Escolar Documentos
Profissional Documentos
Cultura Documentos
by Michael Tse
September 2003
Contents
The Problem Q-factor matching approach Simple matching circuits L matching circuits matching circuits T matching circuits Tapped capacitor matching circuits Double-tuned circuits General impedance matching based on two-port circuits Immittance matrices and hybrid matrices ABCD matrix and matching Propagation equations from ABCD matrix
Impedance Matching
Impedance matching is a major problem in highfrequency circuit design. It is concerned with matching one part of a circuit to another in order to achieve maximum power transfer between the two parts.
max power transfer
Circuit 1
Circuit 2
space
The problem
Given a load R, nd a circuit that can match the driving resistance R at frequency w0.
R R
Obviously, the matching circuit must contain L and C in order to specify the matching frequency.
In general, a circuits reactance is a function of frequency and the Q factor is dened at the resonance frequency w0 .
X As we will see later, the Q factor can be used to modify the overall resistance of a circuit at some selected frequency, thus achieving a matching condition.
w0
Low Q circuit
X X
High Q circuit
w0
w0
Denition:
Q=
w 0 dB w dX = 0 2G dw w =w 0 2R dw w =w 0
It is easily shown that for linear parallel RLC circuits: Q = w0CR = R/(w0L)
6
Resistance () Z IMPEDANCE ()
inductance (H)
jwL = +jX
jwC = +jB
Conductance (S)
susceptance (S)
7
X 1 G Q= = = R RB B
L R
wL Q= R
C R
1 Q= wCR
Parallel:
R L
R Q= wL
R C
R B Q = = RB = X G
Q = wCR
jX
Y=
Z = R + jX
1 jX 1 + 2 Q
R(1+Q2)
1 1 R X = 2 + R 2 X 1+ j + 1 R X 1 1 = + 1 R(1 + Q2 ) jX 2 + 1 Q
or j R' Q
R j (1 + Q2 ) Q
1 jRQ 2 + 1 Q
9
jB
Y = G + jB
G(1 + Q2 )
conductance (S)
1 1 G B = 2 + G 2 B 1+ j + 1 G B 1 1 = + 1 G(1 + Q2 ) jB 2 + 1 Q
or j G' = j G (1 + Q2 ) = jGQ 12 + 1 Q Q Q
10
1 LC
Q factor is Q = R
C L
Bandwidth is Dw = w 2 - w1 =
1 RC
w1 w0 w2
Note: w1 and w2 are called 3dB corner frequencies. Their geometric mean is w0. For narrowband cases, their arithmetic mean is close to w0.
Michael Tse: Impedance Matching 11
Q factor = QC = w0CRC
(unloaded Q factor)
RL
Q factor = QL = RL/w0L
(unloaded Q factor)
1 1 1 = + QLC QC QL
(easily shown)
12
13
Yi n = jwC +
1 R + jwL
R w L = 2 + jw C - 2 2 2 R + (w L) R + (wL)
Obviously, the reactive part is cancelled if we have
L C= 2 2 R + w2 L 0
1 R2 - 2 where w 0 = LC L
(#)
14
(*)
Here, Q is the Q-factor, which is equal to w0L/R (for series L and R). So, we can see clearly that Q is modifying R to achieve the matching condition. Design procedure: -Given R and R, nd the required Q from (*). -Given w0, nd the required L from Q = w0L/R . -From (#), nd the required C to give the selected resonant frequency w0.
15
Shunt L circuit:
L C Zin R
Begin with
Zi n = jwL +
1 G + jw C
G wC = 2 + jwL - 2 2 2 2 2 G +w C G +w C
Reactive part is cancelled when
C L= 2 2 G + w2 C 0
1 G2 where w 0 = LC C 2
(#)
R =
1/ G R = 1 + (w 0 C / G)2 1 + Q2
(*)
Series:
Shunt:
jX2
R = R(1 + Q2 ) 1 jR jX 2 = - jX 1 1 + = Q Q2 X Q= 1 R
R 1 + Q2 jX1 jX 2 = = - j RQ 1 1+ 2 Q B R Q= 1 = G X1 R =
18
jX1
Advantages of L circuits:
Simple Low cost Easy to design
Disadvantages of L circuits:
The value of Q is determined by the ratio of R/R. Hence, there is no control over the value of Q. the bandwidth is also not controllable.
