Vertical Waveguide-to-Microstrip Self-Diplexing Transition for Dual-Band Applications

This letter presents a novel vertical waveguide-to-microstrip self-diplexing transition for dual-band applications. The transition is realized with standard printed circuit board (PCB) manufacturing processing, making it suitable for mass production and practical applications. A standard waveguide is screwed on the topside of the stack-up. Dual-band self-diplexing operation is achieved by coupling two microstrips (one for each band) to two radiating patches through H-shaped slots. The operating bandwidth has been enhanced by adding two parasitic patches above the radiating ones. Metalized via holes are used to form a cage around the rectangular waveguide and the microstrips to prevent power leakage. A prototype has been fabricated to operate at K/Ka frequency band. The experimental results show a −10 dB matching bandwidth of 20% and 14% for the lower and upper bands, respectively. Within these ranges, the maximum measured insertion loss is about 0.6 and 0.7 dB, respectively.


I. INTRODUCTION
R ECTANGULAR metallic waveguides (RWGs) are important components in many microwave and millimeter-wave modules due to their high power handling and low insertion loss characteristics. For this reason, RWGs are widely used in satellite receivers, horn antenna feeding circuits, and high Q filters. To increase the scope of their use and to facilitate the integration with active devices, RWGs are often employed in combination with printed circuit boards (PCBs), thus leading to the integration of heterogeneous highperformance microwave or millimeter-wave systems. In this context, transitions from waveguide to planar transmission lines (e.g., microstrip lines or striplines) play a crucial role in avoiding performance degradation due to high-frequency interconnection losses.
With the emergence of millimeter-wave applications such as 5G or satellite communication, a high number of waveguideto-PCB transitions have been proposed in the literature. In typical configurations like the ones in [1] and [2], the waveguide is orthogonally connected to the microstrip ground plane. The transitions presented in both these works have broadband responses, but an additional short-circuited quarter-wavelength waveguide section is needed on the backside of the PCB to ensure proper operation. More recently, configurations, where the presence of the back-short was avoided, have been proposed by employing a gap-coupled parasitic patch antenna as a waveguide launcher [3]. However, these transitions require a complex 3-D housing (made of brass) to be placed on top of the PCB as reported also in [4]- [6]. A fully integrated configuration that does not require any accessories, and manual assembly is proposed in [7]. However, a complex PCB manufacturing process is required due to the need to produce metalized vias of three different heights. Furthermore, several longitudinal configurations are available in the literature like in [8]- [16] and also in [17]- [19] where multisection impedance transformers have been used. However, the multisection transformers make these configurations bulky. As well, top-side transitions with a vertical configuration are usually preferable over longitudinal ones; in fact, in typical PCB designs, the antenna is commonly located on the back layer, parallel to the RF circuit that is perpendicularly connected to the waveguide. While the larger part of the mentioned transitions provides good performances in terms of insertion loss and bandwidth, many of them require a high degree of mechanical complexity. In addition, none of the mentioned solutions is suitable for dual-band applications.
In this letter, a novel vertical waveguide-to-microstrip selfdiplexing transition [20], [21] is proposed. The solution is fully based on a standard multilayer PCB manufacturing process and does not require any matching structure in the rectangular waveguide. Unlike previous solutions, the proposed configuration can operate in a dual-band scenario. Our transition was optimized for the K/Ka frequency bands [22], [23] and it was conceived to provide good isolation between the two bands showing self-diplexing capabilities. The transition finds application in K/Ka-band satellite communication systems but its principles can be extended to any dual-band application whose frequency ranges lie within the waveguide operating bandwidth. Preliminary results of the proposed solution have been presented by Arnieri et al. [24]. To the best of the authors' knowledge, this is the first microstrip-to-waveguide transition presented in literature having dual-band capabilities.  Layer #4 excite two patches (one for each band) placed on Layer #2 through two H-shaped slots. A parasitic patch is added on Layer #1 above each patch to widen the operating bandwidth. As demonstrated in [25], two rectangular patch antennas in a stacked configuration have a large bandwidth region which occurs when the separation between the two layers is less than 0.15λ, and a relatively high gain region is obtained if the separation exceeds 0.3λ [26]. In our case, we have separated the two patches with a 0.914-mm thin substrate (h1 = 0.914 mm) having Dk = 2.33 (see Layer #1 in Fig. 1). Separation is then 0.09λ0 at 20 GHz and 0.14λ0 at 30 GHz fulfilling the large bandwidth condition at both bands (h1 < 0.15λ).

