Antenna Tuning For WCDMA RF Front End: Reema Sidhwani
Antenna Tuning For WCDMA RF Front End: Reema Sidhwani
Antenna Tuning For WCDMA RF Front End: Reema Sidhwani
Reema Sidhwani
Thesis supervisor:
Prof. Olav Tirkkonen
Thesis instructor:
MSc. Janne Peltonen
Aalto University
School of Electrical
Engineering
aalto university
school of electrical engineering
abstract of the
masters thesis
Language: English
Number of pages:6+64
Code: S-72
iii
Preface
Now that the daunting task of developing an antenna tuner system in ST-Ericsson
has come to a conclusive point, I cant help but wonder how I would have managed
to come this far without the help, encouragement and guidance of my colleagues in
Turku, Lund and Nuremberg offices. Most of all I would like to thank my instructor Janne Peltonen for his useful insights and for sharing his knowledge with me.
Also, my sincere thanks to Jean-Louis Mendes, Michael Hirschmann, Joerg Meissner, Christina Grapsa, Bjorn Gustavsson and Jari Horkko. Next, I would like to
thank my supervisor Prof. Olav Tirkkonen. His suggestions and corrections made
a tremendous difference to the shape this thesis has taken.
I would like to express my heart felt gratitude to my friends in Finland, not just
for their friendship but also for encouraging me to explore the unknown. Without
you I would not have made it. Thanks for making my life here filled with fun and
happiness.
Last but not the least, I want to thank my family for their love and trust and
for constantly supporting me in all my decisions.
Otaniemi, 20.11.2012
Reema Sidhwani
iv
Contents
Abstract
ii
Preface
iii
Contents
iv
Abbreviations
vi
1 Introduction
1.1 Motivation for the research .
1.1.1 Impedance mismatch
1.1.2 Antenna tuning . . .
1.1.3 Impedance Tuner . .
1.2 Objective and goals . . . . .
1.3 Outline . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2 Background
2.1 Theory . . . . . . . . . . . . . . . . . . . .
2.1.1 Impedance matching . . . . . . . .
2.1.2 LC Impedance Matching Networks
2.1.3 Binary capacitance array . . . . . .
2.1.4 Measuring parameters . . . . . . .
2.2 Antenna Fundamentals . . . . . . . . . . .
2.2.1 Antenna Characteristics . . . . . .
2.2.2 Antenna types . . . . . . . . . . . .
2.2.3 Design issues . . . . . . . . . . . .
2.3 Building Blocks . . . . . . . . . . . . . . .
2.3.1 Adaptive Antenna Tuner . . . . . .
2.3.2 Bidirectional Coupler . . . . . . . .
2.3.3 Measurement Receiver . . . . . . .
2.3.4 MIPI RFFE interface . . . . . . . .
2.3.5 Processors . . . . . . . . . . . . . .
2.3.6 Software . . . . . . . . . . . . . . .
3 Design and implementation concept
3.1 Reflection coefficient tracking . . . .
3.1.1 Coexistence with other control
3.2 Impedance calculation . . . . . . . .
3.2.1 Equation based algorithm . .
3.2.2 Gradient Search algorithm . .
3.2.3 Hill Climbing algorithm . . .
3.3 Common algorithm control . . . . . .
3.3.1 Operating modes . . . . . . .
3.3.2 Antenna tuner state machine
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . . .
algorithms
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
3
5
5
6
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
8
8
8
9
12
13
15
16
17
18
18
19
19
21
21
22
22
.
.
.
.
.
.
.
.
.
24
24
26
28
28
31
33
35
35
36
v
3.4
3.5
Timing considerations . . . . . . . . . .
3.4.1 RAT specific timings . . . . . . .
3.4.2 MIPI RFFE timings . . . . . . .
Performance degradation considerations .
3.5.1 Error Vector Magnitude . . . . .
3.5.2 Linearity . . . . . . . . . . . . . .
3.5.3 Insertion loss . . . . . . . . . . .
3.5.4 Reflection coefficient accuracy . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
41
42
44
45
45
45
46
46
.
.
.
.
.
.
.
.
.
.
.
.
.
.
48
48
48
49
49
50
50
50
51
51
51
51
52
52
53
5 Conclusion
57
5.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.3 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
References
62
vi
Abbreviations
RF
RAT
MIPI
GSM
WCDMA
LTE
FDD
TDD
SMA
LO
PA
LB
HB
BB
ADC
3GPP
Radio Frequency
Radio Access Technology
Mobile Industry Processor Interface
Global System for Mobile Communications
Wideband Code Division Multiple Access
Long Term Evolution
Frequency Division Duplexing
Time Division Duplexing
SubMiniature version A
Local Oscillator
Power Amplifier
Low Band
High Band
Baseband
Analog to Digital Converter
Third Generation Partnership Project
Introduction
1.1
1.1.1
A radio front end consists of an antenna and components like power amplifiers,
switches, filters etc. In addition, the connecting tracks between these components
and the RF processing chain of the radio, are also included in the front end. When
tested in isolation or in simulations, these components show accurate results in
compliance with their datasheets. On the other hand, when testing is done in real
life situations where the front end modules are tightly integrated with the device
and the environmental effects are constantly changing, the specifications of these
components can turn out to be grossly inaccurate. Impedance mismatch between
the components, improper termination and losses in the cables and interconnects,
are some of the reasons behind this inconsistent behavior of the front end modules.
One of the key radio performance indicators for mobile phone devices is Over The
Air (OTA) performance. OTA is quantified in terms of Total Radiated Power (TRP)
2
and Total Isotropic Sensitivity (TIS). Both parameters are mainly determined by
the performance of the antennas. Hence antennas are one of the key components of
the front end and even the whole radio system and its design and characterization is
crucial for the performance of the entire device. Antenna is an electrical device that
converts the electric energy from an RF transmitter into an electromagnetic wave
propagating in free space and vice versa. For maximum power transfer between the
antenna and transmitter/receiver (transceiver), the input impedance at the antenna
terminals must be the same as that at the output of the transmitter power amplifier
or the receiver low noise amplifier. When a mismatch occurs in the impedance
between the antenna and that of the radio transceiver, degradation happens in
the antenna performance. Various factors are responsible for this mismatch. Most
often the antenna design compromises the desirable impedance in favor of radiation
pattern, efficiency and the size and shape of the mobile phone. Another factor
is the number of bands and the frequency range that the antenna is expected to
support. A pure resistive load is not frequency dependent but when the impedance
of the antenna starts to have a reactive element, it also becomes dependent on the
operational frequency. Figure 1 shows the variation of impedance of an example
antenna in a frequency range of 600 MHz and 3 GHz. The nearby electromagnetic
environment of mobile phone such as contact with human body, hand and head or
presence of conductive material like metal plates, also de-tune the impedance at
the antenna feeding point. The effect on detuning on the impedance of the same
example antenna is shown in Figure 2.
Antenna tuning
4
cleverly manipulating the configuration of this matching network, it is possible to
make the load impedance almost equal to the source impedance. When used in
the context of antennas with dynamic adjustment of impedance, this mechanism
is referred to as antenna tuning. Antenna tuning system can be Open or Closed
Loop. In Open Loop antenna tuning shown in Figure 3, the matching network
element is fine-tuned to optimize the antenna performance at different frequencies,
modulation schemes and modes of operation (hands-free, slide open, closed etc.).
This configuration is stored in a look-up table in the non-volatile memory of the
handset at the time of production. Based on the information provided by the higher
layer software, the tuning algorithm selects the appropriate setting for the matching
network. However, this mechanism is not able to adapt according to the changing
environment conditions since it does not make any real time measurement. The
usage of a mobile device has a constantly changing environment as the user walks,
drives, moves his or her fingers (commonly referred to as head or hand effect). So the
open loop approach is not as effective in restoring the losses caused by impedance
mismatch.
Impedance Tuner
1.2
This thesis aims at designing and developing an antenna impedance tuner system
that can be introduced at commercial scale in cellular terminals. The broad level
goals are as follows:
6
The antenna tuner shall improve over the air performance (total radiated
power/total isotropic sensitivity).
The antenna tuner shall decrease the VSWR seen by the power amplifier output while improving the efficiency and the overall current consumption.
The antenna tuner shall not influence the complex baseband signal in a way
that the signal processing algorithms are disturbed significantly.
The antenna impedance tuner algorithm shall prioritize closed loop, and shall
transfer to open loop when closed loop is not feasible. The algorithm shall
adapt to the transmitted power and the interference power level.
Besides the above goals, the intention is to keep the system as flexible as possible so
that there is minimum dependence on the topology and technology used in the
antenna tuner and adaptation to a different tuner can be easily done. This is
required because the supplier of antenna tuner hardware is not the same as that
of RF transceiver hence the choice is most probably made by the mobile handset
manufacturer.
In order to achieve the goals set for the performance and design of antenna tuner
system, this thesis aims to implement the following modules:
Reflection coefficient tracking module that characterizes the measurement receiver and sets it up with correct parameters so that the magnitude and phase
of forward and reflected power can be measured.
Tuning algorithm that determines the extent of impedance mismatch and calculates the new settings of antenna tuner to mitigate the mismatch.
Antenna tuner control module that programs the MIPI RF front end bus to
make appropriate settings to the antenna tuner hardware
Common control module that controls the state machine of the antenna tuning
system on the basis of the power level, measurement results of the measurement
receiver and various other factors.
Thereafter measurements will be done in the laboratory where the mismatch conditions of the antenna will be simulated with lab equipment and the performance of
the tuner system, with respect to the RF parameters will be observed and recorded.
1.3
Outline
The rest of the thesis is organized as follows. Chapter 2 presents the basic theoretical
concepts of impedance matching in generic electrical circuits and the measurement
parameters that are used in this subject. The building blocks, both hardware and
software, of a generic antenna impedance tuner are also described in this chapter.
Chapter 3 gives a detailed design of the various modules of the antenna tuner system that are developed as part of this thesis work including a few algorithms for
7
impedance tuning that were designed and evaluated during the course of the project.
