Tunable Bandpass Filters With Co-Integrated Isolator and Switching-Off Functionality

This paper reports on a new type of multi-functional bandpass filtering (BPF) architecture with center frequency tunability, switching-off, and isolator functionality. The proposed filter, isolator, and switch (FIS) combines frequency-tunable passive resonators with unilateral frequency-selective stages (UFSs) to achieve tunable, fully-directional, quasi-elliptic transfer functions. The UFS, consists of a transistor-based path and a tunable feedback network, enabling frequency-selective transmission in the forward direction, RF signal cancellation in the reverse direction, and reconfigurability in terms of center frequency and intrinsic switching-off. It is demonstrated that by cascading UFSs and tunable resonators, highly selective frequency-tunable FISs can be developed. The proposed FIS also offers a reconfigurable switch-off mode capability with high isolation that is obtained by adjusting transmission zeros and reconfiguring the UFSs. In this paper, two types of UFSs are analyzed using circuit schematic simulations and coupling routing diagram (CRD)-based synthesis. Scalability to high-order transfer functions is also demonstrated by CRD-synthesized examples, design examples and manufactured prototypes. For experimental validation, two tunable UFSs and four tunable FISs were manufactured and measured at the L band.

SECTION I. Introduction

Unilateral RF components such as circulators or isolators are important counterparts of many instrumentation, radar and communication systems. They are used to protect high-power RF sources from unwanted reflections or to separate the transmit (Tx) from the receiving (Rx) operation. With the emergence of joint communication and sensing [1], in-band full-duplex [2] and open radio access networks [3], circulators/isolators are expected to be the most required RF component technology for the antenna front-end interface of these systems due to having to deal with high levels of self-interference and Rx-Tx leakage. However, current circulator/isolator technologies aren't suitable for their RF transceivers due to their size and power consumption being prohibitively large. Another limitation is that they are frequency static and don't have the ability to suppress out-of-band interference in interference-dominant communication environments requiring additional RF filtering to be incorporated in their RF front-ends.

Conventional RF isolators/circulator implementations, as for example the ones shown in [4], [5] are based on ferrite materials that require large magnets to be incorporated above and below the RF component. As such, they can't be integrated with the rest of the front-end semiconductor circuitry due to their large size and process manufacturing incompatibility. To address these size constraints and pave the way to IC integration, non-magnetic unilateral components are nowadays being investigated. These include transistor-based [6], [7], [8], [9], and spatiotemporally modulated (STM) [10], [11], [12] isolators/circulators. The transistor-based concepts exploit the unilateral properties of the transistors biased with DC voltage to incorporate directionality within the RF component. In all of the presented STM architectures, the passive resonators are modulated with phase-shifted low-frequency RF signals breaking the time-reversal symmetry. Although these concepts address the size constraints in a way, STM-based approaches, for example the ones shown in [10] are limited to low frequencies of operation (e.g., 200–1000 MHz for STM-based isolators/circulators). Furthermore, many transistor-based approaches are used to functionalize a single RF component (e.g., a circulator [6] or isolator [8]), which isn't in line with the requirement for the RF components to be multi-functional.

For further miniaturization, RF co-design techniques are recently being explored to integrate multiple RF functions within the same RF device, thus reducing the size of the RF front-end. Particularly for RF filters, co-design techniques are explored to co-integrate them with couplers [13], power divider [14], amplifiers [15], [16], [17], isolators [10], [18], [19], [20] circulators [21], [22], [23] phase shifters [24], [25], [26], [27], [28], attenuators [29] and antennas [30], [31], [32], [33], [34]. In particular, the co-integration of isolators and filters has been gaining significant attention due to their enabling capabilities for in-band full-duplex. Alternative design and integration techniques have been presented, including transistor-based configurations and STM resonators. Specifically, the transistor-based filtering isolators have demonstrated decent performance in terms of gain (e.g., 4.2-5.8 dB) and isolation (e.g., 44-71 dB) in [35]. However, their transfer functions in the forward direction are mostly frequency static (e.g., as shown in [35], [36], [37], [38]) and on many occasions they consume large DC power of 91-228.6 mW as shown in [37], [39], [40]. On the other hand, the STM-based approaches can be made tunable without consuming any DC power. They can be implemented utilizing lumped elements [10], [41] or microstrip integration schemes [42], [43], [44]. However, they have only been demonstrated for frequencies between 0.2 GHz [18] and 1.5 GHz [43], with the majority being implemented for frequencies <500 MHz [45]. Furthermore, they exhibit low levels of isolation (8-40 dB) in their reverse direction and high levels of insertion loss between 1.5-6 dB in the forward direction [46], [47].

Given the aforementioned limitations, this paper introduces a new class of frequency-selective unilateral RF components that combine filter, isolator, and switch (FIS) functionalities. Unlike traditional methods that use discrete RF components, this co-design approach integrates multiple functions into a single device, reducing size and complexity. This concept also enables the creation of highly selective, frequency-tunable transfer functions that can be switched off, offering significant advantages over conventional RF front-end designs.

The proposed FIS concept utilizes tunable unilateral frequency-selective stages (UFSs) combined with tunable passive resonators and mixed electromagnetic (EM) couplings. The UFSs offer frequency selectivity in the forward signal direction while it blocks RF signal propagation in its reverse direction with high levels of rejection. While the UFS implementations in [35], [36], [37] focus on frequency-static designs, this work introduces unique UFSs concepts with multi-levels of reconfigurability. A preliminary demonstration of the tunable UFS concept has been briefly introduced in [48] and is further expanded in this work in terms of: i) extended modeling and analysis using a comprehensive filter-based design approach using coupling routing diagrams (CRDs), ii) scalability to the realization of higher-order filter designs with multiple UFSs aiming for improved gain and directivity, iii) realization of tunable quasi-elliptic transfer functions using mixed EM coupling techniques, and intrinsic switching-off reconfigurability and iv) the introduction of a completely new resonator-based UFS with enhanced out-of-band rejection.

The content of this paper is organized as follows. In Section II, the theoretical foundations of two types of reconfigurable UFSs and their co-designed FIS are presented through circuit-based and synthesized examples using CRDs. The experimental validation of the concept is discussed in Section III alongside a comparison with the state-of-the-art. Lastly, the main contributions of this work are summarized in Section IV.

SECTION II. Theoretical Foundations

This section discusses the theoretical foundations and operating principles of the co-designed FIS concept, illustrated in Fig. 1, which includes its CRD [Fig. 1(a)] and conceptual power transmission and isolation responses [Fig. 1(b)]. The FIS integrates tunable UFSs, passive resonators, multi-resonant stages, and mixed EM couplings to reduce RF front-end size and complexity by combining the functions of a tunable bandpass filter (BPF), RF isolator, and RF switch into a single multi-functional component. In this structure, selectivity in the forward direction can be enhanced by increasing the number of resonators, while reverse isolation improves with more UFSs. Out-of-band rejection can be enhanced by incorporating multi-resonant stages or mixed EM couplings to create transmission zeros (TZs). By making each stage tunable, reconfigurability can be incorporated into the FIS. Furthermore, by moving the TZs into the operating band, intrinsic switching off can be achieved. These are discussed in the next subsections. Specifically, Section II-A introduces the generalized UFS concept, Sections II-B and II-C detail the operating principles of two novel tunable UFS configurations (UFS-I and UFS-II) using circuit simulations and synthesized transfer functions, and Section II-D demonstrates the scalability of the concept to high-order quasi-elliptic transfer functions through CRD synthesis and circuit simulations.

Figure 1.

