1. Introduction
Recently, with the development of wireless communication, terminal technology that supports multiple frequency bands simultaneously with a single system is receiving attention. Switchable filters are being presented as an effective alternative in that they can selectively provide multiple modes based on a single structure [
1,
2]. In particular, switchable filters utilizing PIN diodes are being actively developed because they allow for fast switching speeds and relatively simple circuit configurations.
In [
3], a switchable filter with four distinct modes was proposed by integrating a bandstop filter (BSF)/all-stop filter (ASF) structure with a dual-mode resonator (DMR) in a parallel configuration. In [
4], three modes of bandpass filter (BPF), BSF, and DB-BPF were implemented based on coupled-lines. In [
5], a switchable filter capable of implementing two BPFs and one BSF was proposed, while in [
6], a switchable filter designed to implement three modes—BPF, DB-BPF, and BSF—was proposed by combining one main path and an additional path.
In [
7], a PIN diode was used to switch between BPF and BSF, and a varactor diode was used to control their resonant frequencies. Similarly, in [
8], a structure enabling switching between two modes, BPF and BSF, was proposed, with an absorptive circuit added to enhance the matching characteristics of both modes. However, this structure has the disadvantage of being large in size and having many design parameters because it places multiple structures in parallel and selects one of them using a PIN diode.
To address these issues, research on implementing filter structures directly on 50 Ω microstrip lines has been attracting attention [
9]-[
12]. In [
9], a sext-band BPF was proposed by adding a symmetrical short stub outside the 50-Ω line. In [
10], a filter capable of switching between three modes—a broadband filter, a dual-band filter, and a tri-band filter—was implemented by integrating a stub with a PIN diode into the 50-Ω transmission line. However, both studies had the problem of increasing the overall filter size due to the long stub. In [
11], a structure capable of switching between BPF and DBPF was proposed using a switchable J-inverter. However, it still had the problem of large size, with a large IL (Insertion Loss) of 2.9–4.9 dB using 13 diodes. In [
12], a dual-band notch filter with tunable frequency was implemented by applying an embedded split ring resonator (ESRR) structure loaded with a varactor inside a 50-Ω line. In [
13], a simple and compact quadruple-band pass filter was implemented by inserting a C-shaped open-loop resonator inside a 50-Ω line, but the spacing between resonant frequencies is fixed, limiting practical applications.
Table 1 presents the performance comparison of previous switchable filters.
In this study, a switchable filter supporting four operation modes including a wideband filter, two DB-BPFs with different operating bands, and an APF is proposed. Unlike previous works, the proposed filter minimizes the size increase by implementing the bias circuit in the empty space inside the proposed structure.
4. Switchable Filter Design with Bais Circuit
Figure 12 (a) shows a switchable filter with five diodes and a bias circuit to operate them. The configuration of the switchable filter is the same as
Figure 1 except for the bias circuit and diodes. The two
D2s located on the left and right gaps and the two
D3s located on the upper and lower outer gaps are each configured as a pair. The bias circuit consists of an inductor (green), a capacitor (blue), and a resistor (black). As is well known, the thinner and shorter the bias line, the better. Accordingly, the line width was set to 0.1 mm and the length to 5.6 mm. To implement the bias circuit in a limited space, the capacitor is located on the back side where the GND is located.
Li1 = 3,
Li2 = 1mm are set, and other parameters are maintained as optimized values obtained based on the parameter study performed in Section III.
Figure 12b shows the equivalent circuit of the diode used. It acts as a resistor in the On state and as a capacitor in the Off state.
Ron is 5.2 ohm,
Coff is 0.025 pF, and
LS is about 0.5 nH.
When simulating the switchable filter including the bias circuit, the equivalent values of the diode are all considered. The bias voltages used to control the On/Off state of each diode are shown in
Table 2. In the case of
D3, the bias voltage is applied by connecting a bias tee to the ports at both ends of the line. The voltages applied to the bias tees are indicated as
V4 and
V5 in
Figure 12a, respectively.
Figure 12c shows the fabricated switchable filter.
Figure 13 shows the simulated filter characteristics when the values of
R and
L used in the bias circuit are changed. The values of
R and
L used in the bias circuit have a large effect when the proposed structure operates as a dual-band filter, but have a relatively small effect when it operates as a wideband filter. For this reason, only simulation results applying the bias circuit to the dual-band filter are presented in this paper. As the inductor value increases, the insertion loss in the 1.5 GHz band, which is
f1, tends to decrease. At this time, the insertion loss in the 5.5 GHz band, which is
f2, showed almost no change. Since the inductor has a large impedance value as the frequency increases, the impedance is relatively large in the
f2 band, which is a relatively high frequency. Accordingly, the RF signal flowing through the bias circuit can be appropriately blocked. On the other hand, the impedance is relatively small in the
f1 band, so some RF signals flow into the bias circuit. This is because the insertion loss in
f1 improves as the inductor value increases.
