RF-Engineering
1 Introduction 1
1.1 History of wireless communication . . . . . . . . . . . . . . . . . 1
1.2 Wireless Communication Systems . . . . . . . . . . . . . . . . . 2
1.2.1 Terrestrial Mobile Telephony . . . . . . . . . . . . . . . . 2
1.2.2 Wireless local area networks (WLANs) . . . . . . . . . . 2
1.2.3 Satellite Communication . . . . . . . . . . . . . . . . . . 2
1.3 Circuit Technoloy Options for Wireless RF/Microwave Circuit
Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.1 Semiconductor Devices . . . . . . . . . . . . . . . . . . 3
3 Scattering Parameters 29
3.1 Definition of wave amplitudes . . . . . . . . . . . . . . . . . . . 30
3.2 Definition of S-Parameter . . . . . . . . . . . . . . . . . . . . . . 31
3.2.1 Two Port . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.2 N Port . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2.3 Useful Definitions . . . . . . . . . . . . . . . . . . . . . 32
3.3 Lossless Networks . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4 Signal Flow Graphs . . . . . . . . . . . . . . . . . . . . . . . . . 36
I
3.4.1 Definition of graphical elements . . . . . . . . . . . . . . 36
3.4.2 Special devices and their graphical representation . . . . . 37
3.5 Network Analyzer . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4 Microwave Amplifier 46
4.1 Y-Parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.1.1 2-port described by y-parameter . . . . . . . . . . . . . . 48
4.2 Stability of a 2-port . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.1 Stability factor . . . . . . . . . . . . . . . . . . . . . . . 53
4.2.2 Stabilisation of an active device . . . . . . . . . . . . . . 54
4.3 Power gain definitions . . . . . . . . . . . . . . . . . . . . . . . 55
4.3.1 Unilateral transistor . . . . . . . . . . . . . . . . . . . . . 57
4.3.2 Bilateral transistor . . . . . . . . . . . . . . . . . . . . . 58
4.3.3 Optimum input and output matching . . . . . . . . . . . . 59
4.3.4 Maximum available power gain . . . . . . . . . . . . . . 60
4.4 Differnt amplifier concepts . . . . . . . . . . . . . . . . . . . . . 61
5 Plane Wave 65
5.1 Wave Equation of the Plane Wave . . . . . . . . . . . . . . . . . 65
5.2 Weakly damped plane wave . . . . . . . . . . . . . . . . . . . . . 67
5.2.1 Plane Wave in Free Space . . . . . . . . . . . . . . . . . 68
5.2.2 Plane wave in a lossless Material . . . . . . . . . . . . . . 69
5.2.3 Polarisation . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2.4 The Wave Vector . . . . . . . . . . . . . . . . . . . . . . 74
5.3 Plane Wave in a Good Conductor . . . . . . . . . . . . . . . . . . 75
5.3.1 Ideal Conductor . . . . . . . . . . . . . . . . . . . . . . . 77
6 Antenna 78
6.1 Transmitter Receiver System . . . . . . . . . . . . . . . . . . . . 78
II
Chapter 1
Introduction
The rapide development of wireless communication products over the last ten
years has led to an explosion in interest in improved circuit design approaches
in the radio frequency (RF) and the microwave area. The develpoment of the
field is characterized by a shift from purley analog techniques to those employing
digital or hybrid ditigal/analog approaches. This shift is further accelatrated by
the advance in digital integrated circuit technology and monolithic microwave
integrated circuits.
1
1.2 Wireless Communication Systems
1.2.1 Terrestrial Mobile Telephony
GSM has become a worldwide used standard for terrestrial mobile telephony. In
the following table some technical data of the GSM-Standart is given in compari-
sion to IS-54 an U.S. Interim Standard.
IS-54 GSM
Accces methods TDMA/FDMA TDMA/FDMA
Frequency bands 824 - 848 890 - 915
in MHz 869 - 894 935 - 960
Channel spacing 30kHz 200kHz
π
Modulation 4 QPSK GMSK
Transmitted Power 600mW 1W
Bit Rate 8kbps 13kbps
Time Slots per
Channel 3 8
Table 1.1: Comparison of different mobile systems
2
915MHz 2,4GHz 5,8GHz
Frequencies 902 - 928 2,4 - 2,4835 5,725 - 5,825
Bandwidth 26MHz 83,5MHz 150MHz
Availability US/Canada Worldwide US/Canada
Transmission Range Greatest 95% of 915MHz 80% of 915MHz
Technologies Si Si/GaAs GaAS
Interference Celluar Microwave Some Radar
Sources Phones Oven Application
Table 1.2: WLAN in different ISM-Bands
the second is the not yet established mobile satellite communication descirebed
as:
• Geosynchronous mobile satellite service (e.g. Inmarsat) for voice, data and
video services operating with frequencies from one to six GHz (Geosyn-
chronous Earth Orbit, GEO)
• Low Earth Orbit (LEO) mobile satellite service (Iridium system of Mo-
torola) with about 66 satelites in orbit
3
• Minimum necessary suply voltage
• Short circuit current bandwidth ( fT ), important for both analog and digital
circuits
• Maximum power gain bandwidth ( fmax ), important for both analog and dig-
ital circuits
• Minimum noise figure, LNAs, wide dynamic range front ends and mixers
• Yield
4
Chapter 2
Figure 2.1: Electric and magnetic field lines of a two wire line
5
L′ ∆z i(z + ∆z,t)
i(z,t)
✲ s ✞☎
✞☎✞☎
✞☎✞☎ s s ✲
z z + ∆z
2.1 Analyse
Fig. 2.2 refers to the time domain, this means that the function u(z,t) can show
different values for any point z or time t. Considering that all occurring time func-
tions are related to sinusoidal time functions by the Fourier analysis or Fourier
transform, it is sufficient to ristrict ourself to sinusoidal time functions to de-
velop the transmission line theory. If the function u(z,t) has a sinusoidal time-
dependency, we can use the following reprensentation:
u(z,t) = Û(z) sin(ωt + ϕ) = Im Û(z) exp( jϕ) exp( jωt) = Im [U(z) exp( jωt)]
(2.1)
In the last equation, Û(z) represents the amplitude of the function u(z,t) at certain
point z, in addition, the phasor function U(z) = Û(z) exp( jϕ) is introduced. Its
absolute value is equal to the amplitude function Û(z), while its phase is equal to
the phase of the time function sin(ωt + ϕ). Similar considerations can also apply
to the space and time function i(z,t), which also represented by a complex phasor
I(z), beeing a function of the z-coordinate.
6
R′ ∆z L′ ∆z I(z + ∆z)
I(z)
✲ s ✞☎
✞☎✞☎
✞☎✞☎ s s s ✲
z z + ∆z
dU(z)
− = (R′ + jωL′ )I(z) (2.2)
dz
An equivalent equation can be setup for the local decrease of the current:
dI(z)
− = (G′ + jωC′ )U(z) (2.3)
dz
By derivation of the equation 2.2 with respect to z and taking into account equation
2.3 we obtain an homogeneous linear differential equation of second order for
phasor of the complex voltage U(z).
d 2U(z)
2
= (R′ + jωL′ )(G′ + jωC′ )U(z) (2.4)
dz
It is well known that the exponential function is a possible solution for the above
differential equation. So we will use the following function as trial function:
In principal there exist two possible solution. The equation can be fulfilled if there
exists now voltage across the transmission line (U(z) ≡ 0) - a trivial solution of
7
little interest for an electrical engineer. Instead the nontrivial solutions defines the
value of the complex constant of propagation γ.
8
By using equation 2.5 and equation 2.1 to convert the complex voltage distribution
U(z) to the voltage function depending on z and t, we obtain:
u(z,t) = Im[Û0 exp(− jβz) exp( jωt)] = Û0 sin(ωt − βz) (2.10)
u(z,t) ✻
Û0
1
t = ∆t
0 t =0
-1
✲
-1 -0.6 -0.2 0.2 0.6 z 1
λ
Figure 2.4: Voltage distribution along a lossless transmission lne for t = 0 ad
t = ∆t > 0
9
The phase velocity v ph of the lossless transmission line as defined in equation 2.11
is equal to the propagation velocity v of pulses along lossless transmission line.
Moreover its value is identical to the propagation velocity of TEM waves accord-
ing to Maxwell’s theory. We therefore need:
1 c0
vPh = √ = √ (2.12)
L′C′ εr
The last equation shows that there exist a relation between the inductance and
capacitance per unit length of a transmission line according to Maxwell’s theory.
Another important characteristic value of transmission line theory is the wave-
length λ of the transmission line waves. The wavelenght corresdonce to a length
∆z where the wave will show a phaseshift of 2π. From this definition, we obtain:
2π vPh
λ = = (2.13)
β f
which results in the following equation using the results of Maxwell:
vPh c0
λ = = √ (2.14)
f f εr
If we focus on the equation 2.7 in the lossless case, we obtain a real value for the
characteristic wave impedance.
s
L′
ZC = ZC = (2.15)
C′
In the lossless case, the characteristic wave impedance is real and independent of
the frequency.
10
Considering the above conditions, both brackets within the last root can be multi-
plied and as a first-order approximation the product of the small-sized terms can
be negcelcted. s
√ R′ G′
γ ≈ jω L′C′ 1 − j ′ − j ′
ωL ωC
√
The following approximation 1 + ε ≈ 1 + ε/2 is used for further simplification:
√ R′ G′
γ ≈ jω L C 1 − j(
′ ′ + )
2ωL′ 2 jωC′
u(z,t) = Im[Ûo exp(−(α + jβ)z) exp( jωt)] = Ûo exp(−αz) sin(ωt − βz) (2.18)
By comparing the last equation with the result of the lossless circuit (equation
2.10), it is obvious that the losses are reflected in the term exp(−αz). The value
of the exponential function becomes lower with increasing z, so that the ampli-
tude of the wave and as a result the transported power will decrease. Therefore,
the constant /al pha is denoted as damping constant of the transmission line. To
illustrate again the physical process, of a damped wave according to equation for
2.18 Fig. 2.5 shows the space and tine function u(z,t) at time t = 0 and at time
t = ∆t > 0. It becomes evident that the amplitude continuously decreases with
propagating wave in the z-direction. A part of the power transported by the wave
is transformed into heat while the wave propagates along the transmission line.
