Вы находитесь на странице: 1из 6

696 PIERS Proceedings, Hangzhou, China, March 24-28, 2008

Design and Optimization of Microstrip Interdigital Bandpass Filters


with Impedance Matching
Homayoon Oraizi, Nima Azadi-Tinat, and Shahrokh Saeedi
Iran University of Science and Technology, Narmak, Tehran 16846-13114, Iran

Abstract— A least square based method is developed for the design and optimization of mi-
crostrip interdigital bandpass filters. The N coupled transmission line theory is employed to take
into account the couplings among adjacent and nonadjacent conducting strips. The algorithm
also incorporates the load and source impedance matching. The minimization of error function
determines the length, width and spacing of strips. The performance of the optimized filter
design perfectly agrees with the microwave simulation softwares.

1. INTRODUCTION
Interdigital filters were fabricated by inserting metallic round rods in rectangular cavities [1, 2],
and were fabricated to obtain better than 10 percent bandwidth [3, 4]. Since that time, different
types of interdigital filters have been devised and used in microwave circuits. Microstrip interdigital
filters have also been developed, but their design procedures have usually ignored the significant
nonadjacent resonant strip couplings. As reference [5] indicated that the coupling among consecu-
tive strips on nonhomogeneous microstrip structures decreases gradually rather than abruptly, the
nonadjacent strip couplings need to be considered in the design. Furthermore, the N different phase
velocities on the N coupled strips need to be included in the design procedure.
Accordingly, in this paper a numerical procedure based on the method of least squares is pro-
posed for the design of the interdigital bandpass filters, which utilizes the theory of N coupled
transmission lines. The adjacent and nonadjacent couplings among strip lines, their distinct phase
velocities, and dispersion relations are readily incorporated in the design method. Furthermore,
impedance matching of load and source impedances over a frequency bandwidth is also incorpo-
rated into the design algorithm. In this paper two interdigital filter configurations with coupled
feeds for narrow and wide band applications are considered as illustrations for the optimum design
algorithm. The same general development has been applied for the optimum design of microstrip
combline filters [6–8].
2. NUMERICAL PROCEDURE
Consider the general circuit configuration of the narrow and wide band microstrip interdigital
bandpass filters, as shown in Fig. 1 and Fig. 2, respectively.
The common length of strips are L, the width of the ith strip is wi , the gap spacings between
adjacent strips is si , the source admittance is YS and the load admittance is Y1 , as shown in Figs. 1
and 2.
The transmission line equations of the N coupled lines are [9]:
d
[V ] = −(jω[L] + [R])[I]
dz
d
[I] = −(jω[C] + [G])[V ] (1)
dz
where [C], [L], [R] and [G] are the capacitance, inductance, resistance and conductance matrices,
respectively [10–12]. The dispersion effects may be incorporated into the models, by considering C,
L, R and G to be functions of frequency [13]. Now equations may be combined to obtain [14, 15]

(jωk [L] + [R])(jωk [C] + [G]) − γ 2 [U ] = 0 (2)


2
(jωk [L] + [R])(jωk [C] + [G])[Mv ]i,k = γi,k [Mv ]i,k (3)

where [Mv ] and [γ] are the matrix of eigenvectors and the vector of eigenvalues of (jω[L] +
[R])(jω[C] + [G]). We may then calculate the N propagation constants due to the N conduct-
ing strips in the nonhomogeneous medium like microstrips. Consequently, we may determine the
Progress In Electromagnetics Research Symposium, Hangzhou, China, March 24-28, 2008 697

voltages and currents on the N strips as functions of z


£ ¤ £ ¤
[Vi ]k = Mvij k aj e−γj,k z + a0j eγj,k z (4)
£ ¤ £ ¤
[Ii ]k = ([Ya ]k · Mvij k ) aj e−γj,k z − a0j eγj,k z (5)

where [aj ] and [a0j ] are the amplitudes of forward and backward traveling waves on the lines, k is
the subscript indicating the frequency in the band, and [Ya ] is the characteristic admittance of the
lines corresponding to the propagation constants. Here Yij as the entry of [Ya ] is the characteristic
admittance of the ith line for the jth propagation constant.

Figure 1: Circuit model of the wide band mi- Figure 2: Circuit model of the narrow band mi-
crostrip interdigital filter. crostrip interdigital filter.

1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8


0

-10

-20

-30
(dB)

-40

-50

-60

-70
(GHz)

Figure 3: Frequency response of a bandpass filter Figure 4: Frequency response of a narrowband in-
showing stop. terdigital filter before and after optimization.