The timing considerations involved in implementation of antenna tuning system for
different RATs are also treated here. Also the performance parameters that play
a key role in evaluating the performance of a tuning system are mentioned in this
chapter. Chapter 4 describes the test and measurement system used in the laboratory and presents the measurement results. Thereafter the results are analyzed
to check whether and how an antenna tuning system can be a viable solution for a
commercial handset. Chapter 5 concludes with a summary and recommendations
for future research work.
Background
The concept of antenna impedance tuning is not new in the wireless communications world [10]. The traditional approach followed by the antenna designers is to
use extensive simulations and fine tune the antenna after spending long hours of
measurements in the lab. Needless to say this consumes a lot of time, effort and
resources. As the requirement of frequency range to be supported by the antenna
grows larger and the design constraints become higher, this approach is becoming
untenable. Besides, this methodology gives static results that cannot adapt to the
changes in the antenna environment. Hence, the concept of closed loop adaptive
antenna tuning is gathering momentum and most of the mobile phone manufacturers are exploring the possibility of including some form of antenna tuning in their
handsets. Before delving deeper into the design and implementation of such a system, it is important to treat the fundamental concepts behind antenna design and
impedance matching.
This chapter presents the theoretical background on the subject as well as an
introduction to the building blocks involved in developing a practically realizable
antenna tuning system.
2.1
Theory
Matching the impedance of the antenna to the ideal impedance of 50 ohms is one of
the solutions to avoid the consequences of the mismatch. This section presents the
basic theory and mathematical details behind the concept of impedance matching.
The parameters which can measure the quality of match, are also listed.
2.1.1
Impedance matching
Impedance matching can be defined as the task of matching the input impedance
of one circuit or device (source) to the output impedance of another (load) when
both are connected together. The motivation for impedance comes from the fact
that maximum power transfer through the two devices can take place when their
impedance are matched else a large percentage of input power gets reflected. When
both the load and the source are purely resistive then according to the Maximum
Power Transfer Theorem, the load resistance must be equal to that of the source for
optimal power transfer. In case the circuit consists of reactive elements, a conjugate
match between the load and the source is needed for maximum power transfer. In
most of the cases, it is not possible to change the configuration and design of either of
the two devices in order to improve the impedance match. Hence a common solution
is to connect another device in between whose configuration can be controlled or
tuned. Such a device is called the impedance matching network. However, it is
important that such a network should not introduce a significant insertion loss and
must deliver a reasonably good performance over a wide range of frequencies.
The main task of the matching network is to force the load impedance to look
like the complex conjugate of the source impedance in order to ensure maximum
9
transfer of power to the load. For example let us consider a two port network shown
in Figure 5 .
Using inductors and capacitors in series or parallel in an impedance matching network is quite common primarily because these passive elements do not consume
power and do not add any noise to the circuit. Any two resistive terminations can
be matched by introducing two reactive components in between them in the form
of a matching network [11]. Figure 6, 7, 8 and 9 show four possible single section
LC matching networks where (a) and (b) are low pass type while (c) and (d) are
high pass type impedance matching networks. Parallel or Shunt reactance is always
placed on the side of the higher impedance. Depending on the bandwidth requirements and the size of the load and source impedances, a suitable configuration can
be chosen.
The quality of such L-C circuits is quantified in terms of Q (Quality factor).
For a single reactive device, Q is defined as the ratio of the stored power to the
dissipated power. It is a dimensionless quantity and is expressed as a function of
reactance X and resistance R.
Q=
X
R
(1)
10
X1 = QL RS
(4)
RL
(5)
QL
The problem with the above LC networks is that they can either match a smaller
impedance to higher or vice versa. If the load impedance changes such that the earlier assumption about it being lower or higher than the source impedance is not
X2 =
11
1
jYC
Zin
(6)
Z2 =
1
jZB
Y1
(7)
12
1
jYA
(8)
Z2
After the load or antenna impedance is calculated using equations 6, 7 and 8,
the values of the three tunable reactive components has to be estimated for a perfect
match between the source and the load. One approach to simplify the estimation
process is to divide the Pi-network into two LC networks and match the load and
source impedance to a common intermediate resistance. Such matching networks
are not clearly determinable. There is always more than one solution depending on
the choice of the intermediate resistance which in turn determines the value of the
loaded quality factor and thus the bandwidth of the matching.
Yant =
2.1.3
13
Measuring parameters
Quality of impedance match can be characterized in terms of parameters like Reflection Coefficient, Scattering matrix and Voltage Standing Wave Ratio (VSWR).
A brief description of each of these parameters is given below.
Scattering matrix: Scattering matrix is a way of describing the behavior of voltage and current travelling through a transmission line when they come across an
impedance differing from the lines characteristic impedance. Also known as Sparameters matrix, it is a square matrix of unit-less complex numbers. For the
purpose of this thesis work, an S matrix for a 2 port network as shown in Figure 14
is considered.
b2
S21 S22
a2
Each element of the matrix is a ratio between forwarded and reflected voltages
at the respective ports as described below.
14
b2
a2 a1 =0
(9)
Hence S11 is defined as ratio of the reflected voltage at port 1 and the input
voltage at the same port where the output port is terminated by a matched load i.e.
a2 is equal to 0. S11 is the most important S parameter in our calculations and is
also referred to as reflection coefficient.
Reflection coefficient : When applied to a transmission line model, reflection
coefficient is expressed in terms of load and source impedance. Value of reflection
coefficient ranges from -1 to +1 depending on if the load impedance is 0 (short) or
infinity (open) respectively. In a situation of perfect match when load impedance
equals source impedance, there is no reflection along the transmission line and reflection coefficient is 0.
S11 =
b1
a1 a2 =0
S12 =
b1
a2 a1 =0
S21 =
b2
a1 a2 =0
S22 =
ZL ZS
(10)
ZL + ZS
VSWR: When an antenna is not matched to the receiver, power is reflected from
its terminals. This reflected voltage leads to the creation of a standing voltage wave
that has maximas and minimas along the transmission line. VSWR is defined as
the ratio of the peak amplitude of this standing wave to the minimum amplitude.
Hence VSWR is a measure of how much power supplied to the antenna is reflected
back from it. Mathematically, VSWR is expressed in terms of reflection coefficient
.
=
1 + ||
(11)
1 ||
VSWR is a unit-less ratio and is always a real and positive number. The smaller
the VSWR, the better is the impedance match between the antenna and the radio
transceiver and the lesser is the power loss. Ideally the value of VSWR should be 1.0
in a situation where all the power is delivered to the antenna and none is reflected.
The wide range of supported bandwidth and the environment conditions make it
almost impossible to achieve this ideal scenario hence a somewhat higher value is
usually acceptable. However as the VSWR increases, more power is reflected from
the antenna. Table 1 shows the physical significance of different values of VSWR in
terms of actual reflected power.
Smith chart: The value of reflection coefficient can be expressed graphically in
a complex plane such that both the real (resistance) and the imaginary (reactance)
values are shown in the form of circles. Such a plot of circles is called the Smith
chart as shown in Figure 15. The horizontal axis represents pure resistance with 0 at
far left and infinity at far right. The center point is where the reflection coefficient
is 0 and hence the impedance is ideal 50 ohms. Circles on the Smith chart represent
constant resistance curves, while the arcs radiating out from the right side to the
edge of the Smith chart represent reactance curves. The points above the real axis
represent inductive circuits while the ones below represent capacitive circuits.
VSWR =
15
Table 1: VSWR, s11 and reflected power
VSWR S11 Reflected power(%) Reflected power(dB)
1.0
0.000
0.00
-Infinity
1.5
0.200
4.0
-14.0
2.0
0.333
11.1
-9.55
2.5
0.429
18.4
-7.36
3.0
0.500
25.0
-6.00
3.5
0.556
30.9
-5.10
4.0
0.600
36.0
-4.44
5.0
0.667
44.0
-4.02
6.0
0.714
51.0
-2.92
7.0
0.750
56.3
-2.50
8.0
0.778
60.5
-2.18
9.0
0.800
64.0
-1.94
10.0
0.818
66.9
-1.74
15.0
0.875
76.6
-1.16
20.0
0.905
81.9
-0.87
50.0
0.961
92.3
-0.35
2.2
Antenna Fundamentals
Antenna is the integral and the most essential part of any wireless product. It
can be described as a transducer that converts electromagnetic energy to radiating
waves in free space and vice versa or as a band pass filter that operates in a defined
band of frequencies while rejecting all other frequencies. Performance of a wireless
16
device depends to a large extent on the robustness and the design of its antenna.
An expensive and compact radio module can easily lose its benefits because of bad
design or improper integration of the antenna. This section takes a brief look on
the fundamental concepts [14] involved in antenna design, various types of antenna
and their performance characteristics.
2.2.1
Antenna Characteristics
Pr
Pin
(12)
The portion of applied power that is not radiated is lost due to different
reasons. Part of it can be reflected due to impedance mismatch while part is
lost due to absorption by nearby grounded conductors or dielectric materials.
The design constraints of modern handsets where antennas are embedded and
surrounded in plastic casings, can significantly effect the radiation efficiency
of the antenna. On an average the antenna efficiency is between 40 to 75%
where more than 75% is difficult to achieve with embedded antennas and less
than 40% is rejected during certification.
Polarization: Polarization of an antenna is the orientation of its electric field
with respect to the Earths surface. It is determined by the physical structure and orientation of the antenna. For two antennas communicating in Line
Of Sight, it is extremely important that they are polarized in the same orientation for efficient transfer of energy. However in the real world of mobile
communications, the Line of Sight rarely exists. Also the radio wave experiences reflection and scattering from the atmosphere and from the objects in
the path which substantially changes its radiation pattern and polarization.
In addition the usage of a mobile terminal is such that it is used in varying
orientations. Hence any optimization done for antenna polarization in mobile
terminal design is not very useful.
17
Linearity: In ideal conditions antenna can be considered a passive linear device. But when faced with high input power and multiple frequencies, the
metal joints and non-linear materials used in the antenna structure lead to
the generation of intermodulation products with frequencies different from
that of the input signal [15, Chapter 1]. When these intermodulated signals
are reflected back from the antenna, they interfere with the receiver. If the
receiver is highly sensitive, these signals can cause substantial interference.
