See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/291457764

Comparison of S Band Diplexer Design using

Multiplexer Theory with Coupling Matrix
Synthesis

Conference Paper · August 2015

DOI: 10.1109/ISTMET.2015.7359037

CITATIONS READS

0 19

3 authors, including:

Saeed Ur Rahman Daud Mustafa Minhas

Nanjing University of Aeronautics & Astronau… Universität des Saarlandes
3 PUBLICATIONS 0 CITATIONS 6 PUBLICATIONS 5 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Patch Antenna with metamaterials View project

All content following this page was uploaded by Saeed Ur Rahman on 22 December 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 2015 International Symposium on Technology Management and Emerging Technologies (ISTMET), August 25 - 27, 2015,
Langkawi, Kedah, Malaysia

Comparison of S Band Diplexer Design using

Multiplexer Theory with Coupling Matrix Synthesis
Hisham Khalil1, Syed-ur-Rehman2 and Daud Mustafa Minhas3
1
Dept of Technology, The University of Lahore, Pakistan
2 3
Dept of Electronics Engineering, Mohammad Ali Jinnah Univeristy(MAJU), Pakistan

Abstract—This paper presents a comparison between two matrix for individual filter is computed. The input admittance
techniques of S band interdigital diplexer design with i.e. with of both channel filters is added with-out any compensation to
generalized multiplexer theory and coupling matrix synthesis. plot the response of diplexer. Coupling band widths between
The design based on GMT, doubly terminated Chebyshev low
pass filter response is used. The first two resonators of each
two adjacent resonators is also found using MathCAD. The
channel are modified using formulas in GMT. The design is physical parameters for adjacent resonators are found using
verified though MathCAD plots. In coupling matrix synthesis electromagnetic software (HFSS 13). Two group delay peaks
design, the general coupling matrix of each filter and coupling at center frequencies of each channel are found through EM
band width for two adjacent resonators are calculated. A novel software. Finally all parameter are put together to simulate
technique is used for diplexer input design. The group delays are complete diplexer.
measured at the centre frequencies of channel filters. The
physical diplexer design round rod type inter digital filter is
selected. Both designs are simulated in HFSS with 2100 and For GMT design, the theoretical inter channel isolation
2250MHz with 100MHz channel band width. 26dB and designed band width is 100MHz while simulated
isolation is 29.7dB and band width is 80MHz respectively. On
Keywords— Chebyshev response, inter digital diplexer, Generalized other hand CMS deign [9], MathCAD results shows nearly
multiplexer theory, coupling matrix synthesis, Mathcad, HFSS, 16dB in channel isolation and band width is remain same
ripple band width, while the channels are detuned around 20MHz. the ripple and
width 80MHz. The HFSS results show the simulated band
I. INTRODUCTION width is 95MHz and inter channel isolation is 25dB which is
In the microwave industry diplexer are widely used for high 9dB improved my MathCAD results.
power application like Radar and communication system. The
advancement in diplexer design technique in recent years is The presented work is showing the technological benefits of
much significant as compared to classical design process. advanced diplexer design approach. The adapted process
utilize the power of full wave EM solutions while the classical
This paper compare the direct analytical design process design technique has limitation of higher order residuals.
presented by [1] with coupling matrix synthesis design
II. DIPLEXER DESIGN THEORY
process. In generalized multiplexer theory, the independent
doubly terminated low pass prototype with chebychev A. Diplexer design using Generalized multiplexer theory
response is used. According to theory [1] there is freedom of The fifth order chebychev low pass prototype is used as
selection the number of channels, band width and center loss pass prototype [2] design and calculated
frequency. The modifications of first two resonators are
carried out in order to maintain the frequencies of perfect
transmission. The limitation of generalized multiplexer theory (1)
is that the channels cannot place much close. The design is
verified through ABCD matrix plot in MathCAD. Admittance
of each element is calculated using formulation in [1].The
round rod type inter digital filter is selected. For physical (2)
dimension of round rod, the methodology presented in [2-3] is
adopted. The curves of self-capacitance and mutual-
capacitance are plotted. The diameter of rods and inter
resonator spacing is found. The interdigital diplexer is
simulated in HFSS.