19
p matching circuits
jX2
Analysis by decomposing into two L circuit sections: First section (from right): R
R X = X 2 - R Q1 2 1 + Q1 B Q1 = 1 = B1R G R =
jB3
R + jX
jB1
jX
j(X2RQ1)
Second section:
Q2 =
jB3
R + jX
X X 2 - R Q1 X = 2 = Q1 + Q2 R R R
2 R = R (1 + Q2 ) Q Q B = B3 - 2 B3 = 2 R R
20
jB3
R + jX
jB1
R
1 2 1 + Q1
R
2 1 + Q2
Obviously, we have to set Q1 > Q2 if we want to have R<R. Likewise, we need Q1 < Q2 if we want to have R>R.
Michael Tse: Impedance Matching 21
For R > R
1. 2. 3. 4. 5. 6. Select Q2 according to the max Q.
2 Find R using R = R /(1 + Q2 )
Get Q2 using Q2 =
R -1 R
Get Q2 using Q1 =
R -1 R
22
T matching circuits
The analysis is similar to the p case.
jX3 jX1 jB2
R + jX
The difference is that R is rst raised to R by the series reactance, and then lowered to R by the shunt reactance. The design procedure can be similarly derived. (Exercise)
R
2 1 + Q1
1 2 1 + Q2
R
23
For R < R
1. 2. 3. 4. 5. 6. Select Q2 according to the max Q.
2 Find R using R = R(1 + Q2 )
Get Q2 using Q2 =
R -1 R
Get Q1 using Q1 =
R -1 R
24
C1 L C2 R
R 1 + Q2 p
Q factor
Qp = w 0 C2 R
25
C1 R
Q1 = R/w0L
1 + Q2 p C2 Q2 p
R
R R 2 1 + Q2 1 + Q1 p
1 + Q2 p
required
R R = Qp = 2 2 1 + Q1 1 + Qp
R 2 1 + Q1 -1 R
26
2 1 1 1 Qp = + 2 C C1 C2 1 + Qp
R 1 + Q2 p
1 LC
Also, we have the alternative approximation for Q1: Q1 w0RC, which is set to w0 / Dw . Thus, we can go backward to nd all the circuit parameters.
27
28
Disadvantage:
no precise control of the bandwidth
29
wm1
wm2
There is a mid-band dip, which can be made small if the pass band is narrow. Also, large difference in the impedances to be matched can be achieved by means of galvanic transformer.
Michael Tse: Impedance Matching 30
The construction of a double-tuned circuit typically includes a real transformer and two resonating capacitors. M
RG C1 L11
L22 C2 RL
n=
L1 1 k 2 L2 2
31
Equivalent models:
n:1 L22(1k2)
RG C1 L11
C2 RL
ideal transformer
L2 RG C1 L11 C2 RL
1 L11 2 - 1 k
L11 2 R2 k L22 L11 2 C2 k L22
32
Exact match is to be achieved at two given frequencies: fm1 and fm2. L2 RG C1 L11
R1 R2
C2
RL
Observe that: R1 resonates at certain frequency, but is always less than RG R2 decreases monotonically with frequency
So, if RL is sufciently small, there will be two frequency values where R1 = R2.
Michael Tse: Impedance Matching 33
resistance
R2
R1 f fm1 fm2
Our objective here is to match RG and RL over a bandwidth Df centered at fo, usually with an allowable ripple in the pass band.
34
i2
+
v1
v2
Idea: we dont care what is inside, as long as it can be modelled in terms of four parameters.
35
Two-port models
+
i1
i2
port 1
v1
v2
port 2
z-parameters (impedance matrix): y-parameters (admittance matrix): h-parameters (hybrid matrix): g-parameters (hybrid matrix):
: :
36
z11 = z12 =
z21 = z22 =
v2 i1 v2 i2
=
i2 = 0
v2 i1 v2 i2
=
i1 = 0
37
g11 = g12 =
i1 i2 v2 v1 v2 i2
=
v1 = 0
i1 i2
g21 = g22 =
=
i2 = 0
v2 v1 v2 i2
=
v1 = 0
38
Input impedance:
+ v1
i1
i2
[Z]
Zin
ZL
v2
Z in = z11 - Z L + z22
Michael Tse: Impedance Matching 39
z12 z21
Similarly, we can nd the input impedance at any port in terms of any of the two-port parameters, or even a combination of different twoport parameters. We will see that the matching problem can be solved by making sure that both input and output ports are matched.