II. TRANSITION CONFIGURATION
If f 1 and f 2 are the central frequencies of the lower and upper bands, respectively, one patch is designed to resonate at f 1 , while the other one at f 2 . Each patch is excited by a separate feeding line. The operating principle of the transition is illustrated in Fig. 2. The quasi-TEM mode that propagates from one of the two input microstrip ports excites the TM 01 dominant mode of the corresponding patch, thanks to the slot etched in Layer #3. The patch is sized to resonate at frequency f 1 , while the stacked parasitic patch resonates at a frequency slightly different from f 1 . Finally, the patch TM 01 mode excites the TE 10 mode of the rectangular waveguide at f 1 as illustrated in Fig. 2. The same considerations are valid for the second microstrip that excites the corresponding patch at f 2 .
An SIW cavity [27], [28] has been created placing an array of metalized vias around the waveguide profile to avoid leakage from the waveguide and the excitation of traveling parallel-plate modes between the ground planes of the multilayer PCB. The size of the cavity was first estimated following [29]- [31] and then optimized with full-wave simulations [32].
Rogers RT/Duroid 5870 (Dk = 2.33) with a thickness h 1 = 0.914 mm is used under Layer #1. The same material with thickness h 2 = 0.254 mm is used to print Layer #2. Layer #3 is printed on Rogers Ro3035 with thickness h 3 = 0.25 mm. A PCB fusion technique that does not require prepregs has been used to fabricate the whole stackup. Through vias are drilled from Layers #1 to #4. The proposed structure can be viewed as a cavity-backed slot antenna that excites the TE 10 mode into the upper waveguide.

III. SIMULATED PERFORMANCE
The physical dimensions of the two patches in Layer #2 are selected to set the resonant frequency to f 1 for one patch and f 2 for the other. Thanks to the good isolation between the two bands, parameters related to the Tx port (L1, Ls1, Hs1, Ws1, Wp1, Lp1, Wt1, and Lt1) can be optimized independent of the ones related to the Rx port (Ls2, Hs2, Ws2, Wp2, Lp2, Wt2, and Lt2). The impedance matching between the microstrips and the waveguide depends on other parameters related to the slots and parasitic patches. The transitions have been analyzed with the Ansys HFSS simulator [32]. Optimized values are (Fig. 1

IV. EXPERIMENTAL RESULTS
The microstrip-to-waveguide transition has been manufactured and tested. Fig. 3 shows the fabricated prototype realized as a three-port network. Two Rosenberger 02K80F-40ML5 SMD connectors are placed on the lower side of the PCB at the end of the two microstriplines. One WR34 waveguide to coax adapter is directly screwed using its flange on the other side of the PCB. The measurement of the transmission coefficient from the waveguide to the Tx port is performed by closing  the Rx port with a matched load. Similarly, the transmission to the Tx port is performed by placing a matched load on the Rx connector. Results are visible in Fig. 4. The three-port configuration used in this prototype can be used to test the isolation between the two frequency bands. In this case, an S 12 measurement is performed when a matched load is connected to the waveguide port. Results are shown in Fig. 5 where measurements are compared with simulation.
As it can be seen from the above figures, the measured and simulated results are in reasonable agreement except for some differences, which may be related to connectors and the PCB fabrication process. Degradation of the in-band flatness in the transmission responses is mainly related to the used waveguide to coaxial adapters. The losses of the microstrip lines were evaluated by measuring lines of different lengths obtaining a propagation loss of 0.25 dB (Rx) and 0.6 dB (Tx) for 15-mm long lines. The waveguide to coaxial adapter (VT 260WCA2.92 KPC) and the Rosenberger 02K80F-40ML5 SMD connectors have both an insertion loss of about 0.2 dB. Stripping these losses from measured S 12 , it was found that the transition has an insertion loss of 0.6 dB in Rx and 0.7 dB in Tx. These values are the maximum insertion losses measured in the −10-dB matching bandwidth. Table I shows the performance of the dual-band transition proposed in this letter compared with state-of-the-art microstrip-to-waveguide transitions. Notice that dual-band operations are a peculiarity of the proposed solution, and for this reason, the other design included in the table are relevant to only single-band transitions. However, even considering the performance in each one of the two bands, the dual-band design compares well in terms of both bandwidth and insertion losses. A further advantage of the proposed solution is that compared to transitions included in Table I, it eases the integration in hybrid waveguide-PCB antenna designs. As an example, solutions [6] and [7] require a complex PCB manufacturing process due to the need to produce metalized vias of three different heights. Finally, configurations presented in [33] and [34] use a complex 3-D housing that requires a complex metal fabrication process.
V. CONCLUSION In this letter, a novel vertical waveguide-to-microstrip transition for dual-band applications has been proposed. The transition provides good insertion loss and bandwidth on the two frequency ranges considered comparing well to other single-band solutions reported in the recent literature. The isolation at the two printed ports is such that the transitions can be used in a Tx/Rx antenna feeding system. The proposed transition is realized with standard PCB manufacturing processing integrated with standard waveguides, thus making it suitable for mass production and practical applications.