Additive manufacturing can reduce the time and material costs in a design cycle and enable on-demand printing of customized parts. New multi-material 3D printers that can print both metal and dielectric materials enable the additive manufacturing of antennas and RF components. Developments in software are critical to leveraging this capability; good tools allow more effort to go towards creation than implementation.
Three devices are described in detail in this article to demonstrate the 3D printing of RF components. First, a Marchand balun is presented, demonstrating rapid prototyping of a complex, multilayer RF circuit. Next, a monopole array is shown with an integrated beamsteering network and radome to show rapid prototyping of a complete antenna system. Finally, a bowtie antenna with rounded corners is presented, showing good performance in the Kuband.
The devices were made with a Voxel8 printer and materials.
Several existing tools and sites provide the ability to customize mechanical structures; this concept is expanded into the RF domain with software that uses a high-level design parameter to create the circuit, model the performance, and create Computer-Assisted Manufacturing (CAM) files. By intelligently leveraging this process, the design can be readily updated or customized after the initial development. A Computer-Assisted Design (CAD) tool may further modify the structure to customize the mechanical interface and a machine toolpathing code (a slicer) is used to translate the CAM files to a format the printer can use. The monopole array with an integrated beam-forming network and radome is specifically used to illustrate the process that is used for each component presented.
Design tool. A custom frequency-domain circuit simulation code has been developed that uses a schematic input to model a circuit’s S-parameters. The design tool handles variables, calculations, and can perform optimizations; the modeled results for the monopole array are plotted. The tool can automatically generate the layout from the schematic and can create stereolithography (STL) files for both the metal and plastic materials. The overall design process is simplified significantly since the same schematic/tool is both modeling the device and creating all of the CAM files without requiring a CAD-specific tool.
The monopole array schematic has three sets of variables: design, material, and geometry. The design variables are the inputs most likely to be modified such as the center frequency of operation. The material variables are an example of inputs that change infrequently; for example, if the design process shifted to a different plastic. Finally, a series of calculations is needed that defines the geometry, and altering these fundamentally defines a different product; these calculate the dipole positions, line widths, and line lengths.
CAD tool. If needed, the mechanical design can be further modified. Open-SCAD, a free CAD tool, uses code to describe mechanical structures; for example, the location of the mounting holes is easily modified through a few straightforward variables. This simplifies the process of integrating the design to a new platform. While this process simplifies minor changes in form factor, care must be taken not to break the RF performance.
Slicing tool. The final software tool needed is a machine toolpathing algorithm (slicer) to translate the mechanical CAM files into the machine control language of the 3D printer. These prints used Euclid, which is the slicer provided by Voxel8. Euclid dynamically modulates the plastic layer heights to ensure the accurate spatial positioning of the silver ink, which is important for the Marchand balun. Euclid automatically performs subtractive Boolean operations that create cavities in the plastic to ensure proper clearances for printed silver features, electrical components, and printhead geometries. Performing the Boolean operations in the slicer simplifies the requirements of STL generation for the design tools, enabling the use of many different 3D design processes.
Components and Systems
Three example components are presented in this section to validate the performance of the tools and hardware: an L-band balun, an S-band antenna array, and a Ku-band antenna element.
Marchand balun. A Marchand balun is a reasonably broadband device made out of coupled lines and used in many RF applications. This design used broadside coupled striplines (Figure 1) that intrinsically isolate the lines from stray coupling. The printed structure was wrapped in copper tape to provide the upper and lower ground planes. Three throughhole connectors were attached using conductive ink and epoxy. The design had good performance as a balun over the design band of 1.5 GHz – 2 GHz.
Monopole array with integrated beam former. A lightweight Wi-Fi directional array was constructed with mounting holes for a 3DR Solo UAV. Monopoles were used to keep wind loading low. The device was constructed to be directional and to provide a null (Figure 2) that can be useful for improving the link, direction finding, or interference suppression. The location of the four quarter-wave monopoles is calculated based on the design frequency, and the line lengths of the combiner provide a true time delay sum in the desired direction. The device had a good match and pattern. The monopoles for this array are 31 mm tall. The three outer monopoles are a half-wavelength from the center element and 120° in angle from each other. The dielectric height for the stripline section was 4 mm.
Ku-band bowtie. A bowtie antenna was constructed to characterize the printer’s ability to make higher-frequency antennas (Figure 3). The antenna was modeled with ANSYS HFSS and good agreement was found through Ku-band.
Capabilities of the Printer
While the concepts described in this article are by no means restricted to a single process, it is worthwhile to examine the specific process used to validate the designs.
The permittivity and conductivity of the materials used were measured in a variety of ways and all showed agreement. All measurements were taken with printed material (as opposed to raw stock).
The plastic used for these prints is polylactic acid (PLA), which is popular for prototyping in part because it has less toxic fumes than other plastics. The initial measurement used a focused beam system (Figure 4) to measure the permittivity for four colors of PLA, each approximately 3.1 mm thick. These values were validated with a coaxial airline technique and through the measure-model agreement of the various RF circuits.
Voxel8 Standard Silver ink, a room-temperature-curing silver conductive ink, was used for these prints. The vendor lists the DC conductivity as 3.45 MS/m. A 250-m-thick board with a 1.5-mm-wide, 71-mm-long microstrip line constructed and measured to back out the conductivity of the silver ink showed good agreement using 3.45 MS/m in a simulation that agrees with the measurement. For reference, pure silver has a conductivity of 61 MS/m.
When designing printed circuit boards, it is common to consider the trace and space tolerances; that is, the accuracy to which one can maintain a desired trace width and the gap between traces. Because the 3D printer deposits the ink with a 0.25-mm nozzle, it should be expected that small traces might not print as designed. The vendor recommends 0.5-mm lines for general prints (two passes); however, RF circuits generally require more design flexibility.
Three test boards were printed to measure the trace and space tolerances and were characterized with a Keyence digital optical microscope used as a profilometer. The gap measurements were averaged along the lines; at 100 μm, the lines were sporadically shorted. Below 1 mm, line widths had poor accuracy, although from an RF standpoin, it’s acceptable above 800 μm. Gaps at 100 μm were not reliable but accuracy was good above 100 μm.
Useful RF and antenna structures can be 3D-printed, which has the potential to revolutionize the design, supply, and sustainment phases of an acquisition program. A design process was presented that demonstrates how this technology can be utilized to support customization in the field.
This article was written by Gregory Kiesel, Philip Bowden, Kevin Cook, Matt Habib, Jeramy Marsh, David Reid, Cameron Phillips, and Brad Baker of Georgia Tech Research Institute, Atlanta, GA. For more information, visit here .