Imagine if you could have a machine in your home that, like the Star Trek replicator, conjures up a cup of coffee or a hairbrush on command. This sort of personal manufacturing might well be close at hand. Researchers at Cornell University’s Computational Synthesis Lab in Ithaca, N.Y., have come up with a “personal fabrication device” — about as big as a microwave oven — anyone can purchase for a few thousand dollars. Better yet, the device’s open-source firmware, application source code, and parts list are online for those who’d like to build or modify their own device.

Professor Hod Lipson and graduate student Evan Malone named the machine they developed a “fabber” after the “fab labs” (short for fabrication laboratory) that Neil Gershenfeld of the Massachusetts Institute of Technology recently established around the globe. He intended the centers to spark the creativity of ordinary folk by giving them free access to precision manufacturing tools not otherwise readily available, such as 3D CAD, laser cutters, and desktop milling machines. Gershenfeld also hoped to eventually create what he called a “personal fabricator” that would make objects on-demand from computer-generated design specifications. What the Cornell researchers call the Fab@Home project has brought this idea down to earth.

 

Under the hood of a personal fabricator

First, a brief description of the machine. A fabber has a three-axis gantry configuration. Its mostly off-the-shelf parts include shafts, limit switches, cabling, power supplies, sensors, linear motors, and the like. An acrylic housing covers components. On the basic model, these include a carriage, on which mounts a printer head with a replaceable syringe that acts as a deposition tool and builds part layers in the XY plane. Syringes reside in “smart” devices that communicate with the printer head to indicate the material they contain and when they are in position. A second carriage moves the build surface up and down.

In fabricating a part, the syringe tool deposits a thin, continuous bead of material, layer-by-layer onto the build surface. Materials can be almost any fluid or paste. Examples are ceramics such as gypsum plaster, metallic materials such as solder, thermoset polymers such as GE Silicone II, and electrically conductive composites such as SS-26 silver-filled silicone. Part resolution is determined by the needle or nozzle diameter.

The fabber connects to a Windows PC via USB. Ancillary software imports stereolithograpy (STL) files to generate toolpaths for the syringe nozzle to follow, dictates fabrication sequences, calculates deposition rates, and generates tool-exchange commands (some fabbers use more than one syringe tool). The software also provides an onscreen view of the printer that updates a part fabricates.

How to print a battery

One task was to develop and print a battery of an arbitrary shape so a robot, say, could have batteries as legs, says Lipson. Researchers developed a battery with zinc as the anode and air as the cathode because of the design’s simplicity.

First came development of a zinc-powder suspension that did not clog the syringe nozzle. Next came designing the separation layer between the anode and cathode. In commercial batteries, this layer is often paper. In contrast, the layer in the zinc-air battery is made from ceramic slurry or a synthetic resin.

“After a few trial runs, the fabber printed a battery about the size of a coin and with five layers of different material in a plastic case,” says Lipson. “In one test, the device delivered 10 milliwatts for 50 hours to a small dc motor. While not performing as well as a commercial unit, by around a factor of two, the battery frees users from the design constraints imposed by conventional devices.”

Admittedly, items the apparatus builds look crude compared to those made by commercial rapidprototyping (RP) and manufacturing machines from Object, Z-Corp., Stratasys, and other companies. But, unlike commercial equipment, fabbers can build active, functional objects such as batteries and electroactive actuators. Fabbers of the future might generate such consumer items as toys, cell phones, and electric toothbrushes.

Simulating evolution

The idea for the fabbers grew out of a need to quickly produce free-form designs, says Malone. “The Computational Lab focuses on artificial evolution — the simulation of evolution to solve challenging engineering problems, such as robot design,” he says. “The basic idea is to take a large number of candidate solutions to any problem that can be generated in a computer algorithm. Let’s say I want to make a robot walk forward, but don’t know how to control the movement. So I start by just giving the robot, say, 30 random commands. Then I determine which ones make the robot move more forward than the others. I combine these commands and keep iterating cycles in a similar fashion.”

Malone says the lab had been using RP for a few years as a way to make parts for robots. “Our evolutionary searches usually produced nonstandard shapes. For a while, we were outsourcing the 3D printing of unusual robot morphologies. Then, instead of having graduate students assemble every robot iteration with these parts, we decided to build a machine that could spit out complete robot after complete robot with minor and major variations.”

Malone first built a larger test machine to print circuitry, actutors, and batteries. “At the time, commercial machines only printed one material, and you had to use the manufacturer’s special blends,” he says. “We took lessons learned from the test machine and started working on a do-it-yourself version, the fabber. That the machine can print almost any material, as well as more than one material comes from the use of the disposable, standard syringes. And materials with a wide range of viscosities work because a linear motor driving the syringe piston controls the deposition pressure.”

As for the design side: Design software intended for 3D printing does not yet exist (even for commercial equipment), says Malone. So there is no way, for example, to define RP material properties in CAD. Instead, users assign properties to the STL geometry in the fabber software. That said, future design software would need to support a file format more sophisticated than STL — one that would let users specify multiple materials. Malone says developers are working on a format now.

A robot building a robot …

“One of our main goals is to have a robot walk or wiggle out of the machine, its electronic and mechanical parts having been generated seamlessly in one build, battery included. A robot building a robot, so to speak,” adds Lipson. “At this stage of the game, we have demonstrated the free-form fabrication of components such as electromechanical relays, polymer transistors, elastomer strain gages, complete zinc-air batteries, artificial muscle actuators, and inductors and electromagnets. Next comes designing a higher-level way to get larger sub-assemblies built.”

The big hope is that Fab@Home will inspire more people to fabricate integrated systems at home, says Lipson. “Currently, users can’t print something like an iPod. But these kinds of objects will appear as users continue to tweak the source code and further refine the machine’s design. Some have added heaters to melt powders while others have attached a hopper for powder use. One individual even included a feedback mechanism so the machine can self-correct when it is running.”

As for today, suppose someone has an idea for a new electric toothbrush. “Unless there is a way to make it at home, the individual would probably just let the idea go,” says Lipson. “So when fabber use is more widespread, anyone could create a design and post it on the Web as a 99-cent download. This scenario might shake up manufacturing similar to the way the music industry got turned on its head.”

An intermediate step

Fab@Home is useful because it lets a much broader range of users experiment with unusual designs and materials,” says Terry Wohlers of Wohlers Associates, a Ft. Collins, Colo., consultant specializing in the rapid product-development industry. “But one potential problem with personal manufacturing in general, and not just with additive fabrication, is that most people don’t know how to design well. An intermediate step currently has manufacturing involving consumers, but with limited options.” Take for example, NIKEid.com, says Wohlers. “Users can produce a pair of custom shoes for track, football, running, or whatever,” he says. “Customers can change shoe color, add a school mascot, and put their initials on the heel. And, there is no way to make a crappy pair of shoes by, say, making poor design changes to the sole.”

A lot of personalized, on-demand manufacturing is increasingly Web driven, says Wohlers. “For example, jujups.com lets users upload an image and select a picture frame. The company then uses a commercial 3D color printer to make the image and frame, charging around $29.”

Wohlers cites other examples including figureprints.com, where users can order 3D color prints of characters from the World of Warcraft video game, fabjectory.com, which builds avatars from Second Life, and zazzle.com, which provides online tools where users can add designs to objects such as T-shirts and coffee mugs.