Fab@Home: the personal desktop fabricator kit · as the “killer app”; (b, c, d) The general purpose Altair 8800, credited as starting the home computer revolution, came as a $400

Fab@Home: the personal desktop fabricator kitEvan Malone and Hod Lipson

Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York, USA

AbstractPurpose – Solid freeform fabrication (SFF) has the potential to revolutionize manufacturing, even to allow individuals to invent, customize, andmanufacture goods cost-effectively in their own homes. Commercial freeform fabrication systems – while successful in industrial settings – are costly,proprietary, and work with few, expensive, and proprietary materials, limiting the growth and advancement of the technology. The open-sourceFab@Home Project has been created to promote SFF technology by placing it in the hands of hobbyists, inventors, and artists in a form which is simple,cheap, and without restrictions on experimentation. This paper aims to examine this.Design/methodology/approach – A simple, low-cost, user modifiable freeform fabrication system has been designed, called the Fab@Home Model1, and the designs, documentation, software, and source code have been published on a user-editable “wiki” web site under the open-source BSDLicense. Six systems have been built, and three of them given away to interested users in return for feedback on the system and contributions to theweb site.Findings – The Fab@Home Model 1 can build objects comprising multiple materials, with sub-millimeter-scale features, and overall dimensions largerthan 20 cm. In its first six months of operation, the project has received more than 13 million web site hits, and media coverage by several internationalnews and technology magazines, web sites, and programs. Model 1s are being used in a university engineering course, a Model 1 will be included in anexhibit on the history of plastics at the Science Museum London, UK, and kits can now be purchased commercially.Research limitations/implications – The ease of construction and operation of the Model 1 has not been well tested. The materials cost forconstruction (US$2,300) has prevented some interested people from building systems of their own.Practical implications – The energetic public response to the Fab@Home project confirms the broad appeal of personal freeform fabricationtechnology. The diversity of interests and desired applications expressed by the public suggests that the open-source approach to accelerating theexpansion of SFF technology embodied in the Fab@Home project may well be successful.Originality/value – Fab@Home is unique in its goal of popularizing and advancing SFF technology for its own sake. The RepRap project in the UKpredates Fab@Home, but aims to build machines which can make most of their own parts. The two projects are complementary in many respects, andfruitful exchanges of ideas and designs between them are expected.

Keywords Rapid prototypes, Plastics industry, Manufacturing systems

Paper type Research paper

Introduction

Solid freeform fabrication (SFF) technology is capable ofmanufacturing objects with almost arbitrary geometry(Sequin, 2005) – making it in this limited sense a“universal manufacturing technology.” We believe that,manufacturing systems based on SFF technology can be

made capable of the manufacture of complete functionaldevices without compromising the aforementioned geometricfreedom. Such manufacturing systems, dubbed “compactfactories” or “fabbers,” may have the same potential totransform human civilization as another “universal”technology – the digital computer. The ability to directly

fabricate functional custom objects could transform the waywe design, make, deliver and consume products. Moreover,rapid prototyping technology has the potential to redefine thedesigner. By eliminating many of the barriers of resource andskill that currently prevent ordinary inventors from realizing

their own ideas, fabbers can “democratize innovation”

(Burns, 1995; Gershenfeld, 2005; Lipson, 2005).

Ubiquitous automated manufacturing can thus open the

door to a new class of independent designers, a marketplace

of printable blueprints, and a new economy of custom

products.Despite the formidable potential of rapid prototyping

technology, its acceptance over the last two decades has

remained disappointingly slow, with worldwide annual sales

still being measured only in thousands (Wohlers, 2006).

At present, SFF systems remain very expensive and complex,

focused on production of mechanical parts, and used

primarily by corporate engineers, designers, and architects

for prototyping and visualization. These factors are linked in a

vicious cycle which slows the development of the technology:

Niche applications imply a small demand for machines, while

small demand for machines keeps the machines costly and

complex, limiting them to niche applications. Alternatively, if

one could provide either a large market for SFF machines andThe current issue and full text archive of this journal is available at

www.emeraldinsight.com/1355-2546.htm

Rapid Prototyping Journal

13/4 (2007) 245–255

q Emerald Group Publishing Limited [ISSN 1355-2546]

[DOI 10.1108/13552540710776197]

The authors would like to thank to Daniel Periard and Jennifer Yao fortheir contributions to the project. This work was supported in parts by theNational Science Foundation, grant DMI 0547376.

Received: 1 January 2007Revised: 1 April 2007Accepted: 1 May 2007

245

www.emeraldinsight.com/1355-2546.htm

products, or a simple and cheap SFF machine with which

end-users could invent products and applications, then thissame feedback coupling could instead drive a rapid expansionin SFF technology and applications.

