Introduction

Internet of things (IoT) devices are increasing in number over time and penetrate various fields of application [2]. In general, the applications can be separated in indoor and outdoor applications. An important functionality of IoT devices is the submission of data. Part of the transmission system is the antenna. Different antenna designs are available in regard to the requirements of the IoT device. In addition, antennas can be differenced according to their directionality. Omnidirectional antennas exhibit a direction-dependent transmission power, whereas isotropic antennas exhibit a direction-independent behavior. Isotropic antennas transmit and receive power in all directions, thus, the final orientation of the antennas has no influence onto transmission range [1]. For IoT applications in outdoor environments, the orientation of the IoT device might be dynamically (e.g., position determination of mobile objects). Thus, an isotropic antenna might guarantee the transmission of data independent of the orientation of the IoT device. Among the different data transmission technologies, LoRaWAN exhibits certain features which make it to a good candidate for IoT application for transmission intervals of several minutes and low payloads [7]. The advantage of LoRaWAN is the long transmission range in combination with a low power consumption. Additive manufacturing (AM) is a versatile technology enabling new design approaches compared to conventional fabrication processes. Among the additive manufacturing technologies, material extrusion (MEX) is a widespread and straight forward AM technology [3]. Beside new designs, MEX enables also an integrated fabrication approach resulting in one-step fabrication of functional parts. For example, MEX enables fabrication of objects on existing structures [5]. Beside the mechanical combination of different structures, AM enables also the integration of functional elements [6]. Materials for MEX are constantly developed and span a wide area of different polymers. Conductive polymers are also available. These polymers consist of a matrix and embedded conductive particles (e.g., copper). These materials have already been tested for the fabrication of antennas [8, 9]. The aim of the paper is to additively fabricate antennas for IoT devices. The antenna should be finally integrated into the housing, resulting in a functional device.

Materials and methods

The housings were fabricated by material extrusion combining non-conductive and conductive filaments. Different antennas were designs focusing on a frequency of 868 MHz which is the transmission frequency of LoRa in Europe.

Antenna design

Two different antenna designs were evaluated: patch antenna and a near-isotropic arc antenna. The patch antenna was designed according to [4]. In general, the patch antenna consists of a top layer (fed) and a bottom (ground) layer. The dimensions and shape of near-isotropic arc antenna followed the work by to [1]. As described in the publication, the dimensions of the antenna were adapted to the target frequency of 868 MHz.

Antenna simulation

The behavior of both antenna types (patch and arc antenna) was simulated by MATLAB (R2022a, Mathworks). Frequency range was set from 500 MHz to 2.5 GHz. Within this range, 500 discrete frequencies were linearly distributed.

Fabrication of antenna

Reference antennas were fabricated by a standard etching process and by using copper foil which was cut to the dimensions of the different antennas.

Additive manufacturing of patch antenna

Test antennas were fabricated by a material extrusion process on a standard fused deposition modeling printer (Creality, Ender-3). The polymer (Electrifi, Multi3d) was deposited on a non-conductive substrate. The diameter of the print nozzle was 0.8 mm, print temperature was set to 190 °C, layer height was 0.2 mm, and print speed was set to 50 mm/s following the recommendations of the manufacturer. The final height of ground and patch plane were 1 mm. Patch antennas were connected by (i) a 50 Ω connector or (ii) direct soldering to the respective terminals.

Additive manufacturing of arc antenna

The arc antenna was fabricated in a two-step approach: at first, a frame was printed. The U-shaped frame included two separated grooves following the required dimensions. Secondly, the conductive material was deposited in the grooves of the frame. Different deposition strategies were followed: (a) deposition of conductive material (Electrifi, Multi3D) and (b) deposition by soldering. For (a), the branches were fabricated separately and finally inserted into the grooves of the frame. The thickness of the branches was set to 2 mm. In (b), solder (Fittingslot 3, Rothenberger Industrial) was deposited manually in the grooves of the frame. The solid solder wire was heated up to 230° C by a soldering iron (100 W Type, set to 420 °C, 3 mm wedge tip). The filling process is comparable to a hand-welding. Two different feeding points were tested: single-branch feeding and double-branch feeding. In the case of single-branch feeding, ground was connected with one branch. For double-branch feeding, both branches were connected to the feed with an open ground. The connection to the coaxial cable was realized either by embedding the wires into the printed material or soldering to the corresponding branches.

