Dynamic Performance Analysis of T Network Impedance Matching

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS
Rui-Jin Liao, Jian-Wen Tan, Hua Wang, Long Li, Sheng-Ze Qiang
Dynamic performance analysis of T network impedance matching

Rui-jin Liao1, Jian-wen Tan1 2, Hua Wang3, Long Li2, Sheng-ze Qiang2
1
State Key Laboratory of Power Transmission Equipment & System Security and New Technology Chongqing University No. 174 Shazheng Street, Shapingba District, Chongqing, 400044, China Chongqing Communication Institute No. 1 Linyuan, Shapingba District, Chongqing, 400035, China Department of Biomedical Engineering Chongqing University of Medical Sciences, No. 1 Yixueyuan Road, Yuzhong District, Chongqing, 400016, China rjliao@cqu.edu.cn, artan@cqu.edu.cn, wantcop@163.com, lilong@163.com, qshez@163.com
3 2
Abstract: - Harmonic rejection ability and reflection coefficient are the most important factors in the design of impedance matching network. However, dynamic stability of impedance matching should be taken into account in applications existing working frequency drift, load impedance variation and components deviation due to tolerances and process variation. In order to study the dynamic performance of T network impedance matching, theoretical analysis is conducted in accordance with the Q (quality)-based design method. The relationships between the matching result (reflection coefficient) and components deviation, load impedance variation and operating frequency drift are obtained. The analysis results suggest that higher Q value produces worse dynamic stability and that matching result is more sensitive to load impedance variation when the load reflection coefficient of static state is larger. In designing a T network, it is need to make the tradeoff between dynamic stability and harmonic rejection ability. The analysis results provide theoretical basis for the design and characteristics analysis of T network impedance matching.
Keywords: - Impedance Matching, Circuit Analysis, Dynamic Performance, Component Tolerance, Frequency Drift, Circuit Design to the load, the wave power is completely absorbed by the load, and reflection is minimized by matching the impedance of the load to the characteristic impedance of the transmission line [24]. The single-frequency (narrow-band) impedance matching circuits represented by L, T and Pi networks are the common-used passive impedance matching networks by virtue of their simplicity [3, 4]. In practical application, the values of the passive
1 Introduction
Impedance matching using a passive network is very important in the design of RF and microwave circuits to achieve maximum power transfer, minimum reflection, and adequate harmonic rejection [1-4]. By matching the load impedance to the complex conjugate of the source impedance, the load can obtain maximum power from the source. When a transmission line is used to transfer power
E-ISSN: 2224-266X
33
Issue 2, Volume 11, February 2012
components (such as inductors and capacitors) in impedance matching networks often deviate from the theory design value influenced by nominal value tolerances, parasitic and manufacturing process variations [5-12]. For example, capacitance can vary up to 20% for a metalinsulatormetal capacitor due to process variation [13]. These components deviations can cause the impedance matching networks to deviate from perfect matches. In addition to component deviation, variation of load impedance and working frequency drift in applications such as the piezoelectric ultrasonic transducers [14, 15] and cell phone antenna [11] can also cause deviation from perfect matches. Broadband impedance matching and dynamic matching are solutions to this problem [1619]. However, the single frequency matching adopted in L, Pi and T networks is preferred due to its simplicity and manufacturing feasibility in many applications, such as in small type systems. Hence, investigating the dynamic performance of the commonly-used single-frequency impedance-matching networks is of utmost significance. The influence of load impedance variation on the performance of L network is analyzed by Chung [20]. Chen and Weber [11] presented a process variation-insensitive network with matched passive components. Sun and Fidler discussed component tolerance and parasitic sensitivities of frequency response of Pi network in [21]. The most important performance of impedance-matching networks, in many applications, is the resulting reflection coefficient, which represents the effect of impedance matching. In the present work, the dynamic performance of T network impedance matching is theoretically analyzed using the reflection coefficient as the performance parameter. The remainder of this paper is organized as follows. The Q-based design method of T network
impedance matching is introduced in Section 2. Impacts of components deviation caused by tolerances and process variations on matching performance of the T network are analyzed in Section 3. Impacts of load variation and frequency drift on the matching performance are analyzed in Section4 and Section 5 respectively. The discussions and conclusions are presented in Section 6.
2 Q-based design method of T network impedance matching