ZG i1 i2
+
ZIM1
[Z]
ZIM2
ZL
v2
matching: ZG = ZIM1
and ZIM2 = ZL
image impedances
Michael Tse: Impedance Matching 40
i1
i2
+ v2
[ABCD]
Here, voltage and current of port 1 are expressed in terms of those of port 2. So, this is neither an immittance matrix like Z and Y, nor a hybrid matrix like G and H.
v1 A B v 2 = i C D 1 -i2
Note: the sign of i2 in the above equation. This sign convention will make the ABCD matrix very useful for describing cascade circuits.
+ v1
i1
i
+
i2
+ v2
[ABCD]1
v1 A1 B1 v ' = i C D 1 1 1-i'
Since i = i, we have
[ABCD]2
v ' A2 = i" C2 B2 v 2 D2 -i2
v1 A1 = i1 C1
B1 A2 D1C2
B2 v 2 D2 -i2
So, if more two-ports are cascaded, the overall ABCD matrix is just the product of all the ABCD matrices.
A= B= C=
v1 v2
=
i2 = 0
v1 v2
=
port 2 open -circuited
z11 z21
-v1 i2 v i1 v2
=
2=0
=
i2 = 0
i1 v2
1 z21
-i1 D= i2 v
43
Matching problem
ZG i1
+ v1
i2
[ABCD]
ZIM1
ZL
v2
v1 = Av 2 - Bi2 i1 = Cv 2 - Di2
Z in =
ZG
+ v1
i1
i2
+ v2
[ABCD]
ZIM2
v1 = Av 2 - Bi2 i1 = Cv 2 - Di2
Z IM2 =
because AD BC = 1
AZ L + B Z IM1 = ZG = CZ L + D
and
Z IM1 =
AB CD
and
Z IM2 =
DB AC
Alternatively, we have
Z IM1 =
z11 y11
and
Z IM2 =
z22 y 22
46
Note: image impedances are different from input and output impedances. 1. Image impedances do not depend on the load impedance or the source impedance. They are purely dependent upon the circuit.
Z IM1 = z11 y11 and Z IM2 = z22 y 22
2.
Input impedance (Zin) depend on the load impedance. Output impedance (Zout) depends on the source impedance. For example,
Example
i1
+ v1
Za Zb
Zc
i2
+ v2
port 1
port 2
=
port 1 short -circuited
1 Zc + Za Zb
2
ZIM2 = ZIM3
3
ZIM3 = ZIM4
ZL ZIM4 = ZL
Convention
i1
+
i2
+
v1
v2
A wave or signal entering into circuit 1 from left side will travel without reection through the circuits if all ports are matched. Propagation constant g
input power v1i1 v1 e = = = output power v 2 (-i2 ) v 2
g
ZIM1
ZIM2
Z IM2 Z IM1
49
Propagation equations
eg = v1i1 v = 1 v 2 (-i2 ) v 2 Z IM2 Z IM1
v1 e = v2
g
In general,
v1 Av 2 - Bi2 B = = A+ v2 v2 Z IM2
= A+ B
AC = BD
D A
A D
AD + BC
i1 = CZ IM2 + D = -i2
AD + BC
Thus,
eg =
v1i1 = AD + BC -v 2i2
e-g = AD - BC
Michael Tse: Impedance Matching 50
Dene
n=
Z IM1 = Z IM2
A D
We have
For a transmission line, ZIM1 = ZIM2 = Zo, where Zo is usually called the characteristic impedance of the transmission line. Also, for a lossless transmission line, g = jL is pure imaginary, and thus tanh becomes tan, sinh becomes sin, cosh becomes cosh.
v1 Z L + jZ o tan L Z in = = Z o i1 Z o + jZ L tan L
Michael Tse: Impedance Matching 52
This is just the same transmission line equation. In communication, we usually express L as electrical length, and is equal to L = w l / v = 2p l / l wavelength frequency in rad/s
length of transmission line velocity of propagation
So, we can easily verify the following standard results: 1. If the transmission line length is l/2 or l, then the input impedance is just equal to the load impedance. 2. If the transmission line length is l/4, then the input impedance is Zo2/ZL. Impedance value for other lengths can be found from the equation or conveniently by using a Smith chart.
Michael Tse: Impedance Matching 53