11
u(z,t) ✻
Û0
1
t = ∆t
0 t =0
-1
✲
-1 -0.6 -0.2 0.2 0.6 z 1
λ
Figure 2.5: Voltage distribution along a lossy transmission line t = 0 and t = ∆t >
0
α at 200 MHz.
|U(0)| |U(0)|
|in dB = 3 → = 103/20 = exp(α · 100 m)
|U(100m)| |U(100 m)|
1 1
α = ln(103/20 ) ≈ 3, 5 · 10−3
100 m m
To focus on the characteristic wave impedance in the lossy case, we refer to the
equation ??. Provided that the losses are low, we can use following approach:
v
R′ R′
u
u
u1− j ′ 1 − j
Z L = ZL u
u ωL ≈ Z 2ωL′
L
t G′ G′
1− j ′ 1− j
ωC 2ωC′
If we consider the last equation, we see that the numerator of the fraction will
result in a negative phase change when, while the denominator is rotating in the
positive direction. Both rotations will esentially compensation while each rotation
itself will decrease with increasing frequency. A more detailed analysis shows
[?] that for frequencies above 10 kHz, the complex characteristic impedance ZC
practically will become real ZC .
ZC ≈ ZC (2.19)
Therefore, we will assume that for RF- and microwave frequencies the character-
istic wave impedance of a transmission line will always show a real value.
12
2.2.3 The reflection coefficient in the frequency domain
Remembering the results of the lossless transmissin line in the time domain, re-
flections will occure when the load impedance terminating the transmission line
is not equal to its characteristic impedance. To calculate the voltage distribution
U(z) along the line, both waves the incident and reflected wave must be consid-
ered:
U(z) = U i exp(−γz) + U r exp(γz) (2.20)
The constants U i and U r introduced into equation 2.20 define the complexe am-
plitude of forward- and backward-traveling waves z = 0. Therefore, the input
reflection coefficient rin of a transmission line is defined by:
Ur
rin = (2.21)
Ui
In case a reflection occur the same approach must be used for the current distribu-
tion I(z) along the transmission line:
I(z) IA
✲ ✲
s s
U(z) ZL , γ UA ZA
❄ ❄
s s
✲
0 l z
complex load impedance Z A determining the ratio between voltage U A and current
I A . The equation 2.20 can be used to calculate the voltage U A :
13
The current I A is obtained by applying 2.23:
1 h i
IA = U i exp(−γl) − U r exp(γl) (2.25)
ZC
If we form the ratio of the preceding equations, we obtain:
U i exp(−γl) + U r exp(γl)
ZA =
1 h i
U exp(−γl) − U r exp(γl)
ZL i
In the last equation, only the complex amplitudes of the forward- and backward-
traveling waves U i and U r are unknown. For this reason, this term can be used to
determine the input reflection factor rin :
Ur Z − ZL
rin = = exp(−2γl) A (2.26)
Ui Z A + ZL
Considering the last term of equation 2.26, it appears obvious that this term defines
the output reflection coeffcient rA due to the arbitrary load impedance Z A :
Z A − ZL
rA = (2.27)
Z A + ZL
On the other hand, the first term of equation 2.26 describes the transformation of
the output reflection coefficient rA into the input reflection coefficient rin :
Equation 2.28 demonstrates again the interrelation between the output reflection
factor rA and the input reflection factor rin of a transmission line; two effects are
obvious. One of these effects is that the output reflection factor will be rotated in a
mathematical negative sense (clockwise) by 2βl through the transmission line; the
second effect, damping of the transmission line will reduce of the absolute value of
the reflection coefficient. In Fig. 2.7 the impact caused by the transformation line
on the reflection coefficient along the line is depicted. To explain the impact of the
output reflection factor rA on the voltage distribution U(z) along the transmission
line U(z), the equation 2.20 is rearranged as follows:
Ur
U(z) = U i exp(−γz) 1 + exp(2γz)
Ui
The ratio U r /U i can be replaced by the output reflection coefficient:
h i
U(z) = U i exp(−γz) 1 + rA exp(−2γl) exp(2γz)
14
1
Im(r) ✻
0.5 ✒
■
❅
r❅ E ❅ rA
0 ❅
-0.5
-1
✲
-1 -0.5 0 0.5 1
Re(r)
This leads to an equation describing the voltage distribution along the transmission
line: h i
U(z) = U i exp(−γz) 1 + rA exp(−2γ(l − z)) (2.29)
The last term becomes particularly concise if we consider a lossless line (γ = jβ)
and focus on the absolute value of the voltage |U(z)|.
Since the term | exp(− jβz)| is equal to one, the bracket term is particularly rel-
evant. Due to the exponential function and the space coordinate z it will change
its value between 1 − |rA | and 1 + |rA |. This is also illustrated by Fig. 2.8. Here
the voltage distribution U(z) is shown normalised to incident wave amplitude U i
for differnt real values of rA . The diagramm shows that the voltage U(z) shows a
constant amplitude equal to U i for rA = 0 along the transmission line. If reflec-
tions occur, the amplitudes will oscillate between the values Ûmin = |U i |(1 − |rA |)
and Ûmax = |U i |(1 + |rA |). The ratio Ûmax /Ûmin is called Voltage Standing Wave
Ratio (vswr). In the case |rA | = 1 the maximum voltage along the transmission
line can become two times the value of the incident wave amplitude. In this case
15
|U(z)| ✻
|U h | rA = 1
2
1.5 rA = 0.5
1 rA = 0
0.5
0 ✲
0 0.2 0.4 0.6 0.8 z 1
l
Figure 2.8: Voltage distribution along a transmission line for diffferent reflection
coefficients rA
16
matching point • z=1 r=0
short circuit ⊗ z=0 r = −1
open circuit ◦ z→∞ r=1
These points a depicted in 2.9 to show their mapping from the z- to the r-plane. In
addition to the analysis of these discrete points, it is helpful to have a look on the
transformation of entire coordinate lines.
In 2.9, two further coordinate line are represented. We can see that the entire half-
plane Re(z) ≥ 0 on the right side of the imaginary axis is displayed within the unit
circle of the r plane.
17
apply the Smith Chart in order to determine the input impedance of a lossless
transmission line. In this context, we rewrite equation 2.29 for the lossless case.
l
rin = exp(− j2βl)rA = exp(− j4π ) rA (2.33)
λ
As shown by equation 2.33, a transformation of the output reflection coefficient
/ULrA along the transmission line into the input reflection coefficient rin may
be obtained by just conducting a simple rotation (clockwise) round the angle
∆ϕ = −4πl/λ. The only complication may be the calculation of the reflection
coefficient rA and the reconversion of the input reflection coefficient rin , into
the associated input impedance Z in . As pointed out in the previous chapter, the
Smith Chart allows a graphic calculation of the reflection factor. Therefore, an
impedance transformation can easily and accurately be performed by means of it.
To this end, the following method of approach should be applied:
3. Inverse transform of the obtained input reflection coefficient into the nor-
malised impedence zin and de-scaling of the impedance with the character-
istic impedance of the transmission line.
10 Ω
zA = = 0, 2
50 Ω
If we insert this value into a Smith Chart, as shown in Fig. 2.10, it will be
possible to determine the output reflection coefficient to be rA ≈ −0, 66.
18
Figure 2.10: Transformation of a impedance with the help of the Smith Chart
reflection factor rin can be gained by rotating (clockwise) the arrow of the
output reflection coefficient by 1350 .
19
If we examine the last term of the above stated equation, we have the same form
as the equation 2.32 except for the factor −1. It is evident that the Smith-Chart
for impedances also can be used for admittances. However, if we want to deter-
mine the reflection coefficient ry corresponding to the admittance y, the phasor
r′ indicating the position of the normalised admittance must be rotation by 1800
or a mirrorde at the point ”1” of the Smith-Chart. If one reads out the numeri-
cal value one notice that it is equal to normaliesed impedance z = 1y. Mirroring
at point ”1” in the Smith-Char correspond to the transition from the admittance
plane to the impedance plane. To demonstrate our approach, we consider 2.11.
Included was the admittance y = 1 + j2 displayed by the phasor r′ . The reflec-
tion coefficient ry , associated to this admittance, will be obtained by mirroring
of the phasor r′ at point ”1”. The reflection coefficient ry ≈ 0.7 exp(− j1350 ) is
shown in the diagram. Furthermore, we can read the normalised impedance value
z = 1/y ≈ 0, 2 − j0, 4 which corresponds to the admittance y.
20
For the definition of the positive defined directions see Fig. 2.12. For this purpose
IE IA
✲s s ✲
UE ZL , γ UA
❄ ❄
s s
✲z
0 l
Figure 2.12: Definition der Zählpfeile am Ein- und Ausgang einer Leitung
we start from the equations 2.24 and 2.25 and calculate the sum and difference
in order to determine the complex amplitudes of forward- and backward-traveling
waves.