In mobile terminals this problem is not that serious since the input power is
generally kept low to preserve battery life but in base stations where the input
power is high, receiver sensitivity is very low and multiple frequency signals
are commonly transmitted from the antenna, it is important to take care that
the generated intermodulated signals are kept at a sufficiently low level.
2.2.2
Antenna types
Antennas used in wireless devices are either external or embedded inside the device.
An external antenna is connected to the device on one end while the other end is
in free space generally perpendicular to the device. There are several advantages
associated with this kind of construction. The efficiency of such antennas is higher
and their distance from the digital noise sources present inside the device, ensures
better performance. However their susceptibility to damage and the design constraints of modern wireless products have almost completely eliminated their use.
Nowadays most antennas are embedded inside the mobile device. Two commonly
used embedded antennas, Inverted-F and Microstrip patch, are discussed here.
Planar Inverted-F style antenna (PIFA): A basic PIFA antenna is a quarterwave long conductor in the form of a metal plate, a few millimeters above
the ground plane, fed by a 50 ohm feedpoint at one end and grounded by DC
shorting plate on the other end. The input impedance of a PIFA antenna
is controlled by changing the distance between the shorting plate and the
feeding point while the height of the plate above the ground plane is fixed.
The range of supported resonant frequencies depends on the dimensions of the
PIFA antenna. Due to their low cost, easy fabrication, high bandwidth and
efficiency, PIFA antennas are nowadays the most popular antennas used in
wireless devices. They are usually omni-directional i.e. they do not radiate in
a particular direction.
Microstrip patch antenna: Microstrip patch antennas are usually in the form
of a one half wavelength long rectangular structure, fed by a microstrip transmission line, held over a ground plane made of highly conductive material,
separated by a dielectric material of only a few millimeter thickness. The
distance between the patch antenna and the ground plane can be much less
than that of PIFA antenna. The bandwidth and input impedance of such an
antenna depends on the dielectric constant of the substrate and the width of
the antenna. The electric field of microstrip patch antenna is linearly polarized
and its directivity is usually between 5-7 dB. Due to the directive nature and
18
surface area requirement, these antennas are not commonly used in mobile
devices but they are very useful in fixed or mounted devices that radiate in a
particular direction.
2.2.3
Design issues
An antenna cannot be characterized in isolation since the complete reference design of the radio module as well as the front end components, determine the input
impedance and the efficiency of the antenna. Hence picking up an off the shelf
antenna with certain desired characteristics, when used with a pre designed radio
system, may not necessarily give the expected results. Some of the design issues
encountered when designing the front end of a radio system are presented here in
brief.
Connection with the front end components: The front end of a radio system
is comprised of power amplifiers, filters, switches and their connections between the radio module and the antenna. When measured in isolation, each
of these components might be well matched to each other and to the antenna
termination but when interconnect cables and PCB traces come into play, the
impedance matching scenario can change completely. This impedance mismatch detunes the antenna, distorts the filter response and effects the linearity
of the power amplifier.
Self-interference: Improper shielding or the placement of antenna too close
to the digital noise source (processors, clocks, memory etc.), leads to various
spurious emissions from the devices own digital circuitry or from the reflected
signals to enter the antenna and other front end components, thus raising their
noise floor and reducing the sensitivity of the radio module.
Coexistence with other wireless technologies: Modern wireless handsets provide multiple radio technologies like Wi-Fi, Bluetooth, GPS etc. that have
radio transceivers and antennas operating over a varied frequency range. If
enough attention is not paid during the design phase, these antennas might
be quite close to one another interfering with each other and with the cellular
reception. Hence it is important both in the design and the characterization
phase, to test the radio system with other radio technologies.
2.3
Building Blocks
Nowadays, many players have come up in the market promoting different technologies and solutions for antenna tuning. After considering the pros and cons of
different solutions, it was decided in ST-Ericsson to use the antenna tuner part by
STM-Paratek, for developing the first phase of this antenna tuning system. However, the system design will be done considering that there will be customer requests
for other suppliers in the future so that a suitable robust system structure is chosen.
This means that the implementation details of the antenna impedance tuner shall
be well separated and easily identifiable in the antenna tuner system.
19
Many of the hardware components used in developing this system, are already
part of the current modem and RF sub-system project under progress in ST-Ericsson.
The only missing element that needs to be added in this already existing set-up is
antenna impedance tuner. The antenna impedance tuners are controlled via the
MIPI RF Front End Interface (MIPI RFFE IF), which is shared with the other
front end components in the radio subsystem.
The software with the antenna tuner algorithm is to be executed on the CPU
already residing inside the RF sub-system, the STxP70. The drivers for the measurement receiver and the MIPI RFFE IF are also created as part of this thesis work
to support the antenna tuner algorithm. The antenna tuner algorithm needs to interact with the transmitter power control algorithm in order to function properly,
but the antenna tuner algorithm is besides from that modular and can be removed
from the other radio control software.
2.3.1
Bidirectional Coupler
The antenna tuner system includes a bidirectional coupler[16] to sense the transmitted signal and the signal reflected at the antenna. Directional couplers are RF
passive devices that can act as power sensors due to their ability to couple a specific proportion of the power travelling in one transmission line out through another
20
21
Measurement Receiver
The measurement receiver is a radio receiver that samples the acquired transmitter
signal (from the bidirectional coupler) and compares it with a reference signal. From
the comparison, the measurement receiver calculates the phase and the gain of the
acquired signal. It is a part of the radio transceiver and is also used in closed loop
power control. It has a single-ended RF input which can be switched between two
separate input pins, one input for sensing the forward transmitter signal and the
other for sensing the reflected transmitter signal. The input signal is then passed
through an attenuator that has a 24dB control range with 6dB control steps. It is
then down-converted to the baseband and the resulting I and Q signals are passed
through a baseband gain block that can also be controlled till 24dB in steps of 6dB.
The down-converted baseband signal is passed through a low pass filter to reduce
the level of interference. The filter bandwidth is adjustable and it defines the noise
bandwidth of the measurement receiver. The filtered signal is then fed to a 10 bit
Analog to Digital converter (ADC) and then finally to the digital front end. The
digital block has the algorithms to calculate the Root Mean Square (RMS) level and
the phase of the detected signal. The block diagram of the measurement receiver
given in Figure 18 shows its various components clearly.
2.3.4
The HVDACs of antenna impedance tuner are fully controlled through the RFFE
serial interface (DATA, CLOCK, VIO) which is compliant with the MIPI alliance
specification for RF front end control interface[17]. The configuration of MIPI RFFE
IF used in this project operates at 26 MHz where the radio transceiver is the master.
The slave i.e. Antenna tuner is not capable of making any requests to the master.
This interface is shared with the other front end components in the radio subsystem
such as power amplifier and antenna switch. The system also has a 2nd MIPI RFFE
IF which optionally can be used as a dedicated interface to control the antenna
impedance tuners.
In order to meet the precise timing requirements and to avoid RFFE interface
traffic congestion at critical moments, a trigger mode is used to control the HVDACs.
The trigger mode, enabled by default, works such that the register values, containing
the settings for HVDACs are stored temporarily in shadow registers. When the
22
Processors
The radio transceiver used in this thesis work has two processors, and RF sequencer
and a STxP70 CPU. The STxP70 is a 32-bit Reduced Instruction Set Computer
(RISC) based multi-threaded CPU that has separate bus and memory for data and
code. The CPU executes all high level commands and interface control commands
that are not handled directly in the hardware. The commands are either executed
directly in the CPU or it prepares the RF Sequencer (RFS) for the command execution. The RFS is a simple processor that allows simple programs with sequences and
branches. It is able to perform some arithmetic and bit manipulation commands.
The RFS provides three independent threads that are processed in a round robin
manner. It is mainly used to control all radio sequence activities that are time critical and hence are triggered by specific RF timers. Most of the software modules
and algorithms that are developed as part of this project, are loaded and executed
in the CPU. However, since the RFS is responsible for the modules that are to be
executed at certain time instants in the data and control transmission sequence, the
code for measuring forward and reflected power from the measurement receiver is
executed in RFS.
2.3.6
Software
The antenna tuning software can be divided into four logical parts:
Reflection coefficient tracking: This part of the software deals with configuring
the measurement receiver, reading the values of forward and reflected trans-
23
mitter powers and then calculating the corresponding reflection coefficient and
S11.
Impedance calculation algorithm: This part uses the S11 and reflection coefficient measurement results to calculate the new impedance values (and HVDAC
settings) for a better match as per the antenna impedance tuner topology.
Common algorithm control: This is the central controlling block of antenna
tuning software that determines whether the impedance calculation algorithm
works in open or closed loop depending on the antenna condition, accuracy of
the reflection coefficient measurements and other settings.
MIPI RFFE driver: The task of this driver is to control the HVDAC values in
the antenna tuner by programming the MIPI RFFE interface.
The implementation details of all the above blocks are explained in the subsequent
sections.
24
The antenna tuner system is an optional part of the radio sub-system. There is
no remaining cost penalty or PCB area penalty if the antenna tuner system is
removed. In order to achieve this, the antenna impedance tuners and the associated
circuits are situated outside the standardized PCB layout for the radio sub-system.
The antenna tuner system needs support also from outside the radio sub-system.
Information about antenna conditions comes from outside the modem, i.e. the
application part of the mobile phone.
The architecture of antenna tuner system is designed such that there is a central
control block that has the intelligence to control all the sub-modules that encapsulate
the functionality of dedicated peripheral hardware components. These modules
can be easily replaced or modified when the corresponding hardware is changed.
Coexistence with other control algorithms that use the same hardware block is also
taken care inside the respective sub-module. Figure 19 shows the architecture block
diagram of the antenna tuner sub-system. Each element of this architecture is
described in detail in this section.
3.1
The task of this module is the detection and analysis of antenna mismatch. A
measure of antenna mismatch is the complex reflection coefficient which has an
25
amplitude and a phase. Hence this tracking module triggers the measurement of reflection coefficient from the measurement receiver. The hardware details of measurement receiver are explained in section 2.3.3. Before the measurement is triggered, a
sequence of register writes are done by this module for setting the filter bandwidth,
RF, baseband and digital gain blocks, measurement direction and operational frequency band in order to initialize and prepare the different hardware blocks inside
the measurement receiver. The measurement receiver in practice does not return
the amplitude and phase of the reflection coefficient but the power and phase of the
forward and reflected signal. It is the task of this module to calculate the amplitude
and phase of reflection coefficient from these parameters. The following equations
show how the derivation is done.