Antenna characterization

The return loss in dB of each antenna was measured by a vector network analyzer (NanoVNA V2, NanoRFE). Measurement frequencies were logarithmically distributed between 500 MHz and 2.5 GHz resulting in 200 discrete frequencies. Data visualization was done with MATLAB (R2022a, Mathworks).

Data transmission

The antennas were tested in a real-world scenario regarding data transmission. Therefore, a simple IoT system was used. The system consists of a sender (ESP32) and a transceiver (Raspberry Pi with LoRa concentrator module). The receiver was equipped with a fiberglass antenna (#916001, RAK) characterized with a gain of 8 dBi. The distance between sender and receiver was set to 100 m. The signal strength was determined by the measured received signal strength indication (RSSI) value (Fig. 1).

Fig. 1
figure 1

Printed patch antenna consisting of a feed layer (a) and a ground layer (b). Arc antenna printed (c) and soldered (d)

Results

Simulation

The reflection coefficient for both antennas is shown in Fig. 2. Both antennas exhibited a resonance frequency at the target frequency of 868 MHz (patch: 892 MHz, arc: 882 MHz). However, the reflection magnitude at the target frequency differed: the patch antenna exhibited a reflection magnitude of − 12.1 dB, whereas the arc antenna reached − 0.5 dB. In addition, the patch antenna showed a second resonance frequency at 2279 MHz (− 18.2 db). The arc antenna displayed an additional resonance at 1736 MHz (− 0.3 dB).

Fig. 2
figure 2

Simulation of resonance spectrum of path (gray line) and arc (black line) antenna

Fabrication

Fabrication of all antenna was feasible. The conductive filament was processable as described by the manufacturer. In case of the patch antenna, the conductive filament adhered to the substrate after roughening (see Fig. 1). For the arc antenna, the arms were fabricated separately and finally placed within the grooves of the frame. For the soldering process, the frame remained stable. However, a slight deformation occurred due to the resulting heat of the solder.

Measurement

The reflection coefficient of the different antennas was measured and is shown in Fig. 3.

Fig. 3
figure 3

Resonance spectra of all investigated antennas: reference (a) and printed (b) patch antenna, and soldered (c) and printed (d) arc antenna

Patch antenna

The frequency analysis of the different patch antennas is shown in Fig. 3a and b. The reference patch antenna exhibited the expected behavior as shown in the simulation. In the case of patch antenna with the 50 Ω connector, three resonance frequencies were detected (see Fig. 3a): 0.86 GHz (− 13.7 dB), 1.53 GHz (− 199 dB), and 2.25 GHz (− 15.3 dB). The peak at 1.53 GHz was not shown in the simulation. In contrast, the direct wired reference patch antenna exhibited the same peaks without the peak at 1.53 GHz. In contrast, the magnitude of the reflection was higher compared to the patch antenna with the 50 Ω connector (0.86 GHz: − 17.2 dB and 2.24 GHz: − 18.0 dB) (see Fig. 3a). The resonance spectrum of the printed patch antenna is shown in Fig. 3b. The printed patch antenna with the 50 Ω connector exhibited a frequency unspecific behavior without any resonance peak within the measured frequency range. In contrast, the direct wired printed patch antenna exhibited a dominant peak at 0.68 GHz (− 37.1 dB).

