What I'd make with a rapid prototyping/3-D printing machine

[Dave Long]

> Suppose I had a machine that could manufacture any three-dimensional
> shape, down to a few microns, but only out of some single rather
> boring material; what would I do with it?

Since I now have a machine that can manufacture any three-dimensional 
shape, but only down to many tenths of a mm, and out of a rather 
brittle plaster material (we are using it to help produce patterns, 
with the production material being silicone, so quick builds are more 
important than the material properties), I figured it would be good to 
revisit this post.

-Dave

Pretty easy
-----------

> Soma cube --- with faces that fit together.
>
> "Do-nothing" gift-shop toy.  (I don't know if this piece of junk is
> still available for sale or findable on the web.  The one my family
> had was carved from wood; it consists of a crank attached to two
> sliders at right angles.)

can you describe this a bit further?  I've done a geneva stop model, 
and a Peaucellier linkage (which seems to have impressed Kelvin a good 
deal more than any of us), so this device shouldn't be any trouble

> A projective plane, out of mesh.

Hilbert curves, klein bottles, menger sponges

> Pans for things like waxing printouts --- similar to a sumi stone.

although the build costs mean it's probably easier to build these out 
of sheet stock.  Custom cradles and jigs for printouts, on the other 
hand...

> Buttons for clothing.
>
> Parabolic reflectors for sound, light, and radio waves.
>
> Molds to mold clay into shape to cast e.g. metal.

There is also a refractory powder available for our machine, which we 
hope to try out Real Soon Now.

> Carrying cases for cell phones, business cards, earplugs, any number
> of personal items that need a little protection.

somewhat sacrificial protection, at least

> Things with text inscribed on them.  It may be possible to inscribe
> text fairly small --- 10-micron resolution allows 0.7 mm line height,
> so 360 lines of text per inch, or 0.2 points.  Large text allows
> signs.

large text, at least.  In general, this is a very good process for 
surfaces that need complex relief.  (printing out topo maps is a nice 
application -- even for Swiss topography, though, one must exaggerate 
the vertical scale by 4x or so)

Maybe possible
--------------

> Fluidic integrated circuits.  (Maybe.  Surface characteristics may
> make this hard.)

Judging from the Exploratorium's model, these ought to be sufficiently 
macroscopic to work.  I have printed flutes and whistles that work, 
which are the pneumatic equivalent of white noise generators and 
bandpass filters.  (the warbler in the whistle is embedded as printed 
... but the solid warbler in my current .STL are too heavy.   would a 
whiffle-ball warbler warble well, if whiffle-ball warblers wouldn't 
warble worse?)

> Fresnel reflectors.

given the surface accuracy, more suitable for concentrating than 
focusing

> Lego-like toys that fit together.
>
> Lenses, perhaps.  At least, molds to cast lenses of other materials.

At least, molds to get the rough form.  Maybe patterns of lenses, 
allowing one to polish the molds instead of the castings?

> Combs.
>
> Innocuous engines --- a very small "internal combustion" engine that
> runs very slowly on vinegar and baking soda.
>
> Rubik's Cubes.

the spider at the center of the cube might be difficult

> Eating vessels, such as bowls, plates, and cups.

probably best done as molds with fancy patterns

> Handheld mechanical calculators similar to the Curta.  Perhaps even a
> handheld programmable mechanical computer.  Slide rules, of course.

slide rules, anyway, with the advantage of engraved scales.  (although 
I am not sure this is much of an advantage over laser-printed scales)  
For curta-like devices, see the note below about mechanical logics.

How about helical slide rules?

Close But No Cigar
------------------

> Collapsible modular hard-shell suitcases that can be worn as backpacks
> and have a Kragen's-back-shaped side.
>
> Hard shoes.

With one of the more ABS-like machines, yes.  Probably more easily done 
by forming thermoplastics (vacuum helpful but not required).  Digital 
would be more helpful if you didn't have an easy reference for 
Kragen's-back and Kragen's-feet shapes readily available.

> False teeth.  It may be more economical to get new cheap false teeth
> each year than to use a permanent crown.

Osseointegrated false teeth are going in the other direction -- they 
last for much longer than dentures used to.  (dentists have some very 
fancy CAD/CAM systems -- but their build volumes are not so impressive)

Things that would need stronger/flexible materials or higher resolution
-----------------------------------------------------------------------

Pretty obvious.  Also, an unfortunate dichotomy due to resolution 
constraints is that one can print parts already assembled in situ, or 
one can print parts that will fit tightly together, but not both at the 
same time.  Are there mechanical logics that tolerate a great deal of 
slop?