Amplitude [dB] = a Amplituderev [dBm] Amplitudefwd [dBm] + b
Phase [degree] = c Phaserev Phasefwd + d
(13)
(14)
The factors a and c and the additive components b and d have to be determined
by measurements in the lab. These factors take into account the phase shift of
the trace line between the coupler and the tuner as well as the asymmetry of the
bidirectional coupler.
The tracking module also checks if the quality of measurement done is good
enough and returns a quality indicator in addition to the reflection coefficient values.
Although it is quite difficult to determine the accuracy of measurement receiver,
following criteria have been decided after discussion with digital and analog designers
to find out if the measured values are reliable enough for taking into use in the
antenna tuning algorithm.
If the amplitude of reflected power is measured to be higher than transmitted
power, it is likely that the measurement is erroneous and hence should be
ignored.
If the delta between two consecutive measurements of reflected power amplitude is found to be greater than a defined threshold for the same frequency
range and the same antenna conditions, the measurement is probably incorrect.
Reflection coefficient tracking module has interfaces with central control module
and with power control module. It receives the request for triggering the reflection coefficient measurement from central control module and it forwards it to the
power control module. As a result, it receives the amplitude and phase values of forward and reflected power from the power control module and it sends the resulting
complex reflection coefficient values to the central control module.
26
3.1.1
27
28
3.2
Impedance calculation
The approach used in this algorithm is based on the impedance matching theory
described earlier. Maximum power can be delivered from the radio module to the antenna if the impedance of radio module (ZS ) is a complex conjugate of the impedance
of the antenna (Zant ). The impedance of the tuner (Ztuner ) can be altered to achieve
this match. This concept is depicted in Figure 22.
A similar analytical algorithm, based on the equations of a Pi-Network impedance
tuner, is proposed in [18]. The extensive calculation procedures used by the proposed method, make it difficult to implement it in a micro-controller with a limited
computation power. The Equation Based Algorithm proposed in this thesis work,
29
YA = wCA
1
wLA
(16)
ZB = wLB
1
wCB
(17)
1
(18)
wLC
Since the current configuration and topology of the antenna tuner is known, YA ,
ZB and YC can be easily calculated.
The third step is to calculate the antenna impedance. For this the equations
derived in section 2.1.2 are used. The following equations reiterate this calculation.
YC = wCC
Y1 =
1
jYC
Zin
(19)
Z2 =
1
jZB
Y1
(20)
1
jYA
(21)
Z2
Once the antenna impedance is known, the next step is to calculate the values of
matching components in order to achieve the target antenna impedance (50 ohms
in most cases). As mentioned before, this is not a very straight-forward task in
Pi-network and one way to simplify it is to split the Pi-network into two L matching
Yant =
30
networks and match the source and antenna impedances to a common intermediate
resistance (Ri ). Figure 23 shows such a split network.
31
3.2.2
(23)
The vector direction is determined by the information about the gradient which
in turn is based upon information of the previous behavior of the vector or in some
cases, on the uniform distribution pattern of the available data. The step size is
generally determined by the distance from the target condition but can shrink or
expand depending on several conditions. As long as the step direction is close to
the gradient, this kind of algorithm is guaranteed to converge though there is a risk
of running into local minima.
The above approach can be applied to the problem at hand i.e. closed loop
antenna tuning [20]. The current point Xk is the set of capacitor control voltages
(Va , Vb , Vc ) that determine the antenna tuner settings at any instant of time. The
candidate point Vk+1 refers to the antenna tuner settings that produce the desired
impedance that would match the antenna impedance to that of radio module. Step
size Sk is the distance of the current antenna impedance from the desired one and
Dk is the gradient that needs to be calculated. Two possible methodologies of
calculating the gradient are explained here.
Pre-determined Gradient control: The basic idea in this algorithm is to characterize the antenna tuner for different antenna conditions and operational
frequencies and store the results in the permanent memory. In practice, this
means that for each capacitor in the antenna tuner configuration, we record
the change in effective antenna impedance with the change in capacitor voltage while keeping the other two capacitor voltages constant. The data is then
plotted as two separate graphs in the form of real input antenna impedance vs
capacitance voltage and imaginary input antenna impedance vs capacitance
voltage. Based on the data provided by the antenna tuner manufacturer, we
know that the shape of the curve showing the variation of antenna impedance
with capacitor voltage is almost the same irrespective of antenna condition
even if the absolute values are different. Same is true for similar curves plotted for different values of capacitance voltages for other two capacitors. Hence
we can approximate the curve into a polynomial by means of polynomial curve
fitting and determine the coefficients that best fit into the measurement data
in the least-squares sense. Based on the provided data it seems that a third
order polynomial is sufficient to match the curve shapes with sufficient accuracy. The values of coefficients x1 ,x2 and x3 for real and imaginary antenna
impedance for all the three capacitances, are stored in the permanent memory. The effective antenna impedance at any instant for a given input capacitor
voltage VA is given by the following equation:
32
(24)
Since we need the gradient of the curve for the antenna tuning algorithm, a
derivative of the given polynomial is required as shown in the equation below.
Gin = 3 x3 VA2 + 2 x2 VA + x1
(25)
Once the characterization of the antenna tuner is done, the following steps
constitute the antenna tuning algorithm in case of an impedance mismatch:
1. Measure the amplitude and the phase of the reflection coefficient
2. Measure the delta between the measured values and the target values for
a perfect match
3. If the delta is above a certain threshold, the gradient (real and imaginary
components) needs to be calculated using equation for each of the three
capacitor voltages.
4. Compare the calculated gradient with the delta for each capacitance and
rate them.
5. Choose the capacitor voltage whose gradient best reduces the distance
between the measured and the target values in both real and imaginary
planes.
6. Calculate the new HVDAC settings as per the chosen capacitor and its
gradient.
7. Go to step 1.
The advantage with this approach is that it is quite generic and can be applied
to any antenna tuner irrespective of its configuration and without detailed
knowledge of its schematic and layout details. Also its convergence to a tuned
impedance is more or less guaranteed. However, the number of steps and
hence the time taken to converge, and the accuracy of match, may vary. The
main disadvantage of this methodology is that extensive measurements in the
laboratory need to be carried out in order to characterize the antenna tuner
before it is possible to take this algorithm into use. Accurate characterization
of the tuner over all the RATs and the frequency channels is the key to success
for this approach so support from automated tools is also required. This makes
the setup more cumbersome and time consuming. Also, quite a lot of memory
is needed to save the coefficient values for each capacitor and for each RAT
and frequency channel.
Auto-didactic Gradient control: One way to avoid the extensive characterization of antenna tuner is an iterative self-learning approach where the coefficients are not stored in advance, rather the gradient is calculated during the
tuning process. In each iteration of the algorithm, a step is taken to increase
33
one of the capacitor voltages and its impact on the reflection coefficient and
distance from the target value is recorded. A gradient is then calculated for
the real and imaginary components of the impedance as shown below:
slopereal =
G
Vx
(26)
B
(27)
Vx
Once the gradients are calculated for all the three capacitors, they are then
compared and rated and finally the best one is chosen in the same way as
shown in pre-determined gradient control algorithm. Thereafter, another set
of measurement is started and a new set of gradients is calculated with respect
to the HVDAC setting selected in the previous step. This process continues
until a satisfactory impedance match is obtained.
slopeimag =
This algorithm does not need much memory or computational power and its
convergence probability is very high. There is also no dependence at all on the
antenna tuner topology. The downside is that the settling time for finding an
impedance match is longer.
3.2.3
In order to make the algorithm completely independent of the antenna tuner topology, as well as to reduce the dependence on the accuracy of the reflection coefficient
returned by the measurement receiver, it can be useful to adopt an adaptive random search approach. By definition, the random search samples repeatedly in a
feasible range, according to uniform sampling distribution such that each sample is
independently and identically distributed. If pure random search or blind search is
performed, the algorithm is guaranteed to find a global minima but the very long
settling time renders this approach infeasible. Hence an adaptive methodology of
local search [21] is required where the next candidate point is found in a strictly improving direction. It can be described as an iterative improvement algorithm that
tries to improve the current state by making small changes in each iteration until no
further improvement is possible. One such algorithm in the family of iterative local
search algorithms is called Hill Climbing [22]. It attempts to minimize or maximize
a cost function f(x) where x are discrete states.
Xk+1 =
Vk+1
Xk
In some ways, this algorithm is similar to the Gradient Search algorithm described before with the difference that it is not necessary to know the strength or
direction of the gradient. Instead, iterative steps of fixed size are made and new
ones are adopted if they show better performance. On the other hand, it is possible
to sample the performance at all the possible steps from the original solution and
34
then select the best one. Such an approach is refered to as Steepest Ascent Hill
Climbing. Hill climbing suffers from a problem that it can get stuck on a local
maxima or minima (illustrated in Figure 24) depending on the starting point but
in most cases [23] [24], it gives equally good or better results than more complex
algorithms.
35
3.3
Common algorithm control module is the central antenna tuning control block of
the RF sub-system. It is in charge of the antenna tuner state machine i.e. the
state of antenna tuner in different operating modes and the state transitions. It has
interfaces to the reflection coefficient tracking block, the impedance calculation block
and to the algorithm database. The following sub-sections describe the tasks of the
single states of the antenna tuner state machine as well as the state transitions.
3.3.1
Operating modes
A central control block is commonly used in systems that perform repetitive tracking.
A combination of Open and Closed Loop [25] is often used to design this control
module such that the Open Loop works on prior knowledge while the Closed Loop
uses feedback information to regulate the changes and uncertainities in the system.
Since the Open-Plus-Closed-Loop Control combines the advantages of both feedback
and non-feedback based approaches, it is widely used in theoretical [26] as well as
practical applications [27] [28]. This design methodology also meets the requirements
of antenna impedance tuning system and is therefore used in this thesis work.
When the radio sub-system is activated by the baseband from a standby state
to perform transmission or reception (Tx On or Rx On), the antenna tuner system
also gets activated and enters an initial state of Open Loop. At this stage the
antenna condition and its impedance is unknown hence the antenna tuner HVDAC
settings are decided on the basis of current frequency band and RAT. These antenna
tuner settings for different frequency ranges have been determined in advance during
characterization of the hardware in the lab and stored in the permanent memory of
the device. These stored settings are referred to as algorithm database. Besides the
initial state, the antenna tuner system also remains in the open loop mode if the
transmit power level is very low. This is because the dynamic range of measurement
receiver is such that it is not possible to detect signal at a power level below a certain
threshold. When the transmit power is low, the reflected power is even lower which
makes it impossible for the measurement receiver to report sensible data needed for
the calculation of the reflection coefficient.
As soon as the output power level is above the threshold, the system is switched to
Closed Loop. A request is sent to the tracking algorithm to trigger the measurement
of the forward and the reflected signals. The reflection coefficient value returned by
the tracking module is then sent to the impedance calculation module to determine
the new values of the HVDACs to ensure a better match. The Closed Loop remains
operational as long as the resulting VSWR does not reach an acceptable value. The
only exception to this is when the measurement receiver accuracy qualifier returned
by the tracking module is found to be bad. In the absence of reliable feedback
information, the system switches to Broken Loop [29]. Broken Loop means that the
last valid impedance tuner settings from Closed Loop mode remain unchanged. If the
transmitter is turned off or if the transmitted power level drops below the threshold,
the system switches back to Open loop from Closed Loop or Broken Loop. Also, if
the antenna condition changes, Open Loop comes into effect immediately.
36
3.3.2
Initial State Transition to Open Loop: When triggered into the initial Open
Loop state, the antenna tuner algorithm also gets information about the current RAT, frequency of uplink/downlink and antenna condition. Based on this
information, the algorithm looks for valid HVDAC settings from the algorithm
database. If there are no valid settings available for the current frequency, the
algorithm tries to find the data for corresponding sub-band in the same RAT.
Once the settings are available, they are provided to the MIPI RFFE driver
and the execution is triggered.
Open Loop: When in Open Loop state, the algorithm performs the following
sequence of tasks also shown in Figure 25:
1. It checks whether the antenna condition changed since the last time when
HVDAC settings were applied. If this is not the case, the applied HVDAC
settings are up-to-date and need not to be applied again. Else, it gets
the open loop HVDAC settings for the requested sub-band, RAT and
antenna condition from the database.
2. It might be possible that the complete HVDAC settings cannot be applied
in a single execution so that no disturbance is caused to the radio module
functionality. In this case the common algorithm control block partitions
the HVDAC settings into n steps. It stores the number of remaining
steps together with a timestamp indicating when it was used last in the
algorithm database and provides the HVDAC settings of one step to the
MIPI RFFE driver.
3. It checks whether all steps to apply the HVDAC settings are done. If not,
it provides the HVDAC settings of next step to the MIPI RFFE driver.
If the last step is applied, it also stores the now valid HVDAC settings in
the algorithm database.
4. If the transmit power level exceeds the threshold, the state is changed to
Closed Loop else the algorithm remains in Open Loop state.
If the antenna condition changes while the above sequence of steps are being
executed, the Open Loop sequence is triggered to start from the beginning.
Closed Loop: When in the Closed Loop state, the antenna tuner state machine
triggers measurements of the reflection coefficient, calculates the S11 and the
impedances and applies the new HVDAC settings. Closed Loop state can be
segmented into three sub-states each of which have a set of input and output
parameters and are triggered by specific events. Figure 26 shows the Closed
Loop state machine.
The tasks of these sub-states are as follows:
Set-up state: This is the beginning state of the Closed Loop state machine. It has the following tasks (Figure 27):
37
38
39
40
41
3.4
Timing considerations
42
The baseband control interface of the radio module used in this project consists
of control commands and Time Accurate Strobes (TAS) [30]. The control commands contain all information necessary to configure the RF subsystem and the
TAS messages enable precise timing. The execution of the command starts when
the corresponding TAS arrives. The minimum time between the TAS and the start
of reception or transmission is determined by the 150 s synthesizer settling time.
Within this period the tuner software has to find the appropriate tuner setting and
the tuner must be in steady state. RAT specific aspects of this timing constraint
43
are described below.
GSM: GSM is a half duplex FDD (Frequency Division Duplex) system where
transmitter and receiver are never turned on at the same time. The basic unit of
transmission is called a burst that lasts for 0.577 ms. Eight such bursts form one
TDMA frame with a length of 4.615 ms. Within this time the antenna impedance
does not change significantly and there is no need to update the tuner settings
during the burst. The approach is to configure the impedance tuner before the
burst together with the other front end components.
The optimum tuner settings for best impedance match and minimum reflected
power, change with frequency. Since the transmitter and receiver are operating at
different frequencies, it is necessary to have separate open loop impedance tuner
settings for the transmitter and the receiver. However, it will be an overkill in terms
of memory to have one setting for each ARFCN (Absolute Radio-Frequency Channel
Number). On the other hand, accuracy will be compromised if only one setting is
stored per band. The middle path solution chosen in this project is to split each
band into three sub-bands and to store one setting for each sub-band.
WCDMA: WCDMA is a full duplex communication technology with simultaneous transmission and reception. With the exception of the DTX/DRX (discontinuous transmission/reception) [31] and the compressed mode gaps, the receiver and
transmitter are continuously enabled. Each radio frame is 10 ms long and consists
of 5 sub-frames. Each sub-frame has 3 slots and the duration of each slot is approximately 667 s. Since the antenna condition and hence its impedance can change
significantly within 0.5s, it is not efficient to wait for a gap in the transmission. So
the antenna tuner setting has to be updated while the transmission and reception
is ongoing. This update of the tunable capacitance values can cause fluctuations
of power amplitude and phase disturbing the channel estimator and other blocks
in the receiver baseband chain. Such an impact can be mitigated if care is taken
that the tuner settings are changed in small steps. For example, if only a single
DAC is changed by one LSB per slot, the fluctuation can be kept minimal. In order
to make a full sweep on all the three capacitors each controlled by an 8 bit DAC,
this would take 3 256 667s = 512ms. Hence, even in the worst case scenario
where the antenna tuner settings are completely wrong, the small step is sufficient
to change the impedance in a non time critical way. The small steps also ensure
that the change in impedance tuner setting has a minimum impact on the existing
transmitter power control algorithm. As a result the setting can be changed anytime
in the TX and RX slot.
Another aspect of implementing antenna tuning in a WCDMA system is the difference in optimum tuner settings for transmission and reception. Since transmitter
and receiver are operating on different frequencies, it is obvious that they will have
different tuned settings but due to the fact that both are active at the same time,
a compromise has to be made between the best settings for transmission and for
reception. Hence a common target tuner setting is chosen for Tx and Rx for each
sub-band.
LTE FDD: LTE FDD is also a full duplex system like WCDMA, with transmission and reception happening at the same time at different frequencies. The radio
44
frame duration is 10ms and consists of 20 slots with a slot duration of 0.5ms. Two
adjacent slots form one sub-frame of length 1ms. The same concepts of doing tuner
setting updates in small steps and finding a compromise between optimum settings
for transmission and reception, apply here as well.
LTE TDD: LTE TDD is based on time division duplexing where transmitter
and receiver operate on the same frequency but at different times. So Tx and Rx
is never operational at the same time. The frame and sub-frame structure of LTE
TDD is not too different from LTE FDD except for certain special sub-frames like
DwPTS (Downlink Pilot Time Slot), GP (Guard Period) and UpPTS (Uplink Pilot
Time Slot) that are meant for supporting the switching between transmission and
reception and for carrying synchronization and control information. The switching
between transmission and reception is defined by the Guard Period and can take
approximately 20 s to 700 s depending on the configuration and timing advance.
It is best to apply the tuner settings in this time frame since both transmitter and
receiver are inactive and multiple small steps are necessary.
Since the transmitter and receiver are operating at the same frequency, the obvious approach is to use the same impedance tuner settings in open loop mode for
transmission and reception. But the tuner settings that are optimum for Tx are not
necessarily the best for Rx. So an approach similar to GSM is used for storing the
open loop and closed loop target settings.
3.4.2
The MIPI RF front end interface works with a nominal clock speed of 26MHz. A
normal register write consists of 25 Bits and takes 0.96s. A normal register read
consists of 26 Bits and takes 1s. As mentioned before, the antenna tuner consists of
three tunable capacitors where each capacitor has a separate bias voltage controlled
by HVDACs. The bias voltage level is controlled by an 8-bit DAC. By using trigger
mode the values stored in all the three DACs can be activated at once. One tuning
step is composed of three register writes for voltage programming and one register
write for the trigger event. The total programming time is 4 0.96s = 3.84s. In
addition the STM Paratek tuner has a settling time of 300 s when switched on for
the first time and 50 s for each HVDAC setting. The HVDAC settling time is quite
short compared to other timings in RF transceiver for eg. Synthesizer settling time
and hence will have a negligible effect on the RAT timings. However it is necessary
to use the Trigger mode described before so that all the voltage settings are applied
all at once thus minimizing the settling time. The initial settling time when the
tuner is switched on for the first time is rather long and therefore we decided to
switch on the tuner at the time of RF transceiver initialization and then to keep it
powered on all the time. The current consumption of the tuner is very little so it is
not a problem to keep it powered on.
45
3.5
Using an antenna tuner in the transceiver path reduces mismatch which improves the
TRP in the context of transmission and TIS for reception. Reducing the mismatch
loss also improves the current consumption as the power amplifiers output power is
reduced. However, the antenna tuner also introduces some insertion loss. In addition, the control updates during closed loop control introduce dynamic performance
limitations due to changing transfer function of the antenna impedance tuner. How
much the performance is degraded depends on the magnitude and phase response
change of the transfer function vs. time. Depending on the RAT and whether the
transmitter or receiver is considered, this response has different impact on the system
performance. This section focuses on this performance degradation and introduces
parameters that should be measured in the course of developing the antenna tuning
system in order to ensure system performance.