Learning from the history of the computerrevolution

In attempt to break the vicious cycle of expensive equipmentand niche applications, there are many lessons to be learnedfrom the rise and growth of an equally universal technology:

the digital computer. The parallels between universalcomputation technology and universal manufacturingtechnologies are astounding. Though the universal computer

in its modern architecture was realized in the 1940s (Ceruzzi,2003), two decades passed before it reached any significantcommercial acceptance. Early inventors themselves could not

foresee its huge potential, famously anticipating a need foronly a handful of machines (Ceruzzi, 2003). The earlycommercial mainframes of the 1960s were used mostly forniche applications such as payroll and military calculations.

Like today’s rapid prototyping machines, these earlymainframes cost tens and hundreds of thousands of dollars,required hours to complete a single job, had the size of a large

refrigerator and required trained technicians to operate andmaintain.

Though it was clear to early manufacturers that the home

market offered great potential, it was unclear how tosuccessfully capture that market. Early attempts of thecomputer industry to break into the home market through

niche “killer apps” failed miserably: Some brands targetedniche domains such as Honeywell’s “Kitchen Computer”geared towards recipes (Figure 1(a)). Its high cost and narrow

range of application prevented it from achieving success.

Though several other home computers came out in the early

1970s (Unattributed, 2007a, b), the MITS Altair 8800

(Figure 1(b), (c)) is sometimes credited as sparking the homecomputer revolution (Klein, 2007). Designed and sold

through Popular Electronics as a US$400 kit (equivalent to

approximately US$2,000 in 2005 dollars), the Altair 8800

broke the computer free of the “chicken-and-egg” paradox:hobbyists and experts could now afford to dabble with

computers, develop and exchange software and numerous

hardware accessory projects. The availability of computersmade it worthwhile to write software, and the availability of

software made it worthwhile to buy computers. Computer

history had entered its exponential growth era.Based on this history, it seems reasonable to imagine a low-

cost multi-material SFF system in one’s home which could

produce objects or even complete integrated devices from

designs which are shared or purchased online (Lipson, 2005).Should such systems become as available as personal

computers or printers are today, the invention and

personalization of small devices could become as ubiquitousas music sharing is today. MIT’s FabLab project

(Gershenfeld, 2005) provides ample evidence that providing

people with automated fabrication tools serves as an

innovation catalyst; ordinary folk with no formal technicalbackground quickly learn to exploit these tools to design and

realize new inventions. This suggests a large potential impact

for a user-modifiable rapid-prototyper kit.

The Fab@Home project

Inspired by this history, the success of the FabLab project,

and the approach employed by Bowyer (1999) toward thegoal of self-replicating manufacturing systems, we have

developed an open-source, low-cost, personal SFF system

kit, which we call the “Fab@Home Model 1” (Figure 2(a)).

The aim of this project is to put SFF technology into thehands of the maximum number of curious, inventive, and

entrepreneurial individuals, and to help them to drive the

expansion and advancement of the technology. In addition,

taking a page from more modern history, we have developed auser-editable “wiki” web site to publish the designs and

documentation, have set up online discussion forums for

users’ and contributors’ communications, and have shared thesource code for the project via a popular open-source

online project site. Through these media, participants in

Fab@Home have begun to exchange their ideas for

applications and their improvements to the hardware andsoftware with us and each other. We have built six complete

Fab@Home systems, three of which have been delivered to

users outside of Cornell University. The Fab@Home design,

preliminary conclusions about the project methodology, andthese first deployments are presented below.

Overall design

The Fab@Home Model 1 (Figure 2(a)) is a three-axisCartesian gantry positioning system driven by stepper motors

attached to lead screws. Material deposition tools are

modular, and the first tool we have designed is a syringe-based extrusion tool which uses a linear stepper motor to

control the syringe plunger position. The electronics of the

current version of the system provide for four axes of bipolar

Figure 1 Early home computers trying to break into home market (a)The Honeywell Kitchen Computer cost $7,000 and targeted the cookingas the “killer app”; (b, c, d) The general purpose Altair 8800, credited asstarting the home computer revolution, came as a $400 kit ((b)qPoptronix Inc.) ((c) and (d) from (Klein, 2007)

Fab@Home: the personal desktop fabricator kit

Evan Malone and Hod Lipson


Volume 13 · Number 4 · 2007 · 245–255

246

stepper motor control at 24 V, with two limit switches per axis

of positioning, plus an optional one limit switch for the

remaining three axes. The software and firmware support up

to six axes of control in their current incarnation.

A microcontroller controls the positioning of the axes and is

in bidirectional USB communications with the PC.

An application running on the PC displays the real-time

state of the machine numerically and graphically, and allows

the user to manually position the axes, import, assemble, and

perform basic modifications to STL geometry data, apply

specific material properties to each STL, and to generate and

execute tool paths in order to fabricate objects comprising

multiple materials.One concern in developing our design has been that there

probably exists a threshold of quality required in any new

technology kit for hobbyists, below which the excitement of

the new technology will be masked by the malfunctions,

maintenance problems, and poor aesthetics. Users must have

a sense of what the technology is capable of before they can

grasp how to modify it and apply it to their own purposes.