(a) N-stage co-designed reconfigurable FIS CRD based on different types of tunable resonant cells, namely UFSs denoted by a grey circle with a triangle, frequency-tunable passive resonators denoted by a black circle, and multi-resonant stages comprising two frequency-tunable resonators (black circles) and a non-resonating node (grey circle), and (b) its block diagram and conceptual tunable capabilities in terms of center-frequency tuning and switching off.

A. Unilateral Frequency-Selective Stage Basis

The block diagram of the generalized UFS concept is shown in Fig. 2(a). It features two parallel RF signal paths: a transistor-based path (the transistor is reversely placed) with constant S-parameters across frequencies [see the amplitude in Fig. 2(b) and parameters in the caption of Fig. 2] and a feedback path that is frequency-dependent and introduces a phase delay. The feedback path is modelled using a phase shifting element between two attenuating elements, with specific parameters provided in the caption of Fig. 2. At the center frequency, the transistor path has a phase of 84.3° in the forward direction (from P2 to P1), while the feedback path exhibits a phase of −90.5°. Due to the transistor's unilateral properties, the UFS exhibits different power transmission behaviours depending on the direction (from P1 to P2 and from P2 to P1) as discussed in [35]. In the reverse direction (P2 to P1), signal cancellation occurs when the RF signals from the transistor and the feedback path meet at P1 with a phase difference near 180° at the center frequency (f_cen), effectively cancelling each other, as indicated by the red dashed arrows in Fig. 2(a) and the red dashed line in Fig. 2(b). In the forward direction (P1 to P2), the RF signal from P1 reaches P2 through the feedback path (the weak signal passing through the unilateral transistor-based path from P1 to P2 can be ignored here) or a loop formed by the feedback in series with transistor paths. This loop results in a zero-phase difference, causing the signals at P2 to add in phase, as shown by the blue solid arrows in Fig. 2(a) and the red solid line in Fig. 2(b).

Figure 2.

(a) Block diagram of the generalized UFS. Transistor-based path S-parameters at 1.7 GHz: S₁₁ = 0.193∠−1.3°, S₂₁ = 0.007∠36°, S₁₂ = 1.23∠84.3°, S₂₂ = 0.217∠−30.4°. Feedback: Attenuating element loss = 0.2 dB; phase at 1.7 GHz = −90.5°, phase slope = −200°/freq. octave. (b) Comparison of power transmission and isolation responses between the UFS and its single transistor path. (c) Tunability of UFS: the feedback in Case 2 has an 88° phase lead over that in Case 1.

Compared to a single transistor path, the UFS can theoretically achieve infinite isolation at f_cen, as shown in Fig. 2(b). As the operating frequency moves away from f_cen, the balance between the two paths weakens, resulting in a frequency-selective power transmission response in the forward direction. By adjusting the phase of the feedback, it is possible to construct the signal in the forward direction and cancel it in the reverse direction at a different frequency, enabling center-frequency tunability, as demonstrated in Fig. 2(c). Two different types of feedback networks allow for alternative transfer functions, which will be discussed in the next sections for two UFS topologies: UFS-I (coupled-lines-based feedback) and UFS-II (resonator-based feedback).

B. Unilateral Frequency-Selective Stage Using a Coupled-Line Feedback: UFS-I

Design concept: The circuit schematic and corresponding responses for the first UFS (UFS-I) with coupled-line feedback are depicted in Fig. 3(a) and (b) and are designed for an operating frequency of 1.7 GHz. The transistor-based path consists of an NPN transistor (BFU760F) in common emitter configuration, along with resistors R₁ = 63.5 Ω, R₂ = 18 Ω, R₃ = 58 Ω. The initial values of these resistors are chosen to achieve impedance matching, gain attenuation (to prevent oscillation in the transistor-feedback loop), and a phase lead of approximately 90° at 1.7 GHz in the transistor path. Following the setup of the transistor path, feedback is introduced in parallel, consisting of a coupled line loaded with capacitors. This feedback is designed to have an equivalent electrical length of approximately 90° at 1.7 GHz to provide a reverse phase with transistor-based path. With capacitors loaded, the physical length of coupled line is reduced, resulting in the parameters of feedback with Z_E = 163 Ω, Z_O = 64 Ω, l = 36° at 1.7 GHz, C_f1 = 1.14 pF, C_f2 = 1.1 pF.

Figure 3.

(a) Circuit schematic of the tunable UFS-I based on tunable coupled-line feedback. T: BFU760F, V_b = 0.78 V, V_c = 0.8 V, R₁ = 63.5 Ω, R₂ = 18 Ω, R₃ = 58 Ω, Z_E = 163 Ω, Z_O = 64 Ω, l = 36° at 1.7 GHz, C_f1 = 1.14 pF, C_f2 = 1.1 pF. (b) Circuit-simulated responses of UFS-I operating at 1.7 GHz.

With the given parameter settings of UFS, the coupled-line path exhibits a phase of −89°, while the transistor path has a phase of +84° with balanced amplitude at the center frequency (1.7 GHz) [see Fig. 4]. Therefore, RF signals: i) constructively add in the forward direction (P1 to P2) due to the zero-phase loop formed by the feedback and transistor paths, and ii) destructively add in the reverse direction (P2 to P1) due to the nearly 180° phase difference between the two parallel paths in this direction, as discussed in Section II-A. Additionally, the signal transmission (S₂₁) of UFS-I is weaker at lower frequencies than at higher frequencies. This results from the amplitude of the feedback varying more rapidly at lower frequencies, which causes the balance between the coupled-line and transistor-based paths to deteriorate more quickly. This behavior contributes to the observed asymmetry in the transmission coefficient for UFS-I.

Figure 4.

Circuit-simulated amplitude and phase response of the transistor-based and the coupled-line path for the realization of UFS-I at 1.7 GHz.

Design trade-offs: When determining the parameters of the UFS, it's important to consider trade-offs among directivity, impedance matching, stability, gain, noise Figure (NF), and linearity. Various parametric studies, summarized in Figs. 5–​8, address these trade-offs.

Figure 5.

Circuit-simulated directivity (D), power transmission response (|S₂₁|), power reflection response (max (|S₁₁|, |S₂₂|)), and stability factor (μ) of the UFS-I at 1.7 GHz when varying R₁ and R₂, and fixing R₃ with (a) R₃ = 58 Ω and (b) R₃ = 68 Ω.

Figure 6.

Circuit-simulated transmission (|S₂₁|), isolation (|S₁₂|), reflection (|S₁₁|, |S₂₂|) response, stability factor (μ), gain (G), NF, and IIP3 performance of the UFS-I when using the variable datasets in Table 1 around a variable R₃.

Figure 7.

Circuit-simulated transmission (|S₂₁|), isolation (|S₁₂|), reflection (|S₁₁|, |S₂₂|) response, stability factor (µ), gain (G), NF, and IIP3 performance of the UFS-I when using the variable datasets in Table 2 around a variable V_c.

Figure 8.

Circuit-simulated transmission (|S₂₁|), isolation (|S₁₂|), reflection (|S₁₁|, |S₂₂|) response, stability factor (μ), gain (G), NF, and IIP3 performance of the UFS-I when using the variable datasets in in Table 3 around a variable C.