3.5.1
Error Vector Magnitude (EVM) is a measure of the distance between received symbols and the ideal symbols in the I-Q plane. It is used extensively to quantify the
performance of a digital radio system since the deviation of the constellation points
from their correct locations is caused by the imperfections in the implementation.
A large value of EVM leads to poor detection accuracy which can be regarded as
degraded transceiver performance. Mathematically it is defined as the ratio of the
power of the error vector to the average power of the reference constellation. It is
mostly expressed in percentage as follows:
s
Perror
100%
(28)
EVM(%) =
Preference
When the capacitor values are modified in the course of the closed loop tuning
process, it also changes the transfer function of the impedance tuner. Hence a
signal passing through the tuner undergoes a shift of phase and amplitude which
has a direct impact on its EVM. Hence EVM can be used as a reference when
analyzing the impact of changing the antenna impedance tuner settings. Exceeding
the target EVM values cause an unacceptable degradation in BER (Bit Error Rate)
and throughput. 3GPP specifications have laid down strict requirements on the
amount of EVM that can be tolerated in the transmitter and in the receiver for
different RATs. The amount varies between 4% (for WCDMA) and 12.5% (for
LTE).
3.5.2
Linearity
46
Channel Leakage Ratio). ACLR is defined as the ratio of the transmitted power to
the power measured after a receiver filter in the adjacent RF channel. The ACLR
limits imposed by 3GPP specifications are 33 dBc for an adjacent channel frequency
of 5 MHz and 43 dBc for 10 MHz. All the front end components like the power
amplifier, the main Front End Module (FEM), switches etc., exhibit non-linear behavior to a certain extent [32]. Hence most of the linearity budget available in a
radio system is used up by them. That is why it is extremely important that the
linearity of impedance tuner should not significantly worsen platform performance
and therefore must be equal or greater than linearity of main FEM switch.
3.5.3
Insertion loss
Insertion loss for an antenna impedance tuner can be defined as the loss in the output
power when the tuner is inserted between the RF module and the antenna in ideal
conditions i.e. a 50 ohm load. Hence the insertion loss can be quantified in terms of
output power. Since the ideal 50 ohm load condition rarely exists, the improvement
in mismatch loss provided by the antenna tuner mostly offsets the insertion loss.
For example, a VSWR of 7:1 for an antenna means a mismatch loss of 3.6dB. If
the impedance tuner has an insertion loss of 2dB but is able to match the antenna
impedance to that of the transceiver such that the VSWR is 2:1, then there is still a
substantial benefit. Insertion loss depends totally on the tuner topology. The STM
Paratek tuner used in this project has an insertion loss of approximately 0.5 dB in
high band and 1 dB in low band.
3.5.4
The success of closed loop tuning depends to a large extent on the accuracy of measured values of transmitted and reflected power by the measurement receiver. The
operational power range of closed loop antenna tuning depends on the sensitivity
of measurement receiver which in turn depends on its noise figure and used bandwidth. For accurate measurement of power and phase, the measurement receiver
needs a minimum signal to noise ratio of 20dB. In addition to the thermal noise,
the performance of the measurement receiver is limited by the transmitted signal
direct leaking to its input pin. The leaking signal is coherent to the measured signal
and depending on the relative phase angle, it either increases or decreases the measured signal amplitude. This sets high isolation requirements from the antenna, the
power amplifier output and the modulator output to the inputs of the measurement
receiver. Therefore the overall measurement accuracy of the reflection coefficient is
dependent on the following factors:
Bi-directional coupler directivity (power level difference between the coupled
and isolated ports) and coupling difference (power lost to the coupled and
isolated ports) between forward and reflected signal ports
Gain imbalance between forward and reflected power inputs of the measurement receiver
47
Isolation (power level difference) between the forward and reflected power pins
of the measurement receiver (minimum value 30 dB)
Isolation between transceiver output and measurement receiver inputs (minimum value 50 dB)
Isolation from antenna to the input pins of measurement receiver(minimum
value 80 dB)
Interfering signals from WLAN - Third harmonic from WLAN signal can sometimes be down-converted by the local oscillator of measurement receiver to
baseband on top of measured transmitted signal in the MRx mixer.
The above error sources have a worse effect on the reflected power measurements
since power level is generally lower there. That is why it is important to calibrate
the measurement receiver and to determine its dynamic range before starting the
antenna tuning activity.
48
Since the objective of this thesis work is to develop an antenna tuning system that
can be included in a commercial mobile platform, the method chosen to determine
the performance of such a system, is based on actual measurement rather than on
simulation [33]. This section describes the antenna tuning system that was actually
implemented in the lab using the hardware components and measuring instruments
provided by ST-Ericsson. The test set-up is also explained in detail. Finally the
measurement results of this exercise are presented and a discussion on the results
follows.
4.1
49
magnitude until the reference point does not change any more. There is a high
probability that this point would be a local minima but in the absence of information
about the phase of the transmitted and the reflected signal, this approach yields
relatively the best possible results.
4.1.2
Antenna parameters (impedance, VSWR, return loss etc.) depend on the antenna
condition and the frequency of usage. The different types of scenarios that would
determine the antenna condition are as follows:
Free space: Mostly applicable for hands-free mode
Head and hand: The phone is held against the head with a hand
Finger: The phone is held against the head with a hand and with a finger
covering part of the antenna
Only head: The phone is held against the head without a hand but with an
elastic band or tape
Each of the above scenarios can produce different antenna impedance depending
on its design and frequency of usage. Since the intention of this thesis work is
to develop and test an antenna tuning system that is as flexible as possible, we
decided to refrain from measuring the effectiveness of tuning for a specific antenna.
Instead, looking the broad range of data available from different antenna designs,
it was clear that the VSWR of an average antenna varies between 1:2 to 1:7 for
different conditions. Hence it was decided to carry out the measurements for several
mismatch values along this range of VSWR. Also it is important to take the phase
shift into account since it is not necessary that the antenna tuner system is able to
tune equally well on all the angles from 0 to 360 degrees. So we decided to measure
on one central point for each of the four quadrants of Smith Chart.
4.1.3
Chosen frequencies
Once the antenna tuning algorithm is implemented and the antenna parameters are
chosen, it is time to choose the frequencies for which the impedance matching will
be done with the antenna parameters given above. At the time of the execution of
this thesis work there are no automated measurement tools available in ST Ericsson
for the antenna tuning set-up. So each measurement has to be done manually and is
quite time consuming. So in order to keep the time schedule of this project, it was
decided to implement and test antenna tuning for WCDMA only. The radio module
used in our set-up supports five bands out of which three are in high frequency range
and the other two have low frequency. For carrying out the measurements, band 1,
Tx channel 9750 in the high band range and band 8, Tx channel 2788 in the low
band range were selected. In spite of the limited set of measurements, the chosen
bands and channels are such that it is possible to get a fair idea about the behavior
of antenna tuning over a range of frequencies.
50
4.2
Lab set-up
The lab setup used for carrying out the measurements with the antenna tuning
system is shown in Figure 31. The cables and transmission lines in between the
different units of this setup introduce a phase shift which changes the impedance
in an arbitrary way. This is undesired since the key requirement behind antenna
tuning is the reproducibility of its results. Hence care is taken that the same set of
cables are used in the setup from beginning to end and that these cables are also
included while calibrating the measuring devices. A short description of each of the
elements used in the setup is given below.
4.2.1
The radio transceiver used in this thesis work consists of two processors, a CPU
called STxP70 executing at a frequency clock of 208 MHz and a simple Sequencer
executing at 52 MHz. It has a Digital Front End block that interfaces with the baseband and an Analog Front End block that up-converts the signal to RF frequency
for uplink and down-converts to baseband frequency for downlink. Both blocks are
separated by a Digital to Analog converter. The radio module is mounted on a
cradle that also consists of a MIPI connector on which the Antenna tuner provided
by STM-Paratek, is plugged. The tuner module is provided with two SMA (coaxial
RF) connectors for Main and Diversity antenna. The transceiver is calibrated without the antenna tuner so that the insertion loss introduced by the tuner is visible in
the form of reduced output power in case of a perfect match.
4.2.2
Load-pull tuner
51
impedance. Hence the variability in the antenna impedance can be effectively simulated by tuning the load pull tuner to different impedance points on all the four
quadrants of the smith chart. Before starting the usage, the instrument along with
the cables needs to be carefully calibrated with network analyzer for the required
range of frequencies.
4.2.3
Network Analyzer
Radio Tester
The Radio Tester acts as base station for the radio module of the cellular terminal. It
enables the measurement of output power, EVM and ACLR at the antenna interface
of the radio module.
4.3
4.3.1
Measurement results
VSWR and S11
S11 parameter represents the amount of reflected power from the antenna and hence
is a measure of the quality of impedance match between the antenna and the radio
module that the antenna tuner has been able to achieve. In our setup we measure the magnitude of S11, also known as Return Loss, in dB, after the adaptive
antenna tuning process, for a range of load impedances. A Return Loss value of
less than -10 dB is considered as acceptable performance for a radio module since
this indicates a sufficiently low reflected power. S11 can be measured with Network
Analyzer at the output of radio module. Measuring the transmitted and reflected
powers from the measurement receiver (using the firmware inside the radio module)
and then subtracting the two, also yields the S11 or Return Loss. Ideally the two
values should be the same but this is rarely the case due to imperfections in the
front end module design and also due the limited dynamic range of measurement
receiver. Nevertheless it is useful to carry out the measurement in both ways. The
measurement results are given in Table 2 and 3 for 1950 MHz and in Table 4 and 5
for 897.6 MHz.
