Optimization of a Microstrip Tunable Bandpass Filter

Design
2021 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (ElConRus) | 978-1-6654-0476-1/20/$31.00 ©2021 IEEE | DOI: 10.1109/ElConRus51938.2021.9396434

Victoria A. Karpova Nikita V. Ivanov

Peter the Great St. Petersburg Polytechnic University Peter the Great St. Petersburg Polytechnic University
St. Petersburg, Russia St. Petersburg, Russia
[email protected] [email protected]

Abstract— In this paper, an optimization procedure of a II. TUNABLE PARALLEL-COUPLED FILTER DESIGN
microstrip tunable bandpass filter design is represented. The
filter structure consists of parallel-coupled half-wavelength In this paper a parallel-coupled structure of a microstrip
resonators and variable capacitors, which provide the required filter was considered. Filter requirements are as follows:
tuning range of the center frequency. The simulation results were - Center frequency: 1500 MHz
obtained using ADS (Advanced Design System) software. The
method based on ladder type prototype and ADS LineCalc were - Insertion loss: less than 3 dB
used to get the dimensions of microstrip lines. After bandpass
filter synthesis the optimization procedure was realized to - Passband ripple: less than 0.1 dB
improve the filter characteristics. For this purpose, ADS tools - Bandwidth: 80 MHz
were used too. Simulation results confirm efficiency of proposed
optimization approach. - Stopband attenuation (@1580 MHz): more than 40 dB

Keywords— bandpass filter; microstrip lines; ADS; The filter design procedure starts with determining of the
optimization order and coefficients of low-pass filter. The procedure of
ladder type prototype synthesis is described in well-known
I. INTRODUCTION guides, e.g. [4]. In this paper a 7th order prototype and
Filter is a significant component for RF and microwave Chebyshev approximation is used.
applications. The function of a filter is to pass the frequencies After that a ladder into coupled resonators prototype
in a certain range and to attenuate the others. There are conversion should be done. This process consists of two steps.
different ways how filters can be implemented depending on First, characteristic admittances of J-inverters (Jij) and
specifications and requirements. They can be based on lumped characteristic admittances of the terminating lines (Y0) are
elements or transmission lines, such as microstrip, coaxial and determined. Y0 is typically 1/50 Ohm-1. To realize the J-
wave guide. inverters, the even and odd mode characteristic impedances are
One of the bandpass filter applications is using them as a calculated. Design equations can be found in [5, 6].
part of satellite receiver, where they perform a pre-selection Microstrip coupled lines length l, width w and space
function and attenuate out-of-band signals. There are specific between lines s were calculated using ADS LineCalc. It is an
requirements that should be imposed on filters included into analysis and synthesis program for calculating physical and
satellite systems. First of all, environmental conditions should electrical parameters of single and coupled transmission lines.
be taken into account. Furthermore, devices should have good This program computes dimensions of microstrip lines using
performance at ultra-high frequencies (UHF) in the range even- and odd-mode characteristic impedances (Ze and Zo) and
between 300 MHz and 3 GHz. For example, radio navigation information about the substrate. The parameters of substrate
systems such as GPS and GLONASS operate in L-band (1-2 Arlon AD600 and other information are set in MSub controller
GHz). For these purposes, microstrip bandpass filters can be and are as follows:
chosen because of their high tolerance to the environmental
conditions and good performance in microwave region. The - Substrate thickness H=2.286 mm
matching problem is usually solved at the filter design stage. - Substrate relative dielectric constant Er=6.5
Also microstrip technology is quite low cost, compared to the
lumped elements operating in this frequency range. The other - Relative permeability Mur=1
advantage of microstrip filter is possibility to change the center - Metal conductivity Cond=5.8e+7
frequency of the device by connecting elements with variable
parameters, such as varactor diodes [1 – 3]. - Hu=1.0e+33
The connection of tuning elements leads to distortion of the - Metal layer thickness T=0.017 mm
frequency response. To improve S-parameters of designed - Dielectric loss tangent TanD=0.003
filter it is necessary to optimize the schematic circuit. An
optimization procedure is introduced in this paper based on For the center frequency of the bandpass filter to be
ADS (Advanced Design System) software. tunable, elements with variable capacitance are introduced in
the structure. In [7] it is shown that varactor load at both open
ends of microstrip lines provides better results in comparison