Now the voltages and currents of the two terminals of the strips (at z = 0 and z = L) may be
obtained from Eqs. (4) and (5), in terms of the unknown amplitudes aj and a0j . By eliminating
these amplitudes among the equations, we may obtain the terminal currents of strips as a function
of their voltages. Then the boundary conditions at the strip terminals are invoked. They are the
short circuits to ground and open circuits at the end of strips. However, the open end of lines
may be modeled as a capacitor and the shorting vias may be modeled as an inductor [16]. Then,
inclusion of the boundary conditions at the strip terminals may lead to the admittance matrix of
698 PIERS Proceedings, Hangzhou, China, March 24-28, 2008

the input/output two port network relating the source and load terminals.
· ¸ · ¸· ¸
I1 Y11 Y12 V1
= (6)
Is Y21 Y22 Vs

where I1 and Is are the load and source currents and V1 and Vs are the load and source voltages,
respectively. We may then obtain the scattering matrix from the admittance matrix. The scattering
parameter s21 as the transfer function is

2Ys Y21
s21 = − (7)
(Ys + Y11 )(Y1 + Y22 ) − Y12 Y21

1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2 2.5 3 3.5 4 4.5 5 5.5 6
0 0

-10 -10

-20 -20

-30
-30
(dB)
(dB)

-40
-40
-50
-50
-60
-60 -70

-70 -80

(GHz) (GHz)

Figure 5: Comparison of the performance of MLS Figure 6: Frequency response of a wideband inter-
filter design with that of Microwave Office for Fig. 4. digital filter before and after optimization.

2 2.5 3 3.5 4 4.5 5 5.5 6


0

-10

-20

-30
(dB)

-40

-50

-60

-70

-80
(GHz) GHz

Figure 7: Comparison of the performance of MLS filter Figure 8: Comparison of the performance of inter-
design with that of Microwave Office for Fig. 6. digital filter designed by MLS with those obtained
by Microwave-Office and fabrication and measure-
ment.

The insertion loss is


ILk = −20 log |s21 |k (8)

Finally, an error function is constructed for the specified insertion and return losses in the lower
Progress In Electromagnetics Research Symposium, Hangzhou, China, March 24-28, 2008 699

and upper stopbands, transition bands and the passband, as shown in Fig. 3.
K
XSL
¡ ¢
e = (ILk − ILSBk )2 + (RLk − RLSBk )2
k=1
KX
PU
¡ ¢
+ (ILk − ILP Bk )2 + (RLk − RLP Bk )2
k=KP L
K
X ¡ ¢
+ (ILk − ILSBk )2 + (RLk − RLSBk )2 (9)
k=KSU

where ILk and RLk are the insertion and return losses calculated at the kth frequency and ILSB
and RLSB are their specified values in the stopband and ILP B and RLP B are their values in
the passband. The frequency band is divided into K discrete frequencies. wti are the weighting
functions which may emphasize any of the terms in the error function, by adjusting their values.
The error function is minimized with respect to the values of wi , si and L. However, some
initial values are required for them to start the minimization of the error function. On the other
hand, input values of the filter structure including the frequency band and the load and source
admittances should be provided for the incorporated impedance matching.
3. NUMERICAL RESULTS
The design and optimization of the interdigital filter shown in Figs. 1 and 2 are performed as an
illustration for the proposed MLS algorithm. The initial values are selected as follows:
L = λ/4 (center frequency), wi = h (thickness of substrate), si = 2h (or as a function of strip
width), s1 = sn = h/4 (the gap between the first two strips and the last two strips). Then the
minimization of the error function is performed with respect to L, wi and si to obtain the optimum
design of the interdigital filter.
Two examples of filter designs are given for the narrow band and wide band versions. Their
performances are also obtained by commercial softwares, namely Microwave-Office.

Table 1: Data for the narrow band filter design.

εr = 3.38, Loss tangent = 0.0027, h = 20 mil, Zs = 50, Zl = 50, f0 = 2 GHz, BW = 6%,


degree of f ilter = 3
Initial values After Design and Optimization
w1 = 0.508 (mm), s1 = 0.152 (mm) w1 = 1.038 (mm), s1 = 0.219 (mm)
w2 = 0.508 (mm), s2 = 1.016 (mm) w2 = 1.032 (mm), s2 = 1.342 (mm)
w3 = 0.508 (mm), s3 = 1.016 (mm) w3 = 0.621 (mm), s3 = 1.381 (mm)
w4 = 0.508 (mm), s4 = 0.152 (mm) w4 = 0.993 (mm), s4 = 0.254 (mm)
w5 = 0.508(mm) w5 = 1.583(mm)
L = 21(mm) L = 23.23(mm)

In Table 1, the input voltages for the narrow band also the initial values of L, w and s are given
for the filter degree N = 3. Their values for the optimum design of the filter are also given. In
Fig. 3, the frequency response of the filter is given for its optimum design. In Fig. 4, its performance
obtained by the proposed method and that obtained by the Microwave Office are compared which
indicate an excellent agreement. The optimum design of the microstrip interdigital band pass filter
for the wideband version is given in Table 2, Figs. 5 and 6.
4. IMPLEMENTATION
A sample filter is designed and optimized by the proposed method. The filter is fabricated and
its performance is measured by HP 8563E and compared with that of Microwave Office simulation
software and that computed by the proposed method in Fig. 8. The characteristics of the interdigital
bandpass filter are:
700 PIERS Proceedings, Hangzhou, China, March 24-28, 2008

Table 2: Data for the wide band filter design.