However, when the inductor value is 90 nH, S21 ripple is formed in the f2 band. Among the three inductors, the insertion loss could be reduced by lowering the inductor value of the V2 section to 68 nH. When no resistor is used in the bias circuit, the insertion loss is the smallest in the 1.5 GHz band, but resonance occurred at 0.6 GHz and strong ripple occurred in the 5.5 GHz band. When a 0.3 Kohm resistor is used, the insertion loss in the f1 band decreases, but the insertion loss in the f2 band increases.
Figure 14 shows the electric field distribution when the inductor values used in the bias circuit are 29 nH and 90 nH/68 nH, respectively. At this time, the simulation is performed at
f1. In the case of 29 nH, it can be confirmed from the simulation results that the electric field is also formed in the bias circuit. On the other hand, when 68 and 90 nH are mixed, there is almost no electric field formed in the bias circuit. Based on these facts, it can be confirmed that the small inductor value is not sufficient to operate as an RF chock to prevent the electric field from passing through to the bias circuit.
Figure 15 shows the simulation results of the effect of the bias circuit applied position
Li1 on the filter performance. As the bias applied position changes, there is little effect in other modes, but the effect is very strong in the BSF mode where the
D2 diode is turned on. From the simulation results, as
Li1 increases, the insertion loss of
S21 tends to decrease in both resonant frequency bands. However, when
Li1 is 5 mm, there is a problem of
S21 ripple occurring in the 5.5 GHz band. Considering these facts,
Li1 is selected as 3 mm.
The switchable filter operates as two types of dual-band BPF, wideband BPF, and all-pass filter depending on various bias combinations. The simulation and measurement results for these four cases are presented in
Figure 16. It is confirmed that some measured TZ occurred at a slightly lower frequency than the simulation. This is because of the error that occurred during the fabrication of the designed structure and the inaccuracy of the equivalent model of the PIN diode applied to the simulation at high frequencies. Considering this fact, the simulation and measurement results can be seen to be in good agreement overall. When
D1 is off,
D2 is on, and
D3 is off, the switchable filter operates as a DB-BPF, and the results are presented in
Figure 16a. The measured 3-dB fractional bandwidths centered at
f1 = 1.4 and
f2 = 5.1GHz are 23.49 and 14.42%, respectively. The measured insertion losses at this time are −1.33 and −1.3 dB, respectively. On the other hand, when
D1 is off,
D2 is off, and
D3 is off, the switchable filter operates as a DB-BPF, and the operating bandwidths are
f1 = 2.4 and
f2 = 4.2GHz, respectively, and the measured 3-dB fractional operating bandwidths are 23.33 and 13.27%. The insertion losses are −0.88 and −2.04 dB, respectively.
Figure 16c shows the result when the switchable filter operates as a wideband BPF. At this time, the bias conditions are
D1 on,
D2 off, and
D3 off. The measured 3-dB fractional bandwidth is 73.36% (2.05–4.41 GHz) and the insertion loss is −0.69 dB.
Figure 16d shows the result when operating as an APF. At this time, the bias conditions are
D1,
D2 don’t care, and
D3 on. The measurement results show that the APF exhibits an insertion loss of −2.34 dB in the entire operating frequency range of 0–7 GHz, confirming that signal transmission is efficient.
Figure 1.
Basic filter implemented on a simple 50-ohm line.
Figure 1.
Basic filter implemented on a simple 50-ohm line.
Figure 2.
(a) Various filter structures and (b) their corresponding simulation results.
Figure 2.
(a) Various filter structures and (b) their corresponding simulation results.
Figure 3.
Wideband bandpass filter. (a) layout. (b) Transmission line model. (c) even-mode equivalent circuit. (d) odd-mode equivalent circuit.
Figure 3.
Wideband bandpass filter. (a) layout. (b) Transmission line model. (c) even-mode equivalent circuit. (d) odd-mode equivalent circuit.
Figure 4.
(a) Simulation results of type 4 according to stub length s. (b) E-field distribution at the transmission zero frequency.
Figure 4.
(a) Simulation results of type 4 according to stub length s. (b) E-field distribution at the transmission zero frequency.
Figure 5.
Simulation results of Type 4 according to length L.
Figure 5.
Simulation results of Type 4 according to length L.
Figure 6.
Simulation results of Type 4 according to LC length. (a) S11 (b) S21.
Figure 6.
Simulation results of Type 4 according to LC length. (a) S11 (b) S21.
Figure 7.
Simulation results of Type 3 according to length s.
Figure 7.
Simulation results of Type 3 according to length s.
Figure 8.
