It’s a common enough scenario. You work late into the evening to get your latest and greatest design iteration queued up for printing overnight and your client is scheduled to visit you first thing in the morning. You need that print to work first time, even if it involves joints and moving parts. How to guarantee success?
This summer (2015), our intrepid intern Gregor Mackay took it upon himself to make 3D printed assemblies more reliable. In Gregor’s words:
My experience with 3D Printing prior to working at GrabCAD had almost entirely been single solid pieces with no joints or moving parts. Pretty simple stuff! Printed parts naturally don’t come out to the exact measurements as they are intended, which made me think about the effect this would have on how they interact with each other within an assembly. I decided that the ease and practicality of 3D printing could benefit from compiling information regarding this relationship, to ensure that 3D printed assemblies fit together as intended.
Over the course of several weeks, Gregor experimented with a variety of printers, gaining information on how to attain a desired fit. Armed with that information, he designed and built a jointed assembly that would work first time. Or so he thought.
Using 4 printers from Stratasys (Makerbot Replicator Z18, Fortus 250mc, uPrint SE Plus and Objet30), Gregor created a series of round pegs and holes, the peg diameter varying in steps of 0.05mm around nominal hole dimensions of 10, 12.5, 15 and 20mm. He then classified the fit of each pair as follows:
Clearance fit: parts can easily be assembled, moved relative to each other, and disassembled with almost no pressure.
Transition fit: parts can be assembled, moved or disassembled with light hand pressure. They fit tightly without movement, but can be disassembled easily.
Interference fit: parts can be assembled with moderate hand pressure or a light hammer to make a fit that is difficult to move or disassemble.
|Printer||Specified Hole Diameter (mm)||Specified Peg Diameter (mm)||Diameter Delta (mm)||Fit classification|
|Replicator z18||10||9.85||0.15||Clearance / Transition|
|uPrint SE Plus||10||9.8||0.2||Clearance|
|uPrint SE Plus||10||9.85||0.15||Transition|
|uPrint SE Plus||10||9.9||0.1||Very tight interference|
|Fortus 250mc||10||9.85||0.15||Transition / Interference|
|Fortus 250mc||12.5||12.3||0.2||Clearance / Transition|
|Fortus 250mc||12.5||12.35||0.15||Transition / Interference|
It was notable that with all printers tested, even interference fits required the peg to be specified smaller than the hole.
The Objet30 was the most dimensionally-accurate printer in the test. Fairly reliably, a diameter delta of 0.2, 0.15, and 0.1mm on an Obet30 gives a clearance, transition or interference fit respectively. For the other printer types, consult the table.
Layer direction on the Fortus, uPrint, and Makerbot printers was important. If the layers on peg and hole are aligned so that layers must slide over each other, it makes the parts harder to disassemble than if the layers are aligned with the sliding direction. The parts in the experiment had layers aligned with each other, at 90 degrees to the sliding direction.
On the Fortus, uPrint, and Makerbot, a line forms known as the ‘seam,’ where the extrusion of each bead around the peg and hole begins and ends. You can sand off the seam in a post-processing step, or leave it as-is. In the Fortus software, you can set the ‘seam placement method’ option to ‘random’, which makes different layers start and end in different places. Although randomising the seam effectively makes it disappear, this option causes the peg to be slightly bigger and the hole to be slightly smaller; which affects the fit. Further experimentation is required to classify the fits for the various software options. The results in the table are for ‘automatic’ seam placement (which is the default), with no post-processing.
A salutary tale
‘Sashimono’ is the ancient art of Japanese woodcraft. Master practitioners create exquisite furniture, or fashion the beautifully jointed trusses that support the roofs of Japanese temples. When the would-be sashimono master finishes his apprenticeship, he may create a wooden toolbox of several layers, dovetailed together with sublime accuracy. Light pressure with a finger is enough to cause the layers to glide against each other and the toolbox opens.
Armed with his recently-acquired knowledge, Gregor designed in 3D and printed such a toolbox on the Makerbot Replicator z18 (see Figure 2). The dovetail joints should have slid perfectly, and indeed the offsets were correct. But the layers were stuck fast. Why? One word: warping. The parts had warped slightly in the printer due to thermal stresses. Not much, but enough to jam the box tight shut.
Now, there are ways to minimize warping – but that is a whole other story.
What’s the takeaway? There is always something for the designer to learn about manufacturing processes, and 3D printing is no different from conventional manufacturing in that regard. The more we know and share, the better our designs become.
Wouldn’t it be great if one day, this kind of information were made available in CAD systems, for example encapsulated in SolidWorks Custom Features that intelligently applied the right offsets, taking into account printer type and other variables? That way, all of us could make assemblies that work first time, (almost) all of the time.
This eBook is aimed at the engineering professor, new or experienced, that is interested in how their peers are thinking about the challenges associated with modernizing their curriculum.