Fig. 5 illustrates the simulated directivity (D), transmission (|S₂₁|), reflection(|S₁₁|&|S₂₂|), and stability factor (μ) of UFS-I at 1.7 GHz, with variations in R₁ and R₂ while keeping R₃ fixed. Different values of R₃ require specific R₁ and R₂ combinations for optimal directivity, as shown in the first column of Fig. 5. As R₃ increases (datasets in Table 1, performances in Fig. 6), NF and gain deteriorate, while the input third-order intercept point (IIP3) improves. Additionally, the resistor set (R₁, R₂, R₃) affects input and output impedance matching, as reflected in |S₁₁| and |S₂₂| in Fig. 6. The chosen values for a balanced impedance match are R₁ = 63.5 Ω, R₂ = 18 Ω, and R₃ = 58 Ω. Increasing the transistor's control voltage (V_c) (datasets in Table 2, performances in Fig. 7) enhances gain, IIP3, but worsens input and output impedance matching and NF. Increasing the coupling factor (C) of the coupled-line feedback (datasets in Table 3, performances in Fig. 8) reduces selectivity and gain in the forward direction but improves NF and IIP3.

TABLE 1 UFS-I Examples: R3 Varying Dataset (Vb = 0.78 V, Vc = 0.8 V, C, L@1.7 GHz = 0.436, 36°, Cf1 = 1.14 pF, Cf2 = 1.1 pF)

TABLE 2 UFS-I Examples: Vc Varying Dataset (Vb = 0.78 V, C, L@1.7 GHz = 0.436, 36°, Cf1 = 1.14 pF, Cf2 = 1.1 pF)

TABLE 3 UFS-I Examples: C Varying Dataset (Vb = 0.78 V, Vc = 0.8 V, Cf1 = 1.14 pF, Cf2 = 1.1 pF)

Throughout these variations, unconditional stability in UFS is maintained. Furthermore, it is found that IIP3 performance can be improved by raising the transistor's bias voltage, though this increases power consumption. Alternatively, increasing the coupling factor of the coupled-line feedback can enhance IIP3, but requires tighter spacing or narrower transmission lines, demanding higher fabrication precision.

Coupling routing diagrams (CRDs): The response of UFS-I can be modelled by a CRD shaped by one resonating node (node 3’) and three non-resonating nodes (nodes 1’, 2’, 4’) [35], as shown in Fig. 9(a). The lower path (nodes 1’-4’) represents the transistor-based path, where the non-reciprocal coupling element between nodes 2’-3’ represents the non-reciprocity of the transistor path. Node 3’ is set as a resonating node, contributing to the frequency selectivity of UFS-I. The upper path (coupling between nodes 1’-4’) represents the coupled-line feedback network. The initial coupling element values (m_1’2’, m_3’4’, m_2’3’, m_3’2’, R_2’2’, R_3’3’, R_1’1’ and R_4’4’) in the transistor-based path are calculated using the design method in [35], [37]. The Y-parameters for the transistor-based path at a center frequency of 1.7 GHz are obtained through Advanced Design System (ADS) simulation, with values as follows: Y₁₁ = 0.014+j3.067*10⁻⁴, Y₁₂ = 2.13*10⁻⁴-j0.033, Y₂₁ = −1.787*10⁻⁴-j1.5*10⁻⁴, Y₂₂ = 0.014+j0.003. m_1’4’ representing the feedback is chosen to be 1.7 to give the same Y₂₁ response as the actual coupled-line feedback at 1.7 GHz, as shown in Fig. 9(b). With these initial coupling values in UFS-I established and listed in Fig. 9(c), the synthesized S-parameters from CRD are derived using (1)–​(4) [49], with its corresponding responses shown in Fig. 9(c). \begin{align*} {{S}_{21}} =& 2\left[ {{\bm{R}} + s{\bm{U}} + j{\bm{m}}} \right]_{n + 2,1}^{ - 1} \tag{1}\\ {{S}_{12}} =& 2\left[ {{\bm{R}} + s{\bm{U}} + j{\bm{m}}} \right]_{1,n + 2}^{ - 1} \tag{2}\\ {{S}_{11}} =& 1 - 2\left[ {{\bm{R}} + s{\bm{U}} + j{\bm{m}}} \right]_{11}^{ - 1} \tag{3}\\ {{S}_{22}} =& 1 - 2\left[ {{\bm{R}} + s{\bm{U}} + j{\bm{m}}} \right]_{n + 2,n + 2}^{ - 1} \tag{4} \end{align*} where [m] is the coupling matrix, s=jΩ, Ω is the normalized frequency, [R] represents the node impedances, and [U] is similar to the (n + 2) × (n + 2) identity matrix, except that [U]_k,k = 0 for source, load, and the non-resonating node k. These CRD-synthesized responses demonstrate expected signal construction (S₂₁) and signal cancellation (S₁₂) at the target center frequency, though slight discrepancies (such as BW) with the circuit simulation are observed. This difference arises because the amplitude response of the feedback from circuit simulation is frequency-dependent, whereas the CRD-based response is frequency-independent. Consequently, optimization of the coupling values is necessary. This optimization can be performed manually or through an algorithm (Simulated Annealing Algorithm is used for this paper) that minimizes the difference between the CRD-synthesized and circuit-simulated S-parameters at both the center frequency and the 3 dB bandwidth. After optimization, the final CRD-synthesized responses are shown in Fig. 9(d) with its coupling values listed in the caption of Fig. 9(d), forming the foundation for synthesizing later on high-order transfer functions in the co-designed FIS.

Figure 9.

(a) CRD of the UFS-I in Fig. 3(a). (b) Y-parameter amplitude and phase response of the feedback using CRD (m_1’4’ = 1.7) and circuit simulation. (c) Comparison between circuit-simulated and CRD- synthesized responses of UFS-I using the initial coupling values: m_01’ = 1, m_4’5 = 1, m_1’2’ = 1, m_3’4’ = 1, m_2’3’ = 3.1736-j0.62, m_3’2’ = 0.0109-j0.02, m_1’4’ = 1.7, R_1’1’ = −j0.0415, R_2’2’ = 1.4028+j0.0164, R_3’3’ = 1.3506-j0.2423, R_4’4’ = −j0.15. (d) Comparison between circuit-simulated and CRD-synthesized responses of UFS-I using the optimized coupling values: m_01’ = 1, m_4’5 = 1, m_1’2’ = 1.0747+j0.1815, m_3’4’ = 1.223+j0.2771, m_2’3’ = 3.9424-j0.7168, m_3’2’ = 0.0938+j0.0933, m_1’4’ = 1.7866+j0.0697, R_1’1’ = −0.0479-j0.0501, R_2’2’ = 1.3723+j0.0935, R_3’3’ = 1.8896+j0.0411, R_4’4’ = 0.0236-j0.0665.

Tunability: As demonstrated in Fig. 2(c), center-frequency tunability in the generalized UFS is achieved by adjusting the phase of the feedback network when the transistor path is set with constant S-parameters over frequency. However, in practical implementations, the amplitude and phase of the transistor path are frequency-dependent [see Fig. 4]. If only the feedback path is tuned without adjusting the transistor path, the RF signals from the two paths may not cancel each other effectively at the desired frequency, leading to degraded isolation during center-frequency tuning. Therefore, both the transistor path and the feedback need to be tuned for center-frequency tunability. Phase tuning of the feedback is accomplished by synchronously tuning two capacitors (C_f1 and C_f2), which alter the feedback's equivalent electrical length (it should provide approximately 90° phase delay at the desired frequency). The optimal capacitance values can be found by observing the phase of the feedback during capacitance tuning or through calculations using equations in [50], which provide a detailed Z-matrix of the coupled line terminated with arbitrary impedances. Following feedback tuning, the transistor path is adjusted by altering the transistor bias voltages to achieve signal cancellation for S₁₂ of UFS-I at the new desired frequency.

To illustrate this center-frequency tunability, Fig. 10(a) presents the amplitude and phase of the two parallel paths in UFS-I for three tuning cases. It is evident that the phase shift of the feedback (∠S_AB) significantly contributes to the center-frequency tunability of the UFS, while minor adjustments in the transistor path are also required to achieve an optimal balance of phase at the new operating frequency. The tuning parameters and tunable responses of UFS-I are detailed in the captions of Fig. 10 and (b).

