Personal Manufacturing in the Digital Age
David A. Mellis
- “From my point of view, the greatest developments to be expected of technics in the future … will not be, as we are usually led to think, in the direction of universalizing even more strenuously the wasteful American system of mass production: no, on the contrary, it will consist in using machines on a human scale, directly under human control, to fulfill with more exquisite adaptation, with a higher refinement of skill, the human needs that are to be served…. Much that is now in the realm of automatism and mass production will come back under directly personal control, not by abandoning the machine, but by using it to better purpose, not by quantifying but by qualifying its further use.”
- —Lewis Mumford, Art and Technics (1952)
Digital technology is enabling new alternatives to industrial production. Computer-aided design (CAD) tools encode objects as information, allowing their designs to be freely shared online—the practice of open source hardware. Digital fabrication machines turn this information into objects, allowing for precise, one-off production of physical goods. A variety of sophisticated off-the-shelf electronic components enable complex sensing, actuation, communication, and interfaces. Together, these technologies enable individuals to produce complex devices from digital designs, a process we can think of as personal manufacturing.
Because open source hardware involves treating physical objects as digital information, it suggests that we may be able to apply principles and practices from other kinds of online collaboration to the design of hardware. Open source software, Wikipedia, and other digital artifacts incorporate the creativity of many different individuals working without the direction of markets or firms, a process known as peer production. It works because the means of production of digital goods—computers and software—are widely distributed, the Internet makes communication and coordination efficient, and the work can be divided into pieces that individuals can choose to work on based on their own interests, needs, and abilities. The extent to which peer production can apply to hardware will shape the extent to which this approach can provide a viable alternative to mass production for the technology in our lives.
To make electronic devices amenable to these peer production approaches, we need to design with them in mind. This process yields devices that look very different than ones that are industrially produced. Such devices are optimized for translation from the digital design to the physical object. They make use of a variety of processes, from the much-hyped 3D printing to the more prosaic (but potentially more useful) techniques of laser cutting, CNC milling, and circuit board fabrication. They allow for a variety of materials and aesthetics. They can be adapted by individuals for their own needs and interests. They allow for different business models, in which objects can be made on demand or in small quantities to serve specific markets or particular individuals.
Of course, none of this eliminates the need for individual skill, whether in the design process or in the use of the fabrication machines. Good CAD tools can make the process easier, but translating an idea into concrete form requires many decisions and compromises that rely on human skill, experience, and intuition. Similarly, making effective use of a fabrication machine relies on knowledge of its configuration, operation, limitations, and quirks. Technology offers possibilities, but people turn those possibilities into reality. Similar considerations exist in open source software, where peer production doesn’t eliminate the need for expertise on the part of contributors but rather provides new ways of organizing and combining those individuals’ skills and efforts.
The two case studies discussed in this chapter—dealing with Arduino boards and my own consumer electronic devices—illustrate different possibilities and limitations of working with these techniques. Together, they illustrate this new personal manufacturing ecosystem, highlighting its implications for product design, for collaboration, and for business. They show some of the ways that digital technology can transform the production of objects, but also indicate some of the constraints derived from industrial systems that persist in personal manufacturing. They provide some hints of what a peer production ecosystem for electronic devices might look like, yet also point out some of the difficulties to be overcome in creating one.
The next section gives an overview of personal fabrication and the considerations involved in going from an open source hardware design file to an actual physical object. This discussion is followed by the two case studies. The lessons from the case studies are used to derive some general principles for open source hardware and personal manufacturing. Finally, I conclude with some questions and thoughts for the future.
Personal Fabrication, Processes, Parts, and Materials
Digital fabrication machines translate open source hardware designs into actual physical objects. In theory, this process depends only on the digital file and the choice of fabrication machine, allowing for iteration and refinement through successive changes to the file. In practice, though, the constraints and intricacies of various fabrication processes mean that a certain amount of skill is required to use the machine and that the results can vary each time. As a result, open source hardware depends on the selection of appropriate processes and effective use of them. This section discusses some of the considerations involved in various popular fabrication processes.
The purest of these digital fabrication processes are the various forms of 3D printing. These turn digital design into physical objects by gradually adding material in the desired locations, allowing for a wide range of possible geometries. The term 3D printing encompasses a broad range of machines, from personal plastic printers costing a few hundred dollars to industrial machines that sinter metal and cost hundreds of thousands of dollars. Different machines work with different materials and offer different resolutions and tolerances. The materials may have different strengths, optical properties, appearances, finishing possibilities, and so on. Depending on the object being fabricated, some or all of these characteristics may be crucial to creating a useable result. In designing and sharing objects for 3D printing, therefore, it’s important to specify not just their geometries, but also the required tolerances, materials, and other characteristics—most of which are less easily captured in digital form. In addition, many 3D-printing processes need some form of manual post-processing, such as removal of support material, finishing, or curing. These require an operator with appropriate knowledge and skill—and can create variations from one print to the next, even with the same file and machine. Finally, 3D printing technology is evolving and diversifying rapidly. For all these reasons, it’s important not to think of 3D printing as a way to automatically create things from information, but rather as a material process with specific qualities and affordances.