2.1 Design process
The easy-to-use Q-based design method is often adopted in designing impedance matching networks for its simplicity [3, 22]. In order to theoretically analyze the dynamic performance of T network, we introduce the Q-based design method in this section. For simplicity, the source impedance and load impedance of the T network in Fig. 1 were considered as pure resistance, which are denoted by R1 and R2 respectively. Defining Q1=XL1/R1, Q2=XL2/R2 and k=R1/R2, the loaded quality factor Q0 can be defined as Q0 = (Q1+Q2)/2. The design process is as follows [22]:
R1 L1 L2
R2
Fig. 1 A T network for pure resistance matching a. An appropriate Q0 for the network is selected, which meets the designable condition of:
Q0 Q0(min) = 1 k 1 2 for R1 R2 (k 1) ,
(1)
E-ISSN: 2224-266X
34
Q0 Q0(min) =
1 1 k 1 for R1 R2 (k 1) 2
(2)
H ( j ) =
2(k + 1)Q0 2 p 4 (k + 1)2 Q (k + 1) p (k 1)2 Q 2 2 0 0 p 2 3 2 + (k 1) 3 p p + 2(k + 1)Q0 8kQ0 p 3 p
and other conditions, such as harmonic rejection. b. Q1 and Q2 are calculated using Eqs. 3 and 4. (k1) given by:
2 2Q0 4kQ0 (k 1) 2
[ [
) (
1 2
)(
)]
(8) where p = 4kQ02-(k-1)2, p = /0. The n-order harmonic rejection can be calculated by 20 log[|H(jn0)|/|H(j0)|] for specified Q0 and k. Fig. 2 shows the plots of the second and third harmonic rejection performances with respect to the loaded Q0 of the network, respectively when k=0.5. The harmonic rejection ability of T network is proportional to the value of Q0. A suitable Q0 can be selected according to the requirement of harmonic rejection when designing a T network impedance matching.
Q1 =
1 k
and
(3)
Q2 =
2 2kQ0 4kQ0 (k 1) 2
k 1
(4)
c. Components L1, L2, and C of the network are calculated using Eqs. 5 to 7 given by:
L1 = R1Q1
(5)
L2 =
R2 Q2
and
(6)
C=
2Q0
0 R1 (1 + Q12 )
2Q0
2 0 R2 (1 + Q2 )
(7)
where 0 is the matching angular frequency.
2.2 Selection of Q0
As shown in the design process above, the key in designing a T network impedance matching is to select the appropriate loaded quality factor Q0 for each network. In [20] and [22], the guidance of selection considers the harmonic rejection of the network. Defining the voltage transfer function as H(S) = V2/Vs, the complex frequency response H(j) can be obtained using circuit analysis, and then substituting S=j. The magnitude frequency response |H(j)| is derived using Eq. 8 expressed as:
Fig. 2. Harmonic rejection performance of a T impedance matching network
3 Impacts of components deviation on the matching performance

The passive components such as inductors and capacitors are the key components in T network impedance matching. Affected by tolerances and manufacturing process variations, the values of the passive components in practical circuits are different
E-ISSN: 2224-266X
35
from the theory design value. In addition, the small variation of components parameters can be caused by working environment factors such as temperature. These deviations from the theory design values prove to affect the matching effect of T network. The impacts of the deviation of components L1, L2 and C on the matching performance of T network are analyzed in this section. Considering the T network in Fig. 1, the equivalent impedance Ze of matched load at working frequency is demonstrated as
Z e = j 0 L1 + 1 j 0 C + 1 j 0 L 2 + R 2
Eq. 12 suggests that the deviation of inductance L1 from its theory design value can make the matching effect worse and the magnitude of resulting reflection coefficient || is proportional to the relative deviation dL1 and Q1. Fig.3 shows the relationship between the magnitude of resulting reflection coefficient || and Q1 when dL1=0.1. Tolerance of 10% in inductor L1 can cause the resulting reflection coefficient to be 0.5 when the selected Q1 reaches to 11.7. Eq. 12 can be used to calculate the resulting reflection coefficient caused by the deviation of inductor L1 for arbitrary quality factor.
(9)
For the ideal condition, the equivalent impedance Ze at working frequency is matched to be equal to R1 by the T network and the resulting reflection coefficient is zero.
3.1 Deviation of inductor L1

Considering a small deviation L1 from the design value of inductor L1 due to process variation or tolerance in Fig.1, the practical inductance of L1 is given by
L1 = L1 + L1 = L1 (1 + dL1 ) ,
(10)
Fig.3 Relationship between magnitude of reflection coefficient || and Q1 (dL1=10%)
where L1 is determined by Eq. 5 and dL1= L1/L1, the relative deviation of inductor L1. The practical equivalent impedance of matched load at working frequency becomes
Z e = Z e + j 0 L1 = R1 + jR1Q1dL1 .
3.2 Deviation of inductor L2