Figure 10.

(a) Circuit-simulated amplitude and phase response of the transistor-based and the coupled-line path for the realization of frequency tuning of UFS-I. (AB: feedback. DC: transistor-based path). (b) Circuit- simulated responses of the reconfigurable UFS-I. Case 1.5 GHz: V_b = 0.785 V, V_c = 0.56 V, C_f1 =1.49pF, C_f2 = 1.44 pF; Case 1.9 GHz: V_b = 0.775 V, V_c = 1.4 V, C_f1 =0.85 pF, C_f2 = 0.81 pF; Case switching-off: V_b = 0 V, V_c = 0 V, C_f1 =0.4 pF, C_f2 = 2.5 pF.

Furthermore, UFS-I can be intrinsically switched-off by turning off the transistor and increasing the difference between the two loaded capacitances in the coupled-line feedback. When the transistor is turned off, the signal in the transistor path is significantly attenuated. In the coupled-line feedback, increasing the capacitance (C_f1 and C_f2) differential increases signal attenuation, which can be derived from the S-parameter calculations of the coupled-line network modeled in [50]. Ideally, setting one capacitance to 0 pF and the other to infinity—equivalent to one end being open-circuited and the other short-circuited—would fully block signal transmission through the coupled line feedback path. When neither of the parallel paths transmits signals effectively, UFS-I's intrinsic switch-off functionality is realized. However, in practical implementation, the feedback capacitors are realized using varactors, whose limited tuning range constrains the achievable isolation during the UFS switch-off mode. To get closer to an actual implementation scenario, the two capacitances in the coupled-line feedback are asynchronously tuned to 0.4 pF and 2.5 pF in circuit simulations, resulting in the UFS achieving over 10 dB isolation across the 1.1 GHz to 2.3 GHz range, as shown in Fig. 10(b).

C. Unilateral Frequency-Selective Stage Using Multi-Resonant Feedback: UFS-II

Design concept: This section introduces UFS-II, a new configuration featuring a multi-resonant feedback network in parallel with a transistor-based path (similar to the one used in UFS-I), as shown in Fig. 11(a). The feedback comprises two attenuating resistors (R_a), two long transmission line sections (characteristic impedances Z_f and electric lengths l_f), and a multi-resonant cell shaped by two short transmission line sections (characteristic impedances Z₀ and electric lengths l₁ and l₂), an inductor (L_r) and a capacitor (C_r). Similar to the design of UFS-I, the transistor path and the feedback network should achieve the proper amplitude and phase balance. The resistors in the transistor-based path are designed as R₁ = 54 Ω, R₂ = 14 Ω, R₃ = 48 Ω, resulting in a positive phase of +86° at the center frequency of 1.7 GHz [see Fig. 12], which indicates that the feedback network needs to have a corresponding reverse phase response.

Figure 11.

(a) Circuit schematic of the tunable UFS-II using resonator- based feedback. T: BFU760F, V_b = 0.78 V, V_c = 0.9 V, R₁ = 54 Ω, R₂ = 14 Ω, R₃ = 48 Ω, Z_f = 50 Ω, l_f = 22° at 1.7 GHz, L_r = 1 nH, l₁ = l₂ = 17° at 1.7 GHz, R_a = 5 Ω, C_r = 2.8 pF. (b) Circuit-simulated S-parameters of UFS-II at 1.7 GHz.

Figure 12.

Circuit-simulated S-parameter amplitude and phase response of the transistor- and the resonator-based paths for the realization of the UFS-II at 1.7 GHz.

The phase of the multi-resonant feedback is contributed by the phases of LC multi-resonant cell and two transmission line sections. The multi-resonant cell can create one TZ at a frequency (f_TZ) and one pole at a frequency (f_p) that can be specified by solving (5) and (6) respectively: \begin{align*} {{Z}_{in}} =& {{Z}_0}\frac{{\frac{1}{{j{{\omega }_{TZ}}{{C}_r}}} + j{{Z}_0}\tan \left( {{{\beta }_{TZ}}{{l}_1}@{{f}_{TZ}}} \right)}}{{{{Z}_0} + j\frac{1}{{j{{\omega }_{TZ}}{{C}_r}}}\tan \left( {{{\beta }_{TZ}}{{l}_1}@{{f}_{TZ}}} \right)}} = 0 \tag{5}\\ {{Y}_{in}}\!=\!\!& \frac{1}{{{{Z}_0}\frac{{\frac{1}{{j{{\omega }_p}{{C}_r}}} + j{{Z}_0}\tan \left( {{{\beta }_p}{{l}_1}@{{f}_p}} \right)}}{{{{Z}_0} \!+ j\frac{1}{{j{{\omega }_p}{{C}_r}}}\tan \left( {{{\beta }_p}{{l}_1}@{{f}_p}} \right)}}}} \!+\! \frac{1}{{{{Z}_0}\frac{{j{{\omega }_p}{{L}_r}\! + j{{Z}_0}\tan \left( {{{\beta }_p}{{l}_2}@{{f}_p}} \right)}}{{{{Z}_0} + j \times j{{\omega }_p}{{L}_r}\tan \left( {{{\beta }_p}{{l}_2}@{{f}_p}} \right)}}}} \\ =& 0 \tag{6} \end{align*} where ω_TZ and ω_p are the angular frequencies of the TZ and the pole, β_TZ and β_p are the phase constants. The multi-resonant cell is designed with a pole at 1.42 GHz and a TZ at 2.47 GHz, providing a phase delay of 38° at 1.7 GHz. Its design parameters are as follows: C_r = 2.8 pF, L_r = 1 nH, l₁ = l₂ = 17° at 1.7 GHz, Z₀ = 50 Ω. To achieve a phase delay of −82° at 1.7 GHz in the feedback, two transmission line sections of 22° are added in series with the multi-resonant cell. The resulting phase of the feedback is illustrated in Fig. 12. Additionally, the R_a values are set to be 5 Ω to ensure unconditional stability for UFS. Finally, the S-parameters for UFS-II [Fig. 11(b)] show a passband at 1.7 GHz and a TZ at 2.47 GHz. UFS-II follows the similar design trade-offs process as UFS-I; therefore, these details are omitted for brevity.

A key advantage of UFS-II over UFS-I is that the TZ from the multi-resonant cell is transferred to the UFS-II's transfer function (this can be derived using Y-parameters of the parallel paths), enhancing the out-of-band rejection of the UFS. Additionally, the bandwidth of UFS-II can be tuned by simply adjusting the bandwidth of the multi-resonant cell in the feedback, as shown in Fig. 13. In Case 2, where the multi-resonant cell has a narrower bandwidth compared to Case 1, the feedback has more significant amplitude and phase changes, leading to a narrower bandwidth of UFS-II.

Figure 13.

Design cases of UFS based on resonator feedback. Case 1: parameters are listed in the caption of Fig. 11. Case 2: L_r = 0.02 nH, C_r = 3.4 pF, the rest parameters remain the same.

