A Compact Multi-Mode Reconfigurable Filter with Coarse- and Fine-Tunable Pole–Zero Characteristics for Ku/Ka Multi-Band Applications

Abstract

This work presents a compact, multi-mode reconfigurable filter fabricated in a 65-nm CMOS process for Ku-/Ka multiband applications. The filter employs lumped-element LC series and parallel resonant networks to realize two transmission zeros (TZs) and two poles, with the TZs intrinsically clamped between the poles. Coarse-tuning is achieved by controlling voltage-controlled switches, enabling wide-range relocation of the pole–zero frequencies, while voltage-controlled varactors provide continuous fine-tuning. This pole–zero tracking mechanism allows the filter to suppress high-frequency image signals in Ku-band satellite communication terminals and low-frequency image signals in Ka-band applications. Measurements demonstrate 15-dB stopband fractional bandwidths of 58% and 58.3% in the coarse-low and coarse-high modes, respectively, with maximum attenuations exceeding 33 and 40 dB. Fine-tuning shifts the stopband center frequencies continuously from 14.5–19.7 GHz and 24.1–26.9 GHz, respectively, in close agreement with electromagnetic simulations. The measurement results confirm its suitability for multi-band and tunable image-rejection applications.

1. Induction

Recent advancements in 5G millimeter-wave (mmW) Frequency Range 2 (FR2) communications (24.25–52.6 GHz) require RF front ends capable of supporting high-bandwidth communications, while Ku-band satellite terminals (10.7–12.75 GHz for the downlink and 13.75–14.5 GHz for the uplink) remain indispensable in satellite communications. With the integration of these two communication systems, the radio frequency front end (RFFE) faces severe challenges from complex image frequency interference. Specifically, it must suppress both the high-frequency image signals generated in the Ku-band and the low-frequency image signals within the 5G Ka-band. Furthermore, tunable bandstop filters (BSFs), offering continuous frequency tunability, are needed for dynamic spectrum access and spectrum tuning issues. Consequently, investigating multi-mode reconfigurable filters that feature both dual-image suppression and continuous tunability has become a critical endeavor in RFFE design.

Over the past few decades, BSFs have been extensively investigated. In [1], a binary bandwidth-reconfigurable BSF incorporating both lumped- and distributed-element topologies was proposed, where 2^M discrete bandwidth states were achieved using M control voltages, and digital bandwidth reconfigurability was verified at 285 MHz and 2 GHz. In [2], a tunable BSF based on yttrium iron garnet (YIG)-micromachined technology was presented, maintaining a stopband rejection of over 30 dB across a 4–8 GHz tuning range, with a maximum attenuation of 55 dB. However, these BSFs employ off-chip technologies. Compared with off-chip filters, on-chip solutions are much more suitable for integration due to their miniaturized size, reduced fabrication cost, and enhanced frequency tuning precision. Based on standard silicon technologies, several innovative designs have been presented in the literature, which includes lowpass [3,4], highpass [5,6], bandpass [7,8,9,10,11,12,13,14], and bandstop filters (BSFs) [15,16,17,18,19]. Additional examples include a GaAs IPD bandstop filter [20], VO₂-based reconfigurable millimeter-wave filters [21], a SiGe BiCMOS bandpass filter [22], and a MEMS-based tunable bandstop filter [23]. Collectively, these works provide context for evaluating the proposed CMOS implementation across different material platforms and fabrication technologies.

Figure 1 illustrates the block diagrams of the conventional and the proposed multi-band multi-mode filter architectures. To effectively suppress the high-frequency image signals of the Ku-band and the low-frequency image signals of the Ka-band, conventional multi-mode reconfigurable filters typically require the integration of a low-frequency (LF) filter, a high-frequency (HF) filter, and a BSF between the low-noise amplifier (LNA) and the mixer. This approach not only increases the chip area but also leads to hardware resource redundancy. To address these issues, in this article, a compact multi-mode reconfigurable filter configuration is proposed for Ku/Ka multi-band applications using 65-nm CMOS technology. The proposed filter exhibits two transmission zeros (TZs) and two transmission poles (TPs). Additionally, it incorporates a pole–zero tracking mechanism, ensuring that the two TZs are inherently clamped between the TPs based on lumped-element LC series and parallel resonant networks. By controlling the on/off states of voltage-controlled switches, a wide-range coarse-tuning of the pole and zero frequencies is achieved, while the continuous adjustment of voltage-controlled varactors enables dynamic fine-tuning. Leveraging these two flexible TZs, the filter effectively suppresses image signals over a wide frequency range. The measured results are in good agreement with simulations, demonstrating that the proposed filter achieves a compact size, dual-image suppression, and frequency tunable features.