Considering a small deviation L2 from the design value of inductor L2 due to process variation or tolerance in Fig.1, the practical inductance of L2 is given by
L 2 = L 2 + L 2 = L 2 (1 + dL 2 ) ,
(11)
(13)
The changed equivalent impedance will result in a deviation from perfect match. The reflection coefficient caused by the deviation of inductance L1 can be calculated by
= Z e R1 jQ1 dL1 . = Z e + R1 jQ1 dL1 + 2
where L2 is determined by Eq. 6 and dL2= L2/L2, the relative deviation of inductor L2. The practical equivalent impedance of matched load at working frequency is demonstrated by
(12)
E-ISSN: 2224-266X
36
Z e = j 0 L1 +
1
j 0 C +
1 j 0 L 2 + R 2
(14)
Ze = j 0 L1 +
1
j 0 C +
1 R1 j 0 L1
(17)
Substituting Eqs. 5-7 into Eq. 14, the resulting reflection coefficient caused by the deviation of inductance L2 can be derived as
= Z e R1 jQ 2 dL2 A + jB , = Z e + R1 A jB 2 + jQ2 dL2
Substituting Eqs. 5 and 6 into Eq. 17, the resulting reflection coefficient caused by deviation of capacitor can be derived as
=
(15)
(Q1 + j )Q0 dC . ( Q1 + j )( Q0 dC + j )
(18)
where A = (k-1)2, B = 4kQ0-p(k+1). Eq. 15 suggests that the deviation of inductance L2 can make the matching effect worse and the magnitude of resulting reflection coefficient || is proportional to the relative deviation dL2 and Q2. It is obvious to see that Eq. 15 is similar with Eq. 12 and the relationship between the magnitude of resulting reflection coefficient and Q2 is identical with Fig.3. Similarly, tolerance of 10% in inductor L2 can cause the resulting reflection coefficient to be 0.5 when the selected Q2 reaches to 11.7.
3.3 Deviation of capacitor C

For the ideal condition where the passive components are the designed ideal value, the reflection coefficient in front of the capacitor is zero. Such condition indicates that the equivalent load impedance looking into the front of the capacitor C is equal to R1-j0L1 at working frequency. With regard to the small deviation of C in Fig. 1, the inductance of C is given by
C = C + C = C (1 + dC ) ,
Fig.4 Relationship between magnitude of reflection coefficient || and Q0 (dC=10%) Eq. 18 suggests that the deviation of capacitance C can make the matching effect worse and the magnitude of resulting reflection coefficient || is proportional to the relative deviation dC and Q0. Fig. 4 shows the relationship between the magnitude of resulting reflection coefficient || and Q0 when dC=0.1. Tolerance of 10% in capacitor C can cause the resulting reflection coefficient to be 0.5 when the selected Q0 reaches to 5.8.
(16)
where C is determined by Eq. 7 and dC= C/C is the relative deviation of inductor C. The changed equivalent impedance of matched load at working frequency is
4 Impact of load impedance variation on the matching performance

In practical application, the load impedance to be matched varies dynamically due to change of the working environment. This load impedance
E-ISSN: 2224-266X
37
variation can make the matching performance of T network worse or even produces impedance mismatch. In order to analyze the impact of load impedance variation on the matching performance, the load reflection coefficient l should be defined as the reflection coefficient of the load side. The function of T network is to reduce the resulting reflection coefficient to zero. A small change of d in load impedance can result in a changed load reflection coefficient, which is expressed as: l = l + d l 1. (19)
Substituting Eqs. 5-7 into Eq. 21, the resulting reflection coefficient can be derived through analytic derivation, which leads to
(Q1 + j )(Q2 + j )d . (Q1 j )(Q2 j )(1 l d l 2 )
(22)
The changed load impedance Zl can be also expressed in terms of load reflection coefficient l.
1 + l ' 1 l
'
Zl ' =
R1 =
1 + l + d R1 1 l d
(20)
The equivalent impedance Ze after impedance matching is obtained by:

Z e = jL1 + 1 j C + 1 jL2 + Z l '
(21)
Eq. 22 suggests that the dynamic variation of load impedance can make the matching effect worse and the magnitude of resulting reflection coefficient || is proportional to the variation of load reflection coefficient d and the load reflection coefficient l of static state and independent of the loaded quality factor Q0. The quality factor Q1 and Q2 are proportional to the phase angle of resulting reflection coefficient. Fig.5 shows the relationship between the magnitude of resulting reflection coefficient || and the magnitude of static load reflection coefficient |l| when d=0.1. The resulting reflection coefficient is doubled to 0.2 when the load reflection coefficient reaches to 0.66. Larger load reflection coefficient of static state means poorer dynamic stability to load impedance variation. A small variation in the load impedance will cause a large deviation from perfect match when the impedance difference between load and source is quite large. This problem should be specially considered in designing T network impedance matching.
5 Impact of frequency drift on the matching performance