Coupling routing diagrams (CRDs): The response of UFS-II can be modelled using the CRD illustrated in Fig. 14(a). In this CRD: The lower path (nodes 1’-4’) represents the transistor-based path, with a non-reciprocal coupling element between nodes 2’-3’ indicating the transistor's non-reciprocity. The upper path (nodes 5’-7’) represents the multi-resonant feedback, introducing one pole and one TZ. Unlike UFS-I, where a resonating node is present in the transistor path, UFS-II's CRD does not require resonating nodes for the transistor path, which is because the frequency selectivity in the CRD response of UFS-II is already provided by the resonating nodes in the feedback path. The determination of the coupling values for UFS-II follows a similar methodology to that of UFS-I. Initially, the coupling element values (m_1’2’, m_3’4’, m_2’3’, m_3’2’, R_2’2’, R_3’3’, R_1’1’ and R_4’4’) in the transistor-based path are calculated using the design methods outlined in references [35] and [37]. Then the coupling element values for the multi-resonant feedback are selected to achieve the same Y₂₁ response as the circuit simulation at the center frequency, as demonstrated in Fig. 14(b), where the corresponding coupling values are listed in the caption. After establishing the initial coupling values for both the transistor-based path and the feedback network of UFS-II, a comparison between the circuit-simulated and CRD-synthesized responses is presented in Fig. 14(c). This comparison reveals the correct TZ position; however, signal construction in the forward direction and signal cancellation in the reverse direction occur at a slightly deviated frequency compared to the circuit simulation. To enhance consistency between the CRD-synthesized response and the circuit simulation, these coupling values are optimized and then listed in the caption of Fig. 14(d). A comparison between circuit simulations and the CRD synthesized responses, shown in Fig. 14(d), demonstrates good agreement, which confirms that the proposed CRD for UFS-II is effective for synthesizing high-order FIS.

Figure 14.

(a) CRD of the UFS-II in Fig. 11(a). (b) Y-parameter amplitude and phase response of the feedback using CRD (m_1’5’ = −2.5, m_5’6’ = 2.4, m_6’7’ = 1.94, m_4’6’ = 1.6, m_5’5’ = 2.7, m_7’7’ = −5.5) and circuit simulation. (c) Comparison between circuit-simulated and CRD-synthesized responses of UFS-II using the initial coupling values: m_01’ = 1, m_4’8 = 1, m_1’2’ = 1, m_3’4’ = 1, m_2’3’ = 3.0257-j0.7361, m_3’2’ = 0.008-j0.0151, m_1’5’ = −2.5, m_5’6’ = 2.4, m_6’7’ = 1.94, m_4’6’ = 1.6, m_5’5’ = 2.7, m_7’7’ = −5.5, R_1’1’ = 0.45-j0.35, R_2’2’ = 1.2274+j0.0101, R_3’3’ = 1.1897-j0.3579, R_4’4’ = 0.45-j0.35. (d) Comparison between circuit-simulated and CRD-synthesized responses of UFS-II using the optimized coupling values: m_01’ = 1, m_4’8 = 1, m_1’2’ = 0.576+j0.043, m_3’4’ = 1.2-j0.023, m_2’3’ = 2.21-j0.53, m_3’2’ = −0.04-j0.014, m_1’5’ = −2.45-j0.065, m_5’6’ = 2.36+j0.11, m_6’7’ = 1.94-j0.078, m_4’6’ = 1.513+j0.011, m_5’5’ = 2.65, m_7’7’ = −5.76, R_1’1’ = 0.65-j1.42, R_2’2’ = 0.68+j0.001, R_3’3’ = 1.13-j0.076, R_4’4’ = −0.2-j1.54.

Tunability: The f_cen tuning and switching-off capabilities of UFS-II are shown in Fig. 15 alongside corresponding tuning parameters. Similar to UFS-I, both its transistor path and feedback are adjusted to achieve f_cen tuning. Especially, the phase tuning of the feedback is accomplished by modifying the resonant frequency of the LC tank, i.e., tuning the capacitor (C_r). For example, to tune UFS-II to a lower frequency, the capacitor value (C_r) in the feedback is increased to achieve the approximate 90° phase delay for the feedback at the desired frequency. Once the feedback is tuned, the transistor path is adjusted by modifying transistor bias voltages to achieve signal cancellation for S₁₂ of UFS-II at this new frequency. For intrinsic switching off, both parallel paths need to be non-transmissive. This is achieved by first turning off the transistor bias to induce high attenuation in the transistor path, then tuning the center frequency of the multi-resonant feedback by adjusting the capacitor value (C_r) until the TZ moves into the passband. In the intrinsically switched-off mode, high levels of isolation exceeding 20 dB from 1.5 GHz to 1.9 GHz are obtained.

Figure 15.

Circuit-simulated response of the reconfigurable UFS-II. Case 1.5 GHz: V_b = 0.781 V, V_c = 0.59 V, C_r = 3.6pF; Case 1.9 GHz: V_b = 0.775 V, V_c = 1.4 V, C_r = 2.1pF; Case switching-off: V_b = 0V, V_c = 0 V, C_r = 6.2pF.

D. Scalability to High-Order FIS

The UFS concept enables the creation of RF co-designed FISs that offer i) high selectivity and high gain in the forward direction, and ii) strong isolation in the reverse direction. The desired performance metrics can be configured by choosing the appropriate number and type of UFSs, passive resonators, and coupling elements, as illustrated in Fig. 1. For example, if high selectivity is required, then a high number of resonators (UFS or passive) need to be integrated into the FIS. Additionally, mixed EM couplings can be used to add TZs in the out-of-band response. Furthermore, for greater isolation in the reverse direction or enhanced gain in the forward direction, a higher number of UFSs should be incorporated into the FIS. It is important to position the UFSs before the passive resonators to minimize the overall NF of the FIS, as passive resonators tend to be lossy in practical implementations.

To evaluate the co-designed FIS concept, three FIS designs are presented: two three-stage FISs using UFS-I and UFS-II, respectively, and a five-stage FIS using UFS-I. Details of these designs are provided below.

Three-stage FIS using UFS-I: Based on the CRD of UFS-I determined in Section II-B, the CRD for a highly selective FIS, which includes one UFS-I and two tunable reciprocal resonators, is generated as shown in Fig. 16. The coupling values are listed in the caption of Fig. 16. Mixed EM coupling is introduced between nodes 2 and 3, leading to a TZ in CRD-synthesized responses of the three-stage FIS. The TZ's location can be specified using (7): \begin{equation*} m^{\prime }_{23} + {m^{\prime\prime}_{23}}{{\Omega }_{TZ}} = 0 \tag{7} \end{equation*} where Ω_TZ is the normalized frequency of TZ.

Figure 16.

CRD and CRD-synthesized response of the three-stage FIS using one UFS-I and two passive resonators. The coupling coefficients are as follows: m₀₁ = 0.96, m₁₂ = 0.75, m₂₃ = m’₂₃+m”₂₃ Ω = 0.84+0.11 Ω, m₃₄ = 1.

After determining the CRD of the FIS and its corresponding coupling values, the CRD is transformed to a practical circuit with f_cen = 1.7 GHz and 3 dB-FBW of 14%. Using the UFS-I circuit schematic from Fig. 3(a) and two capacitively-loaded (C = 1.7 pF) half-wavelength (λ_g/2) microstrip resonators designed on a Rogers 4003 substrate (h = 0.813 mm, ε_r = 3.55, tan δ = 0.0027), the resulting FIS circuit schematic is shown in Fig. 17(a). The mixed EM coupling m₂₃ is implemented by partially coupling the microstrip resonators (2 and 3) through a section of a coupled line. The characteristics of this coupling are determined from the denormalized inter-resonator coupling coefficients $K^{\prime }_{23}$ and ${{K}^{\prime\prime}_{23}}$ calculated using (8) [51]: \begin{equation*} K^{\prime }_{23} = m^{\prime }_{23}FBW,{\rm{\ \ \ \ }}{K^{\prime\prime}_{23}} = {{m}^{\prime\prime}_{23}}FBW \tag{8} \end{equation*}

Figure 17.

(a) Circuit schematic of the three-stage FIS that comprises one UFS-I and two passive coupled-line resonators. (b) Coupling coefficients K’₂₃ and K”₂₃ as a function of L₁ and S₁. (c) External quality factor Q_e2 as a function of W₁₂. (d) External quality factor Q_e3 as a function of L₅.

Fig. 17(b) shows the extracted coupling coefficients K’₂₃ and K”₂₃ as a function of the length L₁ and the gap S₁ of the coupled-line section, with the total length of the two microstrips kept constant, based on circuit-schematic simulations. Based on these parametric studies, L₁ and S₁ are selected as 12.8 mm and 0.145 mm to achieve the desired coupling coefficients K’₂₃ and K”₂₃ of 0.118 and 0.0154, respectively. The coupling between UFS-I and microstrip resonator 2 is implemented by a quarter-wavelength inverter, with its width determined by the external quality factor Q_e2, which is calculated using (9) [51]: \begin{equation*} {{Q}_{e2}} = 1/\left( {m_{12}^2FBW} \right) \tag{9} \end{equation*}

To achieve the desired Q_e2 = 12.7, the width of the quarter-wavelength inverter (W₁₂) connecting microstrip resonator 2 and UFS-I is chosen to be 2.55 mm, as shown in Fig. 17(c). Similarly, the external quality factor (Q_e3) of resonator 3, is calculated using (9). Parametric studies are performed in Fig. 17(d) to determine the location of the connection between the microstrip resonator 3 and port 2. For the desired Q_e3 = 7.1, L₅ is selected as 6 mm. Besides, the coupling between UFS-I and port 1 is implemented by a quarter-wavelength inverter, with its characteristic impedance determined by (10). \begin{equation*} {{Z}_{01}} = 50/{{m}_{01}} \tag{10} \end{equation*} W₀₁ is chosen to be 1.66 mm to achieve the desired Z₀₁ of 52 Ω.

Using the above specified dimensions, the circuit-simulated S-parameters of the FIS are shown in Fig. 18. The good correlation between the CRD-synthesized and circuit-simulated responses demonstrates the validity of the coupling matrix technique in passive resonator and UFS co-designing process. As noticed, the UFS enhances power transmission from P1 to P2 while effectively canceling the RF signal from P2 to P1. Compared to a single UFS, this configuration offers enhanced selectivity and out-of-band suppression due to the inclusion of passive resonators. Furthermore, the FIS exhibits three TZs (TZ1-TZ3) in the transmission direction: TZ1 arises from the mixed coupling between the two microstrip resonators, while TZ2 and TZ3 are introduced by the open-ended stubs (L₃ and L₅) on the two sides of the microstrip resonators.

Figure 18.

Circuit-simulated S-parameters of the three-stage FIS in Fig. 17(a). W₀₁ = 1.66 mm, L₀₁ = L₁₂ = 26.5 mm, W₁₂ = 2.55 mm, W₁ = W₂ = W₃ = W₄ = W₅ = 1.77 mm, L₁ = 12.8 mm, S₁ = 0.145 mm, L₂ = 5.7 mm, L₃ = 9.5 mm, L₄ = 9.2 mm, L₅ = 6 mm.

Three-stage FIS using UFS-II: To evaluate the usefulness of the UFS-II, the three-stage FIS is re-implemented using UFS-II and coupled-line microstrip resonators. Its CRD and the synthesized responses are shown in Fig. 19(a), and its corresponding circuit-simulated response is provided in Fig. 19(b) exhibiting three poles in the pass band and four TZs (TZ1-TZ4) in the out-of-band response. Notably, it achieves better out-of-band rejection at the higher frequency when compared to the three-stage FIS using UFS-I, due to the additional TZ4 introduced by UFS-II.

Figure 19.

(a) CRD and CRD-synthesized response of the three-stage FIS based on UFS-II. The coupling coefficients are: m₀₁ = 1.1, m₁₂ = 0.83, m₂₃ = m’₂₃+m”₂₃ Ω = 0.68+0.09 Ω, m₃₄ = 0.91. (b) Circuit-simulated responses of the three-stage FIS in (a). W₀₁ = 2.1 mm, W₁₂ = 2.95 mm, L₀₁ = L₁₂ = 26.5 mm, W₁ = W₂ = W₃ = W₄ = W₅ = 1.77 mm, L₁ = 11.9 mm, S₁ = 0.21 mm, L₂ = 6.6 mm, L₃ = 9.5 mm, L₄ = 9.3 mm, L₅ = 6.8 mm.

Five-stage FIS using UFS-I: To further explore scalability, a five-stage FIS comprising two UFSs-I and three capacitively-loaded coupled-line microstrip resonators is implemented for f_cen = 1.7 GHz with a 3 dB FBW of 13%, as shown in Fig. 20(a). Its implementation process mirrors that of a three-stage FIS: i) determine the CRD and the corresponding coupling values; ii) convert the CRD to the actual circuit, with dimensions determined using (8)–​(10). These dimensions are further optimized based on circuit-simulated results and are listed in the caption of Fig. 20 with the UFS-I parameters being listed in the caption of Fig. 3. Its circuit-simulated response, shown in Fig. 20(b), demonstrates four TZs: TZ1 and TZ2 are created by the two mixed-coupling between three microstrip resonators, with their positions controlled by C₁ and C₂, respectively. While TZ3 and TZ4 are created by the two open-ended stubs (L₅ and L₇) on the two sides of the microstrip resonators, with C₃ and C₄ controlling their positions. Center frequency tuning is realized by synchronously tuning all microstrip resonators (by increasing/decreasing C₁-C₄) and the UFSs-I. Furthermore, intrinsic switching-off is obtained by moving the TZs into the passband by altering C₁-C₄ and by turning off UFSs-I. The center frequency tuning and switching-off capabilities are illustrated in Fig. 20(b), with tuning parameters detailed in Table 4. Nota bly, intrinsic switching-off with over 60 dB isolation from 0.5 GHz to 2.5 GHz is achieved by adjusting C₁–C₄ to move all four TZs into the passband, as illustrated in Fig. 20(b), where the red annotations mark the positions of the TZs. Compared to three-stage FIS based on UFS-I, the five-stage FIS has higher gain (4 dB vs 1.9 dB), higher directivity (94 dB vs 47 dB), improved selectivity and better out-of-band rejection.

Figure 20.

(a) CRD and circuit schematic of a five-stage FIS that comprises two UFSs-I and three coupled-line microstrip resonators. (b) Circuit-simulated reconfigurable responses of (a). W₀₁ = 1.3 mm, W₁₂ = 1.2 mm, W₂₃ = 1.9 mm, W₁ = W₂ = W₃ = W₄ = W₅ = W₆ = W₇ = 1.8 mm, L₁ = 10.6 mm, S₁ = 0.1 mm, L₂ = 6.8 mm, L₃ = 12 mm, S₃ = 0.15 mm, L₄ = 8.3 mm, L₅ = 9.1 mm, L₆ = 8.9 mm, L₇ = 7.1 mm. The tuning parameters for the illustrated cases are listed in Table 4.

TABLE 4 Parameters of the Example Responses in Fig. 20

SECTION III. Experimental Validation

To experimentally validate the tunable FIS concept, six prototypes were designed, manufactured, and tested at the L band. They include: i) two test structures of the UFS, namely UFS-I and UFS-II; ii) two three-stage FISs with coupled line resonators, one using the UFS-I and the other one using the UFS-II, iii) two five-stage FISs using UFS-I with different types of microstrip resonators, resulting in various transfer functions with TZs for enhanced out-of-band rejection. The UFS and FIS designs were performed using the guidelines in Section II. Their circuit and post-layout EM simulations using the software package ADS from Keysight and High-Frequency Structure Simulator (HFSS) from ANSYS. All prototypes were built on a Rogers RO4003C substrate with a thickness of 0.813 mm, dielectric permittivity of 3.55, and a dielectric loss tangent of 0.0027. The RF performance of the UFSs and the co-designed FISs were characterized in terms of S-parameters, NF, and IIP3 using N5247B PNA-X Network Analyzer from Keysight. Specially, the frequency spacing (Δf) in IP3 measurement was set to 1 MHz.