129 978-1-6654-0476-1/21/$31.00 ©2021 IEEE

Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 15:02:39 UTC from IEEE Xplore. Restrictions apply.
with single-ended load. The structure of bandpass filter is 1545 MHz. Insertion losses must be less than 3 dB. Attenuation
illustrated in Fig.1. in the stopband is set to be more than -40 dB. Optimization
variables can be set using VAR component. These variables are
parameters to which the circuit performance is sensitive.
Because the designed bandpass filter is sensitive to the change
of the microstrip lines dimensions (l, s and w), these parameters
were determined as variable components. The structure of the
filter based on parallel-coupled half wavelength resonator is
symmetrical. Consequently, only four sets of microstrip lines
parameters should be considered (l1, w1, s1), (l2, w2, s2), (l3,
w3, s3) and (l4, w4, s4). Optimizers define the approach to
adjust the optimization parameters to meet the goal. The
schematic circuit for optimization is presented in Fig. 2.
Simulation results before and after optimization are shown in
Fig. 3 and Fig. 4 respectively.

Fig.1. Schematic circuit of a microstrip tunable bandpass filter

Dimensions of the microstrip lines are presented in Table 1.

TABLE I. DIMENSIONS OF THE MICROSTIP LINES

Length, mm Width, mm Space, mm
1,8 23.689 2.5023 0.7907
2,7 23.1401 3.1116 3.7065
Fig. 2. Components needed for optimization
3,6 23.1018 3.1372 5.2202
4,5 23.0983 3.1395 5.5071

To get good performance it is necessary to use elements

with very low capacity (about 0.3 pF). It can cause problems
with the implementation of the circuit on real elements.
It is possible to use optimization tools provided by ADS
software to improve S-parameters of the microstrip bandpass
filter. This process is discussed in the next chapter.
Furthermore, it should be noted, that capacitors connected to
the input and output microstrip lines have an impact on the
characteristics of the filter, too. Their influence is the main
point of chapter IV.
III. AN OPTIMIZATION PROCEDURE OF A MICROSTRIP TUNABLE
FILTER Fig. 3. Simulation results before optimization
Optimization procedure plays an important role in
engineering process. It helps in finding the optimal design to
meet the design specifications. There is a special tool in ADS
software that can be used to optimize the circuit. Overall,
optimization can be defined as a process of changing an array
of parameters to meet specific performance requirements.
Optimizers match computed and required responses and
modify design parameters to minimize the difference between
computed and desired response.
To start an optimization process it is necessary to place in
the schematic window at least three components: goal,
optimizer and variables [8]. Goal controllers are used to specify
performances to be met. In our case the goal functions were
built for reflection coefficients, insert losses in the passband
and attenuation in the stopband. The maximum value of S11
Fig. 4. Simulation results after optimization
and S22 was set to be -15 dB in frequency range from 1455 to

130

Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 15:02:39 UTC from IEEE Xplore. Restrictions apply.
 The updated dimensions of microstrip lines obtained after
optimization procedure are presented in Table II.

TABLE II. DIMENSIONS OF THE MICROSTRIP LINES AFTER OPTEMEZATION

Length, mm Space, mm Width, mm
1,8 22.7351 0.9767 2.6211
2,7 23.2492 4.9342 3.1484
3,6 23.1447 5.7154 3.3490
4,5 23.2401 6.0916 3.3138