Figure 1.

Block diagrams of the conventional and the proposed multi-band multi-mode filter receiver architectures.

2. Design and Implementation of a Multi-Mode Reconfigurable Filter

Figure 2 presents the schematic diagram of the proposed compact multi-mode reconfigurable filter. To facilitate analysis and design, a symmetrical Π-type topology is adopted, which comprises two sets of LC series resonant networks and a central LC parallel resonant network. To achieve effective suppression of the image frequencies over a wide frequency band, the core mechanism of this circuit utilizes the resonant networks formed by inductors and tunable capacitors to introduce two TZs in the transfer function. Specifically, in this design, two tuning mechanisms, namely coarse-tuning and fine-tuning, are explored to control the pole and zero frequencies. Wherein, the coarse-tuning is realized by adjusting the control voltage V_sw to switch the state of the switch between on and off, allowing for a wide-range adjustment. Furthermore, the fine-tuning, together with the dynamic tracking capability, is achieved by varying the tuning voltage V_tune to dynamically adjust the capacitance of the varactor (C_var).

Figure 2.

Schematic diagram of the proposed multi-mode reconfigurable filter.

2.1. Pole–Zero Generation and Distribution

To reveal the operating mechanism of the multi-mode reconfigurable filter. Based on the schematic equivalent model shown in Figure 2, this article employs Z-parameter analysis to evaluate the system’s pole–zero distribution.

First, To further elucidate the pole–zero distribution of the proposed filter topology, the transfer impedance Z₂₁, neglecting the parasitic resistances, can be obtained as follows:

$$Z_{21}={\displaystyle \frac{C_2{\left(\frac{1}{\mathrm{j}\omega C_1}+\mathrm{j}\omega L_1\right)}^2\left(\frac{1}{\mathrm{j}\omega C_2}+\mathrm{j}\omega L_2\right)}{2C_2\left(\frac{1}{\mathrm{j}\omega C_1}+\mathrm{j}\omega L_1\right)\left(\frac{1}{\mathrm{j}\omega C_2}+\mathrm{j}\omega L_2\right)+L_2}}$$ (1)

By setting |Z₂₁| = 0, the transmission zeros of the circuit can be derived as follows:

$${\omega }_{\mathrm{z}1}={\displaystyle \frac{1}{\sqrt{L_1{C}_1}}}$$ (2)

$${\omega }_{\mathrm{z}2}={\displaystyle \frac{1}{\sqrt{L_2{C}_2}}}$$ (3)

As indicated by (2) and (3), the zero frequencies of |Z₂₁| are determined by the values of C₁, L₁, C₂, and L₂. Specifically, when L₁C₁ = L₂C₂, the topology exhibits only a single transmission zero. Conversely, when L₁C₁ ≠ L₂C₂, two transmission zeros are generated in the transmission response. According to (2) and (3), these two zeros are independent of L₂,C₂ and L₁,C₁, respectively.

Similarly, by setting |Z₂₁| = ∞, the poles of |Z₂₁| can be derived as follows:

$${\omega }_{\mathrm{p}1}={\displaystyle \frac{1}{2}}\sqrt{{\displaystyle \frac{2L_1{C}_1+L_2{C}_1+2L_2{C}_2+\sqrt{-16L_1{C}_1{L}_2{C}_2+{\left(2L_2{C}_2+C_1\left(2L_1+L_2\right)\right)}^2}}{L_1{C}_1{L}_2{C}_2}}}$$ (4)

$${\omega }_{\mathrm{p}2}={\displaystyle \frac{1}{2}}\sqrt{{\displaystyle \frac{2L_1{C}_1+L_2{C}_1+2L_2{C}_2-\sqrt{-16L_1{C}_1{L}_2{C}_2+{\left(2L_2{C}_2+C_1\left(2L_1+L_2\right)\right)}^2}}{L_1{C}_1{L}_2{C}_2}}}$$ (5)