In applications where small frequency drift is inevitable, the deviation of the reflection coefficient from zero should be considered although the bandwidth is not the main concern of single-frequency impedance matching networks. Considering the circuit in Fig. 1, for a small frequency drift , the drifted working frequency can be defined as:
Fig. 5. Relationship between magnitude of reflection coefficient || and magnitude of load reflection coefficient |l| (d=0.1)
E-ISSN: 2224-266X
38
=+=(1+d),
(23)
where d=/, the relative change of working frequency. Similarly, as in the analysis above, the equivalent impedance can be demonstrated as:
Z e = j L1 + 1 j C + 1 j L2 + R 2
and 0.01 respectively. From Fig. 6, we can see that the value of k has a negligible effect on || for Q0 > Q0(min). Hence, k can be taken as a constant to analyze the reflection coefficient caused by frequency drift d. When k is set to 1, Eq. 25 becomes:
jQ0 M M 2 1
(24)
1 + jQ0 M M 1
).
(26)
Substituting Eq. 23 and Eqs 5-7 into Eq. 24, the resulting reflection coefficient in terms of Q0, k and d can be obtained as:
=
( jQ Q Q M + 2Q 2Q )(M 1) ( jQ Q Q M + 2Q )(M 1) 2 jQ M + Q Q
0 1 2 0 2 1 2 0 1 2 0 2 2 0 1
, (25)
where Q1 and Q2 are defined by Eqs. 1 and 2 and M=1+d.
Eq. 26 shows that the magnitude of resulting reflection coefficient caused by frequency drift is proportional to d and Q0. Fig. 7 shows how d influences || when Q0 is set to various values. The magnitude of resulting reflection coefficient || is proportional to Q0 and this is consistent with the circuit theory. A circuit with higher quality factor has higher selectivity and is more sensitive to frequency drift (i.e. poorer dynamic stability). || increases rapidly for a small change of working frequency when Q05.
Fig. 6. 3D graph of magnitude of resulting reflection coefficient caused by frequency drift With the three variables Q0, k and d, Eq. 25 is quite complicated. However, it can be demonstrated by 3D graph for specified d. Fig. 6 shows the magnitude of resulting reflection coefficient caused by frequency drift when d equals to 0.001, 0.005
Fig. 7. Relationship between frequency drift d, Q0, and magnitude of reflection coefficient ||
6 Discussions and Conclusions