E-field distribution at (a) fTZ1 and (b) fTZ2.
Figure 8.
E-field distribution at (a) fTZ1 and (b) fTZ2.
Figure 9.
Simulation results of type 3 according to the length LC of the C-shaped resonator.
Figure 9.
Simulation results of type 3 according to the length LC of the C-shaped resonator.
Figure 10.
E-field distribution (a) f1=3 GHz (b) f2=5GHz.
Figure 10.
E-field distribution (a) f1=3 GHz (b) f2=5GHz.
Figure 11.
Simulation results of Type 3 according to the change in length L.
Figure 11.
Simulation results of Type 3 according to the change in length L.
Figure 12.
(a) Proposed filter with bias circuit included (b) Equivalent circuit of the MADP-000907-14020 PIN diode used (c) Photograph of the fabricated switchable filter.
Figure 12.
(a) Proposed filter with bias circuit included (b) Equivalent circuit of the MADP-000907-14020 PIN diode used (c) Photograph of the fabricated switchable filter.
Figure 13.
Simulation results of switchable filter according to the values of resistors and inductors used in the bias circuit. (a) S11. (b) S21.
Figure 13.
Simulation results of switchable filter according to the values of resistors and inductors used in the bias circuit. (a) S11. (b) S21.
Figure 14.
E-field distribution at f1 when the inductor is (a) 29nH (b) 90nH and 68nH.
Figure 14.
E-field distribution at f1 when the inductor is (a) 29nH (b) 90nH and 68nH.
Figure 15.
Simulated filter performance with different bias positions Li1.
Figure 15.
Simulated filter performance with different bias positions Li1.
Figure 16.
Simulated and measured results. (a) dual-band filter (D1: off, D2: on, D3: off) (b) dual-band filter (D1: off, D2: off, D3: off) (c) wideband filter (D1: on, D2: off, D3: off) (d) all pass filter (D1: don’t care, D2: don’t care, D3: on).
Figure 16.
Simulated and measured results. (a) dual-band filter (D1: off, D2: on, D3: off) (b) dual-band filter (D1: off, D2: off, D3: off) (c) wideband filter (D1: on, D2: off, D3: off) (d) all pass filter (D1: don’t care, D2: don’t care, D3: on).
Table 1.
Comparison with previous switchable filters.
Table 1.
Comparison with previous switchable filters.
|
Ref.
|
Function |
BW[%]
|
IL [dB] |
Number of Diodes |
Circuit Size [λg×λg] |
| [3] |
BSF/ASF/BPF/DBSF |
50.2 / - / 9.7 / 22.1&20 |
1.99 |
4 |
0.21×0.22 |
| [4] |
BPF/BSF/DBSF |
53.1 / 30.2 / 3.8 & 2.7 |
0.9 / 2 / 2.2 |
6 |
0.44×0.29 |
| [5] |
WBPF / BPF / BSF |
95.7 / 25.3 / 106 |
0.94 / 0.8 |
6 |
0.5×0.25 |
| [6] |
BPF/DBPF/BSF |
84.5 / 39.7 & 24.9 / 29.6 |
0.9 / 1.13 & 1.5 / 2 |
6 |
0.61×0.27 |
| [7] |
DBPF/DBSF |
25.6 & 20.4 / 30 & 23.1 |
4.6 / 5.6 |
7 |
0.75×0.4 |
| [8] |
BPF/BSF |
36.75 / 13 |
1.73 / 0.8 |
7 |
0.74×0.6 |
| [10] |
BBPF/DBPF/TBPF |
36.7 / 20.7 & 16.1 /20 & 27.4 & 18.1 |
0.5 / 0.4 / 0.6 |
7 |
0.75×0.4 |
| [11] |
BPF/DBPF |
36.07 / 26.42 & 22.92 |
2.9-4.9 |
13 |
0.25×0.25 |
| This work |
DBPF/DBPF/WBPF/APF |
23.5 & 14.42 / 23.33 & 13.27 / 73.36 / - |
1.33 & 1.3 / 0.88 & 2.04 / 0.69 / 2.34 |
5 |
0.4×0.05 |
Table 2.
Applied bias voltage conditions and operating modes.
Table 2.
Applied bias voltage conditions and operating modes.
| Condition [V] |
Diode State |
OperatingMode
|
| V1 |
V2 |
V3 |
V4 |
V5 |
D1 |
D2 |
D3 |
| 4 |
1 |
0 |
0 |
1 |
On |
Off |
Off |
Wideband |
| 1 |
3 |
6 |
0 |
1 |
Off |
On |
Off |
Dual-band |
| 2 |
4 |
0 |
0 |
1 |
Off |
Off |
Off |
Dual-band |
| - |
- |
- |
1 |
0 |
- |
- |
On |
All pass |