A. UFS-I

The UFS-I prototype was designed for f_cen = 1.7 GHz. Its layout, manufactured prototype and utilized SMD components are shown in the orange frame in Fig. 21(a) and in the caption of Fig. 21. The tunable feedback network is implemented using varactors (D₁&D₂) connected in series with a DC block (C_B) and a DC feed (R_B). A comparison between EM-simulated and RF-measured S-parameters of UFS-I is presented in Fig. 21(b), showing a good agreement, and validating the proposed design. Specifically, the RF-measured performance at 1.7 GHz (with bias state V_b = 0.773 V, V_c = 0.91 V) can be summarized as follows: gain of −0.7 dB gain and directivity of 50 dB. By tuning the bias voltages of the transistor and of the varactors in the coupled-line feedback, center-frequency tunability can be obtained as shown in Fig. 21(d). Specifically, the UFS-I exhibits tuning performance as follows: f_cen tuning: 1.4∼1.9 GHz, gain: −1.8∼0 dB, and directivity: 18∼50 dB. For all tuning states, the NF was measured between 4 and 8 dB, and its IIP3 from −7 to 8.3 dBm. Furthermore, intrinsic switching-off can be realized by removing the transistor's DC bias and by detuning the varactors in the coupled-line feedback path, leading to an isolation of more than 10 dB from 0.5–2.9 GHz, as shown in Fig. 21(c).

Figure 21.

Experimental validation of UFS-I. (a) Layout and photograph of the manufactured prototype. (b) Comparison of RF-measured and EM-simulated response. (c) RF-measured intrinsic switching-off. (d) RF- measured center frequency tuning. Components: T: BFU760F, D1&D2: MA46H071, R_b = 511 Ω, R_c = 0 Ω, L_B = 30 nH, C_D = 220 μF, C_B = 100 pF, R_B = 10 MΩ, R₁ = 57.6 Ω, R₂ = 12.1 Ω, R₃ = 40.2 Ω, coupled line: W = 0.2 mm, S = 0.2 mm, L = 11 mm.

B. UFS-II

Fig. 22(a) illustrates the layout, the manufactured prototype (in the blue frame) and the utilized SMD components of UFS-II, designed for a passband centered at 1.7 GHz and a TZ at 2.44 GHz. Its measured RF performance is provided in Fig. 22(c) and (d) alongside a comparison with its EM simulated response for one tuning state [see Fig. 22(b)]. The measured RF performances at 1.7 GHz (with bias state V_b = 0.811 V, V_c = 0.65 V) are summarized as follows: −2.4 dB gain, 40 dB directivity. By tuning the bias voltages of the transistor and the varactor in the resonant feedback, the center frequency can be tuned from 1.4 to 1.9 GHz, with gain between −3.2 and −1.6 dB and directivity between 28 and 40 dB [see Fig. 22(d)]. For these states, the NF ranges from 4 to 6.2 dB and IIP3 varies between −5.7 to 3.8 dBm. Switching off is achieved by removing the transistor's DC bias and shifting the TZ into the operating band, resulting in more than 20 dB isolation from 1.27 to 2.9 GHz [see Fig. 22(c)], significantly higher than that of UFS-I.

Figure 22.

Experimental validation of UFS-II. (a) Layout and photograph of manufactured prototype. (b) Comparison of RF-measured and EM- simulated response. (c) RF-measured intrinsic switching-off. (d) RF- measured f_cen tuning. Components: T: BFU760F, D: MA46H072, R₁ = 80 Ω, R₂ = 8 Ω, R₃ = 37.4 Ω, R_a = 1 Ω, L_r = 1 nH, W = 0.2 mm, L = 4 mm.

C. Three-Stage UFS-I-Based FIS

The three-stage FIS, comprising one UFS-I and two coupled-line microstrip resonators, is designed for f_cen = 1.7 GHz, 3-dB FBW = 13.3%, and three TZs at 0.9, 2.1 and 2.3 GHz which are generated by the two coupled-line microstrip resonators. Its layout and photograph are shown in Fig. 23(a) with the details of UFS-I listed in Fig. 21. The RF-measured responses are compared with an EM-simulated state in Fig. 23(b), showing good agreement and validating the FIS concept. The FIS exhibits f_cen tunability between 1.4 and 1.9 GHz, with gain between −3.3 and −1.6 dB, directivity between 18 and 45 dB, NF between 4.8 and 7.6 dB, and IIP3 from −3.7 to 6.9 dBm [see Fig. 23(d)]. Additionally, intrinsic switching off is achieved by switching off the UFS and shifting the TZs into the operating frequency band, resulting in isolation greater than 40 dB from 0.5 to 2.22 GHz [see Fig. 23(c)].

Figure 23.

Experimental validation of the three-stage UFS-I-based FIS. (a) Layout and photograph of the manufactured prototype. (b) Comparison between RF-measured and EM-simulated S-parameter response. (c) RF-measured switched-off. (d) RF-measured center frequency tuning. Geometrical parameters: W_inv = 2.4 mm, L_inv = 24.4 mm, W₁ = 1.5 mm, L₁ = 5.9 mm, W₂ = 1 mm, L₂ = 5.3 mm, W₃ = 0.55 mm, L₃ = 7 mm, S₃ = 0.2 mm, W₄ = 1 mm, L₄ = 6.4 mm, W₅ = 1.5 mm, L₅ = 4.2 mm. Varactors: MA46H071.

D. Three-Stage UFS-II-Based FIS

Another experimental validation of the three-stage FIS uses one UFS-II and two coupled-line microstrip resonators. Its layout and photograph are provided in Fig. 24(a). Designed for a passband centered at 1.7 GHz with a 3-dB fractional bandwidth of 10.5%, this FIS features TZs at 2.44 GHz (from UFS-II) and additional TZs at 0.9 GHz, 2.2 GHz, and 2.8 GHz (from the coupled-line resonators). Details of the UFS-II are provided in Fig. 22. The RF-measured performance of the FIS, as shown in Fig. 24, includes: f_cen tuning between 1.4 and 1.9 GHz, gain between −4 to −2.9 dB, directivity between 34 to 35 dB, NF between 5 to 6.2 dB, IIP3 between −5 to 3.4 dBm and an intrinsically switch-off mode of operation with isolation >50 dB from 1.25 to 2.1 GHz.

Figure 24.

Experimental validation of the three-stage UFS-II-based FIS. (a) Layout and photograph of the prototype. (b) Comparison between RF-measured and EM-simulated tuning state. (c) RF-measured switching-off. (d) RF-measured center frequency tuning. Geometrical parameters: W_inv = 2 mm, L_inv = 23 mm, W₁ = 1.1 mm, L₁ = 5.5 mm, W₂ = 3.2 mm, L₂ = 6.4 mm, W₃ = 1mm, L₃ = 10 mm, S₃ = 0.2 mm, W₄ = 2 mm, L₄ = 6.6 mm, W₅ = 1.7 mm, L₅ = 4 mm. Varactors: MA46H071.