U i = 12 (U A + I A ZL ) exp(+γl)
(2.34)
U r = 21 (U A − I A ZL ) exp(−γl)
The last both equations can be used in the equations 2.20 and 2.23. In this way
the voltage and current distribution along the transmission line can be determined
according to the valuse of voltage and current at the ouput of the line. For the
voltage distribution, we obtain:
1 1
U(z) = (U A + I A ZL ) exp(+γl) exp(−γz) + (U A − I A ZL ) exp(−γl) exp(γz)
2 2
If order the terms of the voltage U A and current I A and take into account the
definitions cosh(x) = [exp(x) + exp(−x)]/2 and sinh(x) = [exp(x) − exp(−x)]/2,
we obtain for the voltage distribution U(z) along the transmission line:
Similar considerations lead to the equation for the current distribution I(z):
UA
I(z) = sinh[γ(l − z)] + I A cosh[γ(l − z)] (2.36)
ZL
If we include the parameter in the equations 2.35 and 2.36, we obtain, as wished,
the input voltage U in and current I in as a function of the output voltage U A and
curent I A of the transmission line:
U E = U A cosh(γl) + I A ZL sinh(γl)
U (2.37)
I E = Z A sinh(γl) + I A cosh(γl)
L
21
In case the line shows no losses, (γ → jβ) the above equations change in:
U E = U A cos(βl) + jI A ZL sin(βl)
U (2.38)
I E = j Z A sin(βl) + I A cos(βl)
L
With the help of the equations 2.37 we are able to determine the input resistance
Z in of a transmission line in case it is terminated with the impedance Z A . Fig. 2.13
illustrates the circuit setup. Shown is a transmission line of the length l, which is
ZE
s s
✲ ZL , γ ZA
s s
✲z
0 l
U in U A cosh(γl) + I A ZC sinh(γl)
Z in = =
I in UA
sinh(γl) + I A cosh(γl)
ZC
If we divide the above term by I A cosh(γl), taking into account that the output
impedance Z A is given by the ratio of the output voltage U A and the output current
I A , we obtain for the input resistance Z in of the circuit:
ZA
+ tanh(γl)
Z
Z in = ZL L (2.39)
Z
1 + A tanh(γl)
ZL
In the lossless case the equation changes to:
ZA
+ j tan(βl)
ZL
ZE = ZL (2.40)
Z
1 + j A tan(βl
ZL
22
Figure 2.14: Input impedance of a transmissiion line in a complexe Z plane
23
150 MHz. (l = λ/2).
24
concentrated reactive components. Now we want to examine more particularly
which inductance values can be realized with a short-circuit lossless transmission
line. For the input impedance of this circuit we get:
l
Z Ek = jZC tan(βl) = jZC tan(2π ) (2.42)
λ
Equation 2.42 shows that the input impedance of the circuit for l < λ/4 behaves
like a inductive reactance. Defining fλ/4 as frequency the length of the line be-
comes a fourth of the wave length, we obtain as value of a corresponding induc-
tance:
ZC π f
Lk = tan( ) (2.43)
ω 2 fλ/4
In contrast to ideal lumped inductance, the inductance value Lk ( f ) of a short-
circuit stub becomes a function of the frequency. Its functional dependence is
illustrated in the figure 2.16. We can see that the inductance Lk shows a value
which is for low frequencies approximately frequency-independent but, however,
increases considerably starting at f / fλ/4 ≈ 0.5. If an inductance is realised by
10
Lk ( f ) ✻
Lk (0)
8
0 ✲
0 0.2 0.4 0.6 0.8 1
f / fλ/4
Figure 2.16: Normalised inductance of a short-circuit stub
25
is considered instead of the impedance. According to equation 2.39, following
applies to the input addmitance of the open-circuit stub:
l
Y Eo = jYL tan(2π ) (2.45)
λ
In principle the addmitance of the open-circuit stub is governed by the same equa-
tion already known from the short-circuit stub. By changing the interpretation of
figure 2.15, we can see that an open-circuited stub for low frequencies operates as
a capacitance. Does the frequency further increase frequency it shows a resonance
at l = λ/4. This resonance corresponds to a series resonance circuit. Above this
frequency, the behaviour becomes inductive. A stub open-circuit stub below the
λ/4-resonance may thus be used to realise a capcitance. For its value we obtain:
YL π f
Co = tan( ) (2.46)
ω 2 fλ/4
With the same arguments, a lower limit for the characteristic addmitance of the
stub can be deduced:
YL ≥ π fλ/4 Co (2.47)
RE
RG
s s
✎☞ Anpassungs-
✲
✍✌ netzwerk RA
s s
26
2.3.1 Quarter-wave transformer
An approach to understand the operation principle of a λ/4-transformator is given
by the equation 2.40. If we take the line length l in a wise that it is equal to λ/4 at
the desired frequency, we obtain for the input resistance Rin :
ZL2
Rin = für f0 = fλ/4 (2.48)
RA
As we can see, the input resistance is real at operating frequency. With the con-
verse equation 2.48 we determine the necessary characteristic impedance ZC of
the transmission able to transform the given load resistance RA into the desired
input resistance Rin . √
ZL = RA RE (2.49)
Example: λ/4-Transformator for matching RA (Fig. 2.17)
RA = 200 Ω, RG = 50 Ω und f0 =3 GHz
√
ZC = 200 50 Ω = 100 Ω
27
Figure 2.18: Frequency response of a single step λ/4-transformator
28
Chapter 3
Scattering Parameters
The linear and linearised circuits of RF- and microwave engineering can be sub-
divided in two groups, one group using voltages and currents to describe their
behaviour leading to the well known classical z-, h- or y-parameters [1] and a
second group e.g. waveguide circuits where the description is essentally based
on waves, because it is not possiple to define voltages and/or currents in a unique
way. Here usually s-parameters are used for description [2]. During the last years
the importance of the s-parameter has signifanctly grown due to the availability of
modern vector network anaylyser (VNA).
While below approximately 10 MHz passiv or linearised active two ports are still
described using the classical parameters with increasing frequency this becomes
more and more difficult. We will discuss briefly the situation for y-parameters,
which where traditionally used in the RF-domain.
• Active elementes e.g. transistors tend to become unstable if they are termi-
nated in short-circuits.
To avoid all this problems the characterisation of circuits must be based on pa-
rameters which are easy to measure. At the dawn of microwave engineering it
was noticed that it is more easy to measure powers associated with certain waves.
With the help of directional couplers it is possible to separate forward and back-
ward travelling waves. This was the starting point to develope a measurement
technique based on wave amplitudes and scattering parameters.
29
3.1 Definition of wave amplitudes
As already dicussed in the previous section the concept of waves is of essential
importance for the definition of s-parameter and the assiocated measurement tech-
nique. Instead of using the voltages U 1 , U 2 or the currents I 1 , I 2 defined at the
input or output one introduces wave amplitudes a1 , a2 und b1 , b2 at the input and
output port. With the help of the transmission line shown in Fig. 3.1 their defini-
tion will be explained. In the following we will describe the transmission line in
❡ ✁
a1 ✲ ✲b
2
Uf ✲ ✛ Ur
❄ b1 ✛ ✛ a2 ❄
❡ ✁
✲
z1 z2 z
a1✲ b2✲
✛ (S) ✛
b1 a2
30
waves,
|U f |2 |U r |2
Pf = Pr =
2ZC 2ZC
we will get for the wave amplitudes:
U f (z1 ) U (z1 )
a1 = √ b1 = √r (3.3)
ZL ZL
Using the same argumentation we get for the wave amplitudes a2 und b2 :
U r (z2 ) U f (z2 )
a2 = p b2 = p (3.4)
ZC ZC
b1 = S11 a1 + S12 a2
(3.5)
b2 = S21 a1 + S22 a2
Using the defining equation one already can deduce the meaning of the single
parameter.
b1 Input reflection coefficient of port 1, while port
S11 = |a =0
a1 2 2 is matched
b2 Input reflection coefficient of port 2, while port
S22 = |a =0
a2 1 1 is matched
b2 Transmission factor from port 1 to port 2, while
S21 = |a =0
a1 2 port 2 is matched
b1 Transmission factor from port 2 to port 1, while
S12 = |a =0
a2 1 port 1 is matched
If we now represent the emerging wave amplitudes b1 , b2 of the transmission line
section (Fig. 3.1) as a function of the incident wave amplitudes a1 , a2 we get:
U f (z2 ) U f (z1 )
b2 = √ = exp[− jβ(z2 − z1 )] √ = exp(− jβl)a1
ZL ZL
31
U (z1 ) U (z2 )
b1 = √r = exp[+ jβ(z1 − z2 )] √r = exp(− jβl)a2
ZL ZL
As result a lossless transmission line section of length l is described by the fol-
lowing S-matrix.
0 exp(− jβl)
(S) =
exp(− jβl) 0
3.2.2 N Port
It is easy to extend the equations 3.5 of a two port to an arrangement with an
arbritray number N of ports:
b1 S11 S12 · · ·S1N a1
b
2
S
21 S22 · · ·S2N a2
.. = .. (3.6)
.. . ..
. · · · ..
. . .
bN SN1 SN2 · · ·SNN aN
Like in the case of a 2-port the meaning of the single elements can be deduced
from the defining equations. The element Snn represents the input reflection co-
efficient of port n, while all other ports must be matched. Analogous the element
Snm descibes the transimission factor from port m to port n, while again all other
ports must be matched. To reduce the effort for writing the last equation can be
put into matrix notation. If we define a complex vector of the emerging wave am-
plitudes ~b by (b1 , b2 · · · bN ) and a complex vector of the incident wave amplitudes
~a by (a1 , a2 · · · aN ) and the scattering matrix is denoted by (S), we can write:
~b = (S)~a (3.7)
32
• A N-port is said to be lossless, if the power sum of all incident waves an is
equal to the power sum of all emerging waves bn .
1-port
An arbitrary 1-port is descriped by the following equation:
b1 = S11 a1
According to equation 3.1 and 3.2 the absolute values of the wave amplitudes a1
and b1 are directly related to the incident power Pin and the emitted power Pout of
a 1-port.
1 1
Pin = |a1 |2 Pout = |b1 |2
2 2
If the 1-port is lossless the following equation must hold true:
To be a lossless 1-port the reflection coefficient S11 must show an absolute value
of 1, while the phase may have an arbitray value.
2-port
To go on with our discussion, we now want to examine which conditions are
imposed onto the s-parameters of a 2-port if we know it to be lossless. For the
sake of simplicity we will first discuss the case where only the wave described by
a1 will enter the port 1, while a2 is assumed to be zero. According to equation
3.5 there will still exist the wave amplitudes b1 and b2 which will be emitted from
33
port 1 and port 2. As a consequence we get for the power Pin entering the 2-port
and the power Pout emitted by the 2-port:
1 1 2
|a |2 |b1 | + |b2 |2
Pin = Pout =
2 1 2
If we replace the wave amplitude b1 by S11 a1 and b2 by S21 a1 we find in the case
of a lossless 2-port:
Equation 3.8 is termed the first absolute value condition of a lossless 2-port. If we
now assume only a wave is entering port 2 (a2 6= 0, a1 ≡ 0) we can conclude:
Equation 3.9 is termed the second absolute value condition of a lossless 2-port. Of
course in general waves will enter port 1 and port 2 simultaneously. But to reduce
the effort of writting we will now use matrix notation according to equation 3.7.