Our first design has therefore focused more on ease of use,

reliability, and aesthetics than on minimizing cost.We have tried to assume a modest availability of technical

tools and skills for the end-user, and have tried to facilitate

assembly by providing very detailed assembly documentation

(Figure 2(b)). The builder needs to have a laptop or PC with

a USB port, and basic assembly tools including Allen keys,

screwdrivers, scissors, pliers, and a soldering iron. Assembly

consists of snapping together the acrylic structure, inserting

nuts and screws and threaded inserts, bolting together the

positioning system components and mounting them to

the structure, making cables to connect the microcontroller

to the amplifier boards, and the motors to the amplifier

boards, mounting the electronic boards to the chassis, and

bundling and routing of cables. Soldering and crimping of

cables and connectors are the most challenging assembly

tasks, but the project documentation is exhaustive, offering

advice, images and reference web sites for these and almost

every other task. The user is expected to have some patience

as well: completely assembling a kit from parts to operation

requires roughly 18 h of labor. The parts cost US$2,300 in

2007 dollars, including the cost of having acrylic parts

laser-cut, and not including shipping costs. Remarkably, the

Altair 8800 cost adjusted to 2005 dollars would be $2,015.

Figure 3 summarizes the hardware parts and costs for the

Model 1 kit.

Structure

The structural components of our system are built from laser-

cut acrylic sheet parts held together with snap-fit joinery and

simple “T-nut” style screw/nut fastening, in which a square or

hex nut is inserted into a slot in one part, and a screw is

threaded in through a hole in a perpendicular part. The tight

manufacturing tolerances achievable with laser cut acrylic

enable us to produce good orthogonality and alignment in the

machine base, increasing the ease of setup and reliability of

the machine. Our first three units have been produced

in-house with an Epilog Helix 35W laser engraver (Epilog,

Inc.), cutting parts directly from SolidWorks (SolidWorks,

Inc.) drawing files, which are in turn generated from a

SolidWorks 3D CAD model of the complete system.There are 33 acrylic sheet parts in the chassis, plus an

additional seven acrylic parts for the syringe tool. Currently,

five sheets of 18 £ 24 £ 0.23600 cast acrylic are used to

produce the acrylic parts. Cutting the parts for an entire

machine requires 3 h on our 35 W laser cutter. We have

arranged contract manufacturing service (Koba Industries,

Albuquerque, NM, USA) for these acrylic parts to simplify

the purchasing process for end-users, and costs for this service

will decrease as order volumes increase. An additional benefit

to having the structural parts laser-cut is that there is a large

installed base of laser cutters/engravers in the sign making,

and trophy and gift engraving industries. The custom nature

of the work in these industries should make them amenable to

kit builder’s approaching them to have parts made.

In addition, most laser-cutting equipment used in these

industries does not require specialized CAM software – they

operate from an application printer driver, so almost any

image editing or vector drawing program can be used to

design or modify designs for laser cutting and engraving.

Thus, modifying the structure of a Fab@Home system does

not require an investment in 3D CAD software to make or

publish new hardware designs, and the bitmap engraving

Figure 2 The Fab@Home Model 1 design (a) 3D CAD model of an assembled Model 1; (b) an example of assembly instructions available via the projectweb site (www.fabathome.org)

18.5” (47 cm)

16” (41 cm)

18” (46 cm)

BaseAssembly Step 8 These three 1/2" holes

should line up withthose below

These three 1/2" holesshould line up withthose above

2

1

ITEM NO. PART NUMBER12

QTY.11

Base-Step 7Base-Step 3




Volume 13 · Number 4 · 2007 · 245–255

247

possible with these laser cutters allows them to customize the

appearance of their machine with images and text.Other approaches could also be used to produce the

structural components for the kit. Laser cutting works

particularly well for acrylic, but acrylic, while inexpensive,

attractive, and readily available, is also brittle and susceptible

to chemical attack. Water jet cutting is not well suited to

cutting acrylic with the very fine details required, due to the

material brittleness, but would be quite suitable for tougher

and more chemically resistant alternative materials, such as

polycarbonate or metals.

Positioning

The linear motion components of the positioning system use

off-the-shelf linear ball-bearing pillow blocks running on 1/200

diameter rails (McMaster-Carr, Inc.). The X- and Y-axes are

in a gantry configuration with the deposition tool riding on

the Y axis, which in turn rides on the X. The Z-axis moves the

build surface independently from the X and Y to minimize

acceleration of parts as they are being fabricated. We have

selected HSI Inc. linear stepper motors for our actuators

because of their simple design, high resolution, and the semi-

custom manufacturing focus of the company which permits

specifying precisely the leadscrew and bearing journal

dimensions required for our application, simplifying the

overall design and assembly. For the X-, Y-, and Z-axes we use

NEMA size 14 bipolar motors with rotor-mounted lead

screws. External polymer lead nuts are mounted to the axes

carriages. In the case of the X-axis, a timing belt and pulleys

(Stock Drive Products, Inc.) are used to couple a slave

leadscrew to the motor leadscrew to achieve symmetrical drive

of the gantry. The force, maximum speed, and positioning

resolution all depend upon the lead screw threading selected

– in this case 15.8mm travel per full step, a nominal top speed

of 25 mm/s (1,600 step/s), and a maximum thrust of 120 N.