In the coupling matrix synthesis, the chebychev filter response

Fig. 1chebyshev low pass prototype with ideal admittace inverter and shunt
is selected with unit impedance inverts. The general coupling capacitor

978-1-4799-1723-5/15/$31.00 ©2015 IEEE 243

 inter channel isolation is 24dB. The channel band width is
Where N is the order of filter, k is 1 to 5, γ is calculated as preserved to 100MHz for both channels.
0.636 and ω1 is band edge. The low pass values of g and J are
TABLE I
g1=g5=0.971 g2=g4=2.543 g3=3.144 J0=1 J1= J4=1.361
COMPENSATION PARAMETERS FOR DIPLEXER
J2=J3=1.798.The low pass response is plotted in Mathcad
Correction Channel #1 Channel #2
through ABCD matrix. The band edge is selected unity. parameter
βr11 -0.023 0.023
0 γr02 0 0
InsertionLoss
−3 γr12 0.301 0.301
−6 β113 -4.825×10-3 4.825×10-3
β223 -1.99×10-3 1.99×10-3
−9
− 12
IL( Ω ) − 15
dB

− 18
− 21 0
InsertionLoss
− 24 −3 ReturnLoss

− 27 −6
− 30 −9
0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3
Ω
IL( Ω ) − 12
RL( Ω )− 15

dB
Fig. 2 chebyshev low pass insertion loss for N = 5
− 18
− 21
The given relation [6] is used for low pass to band pass − 24
transformation of kth element for rth channel. − 27
− 30
2 2.04 2.08 2.12 2.16 2.2 2.24 2.28 2.32 2.36 2.4
Ω

, (3)
Fig. 3 Return and insertion loss characteristics of channel filter at fo 2.1
[GHz]
The direct ABCD matrix for nth channel filter given by
0
InsertionLoss
−3 RetrunLoss
−6
0 1 0 −9
∏ ∏ 1 (4) IL( Ω ) − 12
0 RL( Ω )− 15
dB

− 18
− 21
Where Bwn and Ωn are the band width and centre frequency − 24
for nth channel and N is the order of nth channel filter. The − 27
interaction between the channels is depending upon the value − 30
2 2.04 2.08 2.12 2.16 2.2 2.24 2.28 2.32 2.36 2.4
of α. For maximum interaction between channels α is unity Ω

and for minimum interaction α is infinite. The value of α is set

Fig. 4 Return and insertion loss characteristics of channel filter at fo 2.25
in such a way that the product of Ωn and α should be equal to [GHz]
centre frequency of particular channel .In case of diplexer then 0
= 2 and order of channel filter is 5 as mentioned earlier. The −3
ReturnLoss
InsertionLoss
individual channel filters at 2.1GHz and 2.25GHz are plotted −6
in MathCAD and shown in figure 2 and 3 respectively. −9

RL( Ω )− 12
The common port input admittance of diplexer is the sum La( Ω ) − 15
dB

of individual input admittances of both channels filter. − 18

Insertion and return loss characteristics are shown in figure 3. − 21
In band characteristic of channel filter are changed in absence − 24

of any addition compensation. The modification of first two − 27

− 30
resonators of each channel filters are carried out in order to 2 2.04 2.08 2.12 2.16 2.2 2.24 2.28 2.32 2.36 2.4
maintain the frequencies of perfect transmission. Five Ω

correction parameters are computed using formulas in [1] and Fig. 5 Return and insertion loss characteristics of diplexer without
shown in table 1.The modified first impedance inverter of compansation [GHz]
both channels is J0 = 1 and second impedance inverter J1of
both channels is 1.309. The diplexer is plotted in MathCAD
after the compensation. The theoretical return loss is 20dB and

244
 0
(5)
ReturnLoss
−3 InsertionLoss

−6
The general coupling matrix M is composed on self and
−9
mutual coupling of elements of network. The diagonal entries
are impedance inverters and others are mutual coupling
RL( Ω )− 12
between adjacent elements. As the passive network is
IL( Ω ) − 15
dB

