People often claim 3D printers can “make you anything you can imagine.” Dial up the digital model you want, hit “Go,” and the machine hums to work, producing an object, accurately and repeatably. But as an astute 8-year-old pointed out to me when I handed her two of my favorite printed models at Maker Faire Bay Area last year, the results don’t always match your intentions.
“That octopus is red! A TARDIS is not supposed to be yellow!” she wailed and knocked my offerings away.
While overall shape and mechanical fit are valued more highly than surface treatment in today’s desktop 3D printing, it’s sometimes worth judging a print by its cover.
I’m reminded of advice I got from a pair of industrial design professors at Pratt, after I showed them my print of a fluorescent-green clockwork mechanism: “It is worth enormous effort to make prototypes look like they were created from real-world materials.” Even the most creative engineers and businesspeople will have difficulty seeing your prototype as a “machine” when it looks like a toy.
The domain of finishing techniques for 3D printed objects (i.e., everything that takes place after printing) is the craftsman’s workshop, where patience, tools, skills, and experience can transform the raw products of these machines into fully realized models. Like builders of dollhouses and model trains, many 3D printer jocks appreciate a loving and accurate rendering of a miniature world.
The results are impressive, but why should you tackle these craft skills when you could spend that time printing more plastic objects?
Makers who have mastered finishing techniques are granted wizard status by fellow 3D practitioners. Take artist Cosmo Wenman, who creates pieces that accurately mimic distressed metals and stones. And sculptor Jason Bakutis, whose sanded, painted, and polished faux marble and jade prints look remarkably like the real thing. Through careful work, pieces printed in crazy pink, green, and translucent filaments are made to resemble clay, stone, metal, and wood. How do they do that?
The desktop 3D printing community has a lot to learn from the sculptors, model railroad builders, and tabletop gamers now joining their ranks. And as my professors pointed out, these extra steps aren’t just cosmetic. Your capacity to transform your models into “magical” replicas is a crucial means of communicating your inventions.
Desktop 3D printing has yet to spawn third-party finishing services like commercial 3D printing did a decade ago. So, without access to acetone cloud chambers, multi-axis enamel jet robots, agitating chemical baths, and industrial tumblers and polishers, makers have rolled up their sleeves and discovered a host of finishing solutions using inexpensive tools and materials. These methods not only affect a print in post-production, but can often change the way we think about a digital model back in the initial design stages.
In researching my upcoming book Design for 3D Printing (Make: Books, Sept. 2013), I’ve interviewed a wide range of members of the desktop 3D printer community. I’d like to share some of their promising tools and techniques. In turn I hope that those of you refining new methods and sourcing better, safer, and cheaper products and techniques will also share. Post your ideas and thoughts in the comments section.
Friction welding involves the use of high-speed rotating tools and should not be attempted without ANSI-approved safety glasses. Welding and other operations that heat, soften, and melt plastic may release hazardous chemical vapors and should not be attempted without proper ventilation. Sanding and other dust-producing operations should not be attempted without a NIOSH N95-approved particulate respirator. Acetone and other volatile solvents should not be handled without proper ventilation, safety goggles, protective clothing,
and latex or nitrile gloves.
The world may have forgotten the Spin Welder toy sold by Mattel in the mid-1970s, but Fran Blanche of Frantone Electronics did a great job of re-creating the experience in her 2012 video “Build Your Own Friction Welder.” Using an inexpensive rotary tool, Fran was able to spin a styrene rod fast enough to create a strong weld between two pieces of plastic that was difficult to break apart by hand. With the Spin Welder toy, children assembled the frames of helicopters, motorcycles, and other projects by fusing together beams and struts, then used plastic rivets to fasten the outer shell. Sure, it was potentially one of the most dangerous toys of all time, but I agree with Fran’s conclusion: why haven’t tools like these joined the maker’s toolbox?
Unlike adhesives or traditional welding, friction welding fuses metal or thermoplastic objects together by quickly spinning or vibrating one piece against another. Mechanical friction creates a melt zone shared by both parts, fusing them into one solid piece. In friction surfacing — a variant of friction welding — a piece rotated at high speeds is moved across an edge or surface under gentle pressure to weld seams, patch gaps, or smooth surfaces.
These techniques are common for plastics and aluminum in the automotive and aerospace industries, but the tools are expensive. Sophisticated spin welders can spin parts at hundreds of thousands of RPMs for short bursts of even single-digit rotations, parking the fused part at a precise orientation. Where are the cheap, hand-tool equivalents?
As it turns out, many of us already have the equipment to experiment with friction welding. Dremels and similar high-speed rotary tools spin fast enough to melt 3D printer plastics, and printer filament can be used as welding “rod” to solidly fuse parts or close seams. These tools can also spin-weld 3D-printed rivets. And while it takes them a second or two to spin down again, the melting points are comparatively low, allowing for some manipulation after the fact to reposition the joined part.