Material deposition tool

We have selected a syringe deposition tool for inclusion in the

standard Model 1 design because of the broad range of

materials useable with such tools, and for the intuitiveness of

operation. The syringe tool structure (Figure 4(a)) is also

constructed of laser cut acrylic parts with snap fit joinery and

T-nut fasteners. A linear stepper motor controls the position

of the syringe piston. We employ a NEMA size 8 frame motor

with a rotor mounted lead nut. The lead screw, which is not

captive in the motor, has 3.2mm travel per full step, and the

motor can achieve a top speed of 5.8 mm/s (1,800 step/s), and

a maximum thrust of 90 N. For the 10 cm3 syringes we use,

this amounts to a 1.1 cm3/s maximum volume flow rate, and a

maximum syringe pressure of 460 kPa (67 psi). The current

Figure 4 (a) The standard design, single syringe tool, driven by a linearstepper motor; (b) a two-syringe version for a life-sciences laboratory

Figure 3 Breakdown of Model 1 parts list by cost, type, and number of pieces

Con

trol

Sof

twar

e

Har

dwar

e

Ele

ctro

nics

Acr

ylic

Mot

ors

Tra

nsm

issi

on

Tot

al

# of Types

Total # of Items

Materials Cost

$0.00 $158.73$367.78 $493.00 $610.59 $662.52

$2,292.62

3360

24 40 4 53484

3 23 15 27 4 18 90

Fab@Home Model 1 Parts/Cost Breakdown

7%16% 22% 27% 29%

0%




Volume 13 · Number 4 · 2007 · 245–255

248

syringe tool has been designed to allow 10 cm3 disposable

syringe barrels (EFD, Inc.) to snap in and out, and for the

piston to be quickly attached and released from the motor

leadscrew for quick changing of materials. A metal nut fits

tightly inside disposable syringe pistons (EFD, Inc.), and one

end of the motor leadscrew has threading to match the nut.

When firmly threaded into the nut, the leadscrew is prevented

from rotating with the motor rotor, and hence the rotor

motion is converted to linear motion. Manually unscrewing

the leadscrew from the nut allows exchanging syringes,

regardless of how full, without the need to move or remove

the piston. This facilitates the fabrication of multiple-material

objects, and conserves materials.We have also developed a dual syringe tool (Figure 4(b))

which allows two materials to be loaded simultaneously and

independently deposited. As mentioned before, tools are

bolted to the positioning system, and are modular.

Electronics

Personal computers today typically provide several USB

connections, but no longer have RS-232 serial ports or

parallel ports, though these are still heavily used by robotics

and microcontroller hobbyists. As a result, we have opted to

support direct USB connection to our Fab@Home Model 1

system, despite the additional development work and

(internal) complexity that this entails. We chose to use a

microcontroller with an on-chip USB2.0 peripheral, the NXP

LPC-2148 ARM7TDMI (NXP Semiconductors). This is a

very high performance, 60 MHz, flash-memory

microcontroller with a wealth of peripheral functions for

future expansion, including ADC, DAC, PWM, counter/

timers, real-time clock, high-speed GPIO, UARTs, SPI, and

I2C, not to mention the USB2.0 peripheral. In addition, it has

512 kB of flash memory, and 40 kB of RAM. The large

program memory has enabled us to make a very easily

understood and extensible packet data protocol for

communication between the PC application and the

firmware. We use the large RAM space to buffer motion

commands so that real-time motion does not depend on

variations in communication bandwidth. With our current

protocol, we can buffer roughly 670 six-dimensional

path points (for up to six axes of control). The

microcontroller is powered by the USB, and thus can be

communicated with even when the amplifier electronics are

not powered. The high computational performance of the

device enables the system to handle receiving and buffering

path points, sending real-time status and position data, and

controlling step and direction outputs for six axes at 5 kHz.

The microcontroller is available on a 1.5 £ 2.500

(38.1 £ 63.5 mm) board (Figure 5(a)) with header

connectors for all pins and a USB connector for less than

US$50 in single quantity (LPC-H2148, Olimex, Inc.).Currently, we are using a four-axis stepper motor amplifier

board (Figure 5(c)) Xylotex, Inc.) to power the positioning

system and syringe tool stepper motors. This board provides

switch-mode current regulation for four bipolar stepper

motors per board. The current regulation allows us to use a

30 W laptop-style 24VDC power supply and 5 V rated motors

to get much higher acceleration (hence faster builds at finer

resolutions), than would be possible at the nominal 5 V. This

is a design decision which increases cost significantly (relative

to using unipolar stepper motors) for the sake of making the

technology more useable.

Software

The firmware for the LPC-2148 microcontroller was

developed in C language, using the free GNU C ARM

compiler, and Rowley CrossWorks for ARM integrated

development environment (IDE) (Rowley Co. UK).

CrossWorks is not essential – several freeware IDE’s exist

which work with GNU C ARM. The firmware performs the

following main functions:. receiving and parsing of packetized commands from the

PC via the USB;. buffering of motion path segments for fabrication paths;. immediate execution of jog motion and emergency stop

commands;. configuration of limit switches (present/absent for each

axis and direction);. communicating axes positions, limit switch states, and

other system status to the PC via the USB; and. controlling step and direction outputs for up to six axes

at . 5kHz step frequency.

As mentioned before, the microcontroller has additional

resources available for future expansion, and the firmware has

been designed with ease of expansion in mind.A PC application has been written which enables the user

to control the machine, import, position, assign material

properties to, and generate and execute manufacturing plans

for geometry data which is imported in the form of STL files.

In addition, the application has been designed around the

concept that while the core application may eventually be

improved by the user community, in the short-term, the

hardware configuration, the types of materials, and

parameters for depositing materials would be far simpler

and more interesting to explore and share. Thus, the

hardware configuration, material properties, and material

deposition parameters are described in plain text parameter

files. These files describe, for instance, the color and geometry

data used to render the Fab@Home machine, the speeds

and threading of the stepper motors, presence of limit

switches, etc.The PC application is currently targeted only for the

Microsoft Windows (Microsoft, Inc.) operating system. It is

written in Cþþ using the Microsoft Visual Studio.NET

(Microsoft, Inc.) development environment, OpenGL (SGI

Inc.) for graphics rendering, and the Microsoft Foundation

Class (Microsoft, Inc.) library for user interface components.

The application has been designed with the aim of

maximizing the intuitiveness of use. The user interface

(Figure 6) includes a 3D rendering of a Fab@Home machine

which moves synchronously with the real-time position

information sent back by the microcontroller. Dialog boxes

allow importing and assigning material and tool properties to

the part geometry, manual jogging of the axes via buttons and

mouse scroll wheel, including the syringe tool motor, as well

as a numerical view of the real-time position and status data

from the microcontroller. The application also allows a

rudimentary simulation of the fabrication process – the actual

manufacturing plan is executed on a Fab@Home software

emulator, and the motions are displayed in the GUI for quick

checking of toolpaths.




Volume 13 · Number 4 · 2007 · 245–255

249

The workflow for Fab@Home consists of the following:. connecting PC to Model 1 via USB cable, and plugging in

the Model 1’s power supply;. selecting and/or modifying/tuning parameter files to match

the hardware and materials to be used;. starting the PC application;. loading the parameter files;

. loading a syringe with piston, material, and nozzle, and

mounting it in the tool; threading the tool leadscrew into

the piston nut;. importing the geometry of the part to be fabricated;. assigning material and tool properties to part geometry;. automatic generation of a manufacturing plan;. if desired, simulated execution of the plan;. establishing communications with the Model 1;

Figure 6 A screenshot from the PC application displaying a model ready for fabrication and dialog boxes for positioning and real-time statusinformation

Figure 5 The electronics boards of the Model 1: (a) LPC-H2148 microcontroller board at bottom left; (b) DB25 to screw terminal breakout board at topcenter; (c) four-axis stepper motor amplifier on the far right




Volume 13 · Number 4 · 2007 · 245–255

250

. homing of the axes so that GUI and physical positions

match;. jogging axes to the desired origin for fabrication; and. automatic execution of the manufacturing plan.

Building a user community

Ultimately, the success of the Fab@Home project will depend

upon social factors. We cannot stop with the engineering and

publishing of the hardware and software, but rather mustattempt to engineer a self-sustaining user community that will

continue to improve the machines, exploring alternative

implementations, and expanding the range of materials andapplications. To facilitate the growth of this community, we

have established fabathome.org – a wiki-style portal that

allows interested users to access, post and directly shareinformation about the project (Figure 7).

The web site contains detailed instructions for constructingthe Fab@Home machine, material recipes and applications,

and provides user manuals, CAD files, and downloadable

compiled software. To facilitate communications betweencommunity members, we have established an online user

forum (Contributors, 2007) using the free Google Groups

service (Google Inc.), which provides an excellent means forcollaboration and communication with very little

management burden. The Fab@Home application source

code is being distributed and collaboratively developed via aproject page (Malone, 2007) on the free SourceForge

(OSTG, Inc.) open-source software development web site,

which is the de facto standard tool for open source software

collaborations. Making use of services which are external tothe Fab@Home web site may somewhat complicate and

dilute the “online identity” of the Fab@Home project relative

to hosting all of these services directly within the web site,but the project management overhead is drastically reduced,

and the quality of tools and services provided to the user

community is superior.

Results

To date, we have produced six complete Fab@Home Model 1

units, three of which have been delivered to users outside of

Cornell University, while two are used for ongoing testing and

development. The sixth unit is destined for the Science

Museum, London, UK, at their request, for inclusion in an

exhibit on the history of plastics (Unattributed, 2007a, b).

The Fab@Home web site regularly experiences a large

amount of traffic from around the world, and Model 1

machines are beginning to be constructed by individuals and

groups outside of Cornell (Malone et al., 2007a, b). Complete

Fab@Home Model 1 kits can now be purchased online from a

vendor (Ball, 2007), and the project has received a substantial

amount of media attention.The very first Model 1 (Figure 8(a)) was delivered to the

AMTS FabLab in Pretoria, South Africa, on June 25, 2006

(LeRoux, 2006; Mandavilli, 2006). The AMTS FabLab is a

publicly accessible CAD/CAM lab developed by the MIT

Center for Bits and Atoms in partnership with the South

African Government’s Department of Science and

Technology’s Advanced Manufacturing Technology Strategy

and the South African Council for Scientific and Industrial

Research (CSIR). FabLabs, the invention of Professor Neil

Gershenfeld of MIT, are intended to provide simple, highly

automated CAD/CAM facilities and basic training to

members of the general public in order to promote local

invention and micro-enterprise – “Democratization of

Invention.” During the Digital Manufacturing Symposium

held at CSIR during the week of June 25, 2006, we were able

to demonstrate Fab@Home to officials of the CSIR and

AMTS as well as to visitors and staff of the FabLab.

The AMTS has committed resources to integrating

Fab@Home into the CAM suite at the FabLab, and is

using Fab@Home as a prototype for generating additional

examples of kit CAM equipment.

Figure 7 The Fab@Home user-editable wiki web site which provides free explanation, designs, instructions and documentation for the Model 1 fabber,and a central repository for user-contributed ideas, applications, and technology innovations




Volume 13 · Number 4 · 2007 · 245–255

251

At the University of Washington, students of Professor Mark

Ganter have constructed the first Model 1 to be assembled

outside of Cornell University (Figure 8(b)). Professor Ganter

is employing Fab@Home machines in a course on Computer-

Aided Technology (Ganter, 2007).The second Model 1 unit produced by us (Figure 9(a)) was

donated to the life science research laboratory of Professor

Stanislas Leibler at Rockefeller University, New York, NY.

At this laboratory, research is underway which involves the

slime mold organism Dictostylium, which transitions from

single-cell independent life to colony organism with

coordinated motion in response to scarcity of nutrition. The

Fab@Home offers a uniquely simple and customizable

platform for executing experiments on the effect of initial

three-dimensional spatial distribution of Dictostylium cells in

the environment and the spatiotemporal evolution of their

colony aggregation response to starvation stress. A dual

syringe tool was designed and included with this unit to

facilitate the spatial control of both organisms and nutrients in

experimental geometries.Three additional units have been assembled by us. Two

have been retained for use in ongoing debugging and

development of the software and hardware. The third was

loaned to Noy Schaal, a freshman in Manual High

School, Louisville, KY. Noy has collaborated with us in

building a low-temperature heated syringe which is suitable

for depositing chocolate (Figure 9(b), inset). This

modification to the basic Model 1 one-syringe tool offers

the ability to deposit other low-melting point materials as well,

including wax and bismuth metal alloys. Noy has employing

this modified Model 1 for high-school science fair

experiments (Schaal, 2007) involving printing of structures

using chocolate (Figure 9(b)).As shown in Figure 10, a variety of materials can be used

with the Model 1, and thin-walled, multilayer (more than 300

layers, ,10 cm tall), and multi-material parts are achievable,

as are compositions of fabbed with conventionally produced

parts. Part surface quality can be improved by more careful

tuning of the deposition and machine control parameters.The growth of Fab@Home has been accelerated

enormously by media coverage, and we can correlate surges

in web traffic to the Fab@Home web site major media

coverage (Figure 11). An enormous initial wave of interest

was generated when The New Scientist published a brief

electronic article (Simonite, 2007) about Fab@Home which

then led to the popular technology “blog” Slashdot picking up

the story (Dawson, 2007). Since, then, Fab@Home has been

mentioned or covered by: Newsweek Russia (Rotkin, 2007),

Newsweek International (Falby, 2007), The Guardian (Pollitt,

2007), Popular Mechanics (Hutchinson, 2007), and The NewYork Times (Wayner, 2007). Also, Discovery Channel

Canada’s DailyPlanet science/technology television news

Figure 8 (a) The very first Model 1 being demonstrated for staff and users of the Soshanguve FabLab public access CAD/CAM laboratory, SoshanguveTownship, South Africa; (b) the first Model 1 built outside of Cornell University, built by students of Professor Mark Ganter at the University ofWashington for use in a course on Computer-Aided Technology