Arc antenna

The frequency spectrum of the different arc antennas is shown in Fig. 3c and d. The soldered arc antenna with single-feed exhibited peaks at different frequencies (see Fig. 3c): 0.55 GHz (− 13.2 dB), 0.71 GHz (− 16.0 dB), 1.45 GHz (− 23.0 dB), and 2.37 GHz (− 18.1 dB). None of the peaks matched with the target frequency of 0.868 GHz. In addition, the peaks were greater than the simulation results. The double-feed soldered antenna exhibited a different resonance pattern. Five peaks were detected (see Fig. 3c): 0.52 GHz (− 13.4 dB), 0.96 GHz (− 5.3 dB), 1.43 GHz (− 9.3 dB), 1.84 GHz (− 9.8 dB), and 2.25 GHz (− 7.7 dB). The printed arc antenna differed from the soldered antenna. The single-feed printed arc antenna exhibited three peaks (see Fig. 3d): 0.60 GHz (− 10.5 dB), 1.44 GHz (− 8.8 dB), and 2.44 GHz (− 16.75 dB). The double-feed arc antenna exhibited four peaks (see Fig. 3d): 0.54 GHz (− 8.6 dB), 1.04 GHz (− 3.3 dB), 1.48 GHz (− 5.8 dB), 1.91 GHz (− 6.5 dB), and 2.35 GHz (− 6.6 dB).

Measurement of data transmission

The antennas reference patch (direct wired), printed patch (direct wired), soldered arc (double-feed), and printed arc (double-feed) were tested for data transmission in the bandwidth for LoRa (868 MHz). All antennas succeeded in transmitting the payload to the receiver. The reference patch antenna (direct wired) exhibited a RSSI value of − 61.9 ± 0.6 followed by the printed patch antenna (direct wired) with a RSSI value of − 70.3 ± 0.5. The arc antenna exhibited RSSI values of − 72.3 ± 3.7 (soldered) and − 72.6 ± 2.4 (printed), respectively.

Discussion

Patch and arc antennas exhibit different characteristics regarding antenna behavior. Patch antennas are directional, whereas arc antennas are pseudo-omnidirectional. The directionality comes to the prices of the reflection magnitude as it was already shown in the simulation. The patch antenna exhibited a higher reflection magnitude compared to the arc antenna at the target frequency. The fabrication of the different antennas was feasible with a standard MEX printer. The conductive material required a lower printing temperature compared to standard materials (e.g., PLA). The adherence on smooth surfaces was increased by roughening the surface. The soldering process resulted in highly conductive branches of the arc antenna. The thermal influence on the frame is visible and might also cause a variance of the final antenna. Crucial for all antennas was the electrical contact. The reference patch antenna could be contacted by a standardized 50 Ω connector and also by direct wiring by soldering. The conductive filament was used to connect the terminal with the layers of the printed patch antenna. Manual inspection revealed that the connection was not reliable. Consequently, direct contacting by embedding the wire into the layers was more feasible. The arc antennas were contacted by soldering (soldered arc) or embedding of the wire in the branches of the arc (printed arc).

The resonance spectra of the antennas differed regarding design (patch vs arc), material (ref vs printed, soldered vs printed), and contacting method. The reference patch antenna exhibited the same peaks as the simulation. However, an additional peak was measured which might result from the 50 Ω connector. The printed patch antenna with the 50 Ω connector exhibited an unspecific resonance spectrum indicating high losses of the antenna. In contrast, the direct wired printed patch antenna exhibited a resonance frequency which was lower than the target frequency. The magnitude of the detected peak was greater than the simulation result. The soldered arc antenna exhibited peaks at different frequencies which were not matching the target frequency. The single-feed approach exhibited a higher magnitude than the double-feed approach. The printed arc antenna displayed various peaks within the selected range of frequencies. As also described for the soldered arc antenna, the single-feed approach led to higher magnitude of resonance. Compared to the simulation, the magnitude of the peaks of the real antennas was greater in all cases.

In a real-world test scenario, the antennas reference patch (direct wired), printed patch (direct wired), soldered arc (double-feed), and printed arc (double-feed) succeeded to submit the payload. To compare the antennas regarding of signal strength, the RSSI value was measured at a given distance between sender and receiver. The highest RSSI value was achieved by the reference patch antenna (direct wired). The printed patch antenna (direct wired) exhibited a higher RSSI value than the arc antennas which might by caused by the directivity of the patch antenna.

To conclude, additive manufacturing of antennas is a promising approach regarding integration of functionality into the housing of IoT devices. In regard to the antenna specificity, different parameters have to be considered: electrical connection, final design, and feeding approach. In addition, the resonance frequency does not reveal the directionality and transmission distance of the fabricated antennas. As an outlook, additional parameters have to be considered when fabricating antennas by additive means.