Based on the aforementioned pole–zero analysis, the proposed structure is capable of generating two poles and two zeros in its transmission response. To validate this analysis, Figure 3 illustrates the simulated |Z₂₁| curves of the proposed multi-mode reconfigurable filter under various conditions regarding the magnitude and ratio of L₁C₁ to L₂C₂. As shown in Figure 3a, when L₁C₁ = L₂C₂, the two zeros coincide, resulting in a single transmission zero. Conversely, when L₁C₁ ≠ L₂C₂, two distinct zeros emerge, which are consistently situated between the two poles. Furthermore, Figure 3b demonstrates that the frequency spacing between these two zeros is governed by the ratio of L₁C₁ to L₂C₂. Consequently, in practical design, this ratio can be flexibly adjusted to achieve the required bandwidth of the notched band.

Figure 3.

Simulated |Z₂₁| curves of the proposed multi-mode reconfigurable filter for different (a) relationships between L₁C₁ and L₂C₂ and (b) ratios of L₁C₁ to L₂C₂.

2.2. Pole–Zero Tunable Characteristics

Building upon the detailed analysis of the pole–zero distribution in the preceding section, this section focuses on elucidating the relative positioning of the poles and zeros, as well as the continuous tunability of the filter.

Taking pole ω_p2 and zero ω_Z1 as examples, we examine the relative relationship between the pole and the zero within the structure. The expressions for ω_p2 and ω_Z1 are given in (5) and (2), respectively. This section aims to demonstrate that the ratio of ω_p2 to ω_Z1 is consistently less than unity. To this end, the variables are first normalized, and after performing the initial variable substitution, the expression for the ratio of ω_p2 to ω_Z1 is derived as (7):

$$\mathrm{m}=L_1{C}_1,r={\displaystyle \frac{L_2}{L_1}},k={\displaystyle \frac{C_2}{C_1}}$$ (6)

$${\displaystyle \frac{{\omega }_{P2}}{{\omega }_{\mathrm{z}1}}}={\displaystyle \frac{1}{2}}\sqrt{{\displaystyle \frac{2+r+2\mathit{kr}-\sqrt{{\left(2+r+2\mathit{kr}\right)}^2-16\mathit{kr}}}{\mathit{kr}}}}$$ (7)

Subsequently, by performing a second variable substitution, the expression for the ratio of ω_p2 to ω_Z1 is derived as (9):

p = 2 + r + 2kr > 0,q = 4kr > 0 (8)

$${\displaystyle \frac{{\omega }_{P2}}{{\omega }_{\mathrm{z}1}}}=\sqrt{{\displaystyle \frac{p-\sqrt{p^2-4q}}{q}}}$$ (9)

Analytical calculations confirm that Equation (10) holds universally for both p ≤ q and p > q.

$$p-\sqrt{p^2-4q}<q;p,q>0$$ (10)

Based on the preceding analysis, the ratio ω_p2/ω_Z1 is consistently less than unity, indicating that ω_p2 < ω_Z1. By extension, it can be demonstrated that the low-frequency pole ω_p2 is always lies below both zero frequencies ω_Z1 and ω_Z2, while the high-frequency pole ω_p1 remains greater than both zeros, as summarized in Equations (11) and (12).

ω_P1 > ω_z1,ω_P1 > ω_z2;C_i,L_i > 0 (11)

ω_P1 > ω_z1,ω_P1 > ω_z2;C_i,L_i > 0 (12)

The robustness of the proposed mechanism originates from the pole–zero arrangement established by the symmetrical Π-type topology. As demonstrated by the preceding analysis and confirmed by (11) and (12), the two transmission zeros consistently remain between the two transmission poles for arbitrary inductance and capacitance values. Consequently, a wide stopband bounded by the lower and upper passbands is preserved. Although the exact pole and zero frequencies depend on the inductance and capacitance values, the fundamental bandstop mechanism remains unchanged as these parameters vary.