If one defines~a† to be the row vector (a∗1 , a∗2 ) in mathematical terms it will be the
adjoint vector to the column vector ~a. With the introduced vectors one can write
the power entering the 2-port:
1 2 1
|a1 | + |a2 |2 = ~a†~a
Pin =
2 2
Analoguos one gets for the power emitted from the 2-port Pout :
1 2 1 †
|b1 | + |b2 |2 = ~b ~b
Pout =
2 2
The condition for the 2-port to be lossless can now be reformulated :
Bearing in mind the mathematical identity [(S)~a]† = ~a† (S)† and ~a†~a = ~a† (E)~a
with (E) being the unity-matrix. The above equation can be rewritten by:
h i
† †
~a (S) (S) − (E) ~a = 0
34
A matrix fulfilling the condition stated in equation 3.10 is called a unitary matrix.
In detail equation 3.10 reads in case of a 2-port:
∗
S11 S∗21
S11 S12 1 0
=
S∗12 S∗22 S21 S22 0 1
In addition to the already known absolute value conditions 3.8 and 3.9 there exist
two further equations:
S11 S∗12 + S21 S∗22 = 0 und S∗11 S12 + S∗21 S22 = 0 (3.11)
Since the second equation is just the conjugate complex of the first equation, it
is sufficient only to take the firts equation in two account, which imposes further
conditions on the s-paramter. This condition is called phase condition. In the
case of a N-Port to be lossless the N × N s-parameter must fulfill N(N + 1)/2
conditions.
✛ l≈0 ✲
s ✄✄✄✄✄ s
L
ZL = 50Ω ZL = 50Ω
s s
matched.
( jωL + ZL ) − ZL ωL
S11 = ≈ j = j0, 126
( jωL + ZL ) + ZL 2ZL
Due to the setup there is essentially no difference between port 1 and port 2, thus
we have S22 = S11 = j0, 126. To calculate the parameter S21 one can use the fact
that an ideal inductance is lossless and hence the first absolute value condition can
be applied. q
|S21 | = 1 − |S11 |2 = 0, 992
If one chooses the parameter S21 to be real also the phase condition according to
equation 3.11 is fulfilled. A series inductance in a 50 Ω transmission line at a
frequency of a 100 MHz thus can be described by the following s-matrix:
j0, 126 0, 992
(S) =
0, 992 j0, 126
35
3.4 Signal Flow Graphs
3.4.1 Definition of graphical elements
It is allways of advantage if the abstract mathematical operations defined by equa-
tion 3.5 can be represented by graphical elements in a more descriptive way. This
is done with the help of the signal flow graph. In a signal flow graph wave am-
plitudes like an or bm are represented using dots, while a scattering parameter like
Smn will be represented by an arrow starting at the dot with index n and ending
at a dot with the index m. Fig. 3.3 gives some simple graphical representations
and the corresponding equations. Fig. 3.3a shows for example the graphical rep-
al ❍
s
❍❍Sil
Smn
✲s b ak s Sik❍❍❥✲
❍s
an s ✯
✟ bi
✟✟
m
✟✟ S i j
a) bm = Smn an b) aj ✟
s
bi = Si j a j + Sik ak + Sil al
S′mn ❅❅
Skk
an bk bm ❅ an s ❅ b
s ✲s ✲s an s ❅s bm
′′ ❘ ❅s
❘
✲ ✲s m
Skn Smk ❅ ❅
S mn ✒ Skn bk Smk
❍
❍✟✟ ❍
❍✟✟
❍
❍✟✟
ans b
✲sm
an s b an s
✲s m
b
✲s m
′ ′′ Smk Skn
Smk Skn Smn + Smn
a) b) c) 1 − Skk
deduce some simple aproaches to reduce the complexity of a given signal flow
graph. Fig. 3.4a for example shows how the wave amplitude bk can be omitted
by a resulting scattering parameter Smn = Smk Skn . In Fig. 3.4b a signal flow graph
36
is shown representing a wave amplitude bm which is composed via two different
branches from the wave amplitude an . It is obvious that the effektive transmission
parameter Smn is given by the sum of S′mn und S′′mn . A somehow more complicated
flow graph is shown in Fig. 3.4c. Here for the first time a loop occurs. To sim-
plify the structur one can splitt node bk in a further node b′k with a transmission
coefficient of 1 between them. For the node bk we get:
Skn
bk = Skn an + Skk b′k = Skn an + Skk bk → bk = a
1 − Skk n
Taking into account the transmission factor Smk between the nodes bk and bm , we
endup in the result depicted in Fig.3.4c. With the help of the shown examples
all signal flow graphs discussed in this lectures can be understood. In flow graphs
with a lot of nodes the operations will become laborious and it will be of advantage
to use the rules given by Mason [?].
Figure 3.5: Block diagram and signal flow graphs of differnt devices
To analyse circuits of RF- and Microwave engineering with the help of the s-
parameters and their signal flow graph Fig. 3.5 shows the block diagram of some
37
special devices and their signal flow graph. We want to discuss the properties of
these devices using their representation in a signal flow graph without going into
the details of their physical realisaztion.
a) 1-port
Arbitrary 1-port, described by its frequency dependent input reflection co-
efficient S11 .
b) 2-port
Arbitrary 2-port, described by its frequency dependent s-parameters S11 ,
S12 , S21 und S22 .
c) Transmission line
Lossless transmission line of length l. It is free of self-reflections S11 =
S22 = 0. The incide waves suffer a phase change of ϕ = −βl. The kind of
line e.g. circular wave guide, rectangular wave guide, coaxial transmission
line is denoted by a small additional symbol.
d) One-way attenuator
In a one-way attenuator a wave only can propagate in the declared direction,
while the reflected wave is attenuated and finally absorbt in the device.
e) Phase-shifter
With the help of a phase-shifter the phase of a wave is changed. In most
cases it is realised by a transmission line of variable length.
f) Directional coupler
Directional coupler are realised using two coupled transmission lines. It is
free of self-reflections and the coupling is choosen to decouple two ports.
With the help of a directional coupler it is possible to determine the wave
travelleing in different directions. The ideal directional coupler is described
by its transmission-factor and coupling coefficient.
g) Absorber
In the case of a transmission line a resistor with a value equal to the char-
acteristic impedance of the transmission line is an ideal absorber. The wave
entering the absorber will be totaly absorbed, so its input reflection coeffi-
cient is zero.
h) Power meter
The ideal power meter shows no self-reflections and is able to determine the
power of the incident wave.
38
i) Generator
It provides the neccessary power to drive a microwave circuit. The ideal
generator will be free of self-reflections.
As an example to apply the signal flow graph we will have a look on the setup
shown ing Fig. 3.6. It shows a generator which feds its wave using a one-way
attenuator and a directional coupler to a 1-port, which is described by its reflection
coefficient rA . The ports 3 and 4 of the directional coupler are terminated by
ideal power meter, which are able to measure the absolute values of the emerging
wave amplitudes b3 and b4 . Also shown in Fig. 3.6 is the associated signal flow
graph. To enhance the clearness of the flow graph only nodes and branches are
included which are of importance for the measurement setup. To understand the
operation principle of the setup we will calculate the wave amplitudes b4 and b3
in dependdence of the wave a0 generated by the rf generator.
✎☞ ✎☞
✒
✍✌ ✒
✍✌
❄ ❄
✗✔
✲ ✝ ✆
G ❅
■
✒
✠
❅
❘ rA
✖✕
s b4 ✘✘✿✘s b3
S ❨
❍❍✘ ✘
a0s S as1 ✘✘31 ✘ ❍❍ S21 ✲sb2
✲10
s ✲✘ ❍❍
S31❍❍ rA
❍❍s❄a
2
b3 = S31 S10 a0
39
The result is essentially proportional to the reflection coefficient rA . Unfortunately
the measured result is distorted by the unknown transmission factor S21 of the
used directional coupler, which introduces a systematic error. To eleminate this
systematic error in a first stepp a unknown reflection coefficient e.g. a short circuit
can be measured which allows for the determination of the absolute value of S21
(calibration of the measurement setup ) and its value can be used to correct the
measured reflection coefficient rM . In the next section we will discuss this in
more deatil again, but in advance we will discuss a further systematic error which
will occur if a real directional coupler is used in the measurement setup. Fig. 3.7
b
s❍4
❨ S31 ✘✘✘✿
✘s b3
✒ ❍ ✘
S41 ❍✘❍✘
a0 S10 ✘✘✘ ❍❍
s ✲s ✲s✘✘ ❍❍
S21 ✲s b
2
❍❍ r
S31 ❍❍❄
A
s a2
shows essentially the same signal flow graph already given in Fig. 3.6. The only
difference is the inclusion of a further branch termed S41 . It accounts for the finite
isolation between port 4 and port 1 of a real directional coupler. Using the flow
graph of Fig. 3.7 one again can calculate the ratio of the wave amplitudes b4 /b3 .
b4 S
= rM = S21 rA + 41
b3 S31
The finite isolation between port 4 and port 1 results in an additional term. Its in-
fluence becomes dominant for small reflection coefficients to be measured |rA | ≈ 0
and suggested a measured reflection coefficient of the object even its value itself
is zero.
S
|rM | ≈ | 41 |
S31
The abolute value of the ratio S41 /S31 of a directional coupler will thus set a lower
limit to the reflection coefficient which can be measured without calibration. This
ratio is thus a kind of quality factor of a real directional coupler. One denotes the
value aD = −20 log(|S41 /S31 |) as the directivity of a directional coupler.