(a) (b)

Figure 9 (a) A Model 1 unit with a two-syringe tool deployed for Dictostylium Discoideum slime mold research at Rockefeller University, NYC, NY,(inset) a structured alginate hydrogel produced with the Model 1; (b) high-school student Noy Schaal with a personalized, fabbed chocolate bar, and(inset) modified syringe tool for depositing heated materials




Volume 13 · Number 4 · 2007 · 245–255

252

Figure 10 Objects built with a Fab@Home Model 1: (a) a personalized chocolate bar built with a modified Model 1 by Noy Schaal; (b) a mold for amodel airplane propeller fabbed using one-part RTV silicone rubber, and a propeller cast with epoxy from the mold; (c) a watch made by fabbing asilicone watchband and inserting a conventional watch body during the process; (d) a replica of a model car tire fabbed of black silicone rubber

(a) (b)

(c) (d)

Figure 11 Fab@Home web site traffic response to media coverage

Fab@Home Wiki Server Traffic

0

200,000

400,000

600,000

800,000

1,000,000

1,200,000

1,400,000

11/1

2/20

06

11/2

6/20

06

12/1

0/20

06

12/2

4/20

06

1/7/

2007

1/21

/200

7

2/4/

2007

2/18

/200

7

3/4/

2007

3/18

/200

7

4/1/

2007

4/15

/200

7

4/29

/200

7

5/13

/200

7

5/27

/200

7

Week Of

Num

ber

of P

age

Req

uest

s

New Scientist,Slashdot.org

Popular ScienceDaily Planet(Discovery Channel Canada) New York Times




Volume 13 · Number 4 · 2007 · 245–255

253

program has aired a segment on Fab@Home (Belanger,

2007).Initially, the authors and project members at Cornell

University were the only contributors and visitors to the

web site. However, in the first six months after October 2006

(when the web site was first,made publicly accessible), the

web site has had more than 4.3 million requests for pagesfrom more than 150,000 distinct hosts, transferred more than

1TB of data, and averages roughly 18,000 page requests per

day (Malone et al., 2007a, b). Users have begun to make

contributions to the Fab@Home wiki, the Google Group, andthe SourceForge project in the form of new deposition process

ideas, bug reports, questions, feature requests, alternative

vendors, group purchasing arrangements, and more.

Conclusions

Few technologies can truly transform human society to the

degree that computers have. The universal digital computer,from the early experimental machines of the 1950s to the

home computers and embedded devices of today, has affected

almost every aspect of human existence. It is not easy to

foresee such disruptive technologies in advance, but it has notescaped many that SFF technologies bear many of the same

traits. Like universal computation, universal manufacturing,

embodied as compact factories or “fabbers” based on SFF

technology, has the potential to transform human society to adegree that few creations ever have.

In order to accelerate the spread and development of SFF

applications and technology, and to escape the “chicken andegg” paradox, we have developed an open-source, low-cost,

personal SFF system kit, which we call the Fab@Home

Model 1. The current kit design has a parts cost of roughly

$2,300, requires only basic hobbyist tools and skills toassembly and use, and can be used to deposit almost any

room temperature liquid or paste, and to build objects

comprising multiple materials. We have employed a user-

editable “wiki” web site as well as free discussion forum andfree open-source software development services to foster the

growth of a user community and to promote the exchange of

ideas and improvements to the technology. We have thus far

built six complete Fab@Home Model 1 systems, three ofwhich have been delivered to users outside of Cornell

University. Several more Model 1 systems have been built or

are under construction by individuals inside and outside the

USA, including units being used at the University ofWashington in an engineering course. Generating an initial

wave of interest does not seem to be a concern – what

remains to be seen is whether we have provided a system

design cheap enough and simple enough, and a social/technical network complete enough to support a chain

reaction of innovation, application, collaboration, and

commercialization similar to the one that allowed the digital

computer to infiltrate and revolutionize nearly every aspect ofcivilization today.

References

Ball, C. (2007), “Koba industries: Koba store Fab@Home”,

available at: www.kobask8.com/servlet/Categories?

category ¼ Fab%40Home (accessed April 26)Belanger, L. (2007), “Fab@Home”, Daily Planet,

J. Morrison, Toronto, Discovery Channel.

Bowyer, A. (1999), “RepRap”, available at: www.reprap.org

(accessed January 21, 2007)Burns, M. (1995), “The freedom to create”, TechnologyManagement, Vol. 1 No. 4, pp. 157-63.