E. Five-Stage UFS-I-Based FIS: Topology A

The five-stage FIS (topology A) prototype, comprising two UFSs-I and three coupled-line microstrip resonators was designed for f_cen = 1.6 GHz, 3-dB FBW = 9%, and four TZs at 0.66, 1.05, 1.92, and 2.26 GHz which are generated by the three coupled-line microstrip resonators. Its experimental validation is provided in Fig. 25 in terms of its layout, manufactured prototype [see Fig. 25(a)] and measured performance [Fig. 25(b) and (d)]. In this case, the UFS-I was redesigned for higher gain to compensate for the loss of the passive resonators. The comparison between EM simulations and RF measurements confirms the design's validity, as depicted in Fig. 25(b). As expected, the five-stage FIS demonstrates improved selectivity and directivity compared to the three-stage prototype due to more UFSs and passive resonators. Its RF-measured characteristics are as follows: f_cen tuning from 1.4 to 1.8 GHz, gain between −3 and −0.3 dB, directivity between 31 and 71 dB, switching-off isolation > 50 dB from 0.5 to 2.9 GHz, NF between 8.5 and 10.3 dB, and IIP3 between −6.4 and 1.4 dBm.

Figure 25.

Experimental validation of the five-stage UFS-I-based FIS (topology A). (a) Manufactured prototype. (b) Comparison between one RF-measured and one EM-simulated response. (c) RF-measured intrinsic switching-off. (d) RF-measured center frequency tuning. Geometrical parameters: W_c = 0.85 mm, L_c = 4.8 mm, W_inv1 = 1 mm, L_inv1 = 12 mm, W_inv2 = 2.7 mm, L_inv2 = 17 mm, W₁ = 0.49 mm, L₁ = 4.1 mm, W₂ = 2.2 mm, L₂ = 6.1 mm, W₃ = 0.6 mm, L₃ = 6.7 mm, S₃ = 0.2 mm, W₄ = 2.9 mm, L₄ = 4.2 mm, W₅ = 0.6 mm, L₅ = 8 mm, S₅ = 0.2 mm, W₆ = 1.1 mm, L₆ = 5 mm,W₇ = 1 mm, L₇ = 4.2 mm. Varactors: MA46H071, SMV1405.

F. Five-Stage UFS-I-Based FIS: Topology B

For comparison with Topology A, the five-stage FIS (Topology B) comprising two UFSs-I, two half-wavelength microstrip resonators and one multi-resonant stage which creates two TZs at 1.43 and 1.98 GHz was designed for f_cen = 1.6 GHz and 3-dB FBW = 11%. The design can also be achieved using CRD synthesis. The design principles and tunability of the cascaded tunable passive resonator are detailed in [52] and are not repeated here for brevity. The photograph of the manufactured prototype is shown in Fig. 26(a), where the inverters and the multi-resonant stage are implemented by quarter-wavelength transmission lines. The RF-measured response of the FIS is provided in Fig. 26 alongside a comparison with one EM-simulated state. It exhibits the following characteristics: f_cen tuning from 1.43 to 1.75 GHz, gain bet ween −2.7 and −0.4 dB, directivity between 60 and 65 dB, switch-off isolation > 37 dB from 0.5 to 2.75 GHz, NF between 8 and 10 dB, and IIP3 between −6.4 and −1.6 dBm. Compared with Topology B, Topology A using coupled-line microstrip resonators offers better out-of-band suppression and a compact size.

Figure 26.

Experimental validation of the five-stage UFS-I-based FIS (topology B). (a) Manufactured prototype. (b) Comparison between one RF-measured and one EM-simulated response. (c) RF-measured intrinsic switching-off. (d) RF-measured center frequency tuning. Geometrical parameters: W_inv1 = 1.77 mm, L_inv1 = 23 mm, W_inv2 = 0.64 mm, L_inv2 = 28 mm, W₁ = 3.5 mm, L₁ = 30 mm, W₂ = 2 mm, L₂ = 21 mm, W₃ = 4 mm, L₃ = 8.2 mm, W₄ = 1.85 mm, L₄ = 12 mm, W₅ = 3.5 mm, L₅ = 25.7 mm, W₆ = 2 mm, L₆ = 31 mm, W₇ = 1.2 mm, L₇ = 19 mm. Varactors: SMV1236, SMV1405.

G. Comparison With State-of-the-Art

To better illustrate the unique contributions of this work, a comparison with key state-of-the-art unilateral RF components is provided in Table 5. Specifically, the FIS concept is compared with ferrite-based approaches, transistor-based architectures and STM-based concepts as the most relevant to this work. When compared to ferrite-based configurations [53], [54], the proposed FIS approach offers significantly smaller RF components that are reconfigurable and multi-functional, uniquely integrating three component functions into one. Similar RF component functions are realized in the ferrite-based component in [55], but it exhibits lower directivity (<33.1 dB) and larger insertion loss (2.7∼4.5 dB). When compared to active circuit configurations, the majority of them [7], [8], [15], [35], [37], [39], [56] are frequency-static, consume high DC power (100∼228.6 mW in [37], [39], [56], [40]) and have higher NF (13 dB in [8]) than the proposed approach. Furthermore, they are mostly limited to one or two functionalities (e.g., an isolator in [7], [8], [56] and a BPF/isolator in [15], [35], [37], [39]). While [40] has comparable f_cen tuning and directivity, it consumes 10x more DC power than the FIS concept and lacks RF switch functionality. Lastly, the FIS concept outperforms STM-based approaches [42], [43], [45], which are generally limited to low frequencies (100–600 MHz) due to reliance on AC modulating signals proportional to bandwidth, making them difficult to implement at higher frequencies. Additionally, STM BPFs also require complex biasing circuits, have lower directivity (<6.6 dB in [43] and <29 dB in [45]), smaller tuning ranges (up to 1.16:1 in [42]), higher insertion loss (3.9–4.6 dB in [42]), and lack intrinsic switching-off capability [42], [43]. In short, the FIS concept offers high directivity and integrates multiple functions, including frequency-tunable transfer functions with switch-off capability, while reducing size and complexity.

TABLE 5 Comparison With State-of-the-Art Unilateral RF Components

SECTION IV. Conclusion

This paper presented a comprehensive design methodology and the practical validation of a new class of a multi-functional RF component that exhibits the co-designed function of a frequency-tunable BPF, an RF isolator and an RF switch within a single RF device. The proposed FIS concept was built on reconfigurable UFSs (shaped by a transistor-based path and tunable frequency-dependent feedback), tunable passive resonators and multi-resonant stages. The application of UFSs in high-order reconfigurable FISs were presented in detail. The scalability of the concept to high-order quasi-elliptic transfer functions was achieved by integrating the UFSs with frequency-tunable resonators, EM mixed-coupling elements, and multi-resonant cells. The FIS concept was demonstrated through synthesized and circuit-based design examples. For proof-of-concept validation purposes, six prototypes were designed, manufactured, and measured at L-band. These prototypes include i) two UFSs with coupled-line feedback (UFS-I) and multi-resonant feedback (UFS-II), ii) two three-stage FISs using UFS-I/ UFS-II and coupled-line microstrip resonators, iii) two five-stage FISs using UFSs-I and different passive resonator topologies. The proposed devices have potential applications in in-band full-duplex (IBFD) systems and joint communication and sensing (JCAS) applications, where they can effectively suppress self-interference and out-of-band interference, enabling the simultaneous use of the same frequency for TX and RX. They are also applicable in massive MIMO (mMIMO) systems and low-power communication systems, such as IoT devices and cellular communications, where integrated reconfigurable filtering, isolation, and switching functionalities are essential for reducing signal interference, enhancing signal quality, mitigating reflected signals, and improving system agility. To enable further miniaturization and integration into practical systems, the FIS concept can be implemented using MMIC technology, which is why a transistor-based configuration is proposed instead of using ferrite materials for realizing non-reciprocal components.