εr = 4.2, Loss tangent = 0.0027, h = 20 mil, Zs = 50, Zl = 50, f0 = 4 GHz,


BW = 17.5%, degree of f ilter = 4
Initial values After Design and Optimization
w1 = 0.508 (mm), s1 = 0.152 (mm) w1 = 0.287 (mm), s1 = 0.152 (mm)
w2 = 0.508 (mm), s2 = 1.016 (mm) w2 = 1.151 (mm), s2 = 0.484 (mm)
w3 = 0.508 (mm), s3 = 1.016 (mm) w3 = 1.598 (mm), s3 = 0.557 (mm)
w4 = 0.508 (mm), s4 = 1.016 (mm) w4 = 1.505 (mm), s4 = 0.496 (mm)
w5 = 0.508 (mm), s5 = 0.152 (mm) w5 = 0.967 (mm), s5 = 0.146 (mm)
w6 = 0.508 (mm) w6 = 0.282 (mm)
L = 10.78 (mm) L = 10.01 (mm)

Filter degree is three or five microstrip lines, center frequency 2 GHz in the range 1 to 3 GHz,
K = 300 for 300 discrete frequencies for the error function, frequency bandwidth in the passband
is 120 MHz (or 6%), wk = 1, input and output impedances equal to 50 Ω.
5. CONCLUSION
In this paper a design and optimization procedure for the inhomogeneous microstrip interdigital
bandpass filter is developed based on the method of least squares which employs the N coupled
transmission line theory. The proposed method incorporates impedance matching of the load and
source impedances, which leads to shorter and simpler circuit configurations.
The performance of the proposed filter design method agrees very well with that of the com-
mercial softwares for microwave circuits.
REFERENCES
1. Matthaei, G., L. Young, and E. M. T. Jones, Microwave Filters, Impedance-Matching Networks,
and Coupling Structures., Artech House, Boston, 614–647, Nov. 1985.
2. Pozar, D., Microwave Engineering, Third Edition, John Wiley and Sons, 416–438, 2005.
3. Dishal, M., “A simple design procedure for small percentage bandwidth round rod interdigital
filters,” IEEE Trans. Microwave Theory Tech., Vol. 13, No. 5, 696–698, Sept. 1965.
4. Martin, P., “Design equations for tapped round rod combline and interdigital bandpass filters,”
Available: http://www.rfshop.com.au/C&IDES.DOC. Nov. 2002.
5. Milligan, T. A., “Dimensions of microstrip coupled lines and interdigital structure,” IEEE
Trans. Microwave Theory Tech., Vol. 25, No. 5, 405–410, May 1977.
6. Oraizi, H. and N. Azadi, “A novel method for the design and optimization of microstrip
combline filters with tapped-line input,” Mediterranean Microwave Symposium, Genova, Italy,
Sept. 2006.
7. Oraizi, H. and N. Azadi, “A novel method for the design and optimization of microstrip multi-
section bandpass Combline filters,” 36th European Microwave Conference, Manchester, UK,
1217–1220, Sept. 2006.
8. Oraizi, H. and N. Azadi, “Realization of transmission zeros for the improvement of the tran-
sition bands of combline filters by lumped capacitors,” Mediterranean Microwave Symposium,
Genova, Italy, Sept. 2006.
9. Gentili, G. and M. Salazar-Palma, “The definition and computation of modal characteristic
impedance in quasi-TEM coupled transmission lines,” IEEE Trans. Microwave Theory Tech.,
Vol. 43, No. 2, 338–343, Feb. 1995.
10. Amirhossenini, M. K., “Determination of capacitances and conductonce matrices of lossy
shilded coupled microstrip transmission lines,” Progress in Electromagnethics Research, PIER
50, 267–278, 2005.
11. Harrington, R. F. and C. Wei, “Losses on multiconductor transmission lines in multilayered
dielectric media,” IEEE Trans. Microwave Theory Tech., Vol. 32, No. 7, 705–710, July 1984.
12. Wei, C., R. F. Harrington, J. R. Mautz, and T. K. Sarkar, “Multiconductor transmission
lines in multilayered dielectric media,” IEEE Trans. Microwave Theory Tech., Vol. 32, No. 2,
439–450, Feb. 1995.
Progress In Electromagnetics Research Symposium, Hangzhou, China, March 24-28, 2008 701

13. Tripathi, V. K., “A dispersion model for coupled microstrips,” IEEE Trans. Microwave Theory
Tech., Vol. 34, No. 1, 66–71, Jan. 1986.
14. Tripathi, V. K., “Asymmetric coupled transmission lines in an inhomogeneous medium,” IEEE
Trans. Microwave Theory Tech., Vol. 23, No. 9, Sept. 1975.
15. Tripathi, V. K., “On the analysis of symmetrical three-line microstrip circuits,” IEEE Trans.
Microwave Theory Tech., Vol. 23, No. 9, Sept. 1975.
16. Edwards, T. C. and M. B. Steer, Foundations for Microstrip Circuits, 3rd Edition, John Wiley
& Sons, New York, 2000.

Вам также может понравиться