52
Table 2: S11 measured with Network Analyzer before and after tuning for 1950 MHz
VSWR Return Loss before tuning(dB) Return Loss after tuning(dB) VSWR after tuning
1:2
-9.2
-9.8
1:1.95
1:3
-5.9
-8.74
1:2.1
1:4
-4.4
-8.1
1:2.2
1:5
-3.6
-6.5
1:2.7
1:6
-2.8
-7.3
1:2.5
1:7
-2.9
-7.11
1:2.5
Table 3: S11 measured from Measurement Receiver before and after tuning for 1950
MHz
VSWR Return Loss before tuning(dB) Return Loss after tuning(dB)
1:2
-5
-22.5
1:3
-1
-21.56
1:4
0
-24.38
1:5
0
-22.75
1:6
0
-20.45
1:7
0
-19.69
4.3.2
Output power
The output power from the radio module is measured for different antenna impedances
(simulated with different VSWR values on the load-pull tuner) before and after the
adaptive antenna tuning. The measurement of output power before tuning is done
without the antenna tuner so that the insertion loss of the antenna tuner is not included. This is done to give an accurate picture of the deterioration of output power
with increasing VSWR of the antenna and the exact improvement when adaptive
tuning is used. Maximum output power is requested by the radio tester but due to
cable losses, the maximum output power measured, without the antenna tuner and
with an antenna impedance of 50 ohms, is around 22 dBm. The observations for
1950 MHz and 897.6 MHz are given in the Tables 6 and 7 respectively.
In the above cases, the magnitude of load impedance was varied but the phase
was constant so that the impedance of the antenna was always in the same quadrant
of Smith chart. Next, the measurement of output power on 1950 MHz was done for
different phase values of the load impedances while keeping the magnitude of load
impedance fixed to a VSWR of 1:7. The observations recorded are plotted in the
Figure 32.
4.3.3
53
Table 4: S11 measured with Network Analyzer before and after tuning for 897.6
MHz
VSWR Return Loss before tuning(dB) Return Loss after tuning(dB) VSWR after tuning
1:2
-9.2
-10.8
1:1.8
1:3
-5.9
-12.48
1:1.6
1:4
-4.5
-10.86
1:1.8
1:5
-3.5
-21.1
1:1.2
1:6
-2.9
-16.0
1:1.37
1:7
-2.5
-13.4
1:1.5
Table 5: S11 measured from Measurement Receiver before and after tuning for 897.6
MHz
VSWR Return Loss before tuning(dB) Return Loss after tuning(dB)
1:2
-5
-10.8
1:3
-3
-10.0
1:4
-1
-10.86
1:5
0
-12.25
1:6
0
-12.20
1:7
0
-8.725
3GPP requirements. For ACLR, the observations for 1950 MHz are shown in the
Figure 33. It is clear from the plot that there is no major change in the ACLR values
both for 5MHz and 10MHz and the values are well under the limit specified by
3GPP.
4.4
Analysis
Looking at the measurement data given above, it is obvious that antenna tuning
brings about improvement in the radiated power and return loss in impedance mis-
54
Table 6: Output power before and after tuning for 1950 MHz
VSWR Output power before tuning(dBm) Output power after tuning(dBm)
1:1
21.6
20.7
1:2
20.1
20.0
1:3
18.6
19.5
1:4
17.4
19.0
1:5
16.3
18.3
1:6
16.3
17.3
1:7
13.1
17.6
VSWR
1:1
1:2
1:3
1:4
1:5
1:6
1:7
Table 7: Output power before and after tuning for 897.6 MHz
Output power before tuning(dBm) Output power after tuning(dBm)
22.1
20.8
20.4
19.5
19.1
19.4
18.6
19.0
17.9
19.2
17.1
18.9
16.7
19.4
match conditions. However there are some peculiarities in this behavior that are
analyzed here.
One strange observation is that in high band measurements there is a large
difference in the Return Loss that is measured from the network analyzer and the
one returned from the measurement receiver after tuning. Ideally these values should
55
be almost the same as is the case in low band measurements. Also the VSWR value
measured after tuning, with network analyzer, is not all that good. A possible
explanation for this behavior is that the network analyzer measures the quality
of impedance match with respect to 50 ohm impedance. On the other hand, the
front end design of the radio module consists of several switches, filters, resistors
etc. Great effort is made to ensure that the front end is well matched to the ideal
antenna impedance of 50 ohm but at times the layout design introduces errors and
the radio module is terminated at an erroneous reactance instead of the ideal 50 ohm
load. This leads to a condition where antenna tuner system is able to match the
non ideally terminated radio platform to the previously mismatched antenna which
reflects in the improvement of output power level and in the return loss measured
by the measurement receiver but the measurement done by network analyzer still
shows a mismatch.
Analysis of measurement data highlights another important fact that the antenna
tuner module introduces an insertion loss thus reducing the power at a perfect 50
ohm match. Though the documented value of this insertion loss as given by the
manufacturer of the antenna tuner is between 0.5 and 1dB, the measurements done
as part of this thesis work show that in WCDMA low band, the insertion loss can
reach up to 2dB. So there is a definite deterioration in performance of radio module in
perfect match conditions when using the antenna tuner. However, the perfect match
condition rarely occurs not just because of the antenna usage but also because it is
very difficult to design an antenna that would provide a 50 ohm termination across
the wide range of frequencies that are commonly used in mobile handsets nowadays.
The situation can be further improved if the open loop settings are chosen such that
the antenna tuner introduces minimum insertion loss. Thereafter the closed loop can
choose a configuration that minimizes the reflected power. In this way the Antenna
tuning system will deteriorate the performance a little when the impedance match
is almost good but as shown in the results above, it will improve the performance
substantially as the matching condition worsens.
Another aspect that is visible in the above results is that the STM-Paratek
antenna tuner does not tune equally well on all the four quadrants of Smith Chart.
For example, in WCDMA high band, tuning is the best in first quadrant (centered
around 45 degrees), acceptable in second and fourth quadrant and worst in the
third quadrant. This behavior has been deliberately designed by the antenna tuner
manufacturer since according to their research, the effective antenna impedance
in common usage of a mobile device has a higher probability of falling in those
quadrants of Smith chart where the performance of tuning is better than others.
For EVM and ACLR, there is no major change in the measured values before
and after tuning in case of a mismatch. This was contrary to expectations but
analysis shows that there is an explanation for this behavior. On one hand, the
improvement in the antenna impedance match, because of tuning, should improve
the linearity of the power amplifier which in turn should improve the EVM and
the ACLR of the radio system but on the other hand, the antenna tuner itself is
non-linear device. Hence the improvement in PA linearity is probably compensated
by the deterioration in linearity due to the addition of antenna tuner hardware.
56
However, if a radio platform is designed from the beginning with an antenna tuner,
then the bias settings of the power amplifier can be substantially reduced since it
would then be unnecessary to have a large margin for the ACLR compared to the
specifications. This would result in a radio platform that can perform equally well
with a less linear power amplifier, thus saving the current consumption.
57
Conclusion
As the current trend of including more and more functionality in the mobile handsets
continues to grow, the task to fit all the necessary peripheral devices along with
the antennas of different wireless technologies in the least possible size, becomes
extremely challenging. This results in a non-optimal design of the radio module and
compromises in the position of the antenna. Also a single antenna is expected to
cover a wide range of frequency bands. Naturally, these factors lead to a degradation
in the performance of the antenna. On the other hand, the ever increasing demand
for higher data rates, robust communication and minimal power consumption, does
not leave any window for performance degradation.
The performance degradation is mainly caused by the impedance mismatch between the antenna and the radio module. Mismatch might occur due to the design,
shape and position of the antenna but the main cause is the surroundings of the
antenna and its usage. Antennas contact with human body, i.e. when the mobile
device is held between head and hand or when finger is moved around it, can change
its impedance substantially. Also, the environmental factors, i.e whether the mobile
device is held in free space or lying on a metallic surface, contribute to the change
in antenna impedance. This impedance mismatch leads to power loss as part of the
transmitted power is reflected back. Such mismatch losses are undesired as they
shorten the talk time, degrade the receiver sensitivity and cause spurious emissions
to adjacent channels.
This thesis work addresses the issue of impedance mismatch in radio sub-system
and proposes a solution in the form of adaptive or closed loop antenna tuning. The
concept of antenna tuning requires the insertion of a Pi or T shaped LC network
with tunable reactance that can be changed based on the feedback about antenna
impedance. Such an LC network with variable capacitances, based on BST technology, designed by STM-Paratek, is used in this project to develop the antenna tuning
system.
5.1
Summary
After a brief introduction to the impedance mismatch phenomenon in modern wireless devices, its causes and adverse effects, this thesis starts off with a detailed
theoretical background on the subject of impedance matching in electrical circuits.
Some of the commonly used impedance matching circuits like T or Pi shaped LC
networks, binary capacitance array network, are described with mathematical equations and figures. More details are given for the Pi-network that is used in the
implementation of antenna tuning system later on. Measurement parameters that
are used to quantify the quality of impedance match are also introduced here. Next,
a very brief description about antenna fundamentals is included. The commonly
used types of antennas in the wireless handsets nowadays, their design issues and
performance parameters are presented. The thesis then moves on to describe the
basic building blocks, both hardware and software that are used in developing an
antenna tuning system. The hardware elements include bidirectional coupler and
58
measurement receiver to measure the transmitted and reflected power from the antenna, an antenna tuner with variable capacitors controlled by HVDACs, a processor
or micro-controller to execute the tuning control algorithm and the MIPI RFFE interface between the processor and antenna tuner. Information given here is in the
context of the system developed as part of this thesis work but it is in fact quite
general and applicable to any radio module around which an antenna tuning system
is designed.
The software elements consist of a reflection coefficient tracking module that is
responsible for interfacing with the measurement receiver to measure and calculate
the reflection coefficient, an impedance calculation module that decides the new
configuration of the antenna tuner module after receiving the reflection coefficient
update, a MIPI RFFE driver module that is in charge of the physical control of the
antenna tuner module and a common control module that runs the state machine
of the adaptive antenna tuner system. The design and implementation details of
each of the above blocks are described later. The reflection coefficient tracking block
negotiates the availability of measurement receiver with the internal power control
algorithm that has a higher priority because of its timing accuracy requirement.
Once available, the measurement receiver is instructed to measure the magnitude
and phase of the transmitted and reflected power which is then used to calculate the
reflection coefficient. This information is then passed to the impedance calculation
module.
Since the algorithm that calculates the configuration of the antenna tuner for
the best possible impedance match, is the key to the performance of the adaptive
antenna tuning system, a few possibilities were studied and analyzed in this thesis.