40
today. One can distinguish between scalar (SNA) or vector network analyzers
(VNA). While a SNA only able to measure the absolute values of the s-parameters,
with a VNA the complex values of the s-parameter (absolute value, phase) can
be measured. Due to the build in prozessors also the elmination of systematics
errors is possible and post processing of the measured data is possible. So today a
VNA is a indispensiable equipmet in an rf- or microwave lab. Fig. 3.8 shows the
principle setup of a modern NVA. It consists of the following building blocks:
Internal Computer
PP ✏✏
P✏
Signal Source
✂
Forward s❄ Backward ✂
✂
s s ❇
❇
❇
a0 ❄
❥
✲× ❇
❇
b0 ❥ A/D
✲× ❇
a3 ❥
✲× ✂
✂
b3 ❥
✲× ✂
Test-Set
❇ ❇
❇ ❇ Receiver
Port 1 Port 2
• Signal Source
The rf-signal for operation is normally provide by a synthesizers. Due to the
build in phase look loops rf-signal with low phase noise and high frequency
accuracy can be generated. In order not to lose the phase information during
the mixing process the synthesiser will generate a second signal with a fixed
phase relation to the rf-signal for measurement.
• Test-Set
The rf-heart of the VNA is the Test-Set. The rf-signal provided from the
synthesizer can be switched between a forward (S11 , S21 ) and backward
(S22 , S12 ) branch. To provide the reference signal for the forward branch
a0 or the backward branch a3 often a lossy power divider is used. Direc-
tional couplers are used to do the seperation of the forward and backward
travelling waves.
41
• Receiver
The conditioning of the messured signals takes place in the receiver. The
rf-signals are down converted to a intermediated frequency, while the phase
relation of the signal is kept constant. After converting to digital form they
are fed to the internal computer.
• Internal Computer
The controlling of the whole measurement setup is done by the internal
computer. It coordinates the collaboration of the different building blocks,
carries out all error correction and processes the data, so it can be displayed
in various formates.
Today VNAs with coaxial connectors are build covering the frequency range from
10 MHz to approximately 60 GHz. Beyond this frequency range Test-Sets are
usually realised in wave guide technique.
1-port calibration
As already dicussed a real directional coupler shows properties which lead to sys-
tematics errors when measuring e.g. the reflection coefficient of a device under
test (DUT). So for example limits the finite directivity of the coupler the lowest
measureable reflection coefficient. Due to the available data processing capablity
of modern computers it is now possible to setup complicated error models, which
describe the systematic erros and which are used to eliminate them. The concept
a0 b0 a0 b0
✎☞❍❍ ✟
✟ ✎☞❍❍ ✟
✟ error
G rA G rA
✍✌ real ✍✌ 2-port
ideal
a) b)
a0 s e10 ✲s
❅
❅ ✒❅
❅
e00 e11 rA
c) b0 s✠ e
✛ 01 ❅s✠
❅
of a 1-port error model is depicted in Fig. 3.9. In Fig. 3.9 a) the principle setup
of a real reflectometer is shown, which will result in systematic errors of the mea-
surement. As shown in Fig. 3.9 b) one can thing of the systematic errors to be due
to an error 2-port, which is placed between the DUT and the now assumed ideal
42
reflectometer. Fig. 3.9 c) shows the associated signal flow graph. With the help of
a calibration procedure the error parameters of the error 2-port can be determined
and a correction of the measured data can be carried out. The error 2-port will
result in a difference between the real reflection coefficient rA of the DUT and the
measured reflection coefficient rM . Carring out the calculation shows that three
error terms will have an influence e00 , e10 e01 and e11 as shown in equation 3.13.
e10 e01
rM = e00 + r (3.13)
1 − e11 rA A
In the following section we will briefly discuss the influence of the three parame-
ters:
• e00
According to equation 3.13 the value of the parameter e00 becomes domi-
nant when the reflection coefficient of the DUT is small. In the limit rA → 0
we have rM = e00 . Thus it accounts for the finite directivity of the direc-
tional couplers used in the measurement setup.
• e10 e01
The influence of the above product becomes most significant in the limite00 =e11 =
0. In this case we get rM = e10 e01 rA . Comparing this result with equation
3.12 shows that the product e10 e01 accounts for the transmission factors of
the used directional couplers.
• e11
This parameter accounts for the finite input reflection coefficient of the mea-
surement setup.
In the ideal case the parameters of the error 2-port (equation 3.13) should show
the following values: e00 ≡ 0, e10 e01 ≡ 1 and e11 ≡ 0. In this limite the measured
reflection coefficient rM would be identical to the reflection coefficient rA of the
DUT.
43
• ”Load” - Measurement rML
To carry out this measurement the port of the network analyzer has to be
terminated with its characteristic impedance, which results in the ideal case
in a reflection coefficient rA = 0. This is one of the drawbacks of the OSL-
calibration, because with increasing frequency it becomes more and more
complicate really to match the port. As result of an ideal load we will get as
measured reflection coefficient:
✲
✲ C
✲ E
its relection coefficient will have the following form rA = exp(− jδC ). The
frequency dependent phase change δC due to the capacitance must be known
very accurate to use it as calibration standart. Using equation 3.14 we will
get:
e10 e01
rMO = rML + exp(− jδC ) (3.15)
1 − e11 exp(− jδC )
• ”Short” - Measurement rMS
The realisation of a ”Short” calibration standart is less challenging, at least
44
at low frequencies. Again using equation 3.13 we will get for the measured
reflection coefficient:
e e
rMS = rML − 10 01 (3.16)
1 + e11
The equations 3.15 and 3.16 can be used to determine the still unkown error pa-
rameters e11 and e10 e01 :
45
Chapter 4
Microwave Amplifier
The small signal behaviour of active elements like bipolar junction transistor
(BJT) or fieldeffekt transistors (FET) can be described using different physically
based equivalent circuits. But to allow a unified treatment it is of advantage two
use 2-port parameters like y- or s-parameters. Based on the used parameters
equations for some characteristic properties of amplifier e.g. input- and output
impedance, voltage gain or power gain can be deduced without reference to the
physics of the device. Beside the s-parameters also the y-parameters can be used
to deduce some properties of the amplifiers. Here again we want to point out, that
the small signal parameters not only depent on the considered frequency but also
depent on the chosen point of operation. Unfortunatley this is not directly obivi-
ous at the deduced equations.
Even so today amplifiers working upto the GHz-range are available as monolithic
integrated circuits we will focus on simple amplifiers with only a single transis-
tor to explain the basics of amplifier design. Further on we will not discuss how
to establish the neccassary poit of operation, because this is already done in the
lecture of electronics. As an example Fig. 4.1 shows the s-paramters of the BJT
BFR 280 in point of operation with UCE = 5V and IC = 3mA according to a data
set provided by the manufacturer.
4.1 Y-Parameter
Beside the s-parameters specialy in the rf-domain, y-parameter where used to de-
cribe the linearised behaviour of active elements. Equation 4.1 gives the currents
of a 2-port independence of the voltage, when y-parameter are used for its de-
scription.
I 1 = y11U 1 + y12U 2
(4.1)
I 2 = y21U 1 + y22U 2
46
Figure 4.1: S-parameters of the BJT BFR 280
The meaning of the single parameter can be explained by using equation 4.1 under
certain conditions:
I
y11 = U1 |U 2 =0 short-circuit input admittance
1
I
y12 = U1 |U 1 =0 short-circuit reverse transfer admittance
2 (4.2)
I2
y21 = U |U 2 =0 short-circuit forward tranfer admittance
1
I2
y22 = U |U 1 =0 short-circuit output admittance
2
Looking at equation 4.2 it becomes obvious that all currents have to be measured
using an rf-short-circuit. In the lower rf-domain this can be realised using a large
capacitors. But with increasing frequency it becomes more and more complicated.
A further drawback is the tendency of active devices to become unstable when
operating under short-ciruit conditions. These are some reason why today active
devices are in most cases described by s-parameters. Nevertheless y-parameter
are sometimes still of interest because certain situations are easierer to describe.
Fig. 4.2 shows an equivalent circuit just based on the defining equations 4.1. In
principle it is just their graphical representation. A further equivalent circuit is
shown in Fig. 4.3 which is called π-equivalent circuit and which is used to de-
scribe the linearised behaviour of BJTs. Essentially it is equivalent to the physical
motivated equivalent circuit given by Giacoletto [].
47
y12U 2 y21U 1 I2
I1
s✲ s s ✛ s
❄ ❄
✗✔ ✗✔
U1 y11 y22 U2
✖✕ ✖✕
❄ ❄
s s s s s s
−y12 (y − y )U I2
I
s✲
1 s s
21
s12 1
✛ s
❄
✓✏ y
U1 y11 + y12 22 U2
✒✑ +y12
❄ ❄
s s s s s
I0
✲ s s
✗✔
Y G U1 2-port U2 YL
✖✕
❄ ❄
s s
48
Voltage gain
The voltage gain vu ( jω) is defined as the ratio of the output voltage U 2 to the input
voltage U 1 . Using the second equation of 4.1 and the fact that the load admittance
is given by Y L = −I 2 /U 2 we get for the voltage gain of the circuit in Fig. 4.4:
U2 y21
vu ( jω) = =− (4.3)
U1 y22 +Y B
Input admittance
The influence of the 2-port onto the rf-generator is always of interest. It is de-
scribed by the input admittance yin . Starting from the first equation of 4.1 and
using equation 4.3 we find:
I1 y y
yin = = y11 + y12 vu ( jω) = y11 − 12 21 (4.4)
U1 y22 +Y L
Output admittance
The output admittance yout is given by the ratio of current I 2 to voltage U 2 . Es-
sentialy it can be deduce in the same manner as the input admittance. One just has
replace index 1 by 2 and vice versa and the load admittance has to be replaced by
the generator admittance Y G .