− 18
symmetrical the Mij= Mji.in general the order of filter is equal
− 21
to the order of M matrix. The sI matrix has diagonal frequency
− 24 variables while all other entries are zero. R is termination
− 27 matrix with first and last nonzero entries [7].
− 30
2 2.04 2.08 2.12 2.16 2.2 2.24 2.28 2.32 2.36 2.4
Fig. The general coupling matrix fifth order filter is computed
6 Return and insertion loss characteristics of diplexer with compansation using MathCAD and response of individual filter is plotted
[GHz] using direct synthesis of general coupling matrix.
The interdigital type filter is selected for physical
realization with two non-resonant element to maximize input
0 1.007 0 0 0
and output isolation. In distributed domain, the self and
1.007 0 0.697 0 0
mutual admittance of elements are calculated using formulas 0 0.697 0 0.697 0 (6)
[6].To covert the admittance to normalized capacitance 0 0 0.697 0 1.007
multiply by the factor of 7.534. 0 0 0 1.007 0
The physical dimension of round rods is found using The input admittance of each channel is calculated and plot of
methodology presented in [3]. The curves of self-capacitance individual response of filter is shown in figure 7.
(1/2 normalized capacitance) vs s/b (1/2 normalized spacing)
and mutual-capacitance (1/2 normalized capacitance) vs s/b
(1/2 normalized spacing) are plotted. For b=1 inch (centre of
rod to ground plane spacing) the inter element spacing and (7)
diameter of rods are given below in the table
(9)
TABLE II
NORMALIZED SELF AND MUTUAL CAPACITANCE
Self & The input admittance of both channel filters is added with-out
Channel #1 Channel #2
mutual
2.1GHz 2.25GHz any compensation to plot the combined effect of diplexer. At
capacitance the frequencies of perfect transmission for one channel, the
C0 5.441 5.513
C1 5.554 5.617 input impedance is real and for other channel is reactive. The
C2 6.707 6.762 out band attenuation of for one channel is ideally infinite.
C3 6.818 6.865
C4 6.688 6.744 (10)
C5 5.535 5.6
C6 5.441 5.513
C01 2.093 2.021 In figure 8 the insertion loss in preserved and return loss is
C12 0.469 0.438 improved. The theoretical inter channel isolation is about
C23 0.358 0.344
C34 0.358 0.344
18dB and the pass band is little bit detune.
C45 0.488 0.455 0
C00 3.42 3.42 − 12
InsertionLoss1
ReturnLoss2
InsertionLoss2
− 24 ReturnLoss1

IL2( f ) − 36
B. Diplexer design using coupling matrix synthesis RL2( f ) − 48
IL1( f ) − 60
dB

In coupling matrix synthesis design theory, any two port RL1( f ) − 72

network can be presented using matrix of voltage source, − 84
passive network impedance and load termination as the loop − 96

currents circulating in the resonators. − 108

− 120
2 2.04 2.08 2.12 2.16 2.2 2.24 2.28 2.32 2.36 2.4
f

(11) Fig. 7 Return and insertion loss characteristics of diplexer without

compansation [GHz]
The impedance matrix can be expand as

245
 Return Loss of GMT diplexer complete-duplx-design2 ANSOFT
0 0.00 Curve Info
InsertionLoss
dB(S(1,1))
−8 ReturnLoss Setup1 : Sweep
-5.00
− 16
-10.00
− 24

dB(S(1,1))
− 32
-15.00
IL( f )
dB

− 40 -20.00
RL( f )
− 48
-25.00
− 56
-30.00
− 64 1.90 2.00 2.10 2.20 2.30 2.40
Freq [GHz]

− 72
Fig.11 Return loss characteristics of GMT diplexer

− 80
1.95 1.995 2.04 2.085 2.13 2.175 2.22 2.265 2.31 2.355 2.4
f
IV. DIPLEXER DESIGN SIMULATION USING COUPLING MATRIX
Fig. 8 Return and insertion loss characteristics of diplexer without SYNTHESIS
compansation [GHz]
The physical design process of diplexer based on CMS [7]
is much different from GMT design. The input output
III. DIPLEXER DESIGN SIMULATION USING GENERALIZED matching and inter element spacing is obtained my group
MULTIPLEXER THEORY delay and coupling band width method. For input design the
The physical design of diplexer as shown in figure 9is group delay is 6.021ns at centre frequency of each of filter
simulated in HFSS. The length, width and high of box are 14, which are shown in figure 12 and 15. All the coupling peaks
1 and 4 inches respectively. Input and output port impedance are below the 25dB.
is 50 ohms.
The inter element spacing are found by maintaining the
The plots insertion and return loss are shown in figure 10 coupling band widths between two adjacent elements are
and 11 respectively. Design band width is 100MHz and HFSS shown in figure 13 to 17 [4]. The diplexer input is designed
simulated band width is about 80MHz. Insertion loss is by two group delay peaks at the centre frequencies of channel
preserved while return loss is compromised. filters [8]. Insertion loss also preserved and return loss is not
very much high and shown in figure 19 and 20 respectively.
The HFSS simulated band width is about 100MHz.
Group Delay 2.1 GHz M01-2100 ANSOFT