I think both approaches — welding and riveting — are killer tools for 3D print finishing, particularly for “blind riveting” into the side of objects, and for joining parts made of PLA, which is typically much harder to glue than ABS.
I spent some time with Chris Hackett from the Madagascar Institute learning how ideas from traditional metal welding might apply to friction-welding 3D-printed parts. We experimented with the rotary tools in his workshop and came up with the following approach for creating a nice welded seam in plastic, similar to a traditional metal weld. When two printed parts don’t mate perfectly due to warp or poor planning, you can friction-weld them together as securely as if they were a single printed part. Here I’ll demonstrate with ABS parts and ABS filament. It works with PLA too.
1a. Select the collet you need for trapping the filament you’ll be using. For 1.8mm filament, use a 3/32″ collet as shown here (3 rings) and for 3mm filament use a 1/8″ collet (0 rings).
1b. Insert a short length of filament into the collet jaws and tighten down the collet nut to secure it in place.
1c. Trim the filament about 1/2″ from the collet. Short pieces are easier to control, and they spin on a tighter axis. (With experience you can use longer pieces, pressed gently at an angle, to make longer welds. You may need to straighten them by reforming them with
a heat gun.)
2a. After scraping and sanding, these 2 watch body cases meet with a gap that varies between 0.1mm and 2mm around the edges. That’s too sloppy for gluing, so I’ll weld them.
2b. Use a deburring tool or razor blade to bevel the top edges of the seam where the parts meet, forming a narrow, V-shaped channel. Your goal is to create enough room for 3 welding layers, from the bottom of the bevel up to just above the surface of the 2 parts. This method gives a stronger bond than a weld that sits just on the surface.
2c. Warm both parts with low heat from a heat gun. This helps them receive the weld to the same depth. If one part is much larger than the other, focus extra attention on warming the larger piece.
3a. Now you’ll tack the parts together with a series of short spot welds, moving around the joint while holding the parts steady.
Spin up the rotary tool, and lower it until the spinning filament grazes both surfaces of the seam. When the tip of the filament begins to deform, apply a little pressure.
3b. Moving the spinning filament in tight little circles, widen the melt zone slightly into the side of both parts, making a little forward progress with each circuit, until you’ve created a small spot weld.
3c. Tack in 3 or more places along the seam and let the parts cool. They should be tacked tight, difficult to separate by hand.
4a. Gaps that are wider than half the width of your welding filament should be filled before welding a clean seam. Soften a short scrap of filament to use as filler, by using the low setting on a heat gun or by warming to 100°C on your printer’s heated build platform.
4b. Press the softened filament into the widest gaps between the 2 parts, making quick tack-welds if necessary to pin it in place.
Now weld the whole seam in 2 or 3 layers, as shown in the diagram. A single weld would probably bond the parts at the surface only, allowing the seam to be broken if the parts are torqued.
To hide the weld, you can sand it back flush and then paint or seal the surface.
PLA is prone to cracking and splitting, and it’s typically difficult to repair. Solvents such as acetone have little effect. ABS glue or super glue merely cement the parts by surface tension, rather than offering a chemical weld — meaning that the seam can easily be rebroken.
With friction welding, you can form solid joints that are difficult to break.
1. Press the patch or broken part into place and hold it securely.
2. Tack-weld the patch in place in a few different positions.
3. PLA is workable at lower temperatures than ABS, so use a gentle touch to melt and weld the 2 parts. Too much pressure can create a puncture. It takes a bit of practice.
4. Move around the seam, changing direction as necessary for handling and control. For prettier welds, take frequent breaks to let the parts cool.
5. Let the completed seam cool to room temperature before sanding and sealing.
A rotary tool can also be used to permanently fuse a spinning part to a fixed one, using a one-sided “blind rivet.” Blind rivets have one huge advantage over ordinary solid rivets: you don’t need access to both sides of an assembly to rivet its parts in place.
This method works well for attaching plastic panels to the outside of objects when access to the interior is awkward or impossible. It also lets you construct massive objects from multiple panels, each panel printed close to the bed of the printer for optimal printing.
Shown here are two 3D-printed blind rivets, next to a brass solid rivet and 3 aluminum blind rivets. Notice that the printed rivets, like the aluminum ones, have a “mandrel” that extends well beyond the rivet’s head. This is the part that’s gripped in the rotary tool. You’ll clip it off after the rivet is firmly in place. I designed the printed rivets to be gripped by the 1/8″ collet, the standard size for most Dremel accessory bits.
Plastic rivets need not be perfectly cylindrical for friction welding, so I designed them “three-quarters” round, for printing flat on the platform. This way, the horizontal “grain” of the printed rivet helps strengthen it.
I’ll demonstrate by riveting a small panel to the outside of another part. To print your own rivets, get the 3D files at thingiverse.com/thing:61510.
1. With the proper collet in place, loosen the collet nut and insert a plastic rivet.
2. Drill or design in mounting holes in your panel to provide clearance for the shaft of the rivet to pass through to the base part where it will be fixed. The hole should be narrow enough that the rivet’s head will pin the panel in place.
Pre-drilling (or designing) a pilot hole in the base part can help prevent your rivet from mounting off target.
3. Spin up the rotary tool and gently insert the shaft of the rivet through the mounting hole until it contacts the mounting position. Continue spinning until the shaft of the rivet begins to melt and deform — then press it gently down into place.
4. Stop the rotary tool and hold it steady in a fixed position at a right angle to the work, while applying a little downward pressure. It can help to use a piece of cardboard or foam as a friction brake to stop the rotation quickly. (Unlike a professional spin-welding tool, most rotary tools need a second or two to spin down.)
5. Loosen the collet nut and slip the mandrel of the blind rivet out of the rotary tool. If the rivet is still cooling, hold it in position until it’s fully cooled (at which point it should be entirely fused with its mounting point).
6. Using a flush cutter, snip off the mandrel, leaving the head intact.
7. If the rivet head protrudes too far, has a sharp ridge, or seems too narrow to secure the panel in place, warm it with a heat gun and use the head of a steel nail to press it flat.
TIP: It’s possible to fuse ABS rivets to PLA, and vice versa, but you’ll need to find the “feel” for the initial friction stage before pressing down the body of the rivet. Before mounting delicate parts, test-rivet the materials you’ll be using.
People have used rivets since the Bronze Age to fasten together tools, art, bridges, and buildings, so it’s no surprise that 3D printer users are experimenting with riveting techniques. We’ve seen a number of projects using pieces of filament as pins to hold together large assemblies.
Just recently, 3D artist and instructor Jason Welsh demonstrated a method for building his DIY electronics cases that promises to become a new power technique. His Folding Arduino Lab (thingiverse.com/thing:32839, shown here) and Pi Command Center (thing:38965) each use filament “spikes” to create rivets and hinges.
Essentially, Welsh uses heat to reform pieces of filament into straight rivets, flattening one head before inserting the rivet and the other head after the rivet is firmly in place. As with any solid rivet, you need access to both sides of the assembly, but the advantage of this method is the creation of strong fastenings that can be completely removed later using a flush cutter.
While you can make spikes with any filament, I recommend 3mm PLA based on my experiences building Welsh’s project. PLA is easier to soften and work with a heat gun, and 3mm spikes remain straighter and more rigid than 1.75mm spikes after cooling. If you don’t have 3mm filament, you can accomplish the same goal with 1.75mm filament by using more rivets to distribute the load.
1a. With a heat gun set to low, evenly warm a 4″–6″ length of filament until it becomes limp. (Several minutes on a heated build platform works, too.)
1b. While the filament is still hot, straighten it by rolling it on a table, or better yet, on a piece of glass that will quickly cool it. Gently move both hands away from each other while rolling, to keep the filament straight as it cools.
1c. Soften one end using a heat gun on a low setting. (A soldering iron, heated brass nozzle, or heated build platform can work in a pinch.)
1d. Tap the soft end of the spike on a flat, cool surface until it deforms into a flat rivet head. I tend to use a steel nail head, but any flat surface that can cool the filament rapidly will work.
1e. Your rivet should have a nice flat head, wide enough to rest firmly on the edge of the mounting hole. In rivet lingo, this is the “factory head,” as opposed to the second head or “shop head” you’ll create on the other end when installing the rivet.
2b. Use a heat gun to soften the protruding tail of the rivet until it begins to deform.
2c. Use a flat, smooth surface to press down and deform (“buck”) the tail, creating the rivet’s shop head. I find that a large steel nail head works best — it’s easy to handle and it cools the shop head quickly.
2d. Continue to press tight against the shop head as it cools, making sure it doesn’t relax away from the mounting point.
Don’t apply force to the assembly until both the rivet and the parts being assembled cool to room temperature.
Installation of a hinge rivet follows the same procedure, except at the very end.
After pressing the shop head, move the parts gently as they cool to ensure that the joint has enough play for the hinge to open and close easily.
In the completed assembly shown here, the 2 black ABS plates can spin easily on the hinge without coming free.
ABS and PLA plastics have very different physical properties. ABS is printed at a higher temperature (typically 215°C–235°C), is more durable and flexible, and dissolves in industrial solvents like acetone. PLA can print at lower temperatures (starting at 180°C), wears down faster, can be brittle or shatter, and won’t dissolve in acetone. (The chemicals used to dissolve PLA are highly toxic.)
