Radiant Fabrication

Making 3D printing accessible.

[radiantfabrication-blog]

Dovetail joints printed on the same Flashforge Creator 1 printer. The original STL object is on the left and forms the base. The other is the 0.5mm voxel STL exported from Radiant Li.

[radiantfabrication-blog]

An introduction to printing with Radiant Li.

[radiantfabrication-blog]

We've just released a beta-version of our 3D modeling software for 3d printing and 3d scanning. Try it out!

[radiantfabrication-blog]

Meet the new hotness: All-in-one 3D printer-and-scanners

A brief report on the rise of the 3D printer/scanner.

[radiantfabrication-blog]

Changes afoot!

We're transitioning to a new website layout over at http://radiantfabrication.com; let us know what you think.

Even more exciting, we're releasing a beta version of Radiant Li later this week! We're looking for your feedback, so don't be shy in sharing.

[radiantfabrication-blog]

An introduction to 3D scanning with the Lionhead and Li.

[radiantfabrication-blog]

An introduction to Radiant Li.

[radiantfabrication-blog]

Testing at Sector67

Are you in the Madison, WI area and want to see Li and a Lionhead Bunny in action? We’re holding another testing session at Sector67 on Monday, August 19 from 5-7pm. We’ve already filled the initial 10 slots for youth but are only limited by the number of computers available, so if you want to bring your computer with we can give you a copy of Li to test on your own machine.

Read more and register at the event site.

[radiantfabrication-blog]

Changes Afoot

As work continues refining our hardware and software, we've also been planning for a new website. And while development is still ongoing, we wanted to share our new logo:

We're really happy with how it turned out; expect to see further changes on the site in the near future!

[radiantfabrication-blog]

Reproducing Prints

Rabbits are complex symbols with a reputation for cleverness in many cultures. We've been working to make the Lionhead 3D printer smart enough to handle routine tasks to make printing smoother, but thought it should do more. Our goal to make an accessible printer led to another trick:

The Lionhead has an integrated scanner.

With scanning, modeling and customizing many objects becomes even easier. Complicated objects with interior space and bending, such as a PVC elbow joint, are currently problematic. However, we still feel the benefits of scanning make the technology worth pursuing now. Here's an early example of scanning:

This scan shows the general shape of a Lego® minifig. The surface is uneven and the hands hint at being open. After refining our algorithms more, we produced the following image. Note the surface is more uniform than the original image, and the hands and neck are well defined.

Though we're still developing our scanning abilities, we're excited by the possibilities of scanning and printing using the same software and hardware--expect to see more in the future!

[radiantfabrication-blog]

Opening Doors

We’ve been refining the case for the Lionhead Bunny, the 3D printer we’re creating for testing purposes, to better meet our technical constraints and usability goals. The doors on the case illustrate the process as they’re both functional and a hardware interface to the printer.

Determining if a feature is needed is an important first step. Most commercial 3D printers use the frame for the case and don’t have doors, thereby avoiding the issue. We’ve elected to use a case to increase the printer’s reliability and to make it fit into more environments by being quieter. These properties just require a case, though, and not doors—if the mechanics support it, one side of the case could extend as the print completes. The Bunny’s mechanics aren’t really suitable for doing so, though, and hence we need doors at least for the moment.

One of our first attempts used sliding doors on the Bunny’s front panel:

Panels are functional but we felt they’d limit the usability of the case. Sliding small panels such as these could easily cause them to stick, forcing users to awkwardly alternate pushes on the top and bottom edges. Having users focus on the sliding interface is the opposite of what we want to happen; instead, users should be able to open the doors without giving them any thought. Another issue is that the open panels increase the printer’s space footprint as long as they’re open.

The next attempt replaced the sliding doors with a type of French doors:

These hinges addressed both of the previous issues: Opening them would require little work, and when completely open, the panels would fit alongside the case. However, we were concerned about the strength of the hinges, as extending beyond the case made them prime candidates for breaking if the printer is dropped or if something were dropped onto it.