I2 y y
yout ( jω) = = y22 − 21 12 (4.5)
U2 y11 +Y G
49
1’ s s
YG ❄Y 12U 2
✎☞
1s U ′1 Y 11 ✍✌ s s
❄Y U
❄ ✎☞ 21 1
U1 s s s U2
✍✌ Y 11
❄ ❄ YB
s s s s s
Due to the reverse transfer admittance y12 a voltage U ′1 will be excited. For the
ratio U ′1 /U 2 we get:
U ′1 y12
= −
U2 y11 +Y G
The voltage U 1 at the point 1 will result in the voltage U ′1 at point 1’. If both
voltages are equal in value and phase the points 1 and 1’ can be connected and as
a result we will have a selfsustainable oscillation.
y12 y12 y21
U ′1 = − U2 = U1
y11 +Y G (y11 +Y G )(y22 +Y L )
The equation above can be rearranged into two different equivalent forms:
y y21
Y G + y11 − y 12+Y =0
22 L (4.7)
YG + yin =0
y y12
Y B + y22 − y 21+Y =0
11 G (4.8)
YB + yout =0
The new forms also can be used to formulate a condition for stability:
50
One says a 2-port is uncondional stable, if for an arbitrary
generator admittance Y G ( load admittance Y L ) the condi-
tion gin = Re(yin ) > 0 (gout = Re(yout ) > 0) is fulfilled.
A necessary condition for uncondionally stability is g11 > 0
and g22 > 0.
Only a few devices will be uncondionally stable. So it makes sense to discuss for
a given set of y-parameters which load admittance Y L will result in a stable and
which one will result in an unstable circuit. To deduce a expression we want to
assume the necessary conditions g11 > 0 and g22 > 0 to be fulfilled. As a starting
point for the derivation we will use the equation for the real part of the input
admittance yin :
y y
gin = g11 − Re( 12 21 ) > 0
y22 +Y B
Nach einigen Umformungen erhält man:
The denominator of the equation above shows a positive real value. Hence the
product with gin remains positive and can be divided by the posistive parameter
g11 without changing its sign.
51
a circle equation enables to define a admittance to the center of the circle in the
load plane Y CL and the radius of the circle ρYL .
Re(y12 y21 )
Re(Y CL ) = −g22 + 2g11
Im(y y )
Im(Y CL ) = −b22 + 12 21
2g11
(4.11)
|y12 y21 |
ρYL = 2g
11
Up to now the discussion concentrated on the location and size of the stability
circles in the load plane. Using the same argumentation one can also deduce
stability circles in the generator plane. On a more formal base on can get the
equations by interchanging port 1 and port 2 while the load admittance is replaced
by the generator admittance.
Re(y12 y21 )
Re(Y CG ) = −g11 + 2g22
Im(y y )
Im(Y CG ) = −b11 + 12 21
2g22
(4.12)
|y y |
ρYG = 2g12 21
22
52
Figure 4.6: Stability circle of BFR 280 at f = 300 MHz
Since admittances are not in favour of impedances the stability circles of the load-
or generator admittance can also be mapped onto the corresponding impedance
planes. In a mathematical sense this mapping is an inversion and it is well known
[?] that by an inversion circles are again mapped onto circles. So also in the plane
of the generator and load impedance stability circles will exist.
53
Fig. 4.7 shows the stability factor K of the BJT BFR 280 in the frequency range
from 100 MHz to 3 GHz. According to the chart shown in Fig. 4.7 the BJT
BFR 280 becomes stable for frequencies higher than 1.9 GHz. At lower frequen-
cies a circuit using the BJT may become unstable depending on the load admit-
tance Y L used.
54
Figure 4.8: Forward transmission s21 and stability factor K of the BJT BFR 280
in the Fig. 4.8 shown is the stability factor K. At 300 MHz it shows its lowest
value while over the whole freuency range it is greater than one and one can use
the transistor to build up a stable amplifier.
In the previous example an admittance in the load plane was used to stabilise the
transistor. If we start with stability circles in the generator plane we can determine
an addional admittance to be put parallel to the input to reach stability. But also
the stability circles in the impedance can be used as the starting point of the con-
siderations, resulting in series resistors to be put at the output or the input of the
BJT. Thus in general there exist at least four different methods to reach stability,
which can be chosen on the application area of the amplifier to build.
55
ratio between the two powers is called transcuder power gain GT of the amplifier.
PL
GT = (4.15)
Pav
In the following we want to assume that the internal resistance of the generator
Z G as well as the load resistance Z L are equal to the characteristic impedance ZC
of the transmission line system used to define the s-parameters. In this case in
a signal flow graph describing the wave between the generator and transistor as
well as between the transistor and load the reflection coefficient rG and rL have to
be considered. To calculate the available power of the source Pav we can use the
signal flow graph shown in Fig. 4.9. The generator will excite the wave amplitude
1 a
a0 s ✲s1 ✲s b 2
✒ ❅❅
1
rG r∗G
❅
❅✛s 1 ✠
s a
2
Figure 4.9: Signal flow graph to calculate the available power Pav
1 |a0 |2
Pav = (4.17)
2 1 − |rG |2
The power PL entering the load can also be expressed by the power difference of
the waves bL and aL , as shown in Fig. 4.10.
1 1
|bL |2 − |aL |2 = (1 − |rL |2 )|bL |2
PL =
2 2
56
This results in the following general equation to calculate the transducer gain GT
of an amplifier:
b
GT = (1 − |rG |2 ) · | L |2 · (1 − |rL |2 ) (4.18)
a0
a0 s 1 ✲s 1 ✲as1 s21 b✲
2 s 1 ✲s b
L
✒ ❅
❅ ✒ ❅
❅
rG s11 s22 rL
❅
❅✛s ✠
s ❅
❅✛s 1 ✠
s
b1 a2 aL
s21
bL = s21 a1 + s22 rL bL → bL = a
1 − s22 rL 1
Also the wave amplitude a1 can be expressed by the wave amplitude a0 , which
gives us the final result for bL :
1 1
bL = · s21 · a (4.19)
1 − s11 rG 1 − s22 rB 0
Remebering equation 4.18 we finally get in the case of a unilateral transistor:
1 − |rG |2 2 1 − |rB |2
GT = · |s21 | · (4.20)
|1 − s11 rG |2 |1 − s22 rB |2
In the unilateral case equation 4.20 can be split up into the product of there inde-
pendent terms:
the generator gain GG ,
1 − |rG |2
GG = (4.21)
|1 − s11 rG |2
the gain due to the active device,
G0 = |s21 |2 (4.22)
57
and the load gain.
1 − |rL |2
GL = (4.23)
|1 − s22 rL |2
We will get the maximum transducer gain if the input as well as the output of
the active device is matched. For a unilateral transistor it in principle is easy to
realise since the input reflection coefficient s11 does not depend on the load and
the output reflection coefficient s22 does not depend on the internal resistance of
the generator.
rGm = s∗11 rBm = s∗22
In Fig. 4.11 the principle setup of a single stage rf-amplifier is shown. To each
term of equation 4.20 a 2-port can be assigned to. With the help of the input match-
Z G = ZC Z = ZC
s s s s L
✎☞ Input Ouput
✛ active device ✲
✍✌ matching matching
s s s s
s∗11 s∗22
ing network the generator impedance is transformed into s∗11 to match the transis-
tor input, while the output matching network will transform the load impedance
into s∗22 to match the transistor output. This results in a equation for the so called
unilateral transducer gain GTu , which only depence on the properties of the active
device.
1 1
GTu = 2
· |s21 |2 · (4.24)
1 − |s11 | 1 − |s22 |2
a0 + s12 rL rG rL bL
a1 = a0 + rG (s11 a1 + s12 rG bL ) → a1 =
1 − rg s11
58
a0 s 1 ✲s 1 ✲as1 s21 b✲
2 s 1 ✲s b
L
✒ ❅
❅ ✒ ❅
❅
rG s11 s22 rL
❅
❅s✛ ✠
✛
s s12 ❅
❅✛s 1 ✠
s
b1 a2 aL
As a result we get:
s21
bL = a (4.25)
(1 − s22 rL )(1 − s11 rG ) − s21 s12 rG rL 0
Using again equation 4.18 we get for the transducer gain in case of internal feed-
back:
|s21 |2
GT = (1 − |rG |2 ) (1 − |rL |2 ) (4.26)
|(1 − s22 rL )(1 − s11 rG ) − s21 s12 rG rL |2
Into the last equations the following short cut was introduced ∆s = s11 s22 −s12 s21 .
Using the above equations one in principle is able to find the reflection coefficients
rGm and rLm . This leads for example to a quadratic equation for rLm , which only
gives values |rLm | < 1 if the active device is stable. For illustration Fig. 4.13 shows
rGm for the transistor BFR 280 which was stabilised using a parallel resitor of
121Ω at its output. Also shown is the complex conjugated value of s11 . One notice
a strong deviation between the two curves for frequncy values of low satbility
factor K (compare Fig. 4.8)
59
Figure 4.13: rGm und s∗11 des Transistors BFR 280
The above equation can be simplified by introducing the stability factor K []:
|s21 | p
GMAG = (K − K 2 − 1 ) (4.28)
|s12 |
It is obvious that equation 4.28 will only result in meaningfull results for GMAG
as long as the active device is stable (K > 1). On the other hand we learned in
the previous sections how to stabilize an active device using parallel or series
resistors to reach stability. Using equation 4.28 for the case K = 1 thus can give
us a crude approximation of the achievable power gain after stabilisation. It is
termed maximum stable gain GMSG :
|s21 |
GMSG = (4.29)
|s12 |
60
Figure 4.14: GMSG , GMAG des Transistors JS8818A
Fig. 4.14 shows the values of GMSG and GMAG for the MesFet JS8818A in the
frequency range from 1 to 50 GHz. For lower frequencies the transistor is unstable
(K < 1) and only the maximum stable gain GMSG can be calculated. It drops from
approximately 23 dB at 1 GHz to approximately 12 dB at 13 GHz. Above this
frequency the transistor becomes stable and the maximum available gain GMAG
can be calculated. Above about 40 GHz the transistor no longer can be used to
realise an amplification.