IV. INVESTIGATION OF THE IMPACT OF INPUT AND OUTPUT

CAPACITORS ON FILTER CHARACTERISTICS
The change of the capacitance is realized using voltage
Fig. 6. S-parameters of the tunable filter without input and output capacitors
source connected to the tuning elements. Let’s consider the
input signal of the device. Bandpass filters act usually as pre- The other way to solve the issue with DC offset is to
selectors in satellite radionavigation systems and receive replace input and output variable capacitors by elements with
signals from an antenna. These signals have generally zero DC constant values, which can be obtained empirically. These
offset. The connection of the voltage source to the input of the values were set so that they provided maximal tuning range.
microstrip line leads to the DC bias increase, and therefore the Tuning range is defined by the center frequency range, in
filter will not operate correctly in some cases. It is necessary to which S-parameters satisfy the following requirements:
pay attention to the output signals too. The applied DC bias has reflection coefficient is less than -10 dB, bandwidth is equal 80
variable value and depends on the center frequency tuning. In MHz, insertion losses are less than 3 dB and attenuation in the
that case correction system should be included into the filter to stopband is more than 40 dB. Minimum value of variable
provide constant DC level. capacitors is equal 0.6 pF. It is important that input and output
capacitors should have the same values. Otherwise the
There are two ways how to avoid DC voltage application to structure becomes unsymmetrical, consequently S21≠S12 and
the microstrip lines. Removing varactors from the input and S11≠S22. Simulation results in ADS software are represented
output of the circuit (Fig.5) doesn't lead to positive results. S- in Fig. 7, where two set of S-parameters are shown. The 1500
parameters of such filter model are represented in Fig. 6. The MHz center frequency is reached at the capacitance 0.6pF. If
presented simulation results were obtained at 0.6 pF tuning we increase this value up to 1.29 pF the center frequency will
capacity value. Fig. 6 shows that reflection coefficient becomes decrease and reach 1274 MHz. As a result, the tuning range is
higher than required (more than -10 dB). Thus, ripples in the equal to 226 MHz.
passband rises. No matter that the center frequency and the
bandwidth don’t change, as in-band input and reflection losses
are much more important.

Fig. 7. S-parameters of the tunable filter with constant values of input and
output capacitors

Fig. 5. Schematic circuit of a microstrip tunable bandpass filter without input

V. CONCLUSION
and output capacitors In this paper an optimization of a microstrip tunable
bandpass filter based on parallel-coupled half wavelength
resonators for L-band application is presented. Simulation
results confirm that optimization tool given by ADS software is

131

Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 15:02:39 UTC from IEEE Xplore. Restrictions apply.
an efficient addition to the conventional method of microstrip [3] F. Lin and M. Rais-Zadeh, “A tunable 0.6 GHz – 1.7 GHz bandpass
filter design. Impact of input and output capacitors is also filter with a constant bandwidth using switchable varactor-tuned
resonators,” IEEE MTT-S International Microwave Symposium,
discussed in the paper, because of the voltage bias that can Phoenix, AZ, 2015, pp. 1-4.
cause waveform distortion during the signal processing in [4] Saal R., Handbook of Filter design, Berlin: AEG- Telefunken, 1979.
blocks after pre-selector. There are two ways how to avoid this [5] Jia-Sheng Hong, M. J. Lancaster, “Microstrip Filters for RF/Microwave
problem. Simulation results show that the best solution is to Applications”, 1st ed., John Wiley & Sons, Inc., 2001.
replace variable tuning elements by fixed capacitors. The [6] N.V. Ivanov, “A new approach to microstrip coupled-resonator bandpass
proposed tunable filter shows a bandwidth of 80 MHz with a filter design,” Proc. IEEE NW Russia Young Researchers in Electrical
continuous tuning range of center frequency of 1274-1500 and Electronic Engineering Conference, January 28 – February 1, 2018,
MHz. St. Petersburg, Russia, pp. 217-219.
[7] V. Karpova, N. Ivanov, “A symmetrical design of a microstrip tunable
REFERENCES bandpass filter,” Proc. YETI International Youth Conference on
Electronics, Telecommunications and Information Technologies, July 10
[1] A. Zakharov, S. Rozenko and M. Ilchenko, “Varactor-Tuned Microstrip – 11, 2020, St. Petersburg, Russia, pp. 159-167.
Bandpass Filter With Loop Hairpin and Combline Resonators,” IEEE
Transactions on Circuits and Systems II: Express Briefs, vol. 66, no. 6, [8] Yanling Hao, Bingfa Zu and Ping Huang, “An optimal microstrip filter
pp. 953-957, June 2019. design method based on Advanced Design System for satellite receiver,”
2008 IEEE International Conference on Mechatronics and Automation,
[2] W. Tang and J. Hong, “Reconfigurable microstrip combline filter with Takamatsu, 2008, pp. 903-907.
tunable center frequency and bandwidth,” Asia-Pacific Microwave
Conference 2011, Melbourne, VIC, 2011, pp. 1162-1165.

132

Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 15:02:39 UTC from IEEE Xplore. Restrictions apply.