Ceruzzi, P.E. (2003), A History of Modern Computing, MIT

Press, Cambridge, MA.Contributors (2007), “Google groups: Fab@Home forums”,

available at: http://groups.google.com/group/fabathome-

forums (accessed April 26).Dawson, K. (2007), “A 3D printer on every desktop?”,

available at: http://hardware.slashdot.org/article.pl?sid ¼

07/01/09/2239206 (accessed January 21).Falby, P. (2007), “Computers: factory in a kit”, Newsweek,

International Edition, February 5.Ganter, M. (2007), “ME 480 introduction to computer-aided

technology”, available at: www.washington.edu/students/

crscat/meche.html#me480 (accessed April 26)Gershenfeld, N. (2005), Fab: The Coming Revolution on YourDesktop – from Personal Computers to Personal Fabrication,

Basic Books, New York, NY.Hutchinson, A. (2007), “The printed world”, PopularMechanics, Vol. 184, p. 18.

Klein, E. (2007), “MITS Altair 8800”, available at: www.vintage-computer.com/altair8800.shtml (accessed January 21)

LeRoux, H. (2006), “Fabulous fabrication: South Africa sets

aside R15m for digital-fabrication initiative”, CreamerMedia’s Engineering News, Vol. 26 No. 27, p. 16.

Lipson, H. (2005), “Homemade [fabrication technology]”,

IEEE Spectrum, Vol. 42 No. 5, p. 24.Malone, E. (2007), “SourceForge.net: Fab@Home”,

available at: http://sourceforge.net/projects/fabathome/

(accessed April 26).Malone, E. and Periard, D. et al. (2007), “Web statistics for

Fab@Home”, available at: www.fabathome.org/wiki/

FAHServerStats/index.html (accessed April 26)Malone, E. and Vitorassi, L. et al. (2007), “Who is actually

building a Fab@Home Model 1?”, available at: http://groups.

google.com/group/fabathome-forums/browse_thread/thre

ad/2e5929417183d30a (accessed April 26).Mandavilli, A. (2006), “Make anything, anywhere”, Nature,

Vol. 42 No. 7105, p. 862.Pollitt, M. (2007), “The ‘fab’ machine that could spark an

industrial revolution”, The Guardian, March 29,Technology Pages, p. 6.

Rotkin, A. (2007), “

”,NewsweekRussia,Vol. 4.Schaal, N. (2007), Printing Chocolate, Du Pont Manual High

School, Louisville, pp. 1-34.Sequin, C.H. (2005), “Rapid prototyping: a 3D visualization

tool takes on sculpture and mathematical forms”,Communications of the ACM, Vol. 48 No. 6, p. 66.

Simonite, T. (2007), “Desktop fabricator may kick-

start home revolution”, New Scientist, available at:

www.newscientisttech.com/channel/tech/dn10922-desktop-

fabricator-may-kickstart-home-revolution.html (accessed

January 21)Unattributed (2007a), “Plasticity – 100 years of making

plastics”, available at: www.sciencemuseum.org.uk/

about_us/press_and_media/press_releases/2007/04/468.

aspx?keywords ¼ plastics (accessed April 26)Unattributed (2007b), “Pop quiz: what was the first personal

computer?”, available at: www.blinkenlights.com/pc.shtml

(accessed January 21)




Volume 13 · Number 4 · 2007 · 245–255

254

Wayner, P. (2007), “Beaming Up 3-D Objects on a Budget”,The New York Times, p. C7, April 5.

Wohlers, T. (2006), Wohlers Report 2006, Wohlers Associates,Fort Collins, CO, p. 250.

About the authors

Evan Malone, MEng, MS, is a Doctoral Candidate inMechanical Engineering in the Computational SynthesisLaboratory at Cornell University. He received a BA degree inphysics from the University of Pennsylvania after which heworked for two years as an applied physicist at Fermilab inBatavia, Illinois as part of a proton synchrotron conceptualdesign team. Upon completion of that project, he begangraduate studies at Cornell University, earning a Masters ofEngineering degree in mechanical engineering and systemsengineering for work on Cornell’s World Champion RoboCupautonomous robotic soccer project. His doctoral researchinvolves developing systems, materials, and methods for all-

additive fabrication of complete electromechanical devices.

Evan Malone is the corresponding author and can be

contacted at: [email protected]

Hod Lipson is an Assistant Professor of Mechanical and

Aerospace Engineering and Computing and Information

Science at Cornell University in Ithaca, NY. He directs the

Computational Synthesis group, which focuses on novel ways

for automatic design, fabrication and adaptation of virtual and

physical machines. He has led work in areas such as

evolutionary robotics, multi-material functional rapid

prototyping, machine self-replication and programmable

self-assembly. Lipson received his PhD from the Technion –

Israel Institute of Technology in 1998, and continued to a

postdoc at Brandeis University and MIT. His research focuses

primarily on biologically-inspired approaches, as they bring

new ideas to engineering and new engineering insights into

biology. For more information visit: www.mae.cornell.edu/

lipson




Volume 13 · Number 4 · 2007 · 245–255

255

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints

Documents

Fab@Home: the personal desktop fabricator kit · as the “killer app”; (b, c, d) The general purpose Altair 8800, credited as starting the home computer revolution, came as a $400