Having established the relative positioning of the zeros and poles in the transmission characteristics, we apply the substitution method used in Equation (6) to derive the ratios of the two pole frequencies of |Z21| to the two zero frequencies, as follows:

$${\displaystyle \frac{{\omega }_{P1,P2}}{{\omega }_{\mathrm{z}1}}}={\displaystyle \frac{1}{2}}\sqrt{{\displaystyle \frac{2+r+2\mathit{kr}\pm \sqrt{{\left(2+r+2\mathit{kr}\right)}^2-16\mathit{kr}}}{\mathit{kr}}}}$$ (13)

$${\displaystyle \frac{{\omega }_{P1,P2}}{{\omega }_{\mathrm{z}2}}}={\displaystyle \frac{1}{2}}\sqrt{2+r+2\mathit{kr}\pm \sqrt{{\left(2+r+2\mathit{kr}\right)}^2-16\mathit{kr}}}$$ (14)

As indicated by Equations (13) and (14), the ratios of the pole and zero frequencies are primarily determined by the parameter ratios k = C₂/C₁ and r = L₂/L₁. By maintaining k and r at constant values, the poles and zeros can be dynamically tracked through the adjustment of the capacitor values C₁ and C₂, as illustrated in Figure 4. In the implemented circuit, deviations from a constant capacitance ratio and the effects of parasitic elements can modify the frequency ratios.

Figure 4.

Simulated Z₂₁ curves of the proposed multi-mode reconfigurable filter with k = 0.8 as C1 varies in (a) Coarse_LF mode and (b) Coarse_HF mode.

The transmission zeros (|S₂₁| = 0) and transmission poles (|S₁₁| = 0) of the proposed structure are determined by the following equations:

$$S_{21}={\displaystyle \frac{2Z_{21}{Z}_0}{\left(Z_{11}+Z_0\right)\left(Z_{22}+Z_0\right)-Z_{12}{Z}_{21}}}$$ (15)

$$S_{11}={\displaystyle \frac{\left(Z_{11}-Z_0\right)\left(Z_{22}+Z_0\right)-Z_{12}{Z}_{21}}{\left(Z_{11}+Z_0\right)\left(Z_{22}+Z_0\right)-Z_{12}{Z}_{21}}}$$ (16)

where Z₀ denotes the port impedance, it can be derived from Equation (15) that the transmission zero frequencies correspond to the zeros of |Z₂₁|. Although the transmission pole frequencies differ from the pole frequencies of |Z₂₁|, they exhibit a strong correlation. Taking the Coarse_LF mode as an example, Figure 5 illustrates the simulated S-parameters of the proposed filter under varying values of C₁ and C₂. It is evident that as C₁ varies, the transmission poles exhibit a variation trend similar to that of the poles in |Z₂₁|.

Figure 5.

Simulated S₂₁ and S₁₁ characteristics of the proposed multi-mode reconfigurable filter under various C₁ values in Coarse_LF mode with k = 0.8.

2.3. Implementation of Variable Capacitor C

The preceding section analyzed in detail the pole–zero distribution and its tuning mechanism specifically, the continuous tuning of the pole and zero frequencies by adjusting variable capacitance; this section focuses on the actual implementation of the tunable capacitors and delves into their operating principles under both coarse- and fine-tuning modes.

As shown in Figure 6, the tunable capacitor C₁ consists of a varactor C_var1 in parallel with a capacitor C_P1, which is further connected in series with a switching transistor M₁. By adjusting the voltage V_SW to control the on/off state of M₁, a wide-range transition between the low-frequency (LF) and high-frequency (HF) states can be achieved in the coarse mode. Meanwhile, by regulating the voltage V_Tune to vary the capacitance of the varactor C_var1, the pole–zero tracking capability in the fine mode is realized. Compared to C₁, the structural difference in C₂ lies in an additional capacitor C_S2 connected in series at the source of the switching transistor M₂. This capacitor serves as a DC-blocking element to prevent the DC path from interfering with the switching states, thereby ensuring that the switching transistor can be effectively controlled by the external voltage. Notably, The fixed capacitors C_P1, C_P2, and C_S2 are implemented using metal-oxide-metal (MOM) capacitors.

Figure 6.

Schematic of the tunable capacitors.

By comprehensively trading off the parasitic capacitances, on-resistances, capacitance values, and Q-factors of the transistors against the tuning range of the varactors, the optimized parameters of the tunable capacitors are determined, as presented in Figure 7.

Figure 7.

Capacitance values and Q-factors of the proposed tunable capacitors C₁ and C₂ in coarse and fine modes: (a) C₁, (b) C₂.