TIP: If you’d rather use a soldering iron than a heat gun, find a brass tube that fits snugly over your iron. Use the brass tube to work the plastic, and keep your soldering tip clean. Make sure to clean your tube thoroughly so the plastic doesn’t stick to the brass.
TIP: Tape the factory head of the rivet in place when you need both hands to soften and flatten the shop head.
While “super glue” (cyanoacrylate) and plastic model glues do an excellent job of bonding ABS parts, many 3D-printed model builders have switched to using “ABS slurry” for both glue and filler material, because this substance can weld parts together more permanently, and can be exactly color-matched to the printed parts. ABS slurry is simply ground-up ABS filament dissolved in acetone.
Applied in the open air, acetone melts the surface of ABS plastic (and many similar styrene plastics), creates a goopy sludge, and then — after some time — evaporates, leaving behind just the reformed ABS plastic. By sealing up this process in an airtight container that the acetone cannot easily escape, you can prepare a thick, even acetone/ABS mix similar to acrylic gel medium.
There are a variety of methods for preparing ABS slurry. I like ProtoParadigm’s recipe (thing:14490) — 1 part ABS to 2 parts acetone, mixed in fingernail polish containers or similar. Use a cheap coffee/spice grinder to shred ABS filament and scraps as needed. Smaller pieces dissolve faster and make it easier to gauge the mix ratio.
WARNING: Observe proper handling precautions when working with acetone and ABS slurry. Wear gloves and goggles and do not work without proper ventilation or in the presence of open flames. Besides being highly flammable, ABS slurry sticks to anything and burns with a foul-smelling smoke that is widely regarded as toxic. Be very careful or you’ll create a tiny batch of “napalm” that will need to be treated like a chemical fire.
Apply ABS slurry with an inexpensive natural-hair paintbrush (synthetic brushes will dissolve in acetone!) to either fill small cracks or glue 2 pieces together. Leave it to air-dry until the acetone completely evaporates, and your final part will have a joint or patch made only of ABS plastic.
Your exposure to acetone is greatest while applying ABS glue and immediately after, so pin your parts to a piece of cardboard or a tray that you can immediately move to a well-ventilated area away from your workspace. If you move them outside, protect them with a cardboard box to keep leaves, dust, and grime out of the still-goopy slurry.
TIP: While acetone can “weld” the edges of ABS parts to bond them, this joint lacks the shear resistance of parts printed together, because the “melt zone” doesn’t extend deep into the surface. If an assembly needs mechanical strength, design an interlocking joint with lots of surface area — or use hardware.
When I first learned the basics of woodworking, I proved a lazy, inept sander of splintery plywood toolboxes and lopsided Pinewood Derby cars. My father suggested I forget about “sanded” as a goal, and focus on “sanding” as an activity. You cycle from coarse-grit sizes down to finer-grit papers until the surface is as smooth as you intend.
Same goes for sanding 3D-printed plastics. With ABS and PLA, you can work your way down to very fine papers indeed — 3M gem-polishing papers and Micro-Mesh sanding tools with single-digit micron grits that create scoring patterns invisible to the naked eye.
Still, well-sanded 3D-printed projects seem few and far between, for two good reasons. First, ABS and PLA are softer than the wood we’re used to sanding. Second, the tricky horizontal “grain” created by 3D printing reflects light differently than sanded surfaces or the glossy heated base (in ABS printing), and this grain cannot be tooled back into the surface easily — encouraging an all-or-nothing approach to sanding the object.
The basic rule of thumb is to sand 3D-printed pieces like you’d sand a gummy hardwood. Focus on “sanding” and don’t rush toward “sanded”: start with 100- or 150-grit papers or Dremel wheels, then 220, then 320 fine, then 500 super fine, and then tackle the micron-grade grits to eliminate sanding marks. Many 3D makers tend to skimp on the earlier papers to their detriment: these coarser grits are capable of stripping away the peaks of the layer lines. Go too fine too fast and you’ll just round over the peaks without flattening them.
After you’ve sanded a surface to your satisfaction, you can use a heat gun to gently warm the surface until it melts slightly, which will erase many of the smaller scratches and restore the original printed color. Practice on scrap until you get the feel of it.
You can also can use the Novus plastic polishing system to get a remarkably smooth, polished surface on ABS prints, but most 3D modelers opt to just paint their prints instead.
MAKE Volume 34: Join the robot uprising! As MAKE's Volume 34 makes clear, there’s never been a better time to delve into robotics, whether you’re a tinkerer or a more serious explorer. With the powerful tools and expertise now available, the next great leap in robot evolution is just as likely to come from your garage as a research lab. The current issue of MAKE will get you started. Explore robot prototyping systems, ride along with the inventors of the OpenROV submersible, and learn how you can 3D-print your own cutting-edge humanoid robot for half the price. Plus, build a coffee-can Arduino robot, a lip balm linear actuator, a smartphone servo controller, and much more