1. Introduction
Artificial satellites are watching the earth for global-scale wireless networking and scientific data gathering. With telecommunication and weather watches, satellites are indispensable to surveillance and reconnaissance on target regions. Nowadays, LEO satellites are produced for commercial services and deployed in the orbits 800 km to 1000 km above the Earth’s surface; this is sensational for the new concepts such as a short cycle of production, and forming a space-borne network of an overwhelming number of satellites called a constellation [
1,
2]. Transponders are essential to satellites of any kind and any missions for ground-to-space and satellite-to-satellite wireless links.
The transponder is a wireless communication system that receives RF signals and transmits them whether it is a geostationary orbit (GEO) or LEO satellite. Its configuration is represented by the system block diagram comprising plenty of circuits and components from the receiver to the transmitter. The chain of the signal flow meets the components of signal amplification, channel selection, channel splitting and combining, switching, demodulation and modulation and so forth. For successful tasking, system integration (SI) is important, but the functions of the individual components are more prioritized. Active components connected to control blocks have an increased degree of freedom in meeting the requirements even after SI. As for passive components such as input and output multiplexers (MUXes) for channel splitting and combining, and filters for channel selection, functions are not controllable after fabrication and assembly. Great care should be taken of their design and fabrication. While active components such as MMIC amplifiers do not weigh much, passive components of waveguide (WG) filters and WG MUXes tend to be heavy. Along with the feed horn and reflector antenna, WG passive elements are of great concern in terms of size control and weight control to build a satellite transponder. This becomes a critical matter to the cost required in the making and maintaining of LEO and micro satellites.
In the satellite communication, high-frequency signals must be so strong that they can travel hundreds to thousands of kilometers when emanated from the antenna. The filters, like other passive components in the feed assembly of the antenna, are made out of metallic waveguides to ensure high Q-factors and endure high power and the heat [
3,
4,
5,
6,
7]. This is why the structures are bulky and heavy. Size-reduction of them is a crucial factor to cost-saving in developing and launching satellites. Sarun and Somsak showed second- and third-order rectangular WG BPFs [
3]. Typical sized cavities are combined in series through irises for channel selectivity. Angel and Vincente took an action to reduce the size of the WG filter by integrating coaxial lines with rectangular cavities [
4]. The length of their third-order filter is the same as that of the conventional third-order BPFs. Joaquin and Santiago put stepped impedance resonator (SIR) parts with cavity resonators coupled through a mixture of E-plane and H-plane steps [
5]. Their WG BPF is longer than four wavelengths. Fernando and Jon meandered the straight topology of a BPF to reduce the area the filter occupies [
6]. The mitered E-plane bend-added waveguide sections couple half-wavelength cavities and the total length has almost no change. Valencia and Marco changed a straight geometry to a staircase by opening the two spots on the broad wall of each cavity for coupling [
7]. The trace in the longitudinal axis is decreased, but the total length is similar to reference 5. Looking over the latest rectangular waveguide bandpass filters including the reports above, as they follow the design technique of half-wavelength cavities, effective size reduction is nowhere to be found. One might suggest application of metamaterial filter designs seen from microstrip- and CPW-lines like in [
8,
9] to waveguide components, but this turns out to be not possible because of differences in geometries and modes. Motivation arises for setting up a metamaterial resonator which is much shorter than the typical cavity and geometrically appropriate for the waveguide.
This paper proposes a novel design method of realizing the compact WG BPF of a high order for satellite communication system devising and using the waveguide metamaterial resonator. First, the zeroth-order resonator is formed on the cross-section of the standard waveguide, compliant with the composite right/left-handed (CRLH) circuit model. The geometrical parameters are of unit of 100 μm to avoid the problems of fabrication tolerance, whose lower precision makes this design much tougher than others using the unit of 10 μm with more expensive facilities. Second, the ZOR as a building block for the equivalent network of the eighth-order BPF is substituted for conventional cavities. Third, the resonators are cascaded through short waveguide sections as coupling elements to generate the passband, and the stopbands have high attenuation. Fourth, the WG BPF of ZORs and coupling elements is physically prototyped by the CNC milling technique, which is driving the industry to transform into 3D printing as in [
10,
11,
12]. Fifth, the metallic WG filter is adopted to the feeding component of the horn antenna in the wireless communication system to attain channel selectivity. Sixth, the performances of the individual components and the filter-integrated antenna are investigated by way of real experiments. The suggested method is evaluated by circuit modeling, full-wave EM simulation and measurement. Good agreement between the theoretical and measured results is unveiled from the procedural steps. As noted in the required specifications on the WG BPF, the passband has about ≤1 dB and ≤−15 dB as insertion loss and reflection coefficient, the stopband has 40 dB of noise suppression. Regarding size reduction, the proposed PBF of the metallic waveguide is 3.5
for the eighth-order filtering, much smaller than the conventional waveguide filter of the same order. This effect will be obvious for much higher-order filters and MUXes.
3. Fabrication of the WG ZOR BPF and Test of the Prototype
The designed waveguide filter is fabricated and measured to validate the proposed method and geometry. Since the CNC milling technique is conducted for fabrication, taking into account that the end-mill tip cannot realize the aforementioned shapes of the cross-sections of the thin resonators 100%. The round corners appear instead of the sharp right-angle ones. Though the design has been done with the unit of 10
m as a coarse approach to ease the mechanical tolerance, the round corners are inevitable. The secondary procedure of design is done to keep the function of the WG BPF satisfactory, as in
Figure 6. This leads to the modified values for the geometrical parameters and the values are seen in
Table 7.
Despite the geometrical change in the front view of the flat metamaterial resonators, it is necessary to keep the frequency response compliant with the specifications. The corners become round
R and weaken the slot capacitance and the inductance on the edges of the metal patches and their intermediate strip. The rounded corner leads to lower capacitance due to a wider gap and lower inductance due to a shorter current path. This means at a microwave frequency band it degrades the initial performance of the BPF. In the realistic case,
Figure 3c and
Figure 6b are rendered as
Figure 7a–c. Some of the physical dimensions have to be fine to be in the unit of 100
m; there is no more being coarse in order to achieve the required frequency response. Setting up the structure in the electromagnetic analysis software with the values for the geometrical parameters, the transmission and reflection coefficients are obtained as in
Figure 7d,e, which meets the design requirement. This is very different from cavity filters presented by [
15,
16,
17] in terms of shape and length, and is physically realized as follows.
The aluminum ingot is carved into WG metamaterial resonators and coupling sections in the milling process as in
Figure 7a and they are pieced together to the eight-pole BPF as in
Figure 7b,c. The original structure becomes a little longer and wider because it has parts for mechanical assembly with bolts and a flange body for WG port connecting. However, the size of the core has the length of 3.5
for the eighth-order filtering. With this harness, the WG BPF is tested to see the frequency response. This experiment is conducted as in
Figure 8d, which produces S
11, and S
21 is as in
Figure 8e. The insertion loss and reflection coefficients are about −0.9 dB and −19 dB in the passband. The roll-off this manufactured metamaterial filter makes is satisfactory with attenuation of almost −40 dB. There occurs a discrepancy between the simulated and measured results that the frequency is shifted downward a bit. It is inferred that the error in R (round) formed by the end-mill affects the inductive part which causes the frequency shift in the first place, and secondly, connected pieces in the longitudinal direction do not tightly contact each other, making a tiny gap between metal rims with rotational misalignment. This proposed geometry as a guided component is applied to an antenna system as the electromagnetic radiation problem.
Figure 9 provides variation in L3 generated by the error of the rounded corner, which is critical to the change in the frequency response. For a quick simulation, a second-order case of the proposed filter is dealt with. L3 varied from 2.9 mm to 3.3 mm, mainly causing the frequency shift in S
11, S
21, S
11 and S
21 as in
Figure 9a–c, respectively.
The characteristics of the proposed filter and reference BPF structures are compared as in
Table 8. Most of all, the proposed filter has the shortest resonator as the WG metamaterial, which results in a good insertion loss from the complete structure at the length of 3.5
for a relatively low frequency, while [
5,
7] have lengths of around 4.7
and 8.5
, each for a relatively high frequency. If the proposed method is applied to 11 GHz, the total length is expected to be 3.5
, which is shorter than [
5] according to a quick estimation. This work and refs. [
5,
7] are high-order filters, giving high levels of attenuation in the stopbands, but [
17,
18,
19] take four cavities, showing poor noise-suppression effects. If [
15,
16,
17] are elongated to high-order filters, the lengths and insertion loss will be larger. Because the operation frequency of this work is much lower than others’ and has to use a WR-112 cross section as the largest, the 7.5 GHz filter might be the heaviest from
Table 1 when the same order is assumed for all the compared cases. Thus, the total length must be as small as possible, enabled by the metamaterial resonators. The proposed filter is compared with the non-metamaterial filter in terms of size and function.
4. Filter-Integrated Antenna Realized and Tested to Observe Channel Selectivity
The wireless communication system has transmitting (TX) and receiving (RX) antennas. Likewise, the satellite transponder has two antennas that are tested to see the RF-signal transfer and its reception between the opposite sides. The TX antenna is directly connected to the signal generator, which is practiced on a lot for checking the antenna-only function. Horn antennas are adopted for satellite wireless equipment most of the time, and because they have the characteristics of wide bands, frequency channels having several hundred megahertz of bandwidths are not defined for them. As channels are important in any communication, the roles of bandpass filters are highly in demand. Therefore, the proposed bandpass filter is inserted into the TX antenna system as below.
Figure 10a is the scheme of the ordinary RF-link test setup, while
Figure 10b depicts how the TX horn antenna is fed through the BPF from the signal generator. The shape of the horn antenna given in
Figure 10c is commonly used for wireless connectivity observation. It is R&S
®HF907 made by Rohde-Schwarz.
Figure 10d–g show its representative characteristics, in other words, S
11, Φ = 0° plane beam-pattern and Φ = 90° plane beam-pattern. These beam-patterns as the field-strength measured on the angular variation are side-data, and for the RF-link investigation, the RF transmission between the TX and RX sides on a straight line as in
Figure 10h expresses the field-strength. Like
Figure 10d, S
11 is below −10 dB over a very broad band as in
Figure 10i. S
21 as the RF transmission between the TX and RX antennas was measured over the distance ranging from 3 m (denoted as
do) to 6 m (as
dr), and has the feature of a broad band. Signals can be contaminated with noise and interference in that situation. Channel selectivity necessary for wireless communication is made possible by the following method foreseen in
Figure 10b.
Figure 11a includes a gray rectangle denoting the proposed metamaterial BPF as part of the feed of the TX horn antenna. The real experiment was conducted as in
Figure 11b. The field strength from the filter-incorporated antenna was measured by the RX horn antenna, which was plotted as
Figure 11c. Channel selectivity is accomplished as revealed by S
11 as the input reflection coefficient of the TX horn and S
21 as the signal transfer between both the sides. The two s-parameters have clear distinction of the passband and stopband like in
Figure 7. The antenna system is enabled to select the signals as required.
As an extension of the over-the-air test, since the simulation of the three-dimensional structure is time-consuming, simple numerical experiments are conducted.
Figure 12a has the frequency responses of the passband of the filter on the TX-side, which is combined with the antenna.
Figure 12b means the received signal has a plot presenting the frequency shift in S
21 as the result of the frequency shift occurring to the TX side. This kind of test makes the proposed work different from other suggestions [
18,
19], where the over-the-air transmission is not investigated.
Table 9 compares their works with this work from various view-points.
As in
Table 9, the features of the latest articles showing the filter combined with an antenna and this work are dealt with in the aspects of operation frequency, bandwidths, orders of the filter, waveguide cross-section sizes and lengths. The frequency and bandwidth of [
18] is relatively easy to achieve, but that of this work has the highest level of difficulty in design from the stand points of fractional bandwidth and skirt slope. The orders of filtering of [
18,
19] are much lower than that of this work. However, the proposed structure is the most compact in spite of the highest order of filtering. In other words, the size-reduction effect of the waveguide CRLH geometry is superior to that of the references.
5. Conclusions
A novel design method and geometry of the waveguide bandpass filter are suggested. Substantial size-reduction and excellent bandpass filtering functions are made possible by coming up with the waveguide CRLH resonator which leads to a very thin structure much shorter than the half-wavelength for the conventional cavities. The ZOR phenomenon is generated with the transverse geometrical parameters of the waveguide cross-section, unlike other metamaterials utilizing longitudinal line segments. The ZOR as the thin waveguide part does not block the incoming RF signal but passes it to the next ZOR. By cascading the ZORs through transmission-sections as the coupling elements, the passband becomes distinct with the steeper skirt in the stopband. An eighth-order ZOR BPF is designed and simulated, moving to the step of fabrication. It is manufactured into the aluminum waveguide filter. The prototyped BPF is measured and compared with the simulated result. As for the passband, insertion loss and reflection coefficient are around ≤1 dB and ≤−15 dB from simulation to measurement. The attenuation of ≥40 dB at 7 GHz and 8 GHz is achieved as desired in the specifications. The length of the WG ZOR BPF is 3.5 for the eighth-pole, but the conventional one has 5 even for the seven-pole case. Furthermore, the BPF was combined with the horn antenna in order to provide it with channel selectivity. The broad-band horn antenna is made possible to select the signals through the passband. The frequency responses are acceptable for the use of a satellite transponder. The proposed filters make the LEO- and scientific satellites much lighter with the weight of 54 g, which is greatly reduced from the conventional WG filters.