Figure 7a and Figure 7b present the curves of the capacitance and quality factor (Q-factor) versus control voltages for the proposed tunable capacitors C₁ and C₂, respectively, under both coarse- and fine-tuning modes. The coarse-tuning mode, defined as the wide-range switching of the capacitance between the low-frequency (LF) and high-frequency (HF) states, is realized by toggling the switch control voltage V_SW between 0 and 1. Specifically, at V_Tune = −1 V, when V_SW switches from 0 to 1, the value of C₁ abruptly drops from 165 fF to 74 fF, while C₂ decreases from 118 fF to 60 fF. Under a fixed V_SW state, the fine-tuning mode enables continuous capacitance variation by dynamically sweeping the tuning voltage V_Tune from −1 V to 1 V with a step size of 0.1 V. As V_Tune increases from −1 V to 1 V, in the LF state (V_SW = 1), C_{1_LF} and C_{2_LF} continuously increase from 165 fF and 118 fF to 337 fF and 214 fF, respectively; in the HF state (V_SW = 0), C_{1_HF} and C_{2_HF} vary from 74 fF and 60 fF to 95 fF and 77 fF, respectively.

The lower Q-factor observed in the LF mode compared to the HF mode is primarily attributed to the on-resistances (R_ON1 and R_ON2) of the switch transistors. The gate width of the switch MOS transistors is optimized to 6 × 32 μm to mitigate the degradation of image rejection performance caused by large on-resistances.

The capacitance and Q-factor characteristics discussed above describe the steady-state behavior of the proposed tunable capacitors. The tuning speed of the proposed filter is mainly determined by the response speed of the external control-voltage source and its associated bias network. Within each coarse-tuning state, the pole and zero frequencies are tuned by varying V_Tune, while V_SW remains fixed. Therefore, the proposed work focuses primarily on the pole–zero tuning mechanism and the resulting steady-state frequency responses, rather than on the design of the control-voltage driver. The actual tuning speed is mainly related to the charging and settling time of the external control source, together with the parasitic effects of the tuning network.

2.4. Impact of Quality Factor of the Inductors and Capacitors

The previous analysis was conducted under the assumption of lossless circuit elements to elucidate the distribution characteristics of the poles and zeros. To evaluate the impact of parasitic resistance while maintaining analytical simplicity, we assume that inductors L₁ and L₂ have the same quality factor Q_L, and capacitors C₁ and C₂ also maintain the same quality factor Q_C. Building upon these assumptions, this section further investigates the influence of Q_L and Q_C on the S21 responses in the Coarse_LF mode, as illustrated in Figure 8.

Figure 8.

Simulated S₂₁ responses of the proposed multi-mode reconfigurable filter in Coarse_LF mode as a function of (a) Q_L and (b) Q_C.

Figure 8a illustrates the simulated S21 responses as a function of the inductor quality factor Q_L. It is observed that Q_L simultaneously impacts the passband insertion loss and the notch depths of the two transmission zeros. For optimal performance, the inductors are implemented using the thickest metal layer (M₉) in the layout design. Figure 8b shows the simulated S21 curves with respect to the capacitor quality factor Q_C. It is evident that Q_C primarily influences the loss at the high-frequency pole ω_p1 and the notch depth of the high-frequency transmission zero ω_z1. Furthermore, as indicated in Figure 7, when the filter operates in the Coarse_LF mode and the control voltage V_Tune is tuned from −1 V to 1 V with a step size of 0.1 V, the effective quality factors of the tunable capacitor networks C₁ and C₂ gradually degrade, resulting in a corresponding deterioration in the notch depth of the high-frequency transmission zero.

3. Multi-Mode Reconfigurable Filter Implementation and Measurement Results

The proposed multi-mode reconfigurable filter was fabricated using the TSMC 65 nm CMOS process. Figure 9 shows the chip photo of the multi-mode reconfigurable filter, which occupies a core area of 320 μm × 200 μm. The total chip area, including the pads, is 430 μm × 370 μm.

Figure 9.

Multi-mode reconfigurable filter chip photo.