Milling and Cutting
Other fabrication processes work by cutting or removing pieces of a larger stock material. Laser cutters cut 2D shapes out of plywood, cardboard, acrylic, and other flat materials. Vinyl cutters do the same, but with a knife that cuts through thin materials like paper or adhesive-backed vinyl. The water-jet cutter handles stronger and thicker materials like wood, metal, and glass, cutting with a stream of hard particles in a powerful jet of water. CNC (computer-numeric control) machines, like mills or routers, work in three (or more) dimensions, removing material from solid blocks of stock with a variety of cutting bits. They are often capable of very precise operations, albeit only within specific axes of movement. Compared with 3D printers, these cutting and milling tools have the advantage of being able to work with a variety of existing materials, including natural ones with complex structures that are difficult or impossible to replicate with the homogenous stock of most 3D printers. They are more limited in the geometries they can produce, however, and often require more steps in fabricating or assembling the parts.
In addition to specifying the geometry of the design itself, it’s important to be explicit about the nature of the stock material and the characteristics of the cutting process. Whether two parts press-fit tightly together, slip past each other, or don’t fit at all depends as much on the precise thickness of the stock (which can vary even across nominally equivalent materials) and the thickness of the cut as on the shape in the file. Some constructions may be infeasible to achieve given the tolerances of a particular machine. (Laser cutters may yield slightly different cut thicknesses on different sides of their working area; water-jet cutters can give rough, nonvertical edges, for example.) Traditional engineering drawings often capture the required tolerances for various surfaces and the material to be used. A quickly created CAD file used for a prototype and then thrown up on a webpage may not. Parts might be sanded, glued, pounded together, or otherwise tweaked in ways not reflected in the design files. Generating tool paths for a CNC machine is a complex process with a significant impact on the form and finish of the resulting object; this complexity may not be possible to capture in a way that can be easily shared with others, particularly if they are using a different machine. Finishing and assembling parts created with CNC devices requires careful craft, which might be difficult to communicate or learn. All of these factors need to be kept in mind when designing or sharing a digital file for someone else to replicate.
Other Fabrication Machines
A variety of other digital fabrication processes exist, each with its own affordances and constraints. For example, a host of machines are available for working with soft materials: CNC embroidery machines apply custom designs to fabric, knitting machines generate colors and constructions based on digital files, and Jacquard looms are possibly the oldest digital fabrication machines in existence. Industrial production uses a variety of automated machines, including robot arms and other adaptable parts of an assembly line. Furthermore, as digital fabrication becomes more established, more people are creating their own machines for custom purposes of various kinds.
Printed Circuit Boards and Electronics
The production of printed circuit boards (PCBs) can also be considered a digital fabrication process—and a relatively mature one. Digital designs are etched from copper or other materials using a photographic process, then covered with an isolating layer and text and other annotations. While the processes for creating circuit boards in this way are generally toxic and the automated systems for doing so are expensive, many services will produce PCBs on demand for individual customers with small or nonexistent minimums and standard specifications and tolerances. (As a board’s specifications get more demanding, however, costs can increase, sometimes dramatically.) Circuit boards can also be manually etched or milled on a CNC machine, processes that are more directly accessible to individuals but also less robust and precise. While some circuits are sensitive to the precise characteristics of circuit board’s substrate or the exact tolerances of the fabrication process, a great many can be shared with relative confidence that they will work when made on a different machine from a different provider.
In reproducing circuits, then, the main difficulties are typically getting the necessary parts and assembling them. While vast quantities of components are available to individuals—and many distributors specifically target hobbyists—advanced parts with specific functionality may not be accessible. These may be simply impossible to purchase, require an extended procurement process that makes replication infeasible, or be difficult or impossible to assemble with the processes available. As parts are optimized for size and automated assembly, they become harder for individuals to work with. Even easier-to-solder parts rely on manual skill and the knowledge to troubleshoot problems. Different electronic components may be available or preferred in different locations. Parts may go out of stock, become obsolete, or cease being made altogether. All of these factors mean that while making a PCB may be a robust and accessible process, much work must be done to ensure that individuals are able to replicate a complete electronic circuit for themselves. (It’s also worth noting that while the problem may be worse for electronic components, other materials—such as plywood or 3D printer stock—are also industrial products and may not be available everywhere or all the time.)
Access to Fabrication
Access to digital fabrication processes comes in a variety of forms. Some machines, particularly 3D printers and vinyl cutters, are being targeted at individual consumers via low-cost, Easy-to-use models. Local workshops, whether at schools, libraries, community centers, or commercial locations, provide access to larger, messier, and more expensive machines. They also offer opportunities for people to learn how to use the machines and can provide a community of like-minded individuals. Online services offer an alternative for those without local, hands-on access. They can provide a larger variety of processes and materials than those found in a single workshop and obviate the need to learn to operate the machines directly. On the downside, the time required for parts to be produced and shipped—and the lack of direct control over the process—can make it harder to iterate and refine designs when using an online service. Additionally, online services generally involve higher per-part prices than direct machine access, since they need to cover the cost of the machines, labor, and infrastructure required to support the service.