6.25 Curve Info

GroupDelay(1,1)*1e9
Setup1 : Sweep

5.00
GroupDelay(1,1)*1e9

3.75

Fig. 9 Top view of inter digital diplexer 2.50

1.25
Insertion Loss of GMT diplexer complete-duplx-design2 2.00 2.05 2.10 2.15 2.20 2.25 2.30 2.35 2.40
Freq [GHz]
ANSOFT

0.00 Curve Info

dB(S(1,2)) Fig. 12 Input group delay at 2.1 GHz

-20.00 Setup1 : Sweep
dB(S(1,3))
Setup1 : Sweep Inter resonator coupling M12-2100 ANSOFT

-40.00 -20.00 Curve Info

dB(S(2,1))
dB(S(21))

-60.00 Setup1 : Sweep

-30.00

-80.00
-40.00
dB(S(2,1))

-100.00
-50.00
-120.00
-60.00
-140.00
1.90 2.00 2.10 2.20 2.30 2.40
Freq [GHz]
-70.00
Fig. 10 Insertion loss characteristics of GMT diplexer
-80.00
1.90 1.95 2.00 2.05 2.10 2.15 2.20
Freq [GHz]
Fig. 13 Inter resonator coupling between first and second resonators at 2.1
GHz

246
 Inter resonator coupling M23-2100 ANSOFT Inter resonator coupling M23-2250 ANSOFT

-25.00 Curve Info

-25.00 Curve Info

dB(S(2,1)) dB(S(2,1))
-37.50 Setup1 : Sweep Setup1 : Sweep

-37.50
-50.00
dB(S(2,1))

dB(S(2,1))
-62.50
-50.00
-75.00

-87.50
-62.50
-100.00

-112.50 -75.00
1.90 1.95 2.00 2.05 2.10 2.15 2.20 2.25 2.30 2.200 2.220 2.240 2.260 2.280 2.300
Freq [GHz] Freq [GHz]
Fig. 14 inter resonator coupling between second and third resonators at 2.1 Fig. 17 Inter resonator coupling between second and third resonators at
GHz 2.25 GHz

Group delay 2.25 GHz M01-2250 ANSOFT Group Delay of diplexer M01-duplx-input ANSOFT

6.25 Curve Info 7.00 Curve Info

GroupDelay(1,1)*1e9 GroupDelay(1,1)*1e9
Setup1 : Sweep Setup1 : Sweep
6.00
5.00
GroupDelay(1,1)*1e9

GroupDelay(1,1)*1e9
5.00
3.75
4.00
2.50
3.00

1.25
2.00

0.00 1.00
2.00 2.10 2.20 2.30 2.40 2.50 2.60 2.00 2.05 2.10 2.15 2.20 2.25 2.30 2.35 2.40
Freq [GHz] Freq [GHz]
Fig. 15 group delay at 2.25 GHz Fig. 18 Input group delay peaks of diplexer at 2.1 and 2.25 GHz

Inter resonator coupling M12-2250 ANSOFT Insertion Loss of CMS diplexer complete-duplx-design ANSOFT

0.00 Curve Info 0.00

dB(S(2,1))
-12.50 Setup1 : Sweep
-25.00 Curve Info

dB(S(1,2))
-25.00 Setup1 : Sweep
-50.00
dB(S(1,3))
dB(S(2,1))

Setup1 : Sweep
S21(dB)

-37.50
-75.00
-50.00
-100.00
-62.50

-75.00 -125.00

-87.50 -150.00
2.10 2.15 2.20 2.25 2.30 2.35 2.40 1.90 2.00 2.10 2.20 2.30 2.40
Freq [GHz] Freq [GHz]
Fig. 16 Inter resonator coupling between first and second resonators at Fig. 19 Insertion loss characteristics of CMS diplexer
2.25 GHz
Return Loss of CMS diplexer complete-duplx-design ANSOFT
Curve Info
5.00
dB(S(1,1))
Setup1 : Sweep
0.00
V. COMPARISON OF CMS AND GMT DIPLEXER
-5.00
The simulated comparison of classical and modern design -10.00
dB(S(1,1))

technique is carried out. The design process complexity in

-15.00
CMS is much less then GMT design. In the mesh GMT has
theoretical error term of α-4 .The novel two peak input -20.00

design technique is verified as the band width in return loss -25.00

characteristics in CMS design better. -30.00

-35.00
1.90 2.00 2.10 2.20 2.30 2.40
Freq [GHz]
Fig.20 Return loss characteristics of CMS diplexer

247
 Insertion Loss of CMS and GMT diplexer complete-duplx-design ANSOFT
characteristics,” IEE Proc.-Microw. Antennas Propag., 146,1,
0.00
Feb. 1999, 84–90.
-25.00
[6] J.-S. Hong and M. J. Lancaster, “Design of highly selective
-50.00 Curve Info microstrip band pass filters with a single pair of attenuation poles
dB(S(1,2))
S21(dB)

Setup1 : Sweep at finite frequencies,” IEEE Trans., MTT-48, July 2000, 1098–
-75.00
dB(S(1,3))
Setup1 : Sweep 1107.
-100.00 dB(S(1,2))_1
Imported [7] R.J.Cameron, "Advanced Coupling Matrix Synthesis Techniques
dB(S(1,3))_1
-125.00 Imported for Microwave Filters," IEEE Transactions on Microwave Theory
dB(S(1,2))_2
Imported and Techniques, Vol. 51, No. 1, January 2003, pp. 1-10
-150.00 dB(S(1,3))_2
1.90 2.00 2.10 2.20 2.30
Imported 2.40
Freq [GHz]
[8] S.Shin and S.Kanamaluru, “Diplexer Design Using EM and
Fig. 21 Insertion loss characteristics of CMS and GMT diplexer
Circuit Simulation Techniques,” IEEE Microwave Magazine, Vol.
Return Loss of CMS and GMT diplexer complete-duplx-design ANSOFT
8, No. 2, April 2007, pp. 77 – 82
5.00

0.00 [9] Cameron, R.J., Ming, Y.: ‘Design of manifold-coupled

-5.00 multiplexers’, Microwave Magazine, IEEE, 2007, 8, (5), pp. 46-59
Curve Info
-10.00 dB(S(1,1))
Setup1 : Sweep
Y1

-15.00 dB(S(1,1))_1
Imported
-20.00

-25.00

-30.00

-35.00
1.90 2.00 2.10 2.20 2.30 2.40
Freq [GHz]
Fig. 22 Return loss characteristics of CMS and GMT diplexer

VI. CONCLUSIONS
This paper has dealt with comparison of analytical design
process presented by generalized multiplexer theory and
physical design technique using coupling matrix synthesis.
The theoretical diplexer based on GMT is simulated in
MathCAD with and without compensation and compared with
CMS design technique. A dual peak group delay input
matching technique is also presented for common transformer
design. Final simulation of physical inter digital S band
diplexer is simulated HFSS and results are compared. The
theoretical and simulated inter channel isolation for GMT is
24dB and 29dB respectively while CMS design is 18dB and
24dB. HFSS results show the simulated band width in GMT
and CMS is 80MHz and 100MHz.

References
[1] J. D. Rhodes and R. Levy, “A generalized multiplexer theory” this
issue,pp. 99–111

[2] CAMERON, R.J.: 'General prototype network synthesis methods

for microwave filters', ESA Journal, 1982, 6, pp. 193-206.

[3] CAMERON, R.J.: 'Fast generation of Chebyshev filter prototypes

with Asymmetrically -prescribed transmission zeros', ESA
Journal, 1982, 6 (1),pp.83-95.

[4] HUNTER, I.C., and RHODES, J.D.: 'Electronically tunable

microwave band stop filters', IEEE Transactions on Microwave
Theory and Techniques,1982, 30, (9) (Special Issue on Microwave
Filters)pp. 1361-67

[5] J.-S. Hong and M. J. Lancaster, “Microstrip cross-coupled

trisection bandpass filters with asymmetric frequency

View publication stats

248