Figure 10a,b show the measured and simulated S-parameters of the proposed multi-mode reconfigurable filter under two representative states. In Coarse_LF mode (V_SW = 1 V, V_Tune = 0 V), the filter exhibits a stable bandstop response with a >15 dB rejection stopband covering 11.5–20.8 GHz (fractional bandwidth, FBW ≈ 57.6%), flanked by lower and upper passbands of 2.7–9.7 GHz and 24–29.5 GHz, respectively. In Coarse_HF mode (V_SW = 0 V, V_Tune = −1 V), the stopband is reconfigured to 18.6–35.2 GHz (FBW ≈ 61.7%) while maintaining >15 dB rejection, with the lower and upper passbands expanded to 2.7–13.8 GHz and 40–50 GHz. The excellent agreement between measured and simulated results validates the reliability of the proposed design in stopband reconfigurability and frequency selection. Within each selected coarse-tuning state, the proposed filter supports simultaneous signal transmission through the lower and upper passbands while rejecting signals in the intervening stopband. The Coarse_LF and Coarse_HF states are selected sequentially by V_SW, whereas V_Tune provides continuous fine-tuning within each state. Thus, simultaneous dual-passband operation is supported within a fixed state, but the two coarse-tuning states cannot operate simultaneously.

Figure 10.

Simulated and measured S-parameters of the proposed filter under different tuning conditions. (a,b) Comparison of simulated and measured S11, S21, and S22 in the Coarse_LF mode with V_SW = 1 V and V_Tune = 0 V, and in the Coarse_HF mode with V_SW = 0 V and V_Tune = −1 V, respectively. (c,d) Measured S21 responses under fine-tuning in the Coarse_LF mode (V_SW = 1) and the Coarse_HF mode (V_SW = 0 V), respectively. (e,f) Measured S11 responses under fine-tuning in the Coarse_LF mode (V_SW = 1 V) and the Coarse_HF mode (V_SW = 0 V), respectively. In (c–f), V_Tune is swept from −1 to 1 V in steps of 0.2 V.

Figure 10c–f illustrate the performance of the proposed reconfigurable filter under continuous tuning (Fine Mode), where the control voltage V_Tune is stepped from −1 V to 1 V with an increment of 0.2 V. In the Coarse_LF mode, the stopband rejection range (with >15 dB rejection) is continuously tuned from 14–25.3 GHz down to 10.2–18.7 GHz. Consequently, the lower passband is dynamically adjusted from 2.7 to 11.5 GHz, while the upper passband remains centered at 24–29.5 GHz. In the Coarse_HF mode, the tunable stopband is reconfigured from 18.6–35.2 GHz to 17.1–31.2 GHz, accompanied by the lower passband expanding up to 14.4 GHz and the upper passband tuning between 37 and 43.5 GHz. These results successfully demonstrate the high spectral flexibility and stopband reconfigurability of the proposed design. Simulation and experimental results demonstrate that the proposed reconfigurable filter has zero DC power consumption.

As shown in Figure 11, the measured input 1-dB compression point (IP1dB) of the proposed filter varies with both frequency and tuning voltage in Coarse_HF mode (V_SW = 0 V). As shown in Figure 11a, in the Ku-band, the IP1dB decreases with increasing frequency. For V_Tune = 1 V, it decreases from 16.6 dBm at 2 GHz to 5.4 dBm at 13 GHz, whereas for V_Tune = −1 V, it decreases from 28.5 dBm at 3 GHz to 9.3 dBm at 13 GHz. In the Ka-band, the IP1dB generally increases with frequency after reaching a minimum value. It varies from 7.1 to 15.5 dBm over 34–50 GHz for V_Tune = 1 V, and from 8.4 to 13.9 dBm over 38–50 GHz for V_Tune = −1 V. Figure 11b further illustrates the influence of fine-tuning on the IP1dB. At 10 GHz, the IP1dB decreases from 12.2 to 8.6 dBm as V_Tune increases from −1 to 1 V. At 40 GHz, it varies within a relatively narrow range of 7.8–9.1 dBm, reaching its minimum near V_Tune = 0 V. These results demonstrate that the IP1dB is strongly dependent on the operating band and frequency, while its variation with fine-tuning voltage is more pronounced in the Ku-band than in the Ka-band. In addition, in Coarse_LF mode (V_SW = 1 V), the IP1dB exceeds 20 dBm over both 3–12 GHz and 27–50 GHz at V_Tune = 1 V.

Figure 11.

Measured input 1-dB compression point (IP1dB) of the proposed filter in Coarse_HF mode (V_SW = 0 V): (a) IP1dB versus frequency in the Ku and Ka-bands at V_Tune = −1 V and 1 V; (b) IP1dB versus fine-tuning voltage at 10 GHz and 40 GHz, with V_Tune swept from −1 V to 1 V in steps of 0.2 V.