> Tripods.
>
> Screwdrivers, wrenches, pry bars, pliers, putty knives, paint scrapers
> (?).  Drill bits?  Hand drills: both brace-and-bit and
> eggbeater-style.  Leatherman multitools.  Saws.  Shovels.
>
> Screws.  And tapped things.
>
> Merkle buckling-spring logic devices.
>
> Bill Beaty's scratch holograms might work, depending on the particular
> type of error in the surface.  Roughness or some kinds of systematic
> errors larger than a wavelength of light would cause problems.
>
> Mechanical clocks and watches.  Perhaps I could make a 28-hour-day
> mechanical watch.
>
> Knitting needles.  Crochet hooks.
>
> Knives, if the material was suitable.
>
> An air-driven centrifuge.
>
> Fans.  The kind with blades.
>
> Boats.  Floats.
>
> Chopsticks.
>
> Shelves.  Other furniture: chairs, (hard?) pillows, folding tables,
> stools.  Cabinets.
>
> Bottles.  Other food containers.
>
> Buildings that regulate their temperature by breathing.
>
> Jacks.  Tops.  Carts.  Other toys.
>
> Tent poles.
>
> Locks.  (And their keys, of course.)
>
> Anisotropic structured 3-D composite materials --- like honeycombs,
> but simpler and more nearly isotropic.  Closest-packing-of-spheres
> lattices might be stronger than foams of similar density.  This is
> most interesting if the strength (or stiffness, etc.) and cost of the
> resulting material is an improvement over already available materials,
> but it's probably a good technique for interior parts of a wide class
> of objects to be made in this fashion.  By deforming the tessellation
> or using a different crystal structure, it may be possible to produce
> materials with desired anisotropies; by doing this in a way that
> varies throughout the substance, it may be possible to make a material
> stronger in some places than in others.
>
>
> Resolution close to 1 micron would enable diffraction gratings and
> perhaps even (non-scratch) holograms.
>
> Airfoils.
>
> Pipes and, perhaps, pipettes.  You could make a line up one side thin;
> Very many materials can withstand water's surface tension when thin
> enough to be transparent.
>
> Along the same lines, a flat piece of material with many pinholes can
> let light through; if they are only near-pinholes, it won't let quite
> as much light through, but it will be airtight.
>
> Chains, perhaps with very small links, can make stiff materials
> flexible.  Thin fibers also do that, and they can be soft, but they
> require finer resolution.  A mesh of chains with overlapping scales on
> one side might work as a smooth, flexible surface that requires less
> precision than fibers do.
>
> There are several interesting variants of the chain --- circular links
> that can rotate to reduce friction as the chain slides lengthwise
> (which has the lowest inter-link pressure under tension when the two
> neighbors of each link must nearly touch inside it), chains of spheres
> or end-linked cylinders that can roll, links with sharp edges to
> maximize friction against sliding lengthwise.
>
> Such a chain with certain kinds of sharp edges can serve as a small,
> foldable saw or knife.
>
> If you can make fibers, you can make them pre-woven into a shape, such
> as cloth, or a three-dimensional solid wave.
>
>
> If the resolution is closer to 50 nm, perhaps you can make colors by
> means of interference, much as Iridigm's product does dynamically.  A
> flat surface reflects; two nearby flat surfaces reflect wavefronts
> that interfere at the edge; alternating frequently between two heights
> of surface will work to create mixed colors.  If a cross-section of
> the surface looks like a square wave, this will work reasonably well
> when the light is exactly perpendicular to the surface, but not
> otherwise.  If the "cliffs" between the two heights are not
> perpendicular, but rather overhang somewhat, it will work better at
> angles off the perpendicular, and if the material is transparent
> enough that light can travel through the sharp edges of the upper
> layer, it will work better still.  Such a surface would be fragile to
> the touch and very difficult to clean, however, and this may prevent
> it from being manufactured.  It might also be dangerous or harmful to
> touch, like fiberglass.
>
> Posable action-figure-like models with lockable joints, as a sort of
> construction kit that requires less assembly --- ideally you can lock
> all the joints at once.  Such a device could even be spring-powered
> and self-winding, driven from some resistance to being repositioned in
> its movable state, which resistance is necessary anyway to allow it to
> be posed in the presence of gravity.
>
> Generally, clockwork things that would be unacceptably expensive if
> assembled by hand.  Every rotating joint, wheel, and gear can run on
> ball or roller bearings to lengthen its life.  Every nut can be a
> ballnut (with ball bearings inside) if it's advantageous.  With some
> 3-D printing techniques, it's possible to make permanentlyhermetically
> sealed enclosures, which can be useful for extending part life in
> grotty environments.
>
> A reflective display like those used on highway signs, but mostly
> mechanical: a plunger to rotate a pixel, some column-select inputs to
> select which column of pixels, and some row-select inputs to select
> which row of pixels.  For a 1024x1024 display, you'd need at least 21
> inputs for a 1024x1024 display, if each input had only two positions,
> and each input could be connected to a solenoid.  So far I haven't
> hypothesized any kind of coloration generated by the machine (except
> in the far-off case that it can shape them to within nanometers) but I
> think I could probably paint the screen by hand when all of its pixels
> are in the "off" position, so that the "off" position was the color of
> the paint, and the other position was the color of the material.
>
> Likewise, a mechanical keyboard could contain a mechanical, hydraulic,
> or fluidic key encoder.
>
> Mechanical digital memory, which works exactly like the display above,
> but without the paint.  Each bit probably requires at least four
> resolution units on a side, e.g. 4 microns if we can fabricate with
> one-micron resolution.  They can presumably fill space rather than
> being limited to covering a surface as printed semiconductor memory
> is.  The cuberoot(2 GiB) is cuberoot(2) * cuberoot(B) * 1024, so
> space-filling memory technology becomes space-competitive around 1-10
> micron resolution, which enables us to make 4000-unit-on-an-edge cubes
> in 4mm-4cm.
>
> Tripods.
>
> Crutches, mostly.
>
> Rebar bending machines.
>
> Capsela.
>
> Springs are an important enabling technique for many of these things.
> You can fabricate something like a watch mainspring out of any
> material, and you can corrugate it lengthwise to enable it to be
> curled more tightly for a given effective thickness.  It may not have
> the energy density of the watch mainspring, but it will still be
> useful.
>
> Telescoping and folding items, such as tables, chairs, drinking cups,
> walking canes, backpack frames, sleeping cots.
>
> Hoberman expanding spheres.
>
> Portable scales for weighing things, such as myself.