherwood shared it in a five-minute standup presentation at Ignite TAO PDX at the Alberta Rose Theater in Portland, which they recorded for posterity...
http://www.youtube.com/watch?feature=player_embedded&v=Tim3PSv5GrQ#t=305