a) In practical application, the components deviation, load impedance variation and working
E-ISSN: 2224-266X
39
frequency drift can make the matching effect worse (i.e. make the reflection coefficient deviate from zero). b) For a T network with larger quality factor value, the matching effect is more sensitive to components deviation, load impedance variation and working frequency drift and the dynamic stability is worse. c) In the application of load impedance dynamically varying, if the static load reflection coefficient is larger (i.e. the impedances of source and load are mismatched significantly), the matching effect is more sensitive to load impedance variation (i.e. the dynamic stability is worse). d) In designing T network impedance matching, the quality factor can not be too high in order to ensure good dynamic performance. However, the harmonic rejection ability is proportional to quality factor. Therefore, it is need to make the tradeoff between dynamic stability and harmonic rejection ability. e) The formulas to calculate reflection coefficient presented in this paper can be used to determine whether the T network is suitable for dynamic application.
7 Acknowledgements
This work has been supported by the Funds for Innovative Research Groups of China (51021005).
8 References
[1] Zhang S W. High frequency electronic circuits. Beijing: Higher Education Press, 2004: 44-64. [2] Matthaei, G L, Young L, Jones E M T. Microwave filters, impedance matching network and
coupling structure. Norwood: Artech House, 1980: 255-272. [3] Bowick C. RF circuit design. London: Newnes, 1997: 63-102. [4] Motorola application note AN-721. Impedance matching networks applied to RF power transistors. Motorola Inc. 1993. [5] Thompson M, Fidler JK. Determination of the impedance matching domain of impedance matching networks. IEEE Trans. Circuits Syst. I: Regular Papers 2004; 51(10):2,0982,106. [6] Gao S J, Wang Z B, and Park C W. A novel RF tunable impedance matching network for correcting the tested result deviation from simulated result. IEEE Asia-Pacific Conf. App. Electromag. (APACE). 2010(11):1-4. [7] van Bezooijen, A., de Jongh, M.A., Chanlo, C., et. al. RF-MEMS based adaptive antenna matching module. IEEE Radio Frequency Integr. Circuits (RFIC) Symp., 2007(7):573-576. [8] van Bezooijen, A., de Jongh, M.A., Chanlo, C., et. al. A GSM/EDGE/WCDMA adaptive series-LC matching network using RF-MEMS switches. IEEE Journal of Solid-State Circuits, 43(10), 2259-2268. [9] Serban A, Karlsson M, Gong S. Component tolerance effect on ultra-wideband low-noise amplifier performance, IEEE Trans. Adv. Packag. 2010, 33 (3): 660-668. [10] Nieuwoudt A, Ragheb T, Nejati H, et al. Variation tolerant design methods for wideband low noise amplifiers. Analog Integr. Circ. S., 2009, 58(1): 49-54. [11] Chen F, Weber R J. A novel process-variation insensitive network for on-chip impedance matching. communications and information technology. IEEE Int. Symp. Commun. Inf. Technol. (ISCIT), 2004(1): 43-46. [12] Liao R J, Tan J W, Wang H. Q-based design method for impedance matching network considering load variation and frequency drift. Microelectron. J. 2011, 42 (2): 403-408. [13] Hastings A. The art of analog layout. New Jersey: Prentice-Hall Inc.; 2001: 200203. [14] Ramesh R, Ebenezer D D. Equivalent circuit
E-ISSN: 2224-266X
40
for broadband underwater transducers. IEEE Trans. Ultrasonics, Ferroelectrics and Frequency Control. 2008, 55(9): 2079-2083. [15] Dogra VS, Zhang M, Bhatt S. High-intensity focused ultrasound (HIFU) therapy applications. Ultrasound Clin 2009; 4(3):30721. [16] Bezooijen A, Jongh M A, Straten F, et al. Adaptive impedance-matching techniques for controlling L networks. IEEE Trans. Circuits Syst. I, 2010, 57(2): 495-505. [17] Kong Q, Zhao C L, Tan J W. On-line detection of ultrasound transducer impedance angle. Measurement & control technology, 2009, 28(3): 16-19. [18] Bao S H. Dynamic matching for a piezoelectric transducer. Applied acoustics, 1998, 17(2): 16-20. [19] Bennett B J. Broad-band impedance matching of ladder loads. IEEE Trans. Circuits Syst., 1985, 32(12): 1201-1208. [20] Chung B K. Variability analysis of impedance matching network. Microelectron. J. 2006, 37(11): 1419-1423. [21] Sun Y, Fidler J K. Design method for impedance matching networks, IEE Conf. Proc. Circuits Devices Syst. 1996, 143(4): 186-194. [22] Chung B K. Q-based design method for T network impedance matching. Microelectronics J. 2006, 37(9): 1007-1011.
with college of electrical engineering, Chongqing University. He is currently professor and the general secretary of college of electrical engineering and deputy director of the State Key Laboratory of Power Transmission Equipment & System Security and New Technology, Chongqing University. He has been engaged in intelligent electrical apparatus insulation online monitoring, fault diagnosis technology and high-voltage testing techniques work. In areas of online intelligent monitoring and fault diagnosis technology of transformer, capacitive equipment insulation online monitoring and intelligent fault diagnosis technology, state maintenance of electric equipment, electrical equipment insulation aging and remaining life expectancy theory and technologies, such as steep pulse wave measuring research direction, he has made the international advanced, part of the outcome of the international leading level. He has directed and participated in over more than 40 projects, which presided over 9 projects in recent 5 years. The project funding amounted to more than 5 million. He has published more than 110 papers, 52 of which were cited by SCI/EI.
9 Biographies
Liao Ruijin was born in 1963. He received his masters and doctors degrees of Electrical Engineering from Chongqing University and Xian Jiaotong University respectively. In 1988 he began his career
Tan Jianwen was born in 1980. He received his bachelors and masters degree of Electrical Engineering from Science and Technology University of PLA. His masters thesis was awarded the outstanding masters thesis of PLA. In 2005 he began his career as lecturer with department of electrical engineering in Chongqing Communication Institute. Currently, he is doing his doctoral studies by research in Electrical Engineering of Chongqing University. His research interests are in the field of power electronics,
E-ISSN: 2224-266X
41
electromagnetic compatibility and new electrical technology of bio-medical. He has participated in over more than 5 research projects. He has published 15 papers, 5 of which were cited by SCI/EI.
E-ISSN: 2224-266X
42

Dynamic Performance Analysis of T Network Impedance Matching

Uploaded by

Copyright:

Available Formats

Dynamic Performance Analysis of T Network Impedance Matching

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dynamic Performance Analysis of T Network Impedance Matching

Uploaded by

Copyright:

Available Formats

WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS