Introduction: 3D Printing Threaded Features With Autodesk Ember and Inventor
One of the most exciting prospects for 3D printing is creating usable, functional parts on-demand that might otherwise take days or weeks to acquire. In an effort to explore a pragmatic use case for 3D printing, I used Autodesk’s Ember printer to print a variety of parts with threaded features. Although the material ultimately limits the usability of the parts, the results provide insight into the potential of 3D printing mechanically functional features and the precision and limits of Ember. I also cover the use of Autodesk Inventor to prepare models for printing on Ember.
I started by downloading solid models of nuts, bolts, and pipe fittings from McMaster-Carr’s online catalog. This website has a large library of 3D, dimensionally accurate cad drawings and models available for download. Here is a list of the parts I selected:
- 000-120 hex nut and machine screw
- 00-90 hex nut and machine screw
- 0-80 hex nut and machine screw
- 1-72 hex nut and machine screw
- 2-56 thumb nut and machine screw
- 4-40 hex nut and machine screw
- 6-40 hex nut and socket head cap screw
- 8-36 hex nut and socket head cap screw
- 10-32 thumb nut and socket head cap screw
- 1/4"-20 wing nut and socket head cap screw
- 1/8" NPT plug and cap pipe fittings
- 1/4" NPT plug and cap pipe fittings
In an attempt to get an idea of the limits of the printer, I chose parts featuring the smallest thread size listed up to a ¼-20 fastener that has features that I expected Ember to resolve without issue. Additionally I opted for fine threads when possible to further highlight the high-resolution capabilities of the machine.
Step 1: Preparation
After downloading the models in 3-D Soldworks format, each file was imported into inventor to create a corresponding Inventor part file. When creating the part in Inventor, it is important to start with the mm template to ensure compatibility with Ember’s online slicer.
As the z dimension of the part increases, additional torque is generated at the build head-part interface due to the separating action of the printer. Subsequently, I added support material to further anchor the bolts to the build head. I did not need any support geometry for the nuts since the “aspect ratio” of those parts led me to believe that the surface area of the part is sufficient to enable successful printing. While it is possible to use MeshMixer to automatically generate supports for Ember, I manually created support geometry for the bolts in Inventor. The support geometry was added to each bolt model by creating a sketch referenced against the bolt geometry that was then extruded and patterned.
After preparing all the individual parts, I leveraged Inventor assemblies to layout the parts for printing. I started by creating an Inventor part to represent Ember’s build area. I sketched a rectangle constrained to the dimensions of the build area (64mm in the x direction and 40mm in the y direction) and extruded it by 0.5mm. The z dimension is not important since this is only used to position the parts in the x-y plane. I imported the stand-in build area and all parts into an Inventor assembly initiated from the mm template. I constrained the surface of each part I wanted to be in contact with the build head to the x-y plane of the stand-in build area. This effectively placed all the parts against the same z datum that represents the build head surface. Then I simply dragged the parts around until they were nicely packed within the build area.
It is only important that the parts be positioned correctly relative to each other. I.e., they all have a common surface at the same z dimension and the bounding box encapsulating all the parts fits within Ember’s build volume. Ember’s online slicer does not look at the absolute position of the model but rather centers the extents of the final model within the x-y plane of the build area and places the x-y plane with the smaller z dimension against the build head.
As a result of the relative positioning performed by the slicer, if one desires to position a part at a specific location in the x-y plane of the build area, a small feature must be added to effectively increase the bounds of the final model. In the case of the bolts, I wanted to minimize distance covered by the smallest bolts relative to the PDMS window during the sliding action due to their delicate nature. To accommodate this constraint, I positioned the smallest bolts on the left side (when looking at printer with model upside-down) of the build area and placed a tiny cylinder at the opposing corner to ensure that the position of the bolts relative to the build head is preserved during slicing.
After placing the parts, I disabled the build area stand-in part to prevent it from appearing in the final output and exported the assembly in STL format. When exporting, it is necessary to change the units of the exported STL to millimeters by editing the options accessible from the export dialog box after choosing STL as the export format. Additionally, I chose the high detail setting to ensure that accurate slices are generated.
Step 2: Printing
I used emberprinter.com to prepare, start, and monitor the print. I printed the parts in Autodesk’s standard prototyping resin (PR-48) and left all settings to their default values except for the "First Layer Wait (After Approach)". Adding a non-zero value (I used 1.5 s) for this setting ensures that the position of the resin tray equilibrates after the build head approaches the PDMS window at the end of each separation cycle. For prints with a large amount of surface area, this can help reduce jamming.
After the prints completed, I carefully removed the parts from the build head using a thin metal scraper and rinsed them in isopropyl alcohol. Then I let the parts sit out to dry.
Step 3: Results
Upon initial inspection, the parts appear to have excellent feature resolution. However the thread dimensions do not seem to fall within the acceptable tolerance range. The nuts appeared to be “too small” when I attempted to thread the printed parts together. I tried threading an off the shelf ¼-20 bolt into the appropriate 3D printed nut and was also unable to successfully mate the parts. I ran a tap through some of the fasteners with thread sizes 4-40 and larger and was able to then thread the 3D printed nuts and bolts together with varying degrees of success. The bolts did not require any post-processing of their threads however the fit was somewhat tight when first assembled. The 4-40 3D printed nut and screw seized together eventually, perhaps due to cross-threading. The fasteners smaller than #4 were too delicate to attempt assembly. As expected, the material strength limits the feasibility of creating mechanical features below a certain size. I was also not able to assemble the pipe fittings.
Other experience indicates Ember is limited in its ability to print features with negative space accurately below some dimensional threshold. Essentially, the thread grooves may not be "sharp" enough, which prevents the parts from mating without post processing. It's possible that varying the settings may improve this aspect of the prints.
Additionally I measured the outside diameter of the printed pipe caps, which indicated that they printed as ellipses. The measurements for the 1/8" cap, which has an OD modeled at 15.997 mm, ranged from 16.10 mm to 16.15 mm. For the 1/4" cap modeled with an OD of 18.999, the measurements ranged from 19.03 mm to 19.17 mm. The diameter in one direction seemed to measure consistently larger than in other directions. These parts are still impressively accurate given that they were 3D printed. I think the limitations of the material will come in to play before dimensional variations on the scale I observed become significant. However, this deviation could contribute to the difficulty in mating. The DLP projector may introduce these errors as a result of the way it maps the input image to its pixel arrangement.
It is also obvious that the material used does not wear very well. As the parts are threaded together, the assembly continues to produce “dust” as the contact surfaces wear away on each other
In conclusion, threaded features printed with Ember look quite impressive but are limited in functionality. Normally a lathe or other tooling is required to produce these sort of features. It is pretty cool that Ember can get as close as it does. As 3D printing advances I look forward to seeing more technology and results that fulfill real, functional need.