61
Single ended amplifier
Its principal setup is shown in Fig. 4.11. It consist of three different parts, the
input matching network, the activ device and the output matching network. As
already discussed in the previous section input and output matching network are
use to maximize the generator gain GG and the load gain GL as defined in equation
4.20. To design the matching networks we can use the technique discussed in the
section of transmission line theory and its application with the help of the smith-
chart. To demostrate the prorcedure we will start with the stabilized transistor
BFR 280 discussed in section 4.2.2 and design the matching networks to operate
at a frequency of 500 MHz. At this frequency the transistor is described by the
following set of s-parameters:
With the help of a simulation tool it is posible to calculate the optimum laod re-
Figure 4.15: Constant gain contours in the generator plane. stabilized transistor
BFR 280, f = 500 MHz, GA = 18 dB, ∆G = 1 dB
62
4.15 shows contours of constant gain in the generator plane for a frequency of
500 MHz. Also the values for the optimum generator and load reflection coeffi-
cient are displayed.
With the help of the Smith-chart its possible to design the input and output match-
ing networks. One possible realisation example is shown in Fig. 4.16.
Fig. 4.17 shows the computed absolute values of the parameters s21 , s11 and
19nH
s s ✞☎
✞☎✞☎ s
BFR280
50Ω 40nH
s ✞☎✞☎
✞☎ s 121Ω ☎
☎
✆
☎ ❅
❘
❅ ☎106nH
✆ 50Ω
✎☞ ☎
✆ ✆
✍✌ 177nH ✆☎
✆
Figure 4.16: Single ended amplifier using the transistor BFR 280
Figure 4.17: s21 -, s11 - und s22 of single ended amplifier with BFR 280
s22 using the circuit according to Fig. 4.16. At the frequency of operation the
amplifier will show an gain of approximately 17 dB. The frequency dependence
63
of the input and output matching network is clearly displayed in the behaviour of
the parmeters s11 and s22 , which show a small bandwidth characteristic.
64
Chapter 5
Plane Wave
To find a simple solution for Maxwell’s equation we use the differential form of
Faradays and Amperes law []. The electromagnetic field will be represented using
the electric field strength ~E and the magnetic field strength H. ~ All other fields
will be duduced from this basic fields. In general the phasor of the electric field
will be descriped by three independent functions E x , E y , E z which are themselves
functions of the three space coordinates x, y, z. To find a simple solution for
Maxwells equation we assume the existens of only the x-component of the electric
field E x which only depents on the z-coordinate.
~E(z) = E x (z)~ex (5.1)
The electrical properties of space will be described by its conductivity κ and its
permitivity ε = εr ε0 , which are assumed to be constant in the considered space.
In this case we will get the following equations for the conduction current denisty
J~C and the electric displacement density field ~D:
J~C = κE x (z)~ex ~D = εE x (z)~ex (5.2)
The magnetic properties of space are described by the scalar permeability µ = µr µ0
which is also assumed to be constant in the considered space. Hence the following
equation gives the relation between the magnetic flux density ~B and the magnetic
~
field H:
~B = µH ~
65
According to the equation above the electric field with only one x-component with
a z-dependents in space will result in a magnetic field having only a y-component:
~ 1 ∂E x (z)
H(z) = − ~ey (5.3)
jωµ ∂z
After realising that the considered electric field results in a magnetic field with
y-component and z-dependence, we now use the magnetic field and Amperes law
to calculate the associated electric field.
∂
J~L + jω~D = (κ + jωε)~E = (~ez ) × (H y (z)~ey )
∂z
It makes sense to commbine the conduction current density J~C = κ~E and the dis-
placement current density J~V = jωε~E by the introduction of a complexed valued
permitivity εr :
κ
κ + jωε = jωε0 (εr − j ) = jωε0 εr (1 − j tan(δε ) = jωε
ωε0
Comparing the coefficients we find for the complex valued permitivity ε:
κ
ε = ε− j (5.4)
ω
This results in the follwoing equation for the electric field:
∂H y (z)
~E(z) = − 1 ~ex (5.5)
jωε ∂z
According to equation 5.5 the y-orientated magnetic field with z-dependence will
result in a electric field with x-component and also a z-dependence. One observes
that both field excite each other. As a result equation 5.3 can be combined with
equation 5.5 and one ends up in an ordinary differential equation for the electric
field:
1 d 2 E x (z)
E x (z) =
−ω2 εµ dz2
As final result we find the wave equation of the plane wave:
d 2 E x (z)
2
+ ω2 µε E x (z) = 0 (5.6)
dz
Comparing equation 5.6 with the differential equation for a wave on a transmis-
sion line shows a correspondens between the voltage distribution U(z) and the
electric field E x (z), while the indutance per unit length L′ and the capacitance per
66
unit length C′ correspond to the space constants µ and ε. Due to this analogy one
can use the same function to solve the differential equation:
One can introduce the same complex valued propagation constant γ, the attenua-
tion constant α and the phase propagation constant β. In case of the plane wave
their numerical values depent of the space constants µ and ε.
κ
r
√ √ √ p
γ = jω µε = jω µε 1 − j = jω µε 1 − j tan(δε ) (5.8)
ωε
Due to the square root in equation 5.8 the sign of the complex valued propagation
constant γ can be positive or negative, corresponding to two different waves, one
travelling in the positive z-direction and one travelling in the negative z-direction.
According to equation 5.3 we get for the magnetic field H y (z) in dependence of
the electric field E x (z):
1 d(E x0 exp(−γz)) γ ε
r
H y (z) = − = E x (z) = E (z)
jωµ dz jωµ µ x
It makes sense in analogy to the characteristic impedance of a transmission line to
introduce the complex valued field impedance Z F .
r
µ
ZF = (5.9)
ε
Using the field impedance we get the following equation for the magnetic field
H y (z):
1
H y (z) = E (z) = H y0 exp(−γz) (5.10)
ZF x
According to equation 5.8 the complex propagation constant becomes imaginary
in the limit κ → 0 which corresponce to a lossless wave propagation. For a certain
value of conductivity κ the wave progation will now longer be lossless. Essentially
for the amount of damping is the relationship between conduction current density
|J~C | ∼ κ and displacement current density |J~D | ∼ ωε.
67
Using this limit we get for the complex valued propagation constant 5.8 the fol-
lowing approximation:
√ κ
γ ≈ jω µε(1 − j )
2ωε
For the phase constant β and the damping constant α we get in this case:
√
β = ω µε
(5.11)
κ tan(δε )
q
µ
α = 2 ε = 2 β
If one wants to answer the question with what velocity a point of constant phase
φ0 is propagating, we can in analogy to transmission line theory formulate the
equation for the phase velocity vPh :
ω 1
vPh = = √ (5.12)
β µε
For the field impedance Z F we get for the case of dominent displacement current
density: r
µ
ε
v
u µ ZF
ZF = u κ ≈ 1− j κ = 1− j κ (5.13)
ε(1 − j )
t
ωε 2ωε 2ωε
The phase propagation constant β and the damping constant α will reduce to:
√
β = ω µ0 ε0
(5.15)
α ≡ 0
According to the last equation the plane wave in free space will not be attenuated.
The length ∆z in which the phase is changed by 2π is know as the wavelength λ0
of the plane wave in free space. Using equation 5.15 we find:
2π 1 c0
λ0 = = √ = (5.16)
β f µ0 ε0 f
68
The equation for the field impedance reduces to:
r
µ0
ZF = = Z0 (5.17)
ε0
In free space the field impedance shows a real value Z0 ≈ 377 Ω. Fig. 5.1 shows in
its lower part the space function of the electric field at t = 0, what can be deduced
from the complex valued equation 5.7 (α = 0). In the upper part are shown the
correponding field lines of the electric field. Since the electric field consists only
a x-componenet the electric field lines are parallel to the x-axis. In principle they
have an extension from x = −∞ to x = ∞.
x
✻
✻✻✻✻✻ ✻✻✻✻✻
~E
✲
z
❄❄❄❄❄
Ex (z,t = 0)
✻
✛ ✲
λ0 /2
✲
z
Figure 5.1: Field lines and space function of the eletric field of a plane wave
2π λ0
λ = = √ (5.19)
β µr εr
69
For the majority of materials used in the rf-domain we have µr ≡ 1 und εr > 1 as
a result also the field impedance reduces compared to its value in free space:
r r
µ µr
ZF = = Z0 (5.20)
ε εr
From a physical point of view the real value of the field impedance results in the
fact that the real space and time functions of the electric field Ex (z,t) as well as
the magnetic field Hy (z,t) are in phase, as illustrated in Fig. 5.2.
Ex (z,t = 0)
✻
✙ ③
z
Hy (z,t = 0)
Figure 5.2: Space and time function of the electric and magnetic field of a plane
wave
To develop an idea how a wave does transport power, we will calculated the time
averaged poynting vector < ~S >. We start with the solution given in equation 5.7
for the case without losses (α ≡ 0).
1 ∗ 1
< ~S > = Re(~E × ~H ) = Re[E x0 exp(− jβz)E ∗x0 exp(+ jβz)]~ez
2 2ZF
For the time averaged poynting vector we get:
1 1
< ~S > = |E x0 |2 = ZF |H y0 |2 (5.21)
2ZF 2
According to 5.21 the discussed plane wave in a lossless media will transport a
constant mean power density in the direction of propagation (z-direction). For il-
lusstration Fig. 5.3 shows the electrical and magnetical field lines at a certain time,
showing their negative and positive maximum. Easy to recongnise are the planes
in threee dimensional space, which are reponsible for the name of the wave and
which are also planes of constant phase. Also shown is the instantaneuos poynt-
ing vector which is perpendicular to the planes of constant phase. It is worth to
70
✻
✲ ✒
✲
✻x ✻ ✒
~E ✲ ✲ ✲ ✲
✻✲ ✲ ✲ ✒
✲ ✲ ✲
✻
✲ ✲ ✲ ✲ ~ ✒
✲ ✲ ✲ ✲
S
✻
✲ ✲ ✲ ✲ ✲ ✒
✲ ✲ ✲ ✲ ✲
✻
✲ ✲ ✲ ✲ ✲ ✲ ✒
✲ ✲ ✲ ✲ ✲ ✲
✠ ✲ ✲ ✲ ✲ ✲ ✲ ✲ ✲ ✲ ✲❄
✠ ✲ ✲ ✲ ✲ ✲ ✲ ✲ ✲❄ ✲
✠ ✲ ✲ ✲ ✲ ✲ ✲❄ z
~
H ✠ ✲ ✲ ✲ ✲❄
✠ ✲ ✲❄
✠
y ✠ ✛ λ/2 ✲ ❄
mention that the poynting vector always points in the same direction even when
the directions of the field lines changes.