To evaluate its performance, Table 1 summarizes the key characteristics of the proposed multi-mode reconfigurable filter in comparison with other state-of-the-art designs reported in the recent literature. Compared to references [13,14,15,16,17], which generally utilize fixed or single-stage tuning mechanisms, the proposed circuit uniquely incorporates a dual coarse- and fine-tuning mechanism. This approach enables enhanced dynamic reconfigurability of the stopband frequencies, allowing precise and flexible control over filtering characteristics. Furthermore, when compared to the design presented in [19], the proposed filter demonstrates superior performance by achieving a wider 15-dB stopband fractional bandwidth (15-dB SB FBW) and greater maximum attenuation (Max. Att.), which are critical for effective interference suppression. Benefiting from the distinctive transmission pole and zero distribution inherent to the filter topology, the structure attains exceptional stopband rejection across the targeted Ku- and Ka-band frequencies, making it highly suitable for advanced RF and millimeter-wave applications.

Table 1.

Performance comparison with other state-of-the-art filters.

Ref.	Technology	Mode	fcs/T.R. (GHz)	15-dB SB FBW	Max. Att. (dB)	fcp (GHz)	3-dB BW (GHz)	IL (dB)	RL-Max dB)	Size (mm2)
[13]	0.13-μm BiCMOS	-	59.4 /Static	6.7%+	20	35.3	7.7	1.7	25+	0.039
[14]	0.13-μm BiCMOS	-	67 /Static	18.2%+	25	40	8.2+	1.7	19	0.012
[15]	0.13-μm BiCMOS	-	24.5 /Static	6% *	23.1	45	-	0.95	24	0.038
[16]	0.18-μm CMOS	-	53.2 /Static	5.6% *	14.2	39.8	-	0.5+	23.1	0.024
[17]	0.18-μm CMOS	-	30, 85 /Static	37%, 11%	25,18	55	-	4.6+	20	0.6
			30, 80 /Static	39%+, 10%+	35,25	55	-	2.5	50
[19]	65-nm CMOS	-	24.9–28.7 /14.2%	40.2%	32.1–35	36.5–38+	4–7	2.7	17	0.036
			24.6–29.8 /19.1%	38.2%	21.6–29.5	36.5–38.8+	2.5–7	3.5	33
[20]	GaAs IPD	-	29.5 /Static	16%	42	21/36	-	0.5	20	0.9
[21]	VO2	Work1	-	-	-	30.1–36.5	22–22.3	2.7–4.4	20	0.184
		Work2	-	-	-	38.5–49	4.7–6	1.4–3	30	0.51
This work	65-nm CMOS	LF	14.5–19.7 /30.4%	58%	31.4–33.3	5.7–7.2@Ku 35.4–39@Ka	5.9–8.7@Ku 22.7–29.2@Ka	2.3–6.3@Ku 5.1–9.1@Ka	24.2@Ku 27.6@Ka	0.064
		HF	24.1–26.9 /11%	61.7%	40.1–58.2	7.9–8.6@Ku 42–44@Ka	10.4–11.7@Ku 12–16@Ka	2–5.2@Ku 5.8–10.6@Ka	44@Ku 42.3@Ka

T.R.: tuning range; SB: stopband; Att: attenuation; fcs: stopband center frequency. fcp: passband center frequency; RL-Max: maximum passband return loss value; 3-dB BW: 3-dB passband bandwidth. *: 10-dB stopband FBW; +: evaluated graphically.

4. Conclusions

A compact, multi-mode reconfigurable filter implemented in a 65-nm CMOS process is presented for Ku/Ka multi-band applications. A pole–zero tuning scheme combining coarse and fine control is utilized. By employing lumped-element LC series and parallel resonant networks, two poles and two transmission zeros are generated, with the zeros inherently clamped between the poles due to the filter topology. The proposed design features a compact footprint, a wide stopband, and pole–zero tracking capabilities. Good agreement is obtained between the simulation and measurement results. The proposed filter structure can be a candidate for receiver systems that require wideband interference rejection, compact footprint, and continuous tunability.

Author Contributions

Methodology, J.F. and X.W.; data curation, X.W. and J.F.; Conceptualization, X.W. and J.W.; writing—original draft, J.F.; supervision, J.W. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

This work was supported by the Zhejiang Provincial Natural Science Foundation of China under Grant No. LD25F040005.