We’re still improving our printer doors as we refine the case and internal mechanics; we’ll leave you with the latest version we’re trying out:

[radiantfabrication-blog]

Material Limits

Information on 3D printing is in a curious state for people who aren’t familiar with the technology. Journalists report that 3D printers will be creating everything from pizza to guns, and yet the material constraints aren’t clear. In 2D printing, users insert sheets of paper and generate documents that are fundamentally the same with some differences such as glossiness determined by the paper.

Applying the analogy to 3D printing suggests that a similar source medium (the paper in 2D, plastic filament in 3D) will be used to generate the whole range of prints (photos and text in 2D, food and complex mechanical parts in 3D). This may be true in the future, but current 3D printers are constrained to their input materials: 3D printers using plastic generate plastic objects, and those using metals generate metal objects.

Creating complex objects out of multiple materials is still difficult for 3D printing. 3D printers using multiple printheads can print objects using the same number of or fewer materials. The available printheads therefore limit the material complexity of printed objects. With industrial 3D printers currently supporting two printheads, printed objects are often limited to their main material and a support material.

Another option is to break the object into parts based on the number of printheads available and then assemble the printed parts to create the final object. Doing so increases the time required to create the final part; further, not all parts lend themselves to later assembly. For those parts that can be assembled post-printing, this method is limited by software interface complexity and printing time.

Yet another method involves manually changing materials while printing. Doing so requires strict attention to detail and ensuring that material changes don’t impair the final print. As such, this method is the least useful for general audiences and will likely remain only until multiple printheads and materials become more common.

The above methods to print complex objects using multiple materials avoids the other large problem: the available materials to print with. Together, this means that developing a 3D printer capable of printing any object is a long ways off, but we’ll hopefully see users making more complex parts as the multiple material problem is addressed.

[radiantfabrication-blog]

Leakage

Traditional solutions to problems in an established technology may not apply to new, but related, technologies. Take, for example, that embarrassing problem known as leakage. Leakage occurs when the plastic filament melts and either flows out where it shouldn't, such as between the heater barrel and the heat barrier, or it continues to flow out of the nozzle when it's meant to be stopped. In the first case, the print may proceed well until the leakage breaks apart or even damages the extruder; in the second, prints will exhibit wisps of plastic that need to be removed to finish the part.

Leakage in unintended locations is largely a matter of extruder design. Using PEEK instead of PTFE is one such change, as is ensuring that extruder components are tightly joined. And while some maintenance is still required, modern extruders have largely addressed this problem.

Nozzle leakage has also improved but is still problematic. One reason is that traditional extrusion algorithms use one filament to create the entire part: After drawing the part's outline for a given layer, the same unbroken line is used to create the infill. After completing the current layer, the line continues as the platform or extruder moves vertically to create the next layer. With simple parts, the only place that leakage would occur is when the part is finally finished, and removing a single wisp of leaked filament is relatively minor compared to all the other work required to make the print turn out well.

As 3D printers become faster and print multiple parts simultaneously, the nozzle leakage problem becomes more apparent. Instead of using a single filament, printing algorithms are forced to stop extruding while moving to the next unconnected location of the current printing layer. This is most apparent in consumer-grade 3D printers while industrial printers have already created solutions that work for them and their clients.

The big question that remains is whether existing nozzle leakage solutions scale to faster print speeds especially when printing multiple, small parts. We're working on an answer to this question in part because it becomes more salient in our printing algorithms, printing mechanism, and anticipated use-cases. And with some luck, it'll be something that users never notice because printed objects just turn out right the first time.

[radiantfabrication-blog]

Hardware Reliability

Though 3D printing has been around since 1984, it wasn’t until patents surrounding printing with plastic filament expired in 2009 that consumer 3D printing became a reality. Since then, four years have passed with multiple parties refining the technology until today, when 3D printers range in price from a few hundred dollars to a few thousand. As we’ve previously discussed, most of the refinement has taken place on hardware with software lagging behind. But despite the advances in hardware, significant barriers still remain before most people can set up a factory on their desktop.