5.2.3 Polarisation
To discuss the terms poalrisation and wave vector of a plane wave we again use
the solution for a plane wave in a lossless media (γ = jβ, Z F = ZF ). To find a
simple solution of Maxwells equation we arbitrarily started with a electric field
in x-direction, which only shows a z-dependence. As a result we found a plane
wave, which propagates in the z-direction and which has a y-orientated magnetic
field also with z-dependence. As starting point we also could have choosen a y-
orientated electric field with z-dependence. As a possible solution for Maxwells
equation we would have found a plane wave, but now with a magnetic field in
the negative x-direction. The values for the other parameters describing the wave
e.g. phase constant β or field impedance ZF do not change. Extending our knowl-
edge we could also use the following more general field distribution which solves
71
Maxwells equation and which describes a plane wave in a more general form:
~E(z) = E x0
exp(− jβz) = ~E 0 exp(− jβz) (5.22)
E y0
The values E x0 and E y0 defined in equation 5.22 are arbitrarily choseable which
define the complex electric field vector ~E 0 at the space coordinate z = 0. Using
the law of induction we find for the magnetic field:
~E 2 = |~E 2 |
0
Beeing a complex vector ~E it can be decomposed in its real and imaginary part:
~E = ~E r + j~E j
Using its real- and imaginary for forming the square of the vector ~E , shows that
the vectors ~E r and ~E j are perpendicular to each other.
The equation of an arbitrary plane wave (5.22) may thus be writen in the following
form:
~E(z) = (~E r + j~E j ) exp(− jβz − jδ) with ~E r~E j = 0
In the last equation the electric field consists of two field components ~E r and ~E j ,
which are perpendicular to each other and which show a phase shift of 900 . If a
unit vector ~eξ is introduced, which has the same direction like the vector ~eξ and a
further unit vector ~eζ which is parallel to ~E j , the equation of the electric field can
be rewritten as:
~E(z) = (E r~eξ + jE j~eζ ) exp(− jβz − jδ) (5.24)
72
For the real time functions of the electric field we now will get:
Eξ cos(ωt − βz − δ)
~E(z,t) = Re[~E(z) exp( jωt)] = (5.25)
Eζ sin(ωt − βz − δ)
Equation 5.25 gives the space and time function of the electric field strength in
a general form. To discuss the different forms of palarisation of a plane wave
Fig. 5.4 shows some spezial cases. In Fig. 5.4a the direction of the unit vector
~eξ coincidence withe x-direction of the coordinate system, while the ζ-component
does not exist. In this case it is said the wave shows a linear polarisation in x-
direction. If the directions of the vectors are given as shown in Fig. 5.4b one
speaks of a linear polarised wave in y-direction. If the shown x-direction coin-
✻x ✻x
✻
~E ~E
✲ ✲ ✲
y y
a) b)
✻
x ✻
x
✒
✒
~E ~E
✲ ✲
y y
c) d)
cidens with the vertical axis and the y-direction coincidents with the horiszotal
axis, one also speaks of vertical or horizontal polarisation. The existens of two
polarisation which do not influence each other can be used to double the chan-
nel bandwidth in radio link systems [?]. The most general case of polarisation is
shown in Fig. 5.4c. Here the end of the vector of the electric field strength will
move on a ellipse, one speaks of a wave with eleptical polarisation. In the special
case when the absolute values of the ξ and ζ are equal the tip of the vector of the
73
electric field strength will move on a circle, as shown in Fig. 5.4d. In this case
one speaks of a circular polarised wave.
In the last equation a real unit vector ~eE was introduced to denoted the direction
of the electric field ~E 0 = E 0~eE . Since the direction of the maganetic field is
perpendicualr to the electric field and the direction of propagation one gets the
✂✍ ❅
✂ ❅
✂ ~E❅
~r ✂ ✒
■
❅❅ ❅ ~k
✂
✂ ❅ ❅ ❅
❅
z′ ❅ ❅
✻ ✂ ~ ❅
✂ ❅❄H Fläche
✂ ❅ konstanter
✂ ❅
❅ Phase
✂
✂ ❅
✂ ✲ y′
❅
❅
❅
❘
❅ x′
following equation:
~
~H(z) = k ×~eE E 0 exp(− j~k~r) (5.28)
|~k| ZF
74
Fig. 5.5 shows a plane wave, which propagates in the coordinate system x′ , y′ , z′ in
the direction given by the wave vector ~k. For illustration a plane of constant phase
is shown, which is located at a distance of~r~k/|~k| from the origin of the coordinate
system. Within this plane are located the vectors of the magnetic field H ~ and the
electric field ~E, which are perpendicular to each other and also perpendicular to
the direction of propagation.
γ ≈
p
jωκµ (5.29)
This results in the following value for the phase and damping constants:
ωκµ
r
β = α = (5.30)
2
Since the displacement current density is neglected ε is not present in equation
1.2
|H y (z)| ✻ y
✻
|H y0 |
0.8
0.6
0.4 ✲
❥ z
0.2 x
0 ✲
0 1 2 3 z 5
δ
Figure 5.6: Penetration of a plane wave into a conductor
5.30. To further explain the different behaviour we have a look to Fig. 5.6. It
75
shows a space described by a cartessian coordinate system. In the region z > 0 a
metallic conductor is present. We assume the existants of the magnetic field H y0
at the boundary. Since both the magnetic as well as the electric field is describe
by the same propagation constant, we find according to equation 5.7:
H y (z) = H y0 exp(−γz)
Fig. 5.6 shows the normalised absolute value of the magnetic field H y (z) as a
function of the normalised space coordinate. The length ∆z resulting in a drop of
the magnetic field at the surface |H y0 | by a factor of 1/e is called the skin depth δ
of the plane wave for a given conductor.
s
1 2
δ = = (5.31)
α ωµκ
With the help of the calculated value we can setup an equation to calculate the
skin depth for copper δCu with a specific set of units:
65 µm
δCu = r
f
MHz
The last equation essentially shows that a rf or microwave field can not penetrate
into a good conductor. This effect is called skin effect, because all fields will only
exist next to the surface of the conductor.
The complex valued field impedance Z F of a conductor posses real and imaginary
parts with the same value, as also does the complex valued propagation constant
γ. Starting from equation 5.9 wie get:
u µ
v
r
t ε 1 + j µω
u
ZF = κ = √2
− j ωε κ
(5.32)
76
With the help of the field impedance Z F we get for the electric field inside the
conductor: r
µω
E x (z) = (1 + j) H exp(−γz) (5.33)
2κ y0
The electric field inside the conductor will according to 5.2 result in a conduction
current density J x (z). Its absolute value will also exponentially decay:
~n × ~E = 0 (5.35)
The behaviour of the current density will be more complicated. Using equation
5.34 it can be deduced using the magnetic field at the surface:
r
µωκ 1+ j
J x (z) = (1 + j) H y0 exp(− z)
2 δ
√
At the one end its aboslute value will tend to infinite as κ and one the√other hand
the thinkness of the current carrying layer will decay according to 1/ κ. It is of
advantage to introduce a so called surface current density J Ox with the unit A/m,
which can be defined by a one dimensional integration of the current density J x :
Z ∞
J Ox = J x (z) dz = H y0
0
The surface current density thus shows the same value like the magnetic field,
but its direction is perpendicular to the magnetic field. Again using ~n to define a
normale to the surface, we get the following relation between the surface current
density and the magnetic field.
~n × ~H = ~J O (5.36)
77
Chapter 6
Antenna
78
✻
■ ✒ G(ϕ, ϑ)
✶
✛ ✉P ✲ P1 ✉ ✲
1
q
✠ ❘
a) ❄ b)
Figure 6.1: Radiated power density of a isotropic point radiator and a antenna
described by the gain function G(ϕ, ϑ)
79
✲ ✻
✲
S2
P2 ✞ ✎✴
✝❖♦
r D
✲
❄
According to Fig. 6.2 the effective receiving area of the parabolic antenna will
be in the order of its cross-sectional area πD2 /4. Using equation 6.2 and 6.4 it is
easy to describe the properties of a Transmitter-Receiver-System.
Aw2 G1
P2 = P1 (6.5)
4πr2
Of course the transmitting antenna can be used as receiving antenna and vice
versa. This will lead to the following equation:
Aw1 G2 ′
P1′ = P2
4πr2
If we assume in both case the same transmitting power P1 ≡ P2′ due to the reci-
procity law [?], we must have in both cases the same value of the received power.
This leads to:
Aw1 G2 Aw2 G1 Aw1 Aw2
2
= → = = konst.
4πr 4πr2 G1 G2
According to the last equation the ratio between effective area and gain of a an-
tenna in direction of the main beam shows a constant value. Accoding to Maswells
theory this ratio is given by:
Aw λ2
= 0 (6.6)
G 4π
This results in the following relation for the received power of a transmitter re-
ceiver system:
Aw2 Aw1 G2 G1
P2 = P1 = r P1 (6.7)
λ0 r
2 2
(4π )2
λ0
80
Example: Received power of a satellite receiver station
The satellite DFS-Kopernikus transmits at a frequency of 12 GHz with an EIRP
of 56 dBW with its main beam to Europe. (Diameter of the receiving antenna
D = 80 cm)
EIRP
56 = 10 log( ) → EIRP = 1 W · 105,6 = 398 kW
1W
For the power density S2 at the receiver side we get:
398kW pW
S2 = 2
= 24, 4 2
4π(36000km) m
24, 4pW/m2
P2 = = 12, 3pW
π(0, 4m)2
81
Bibliography
[1] Pozar, M. D.: Microwave Engineering, John Wiley & Sons, 2011.
[2] Collin, E. R.: Foundations for Microwave Engineering, McGraw Hill, 1992.
82