CAPACITOR PI NETWORK FOR IMPEDANCE MATCHING

PI Matching Networks

Designing matching networks is one of the key aspects of RF/Microwave design. A lossless network that
matches an arbitrary load to real impedance has to have at least two reactive elements. However, two elements
do not give control over the bandwidth and the degree of match simultaneously. Three-element matching
networks, i.e. Pi- and Tee-networks, provide additional control of the frequency response.

In this application note we explore the idea of designing an all-capacitor Pi matching network by using one of the
elements beyond its self-resonant frequency, when it has become inductive rather than capacitive. Essentially,
the effective series inductance (ESL) of the series element in the Pi network is utilized. The design process is
enabled via the broad-band accuracy of the Modelithics Global capacitor models, in particular those for ATC
600L 0402 capacitors. The use of method of moments co-simulation, wherein numerical electromagnetic
simulation is used to represent the distributed interconnects and discrete models are used for the lumped
components, is also demonstrated.

Capacitors Used as Inductors after SRF

Port
P1

L Capacitive
Ls
SRF

C Inductive
S(2,1)

S(1,1)

C C
Cp -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

R
Rs Inductive
Capacitive
Port
P2
freq (1.000MHz to 4.000GHz) freq (1.000MHz to 4.000GHz)

Figure 1 Simple circuit model. Figure 2 Frequency response of simple model (S21 on left, S11 on right).

Figure 1 shows a simplified two port equivalent circuit of a capacitor; C – Capacitance, Ls – Series Inductance,
Cp – Parallel capacitance, Rs – Series resistance. As there are three reactive components in the circuit model,
there are at least two resonant frequencies – the series resonant frequency (SRF) and the parallel resonant
frequency (PRF). At the SRF the capacitor’s impedance is a minimum, while at the PRF it is a maximum.
Usually, the parallel capacitance (Cp) is much smaller than the nominal component value (C) and the PRF
occurs at a higher frequency than the SRF.

Figure 2 shows a 2-port S-Parameter simulation of the circuit shown in Figure 1. It can be seen that at above the
resonant frequency the capacitor acts like an inductor. In general, the SRF sets the frequency limit on the useful
operating range of a capacitor. But in this application, we exploit this phenomenon and use a capacitor as an
inductor after SRF to match a complex load to a real load.

PI Network

© American Technical Ceramics, 2007

1 of 5
 C
C1 Load
C=27 pF Series L
Term Shunt C
Term C C Term3
Term1 C2 C3 Num=2 Shunt C
Num=1 C=1.0 pF C=1.0 pF Z=50
Z=50 Ohm
C
C4
C=27 pF
Source
Term
Term C C Term2
Term4 C6 C5 Num=2
Num=1 C=1.0 pF C=1.0 pF Z=25+j*12
Z=50 Ohm

Figure 4 Typical path traced on the Smith Chart at

Figure 3 Simplified equivalent circuit. the center frequency.

The simplified schematic of the matching network is shown in Figure 3. The design specification is to match an
impedance of 25+j12Ω to a 50Ω load at 1.6GHz. As mentioned earlier, the basic idea is to use the capacitor C1
(27pF) in the series arm of the PI network, as an inductor after SRF. The new pad-scalable model for ATC 600L
capacitors (CAP-ATC-0402-101) and Rogers 60 mil-thick R04003 (εr = 3.6) substrate were used in the design.
The impedance transformation at the center frequency is illustrated in Figure 4. Figure 5 shows the layout of the
PCB. It is important to note that the inductance of interconnects does play a role in determining the SRF and
hence the effective equivalent series inductance.

CAP_ATC_0402_001_MDLXCLR1
ATC_600L_C1
C=27 pF
Subst=Sub
MTAPER MTEE_ADS MTEE_ADS MTAPER
Sim_mode=0 MLIN
Taper7 Tee1 MLIN Tolerance=1.0 Tee2 Taper8
Subst=Sub Subst=Sub TL2 Subst=Sub Subst=Sub Term
TL1 Pad_Width=W1 mil Subst=Sub
W1=138.23 mil W1=14.1 mil Subst=Sub Pad_Length=L1 mil W1=14.1 mil W1=138.23 mil Term2
W=14.1 mil
W2=14.1 mil W2=14.1 mil W=14.1 mil W2=14.1 mil W2=14.1 mil Num=2
Pad_Gap=G1 mil L=25 mil
L=72 mil W3=14.1 mil L=25 mil W3=14.1 mil L=72 mil Z=25+j*12

Term
Term
MLIN MTAPER MLIN Term3
Term1 MTAPER MLIN
TL5 Taper2 TL11 Num=2
Num=1 MLIN Taper1 TL8
Subst=Sub Subst=Sub Subst=Sub Z=50
Z=50 Ohm TL3 Subst=Sub
W=14.1 mil W1=W1 mil Subst=Sub W=14.1 mil
Subst=Sub W1=14.1 mil
L=18.6 mil W2=14.1 mil W=14.1 mil L=18.6 mil
W=14.1 mil W2=W1 mil L=25 mil
L=25 mil L=24 mil
L=24 mil
MTAPER CAP_ATC_0402_001_MDLXCLR1 MTAPER CAP_ATC_0402_001_MDLXCLR1
MSub
Taper3 ATC_600L_C2 Taper6 ATC_600L_C3
MSUB Subst=Sub C=1 pF Subst=Sub C=1 pF
MSub3 W1=14.1 mil Subst=Sub W1=14.1 mil Subst=Sub
H=60 mil W2=W1 mil Sim_mode=0 W2=W1 mil Sim_mode=0
Er=er L=24 mil Tolerance=1.0 L=24 mil Tolerance=1.0
Mur=1 Pad_Width=W1 mil Pad_Width=W1 mil
Pad_Length=L1 mil S-P ARAM E TERS Pad_Length=L1 mil
Cond=1.0E+50
Hu=1.0e+033 mm Pad_Gap=G1 mil Pad_Gap=G1 mil
S_Param MLIN
T=1.7 mil SP1
TanD=0.0021 MLIN TL10
Start=100 MHz Subst=Sub
Rough=0 mm TL9
Stop=5 GHz W=24 mil
Subst=Sub
Step=50 MHz L=len mil
W=24 mil Var
VAR
VIAHS Var VAR Eqn VIAHS
L=len mil Eqn
Size402
V3 VAR1 V4
D=15 mil W1=28 D=15 mil
er=3.62
H=60 mil L1=15.5 H=60 mil
tand=0.0027
Sub="MSub3"
T=1.7 mil len=4.5 T=1.7 mil
G1=16

Figure 5.ADS schematic of the layout used.

 (-Y21) 3

Ls (nH)
MTAPER MTAPER
Taper1 Taper2 MLIN 1
Subst=Sub Subst=Sub TL2
(Y11+Y12) (Y22+Y21) W1=14.1 mil W1=W1 mil Subst=Sub
W2=W1 mil W2=14.1 mil 0
W=14.1 mil
L=24 mil L=24 mil L=25 mil 50
40
Term Term
Term1 MLIN Term2

C (pF)
30
CAP_ATC_0402_001_MDLXCLR1
Num=1 TL1 Num=2
Z=50 Ohm Subst=Sub ATC_600L_C1 Z=50 Ohm 20
W=14.1 mil C=27 pF
Subst=Sub
Var
Eqn
VAR 10
L=25 mil Size0402
Sim_mode=0 0
Tolerance=1.0 W1=28
Pad_Width=W1 mil L1=15.5 0 1 2 3 4 5
Pad_Length=L1 mil Sub="MSub3"
freq, (GHz)
Pad_Gap=G1 mil G1=16

Figure 6.ADS schematic used to extract Ls (right). Extracted values for nominal Capacitance (C) and series
Inductance (Ls) (left)

Figure 6 shows the schematic of the circuit used to extract the equivalent capacitance (C) and inductance (Ls)
of the series element in the matching network; these values are found by using Y21. When -Imag(Y21) > 0, C=-
Imag(Y21)/(2πf), and when -Imag(Y21) < 0, Ls=1/(Imag(Y21)2πf). The microstrip and taper sections are
included with the capacitor effects in this calculation. It is interesting to note that the Ls value is fairly constant at
~2.2 nH after the SRF.

Results

Figures 7 and 8 compare the measured result with circuit simulation and method-of-moments (MoM) co-
simulation results. The layout of the PCB used in the MoM simulation is shown in Figure 9. The taper at the
input and the output side is used to match the 50Ω line (not shown in the layout) with the mounting pads for the
capacitor. The MoM co-simulation results provide improved comparisons to the measured data, as the
interconnect and ground-via effects are more accurately represented.

0 0
Circuit
-5 MoM -10 Simulation
Cosimulation Circuit
S21, dB
S11, dB

Simulation
-10 -20 MoM
Cosimulation

-15 -30
Measured Measured
-20 -40
0 1 2 3 4 5 0 1 2 3 4 5
freq, GHz freq, GHz

Figure 7 Comparison between simulated and measured response when port 2 is terminated with 50Ω.

The properties of the substrate (thickness and εr) used are critical in obtaining a good correlation between
measured and simulated data. The substrate-scaling features of the Modelithics Global models enable all
substrate-related parasitics to be taken into account. To illustrate this feature, a simulation comparing two
substrates is shown in Figure 11.
 0 0
Circuit Circuit
-5 Simulation -10 Simulation

S21, dB
S11, dB

MoM
-10 -20 MoM
Cosimulation
Cosimulation

-15 -30 Measured

Measured

-20 -40
0 1 2 3 4 5 0 1 2 3 4 5
freq, GHz freq, GHz

Figure 8 Comparison between simulated and measured response when port 2 is terminated with 25+j12Ω.

Figure 9 Layout of the PCB used in the MoM Simulation. Figure 10 Photograph of the matching circuit.

0 0
RO4350, 4mils
-5
-10 RO4350, 4mils
S11, dB

-10
S21, dB

-20
-15 RO4350, 60mils
RO4350, 60mils
-20 -30

-25 -40
0 1 2 3 4 5 0 1 2 3 4 5
freq, GHz
freq, GHz

Figure 11.Comparison of simulated results Case1: RO4350 60 mils substrate, Case2: RO4350 4 mils substrate
and Case3: RO4350 60 mils substrate using ideal components. Port 2 is terminated with 25+j12Ω
About this work

This work was performed as a collaboration between American Technical Ceramics and Modelithics,
Inc, funded by ATC. University of South Florida MS student Aswin Jayaraman assisted with the
development of this material under grant funding provided by Modelithics, Inc..

Contact information

For information about ATC products and support, please contact American Technical Ceramics, One
Norden Lane, Huntington Station, NY, 11746 • voice 631-622-4700 • fax 631-622-4748 •
sales@atceramics.com • www.atceramics.com

For more information about Modelithics products and services, please contact Modelithics, Inc. , 3650
Spectrum Blvd. , Suite 170, Tampa, FL 33612 • voice 888- 359-6539 • fax 813-558-1102 •
sales@modelithics.com or visit www.modelithics.com