Compact Dual-Mode Triple-Band Bandpass Filter

Article information

J. Electromagn. Eng. Sci. 2025;25(4):397-399

Publication date (electronic) : 2025 July 31

doi : https://doi.org/10.26866/jees.2025.4.l.28

Youngje Sung^a

Department of Electronic Engineering, Kyonggi University, Suwon, Korea

^*Corresponding Author: Youngje Sung (e-mail: yjsung@kgu.ac.kr)

ahttps://orcid.org/0000-0002-9310-7450

Received 2024 August 17; Revised 2024 November 1; Accepted 2024 December 2.

Abstract

I. Introduction

Addressing the need for multi-band microwave devices to support various frequency services, stepped impedance resonators (SIRs) or uniform impedance resonators are widely employed in the design of multi-band filters [1–3]. Notably, since these structures implement separate paths for each resonance band, at least two or more SIRs are used in the case of a triple-band filter. Therefore, drawing on the composite right/left-hand structure, [4] proposed a triple-band filter with an independent current path for each resonance band. However, its applications are limited by the disadvantages of large insertion loss and size.

Recently, stub-loaded resonators (SLRs) have been proposed for the design of triple-band bandpass filters. For instance, a triple-band filter was implemented using a T-shaped SLR in [5]. In [6], the structure of the filter was further improved by employing a stub-loaded SIR to enable triple-passband responses. Meanwhile, [7] presented a compact triple-band bandpass filter (BPF) design using mixed-coupled resonators. In addition, balanced triband BPFs that use SLR [8] or embed multi-band coupled complementary split ring resonators [9] have also been proposed.

Table 1 presents a comparison of the proposed filter with previously reported triple-band BPFs with regard to operating frequency, insertion loss, bandwidth (BW), and number of transmission zeroes (TZs), revealing that the former offers the advantages of good in-band performance, wide stopband, and compact size.

Table 1

Comparison with previously reported triple-band bandpass filters

This work presents the design procedure for a tri-band filter operating at 1.93 GHz for digital enhanced cordless telecommunications (DECT) and at 3 GHz and 3.3 GHz for WiMAX.

II. Triple-Band BPF Configuration

As shown in Fig. 1(a), the length L₁ of the slot in the proposed triple-band BPF can be divided into L_1,long and L_1,short, signifying its short part. Therefore, L₁ = L_1,long + L_1,short + W₁. Furthermore, the cross-shaped slot is moved along the right diagonal by d₁ in the –x axis and d₁ in the –y axis. A pair of slits (length L₂ and width W₂) is placed at distance d₂ from the corner, while a stub of length L₃ and width W₃ is applied to the feed line. Notably, a substrate of ɛ_r = 2.2 and h = 0.508 mm was used. The simulation was carried out using the High-Frequency Structure Simulator (HFSS).

Fig. 1

Proposed triple-band BPF: (a) top view and (b) fabricated structure.

III. Operating Principle

The design process of the proposed filter, as depicted in Fig. 2(a), involved the following steps:

Fig. 2

Design evolution of the proposed triple-band BPF: (a) geometry, (b) simulation results at low frequency, and (c) simulation result at high frequency.

• First design: Since the filter features a symmetrical structure with respect to the right diagonal, f₁, f₂, and f₃ were formed at 2, 3, and 3.5 GHz, respectively.

• Second design: The feed line was moved by 1 mm. Accordingly, f₃, which was weakly formed in the first step design, was now observed to be strongly formed. However, the BW at f₃ could not be implemented properly.

• Third design: On setting L_1,long and L_1,short in a different way in the slot, f₂ and f₃ were formed more clearly than when using the second step design. However, S₂₁ was very weakly formed at −5 dB in the 3.1 GHz band, due to which the TZ for distinguishing the two bands (f₂ and f₃) was very weak.

• Fourth design: The cross-shaped slot was moved. Accordingly, S₂₁ improved from −5 dB to −10 dB.

• Fifth design: The slits were moved. Accordingly, S₂₁ reached - 20 dB, and good TZ characteristics were realized. Furthermore, L₁ = 24.5 mm, △(= L_1,long – L_1,short) = 4 mm, W₁ = 0.5 mm, d₁ = 1.2 mm, L₂ = 15.5 mm, d₂ = 2.3 mm, W₂ = 1 mm, L₃ = 14 mm, d₃ = 2.5 mm, and W₃ = 0.7 mm.

IV. Parameter Study

Fig. 3 illustrates the simulation results obtained on changing the length L₂. Notably, to exclude the influence of other asymmetries, the first design was employed, and only d₃ was changed, with △ = 0 mm and d₁ = 1.2 mm. The simulation results revealed that f₃ was formed by the stub of the feed line. This finding was verified based on two reasons. First, f₃ disappeared when the stub was removed. Second, f₁ and f₂ hardly changed with a decrease in d₃, whereas f₃ was both strongly influenced and strongly formed.

Fig. 3

Simulation results based on changes in length d₃.

Fig. 4 presents the simulated performance based on the presence and absence of the slits and slot. Notably, this analysis was conducted to accurately grasp their influence. The fifth design in Fig. 2 was employed for this purpose.

Fig. 4

Simulation results of the filter with and without the cross-shaped slot and the slit.

The filter with a slit at the edge exhibited triple-band performance, but the TZ at 2.6 GHz disappeared when the slit was removed, leaving only f₁ and f₃. Moreover, although the characteristics of the two filters (5th filter and 5th filter without slit) in the f₁ band were almost similar, when the slit placed at the edge was removed, the current path decreased, and f₁ increased slightly. Also, the characteristics in the f₃ band are almost similar. Based on these observations, it became apparent that f₂ was formed by the slit at the edge.

When the slot placed at the center was removed, f₁, f₂, and f₃ were all affected, although f₁ changed the most. Owing to the slot’s location at the center, the current path increased, while the resonant frequency of the patch decreased. Furthermore, as shown in the simulation results, in the case of the structure without a cross slot, the first band of the proposed structure formed at 2.5 GHz, which is considerably higher than 2 GHz.

Fig. 5 shows the simulation results obtained for the fifth design on changing L₂. Here, the slit at the edge played an important role not only in the formation of f₂ but also in ensuring that the proposed structure exhibits triple-band characteristics by forming a TZ between f₂ and f₃. The simulation results clearly show that as L₂ increased from 15.5 mm to 17.5 mm, the TZ between f₂ and f₃ declined from 3.1 GHz to 2.8 GHz.

Fig. 5

Simulation results of the filter at different L₂.

Fig. 6

Simulated and measured results.

V. Simulated and Measured Results

In this study, a triple-band BPF was successfully designed at center frequencies of 1.96, 2.94, and 3.34 GHz, exhibiting insertion losses of 1.24 dB, 1.35 dB, and 1.62 dB at f₁, f₂ and f₃, respectively. Furthermore, the measured BWs for f₁, f₂ and f₃ were 14.2%, 7.5%, and 10.9% respectively. Notably, the measured resonance frequency decreased slightly due to the fringing effect at the edge.

VI. Conclusion

In this letter, a novel microstrip tri-band BPF is proposed and designed using a square patch. f₁, centered at 1.96 GHz, was achieved using a cross slot, while f₂ and f₃, located at 2.94 GHz and 3.34 GHz, respectively, were achieved using a pair of slits and stubs. Moreover, each passband can be controlled individually.

Article information Continued