Industrial-grade 3D printers continue to outperform commercial-grade 3D printers in part due to the pre-print process and the print monitoring issues. Before printing, many commercial-grade printers still require manual calibration. Automating calibration requires some additional hardware and software support, potentially leading to increased cost. Manual calibration depends on the user to be capable of accurately calibrating the machine and remembering to do so before each print.

During a print, industrial 3D printers clean their nozzles regularly. Doing so allows the filament to flow properly throughout the print. Without automating the process, users often need to watch the extrusion and clean the nozzles using pliers or tweezers while printing. Failure to do so can require aborting and restarting the print.

Both of these issues are addressable now; other hardware changes are still being hindered by patents. Using a heated build area, almost a necessity for ABS prints, and using cartridges with chips, needed to automatically configure software for materials and to report the material remaining, are still under patent. Simplifying extruder movement with a belt driven by two motors, which would significantly reduce costs in some cases, is covered by a pending patent application.

Though the above concerns specific features of 3D printers, manufacturers can more generally indicate their confidence in their hardware to survive everyday life by offering warranties. But until these concerns are addressed, it will be prohibitive for commercial-grade 3D printers to do so.

[radiantfabrication-blog]

Madison goes 3-D: City a potential hotbed for 3-D printing

"We’re trying to make the process more reliable and easy to use."

[radiantfabrication-blog]

The Radiant 3D printing circuit boards arriving and being assembled on Parallax's pick-and-place machine.

[radiantfabrication-blog]

Realizing Ideas

The promise of 3D printing is personal manufacturing whereby individuals realize their ideas by creating a physical object. What's often overlooked is that this is already happening today with tools ranging from personal CNC mills to crochet hooks. Whether creating hand-crafted bowls, custom electronics, or embodiments of funk, people are creating wonderful products without the use of 3D printing.

Implied in 3D printing's promise are the ideas that models will be customizable and printable by others and that most individuals will be able to both design and customize parts to print. So far, 3D printing companies have focused on the first implied idea, using a common file format (STL files) to exchange models. Many 3D modeling packages can import and export STL files, and so models are customizable by individuals with access to relevant software.

The second idea, that most people will be able to create and customize models, has been more problematic. The most common software packages including Blender, Maya, and SolidWorks use boundary representations for models. Designing an object requires creating and manipulating faces made up of edges and vertices, and skilled users can create intricate models quickly and easily. Unskilled users, though, have to learn a complex interface and the non-intuitive restrictions upon such models including intersecting and coplanar faces. Creating well-formed boundary representation models requires significant experience and skill, and requiring individuals to do so breaks the implied idea that anyone can create their own models to 3D print.

An alternative that holds some promise is using parametric models. For these models, designers create equations including constraints that describe an object and expose variables for others to manipulate. Importantly, this means that changing one aspect of a model automatically changes related features. Adjusting the height of a chair, for example, may simultaneously increase the area of the seat or the thickness of the legs depending on how it was constructed. A clear user interface can enable non-technical users to customize the variables and partially achieve the implied idea of customization. However, if the customization values aren't already exposed, then users either need to use either a CAD interface or even work with the mathematical models directly through programming; in either case, these likewise require sufficient technical skill to implement.

At Radiant, we believe that an embodied 3D modeler has the best chances of providing expressive power to users with the minimal amount of additional learning. By embodied, we mean that designers should be able to use their body to analogously understand how to create a model. A great example of this is the Logo turtle: Instead of trying to draw a circle on screen by deriving or looking up the standard equation (x - a)2 + (y - b)2 = r2, programmers could imagine how they could walk to make a circle by taking a step, turning, and repeating. By telling the on-screen turtle to replicate those steps, they could quickly create the circle on-screen. The Radiant Li is our attempt to bring body-based reasoning to 3D modeling for 3D printing.

Without addressing users' abilities to develop their ideas as 3D models, 3D printing will remain consigned to the hands of technicians or people satisfied with reproducing existing models. The technology has already reduced costs for professionals already, leading to indirect benefits for consumers, but much work remains before a 3D printer can claim to be usable by everyone.