One such algorithm is based on the equations of a Pi-network with three variable
capacitors. This approach is very efficient since it finds the new configuration of
antenna tuner in only a few iterations but it demands highly accurate measurement
of reflection coefficient and is not flexible with respect to the changing track lengths
and PCB traces that are difficult to avoid in a real life hardware design. Another
algorithm that was studied in this thesis is based on gradient control. There are two
possible approaches that can be followed in this algorithm. One approach requires
characterization of the behavior of the three tunable capacitors with respect to
the impedance match, calculating the coefficients of their respective gradients and
storing them in permanent memory. The other approach tries to avoid the extensive
measurement session in the lab by calculating the gradient adaptively during the
tuning process. Both approaches are quite flexible and are more and less guaranteed
to converge to a good match but there is a risk of running into a local minima. The
third algorithm studied in this thesis is a version of adaptive local search algorithm
also known as Hill Climbing. The idea followed in this case is to take small steps
in the three dimensional coordinate space and move to the point that gives the
best results compared to the other options. The simplicity and flexibility offered
by this approach is immense. It is also very accommodating for the inaccuracies
of measurement system and faults in the radio module design. Downside is longer
settling time and high probability of converging to a local minima.
The Common control block is responsible for deciding whether the antenna tun-
59
ing system works in Open loop, Closed loop or Broken loop mode and for controlling
the transitions between these modes. The decision is based on the power level, the
extent of impedance mismatch and the accuracy of measurement receiver results. At
the starting point, the antenna tuner is always configured on the basis of Open Loop
settings relevant for the operational frequency, retrieved from the permanent memory. If the power level is high enough and the impedance mismatch as determined
from the reflection coefficient measurement, is worse than a benchmark, transition
is made to Closed Loop i.e. the antenna tuner configuration is changed adaptively
until a satisfactory impedance matching is obtained. Broken Loop mode comes into
effect if the accuracy of measurement receiver results is unacceptable so that the
antenna tuner configuration is not changed any more until the measurement results
are reliable.
After describing the design of the software modules, the thesis then goes on
to discuss the implementation intricacies of antenna tuner system in different RATs
with respect to their timing constraints and impact on other blocks of the transceiver
chain. The steps taken to mitigate the possible adverse effects of introducing antenna tuning on the functionality of RATs are also discussed. Also included here is
a description of the performance degradation scenarios that are associated with the
use of antenna tuner. Examples of such scenarios are loss in the output power in a
perfect match due to the insertion loss introduced by the antenna tuner hardware,
degradation in EVM and ACLR of the signal due to the non-linear effects of the
capacitors in antenna tuner and the possibility of converging to an incorrect configuration due to inaccurate results from the measurement receiver. None of these
effects can be completely avoided or compensated but an effort is made in this thesis
work to measure these parameters carefully and take into account when analyzing
the results.
During the actual implementation of the antenna tuner system, several limitations came up due to the performance of the radio module available at hand. It
was realized during the characterization of measurement receiver, that the phase
measurements were too noisy i.e. there was a lot of variation between two consecutive measurements of phase values of transmitted and reflected powers. This lack
of reliability of phase measurements forced the development of the antenna tuner
system based on just the magnitude of transmitted and reflected power. Hence Hill
Climbing algorithm was chosen for this thesis work. In addition, a VSWR of 2 to 7
and WCDMA band 1 (mid-channel) and 8 (mid-channel) were chosen to carry out
the measurements in order to complete the project in a realistic time-frame. Thereafter, the lab setup for carrying out the measurements is described. Load-pull tuner
is used to simulate the antenna impedance and Network Analyzer is used to measure
the S11 parameters after the antenna tuner has converged to a configuration that is
optimum for the given mismatch while a Radio tester is used to measure the output signal for power, EVM and ACLR. The measurement results that presented in
the thesis later show a clear improvement in output power and PA efficiency when
adaptive antenna tuning is used. This improvement gets more and more substantial
as the impedance match worsens. However, there was not much effect on the EVM
and ACLR measured values with or without antenna tuning.
60
5.2
Discussion
The idea of adaptive antenna tuning is now gaining momentum with most mobile
phone and chipset manufacturers. The availability of tunable reactive components
that can be integrated in electronic circuits along with better accuracy in the measurements done by bidirectional couplers, have cleared the way for introducing this
technology at commercial scale. As shown by the results of this thesis work, the
antenna tuner does introduce an insertion loss so the performance deteriorates by
a small measure in perfect match case but in most other cases when the antenna
impedance varies and mismatch increases, the antenna tuner improves the performance more than it deteriorates. The most obvious improvement in performance
is in the form of output or Total Radiated Power. The improvement in output
power has a direct relationship with PA efficiency and is reflected in lower current
consumption by the power amplifier which in turn results in longer talk time. Moreover, the remarkable reduction in amount of reflected power with antenna tuning,
protects the front end components from being damaged thus removing the necessity of an isolator. The noise floor of the receiver also improves as standing wave
magnitude gets reduced thus enhancing its Total Isotropic Sensitivity. In addition,
the design of front end module including the antenna can be much more relaxed as
the stringent requirements of preserving a reasonable impedance match over a large
range of frequencies is no longer required.
5.3
Future Work
61
Compromise between Tx and Rx: This thesis work has concentrated on getting
optimal adaptive antenna tuning performance for radio transmission. Hence
the closed loop tuning algorithm works on transmitting frequency. But in
WCDMA where transmission and reception are active at the same time on
different frequencies, the antenna tuner configuration that is best for transmission might not be the best for reception especially in case of bands where
the two frequencies are separated by a large gap (WCDMA band 4). This
aspect should be tested and a compromise on antenna tuning configuration
should be found if the reception performance is found unsatisfactory.
MIMO and Carrier Aggregation: Multiple antenna techniques like MIMO
(Multiple Input Multiple Output) and carrier aggregation are going to be
a part of commercial handsets soon. Integrating antenna tuner with multiple
antennas would be rather challenging as the system will become quite complex. It needs to be studied if it would be better to use separate antenna
tuners for different antennas or to implement antenna tuner system for only
the main antenna or to design the front end module in such a way that the antenna tuner can be dynamically switched between the antennas with the help
of switches. Each of these options has its own advantages and disadvantages.
So a systematic study is required to find a good solution.
62
References
[1] M. Feng, Shyh-Chiang Shen, D.C. Caruth and J.J. Huang, Device Technologies
for RF Front-End Circuits in Next-Generation Wireless Communications, in
Proc. IEEE, Vol. 92, No. 2, pp. 354-375, February 2004.
[2] Andre van Bezooijen, Reza Mahmoudi and Arthur H. M. van Roermund, Adaptive Methods to Preserve Power Amplifier Linearity Under Antenna Mismatch
Conditions, IEEE Transactions on Circuits and Systems, Vol. 52, No. 10, pp.
2101-2108, October 2005.
[3] Qizheng Gu, RF System Design of Transceivers for Wireless Communications,
Springer, New York, 2005.
[4] Jesus de Mingo, Antonio Valdovinos and Alfredo Crespo, Input Impedance Antenna Automatic Matching System, in Proc. IEEE International Symposium
Personal, Indoor and Mobile Radio Communications, pp. 1872-1876, September 2002.
[5] Fanfan Meng, Andre van Bezooijen, and Reza Mahmoudi, A Mismatch Detector for Adaptive Antenna Impedance Matching, in Proc. European Microwave
Conference, pp. 1457-1460, September 2006.
[6] Papapolymerou. J., Lange. K., Goldsmith. C., Malczewski. A. and Kleber. J.,
Reconfigurable double-stub tuners using MEMS switches for intelligent RF front
ends, IEEE transactions on Microwave Theory and Techniques, Vol. 51, No. 1,
pp. 271-278, January 2003.
[7] Amro M. Elshurafa and Ezz I. El-Masry, Tunable Matching Networks for future MEMS based transceivers, in Proc. International Symposium Circuits and
Systems, Vol. 4, pp. 457-460, May 2004.
[8] L.Y. Vicki Chen, Roger Forse, D. Chase, and Robert A. York, Analog Tunable
Matching Network Using Integrated Thin-Film BST Capacitors, IEEE International Microwave Symposium Digest, Vol. 1, pp. 261-264, June 2004.
[9] Peregrine Semiconductor, DuNE Digitally Tunable Capacitors, June 2012, available http://www.psemi.com/pdf/sell_sheet-psg/73-0039.pdf.
[10] Luyi Liu and Richard Langley, Electrically small antenna tuning techniques, in
Proc. Antennas and Propagation Conference, pp. 313-316, November 2009.
[11] Andre van Bezooijen, Maurice A. de Jongh, Freek van Straten, Reza Mahmoudi
and Arthur H. M. van Roermund, Adaptive Impedance-Matching Techniques for
Controlling L Networks, IEEE Transactions on Circuits and Systems, Vol. 57,
No. 2, pp. 495-505, February 2010.
63
[12] Ronald Davidson, Transfer Unit for Radio Frequency Signals and Method for
Alternatively Using an Electrical Antenna or a Magnetic Antenna with a Classic
Antenna Tuner, United States Patent Application 20110273354, http://www.
freepatentsonline.com/y2011/0273354.html.
[13] Andre van Bezooijen, Maurice A. de Jongh, Christophe Chanlo, Lennart C. H.
Ruijs, Freek van Straten, Reza Mahmoudi and Arthur H. M. van Roermund,
A GSM/EDGE/WCDMA Adaptive Series-LC Matching Network Using RFMEMS Switches, IEEE Journal of Solid-State Circuits, Vol. 43, No. 10, pp.
2259-2268, October 2008.
[14] Constantine A. Balanis, Antenna Theory: Analysis and Design, 3rd edition,
Wiley, 2005.
[15] Ahmed A. Kishk, Fundamentals of Antenna, Mc Graw Hill, 2009.
[16] Agilent Technologies, Directional Coupler Overview, http://www.home.
agilent.com/upload/cmc_upload/All/Direct_CouplerOverview.pdf.
[17] MIPI Alliance, MIPI RF Front end specifications, http://www.mipi.org/
specifications/rf-front-end.
[18] Qizheng Gu, Javier R. De Luis, Arthur S. Morris and Jeff Hilbert, An Analytical
Algorithm for Pi-Network Impedance Tuners, IEEE Transactions on Circuits
and Systems I, Vol. 58, No. 12, pp. 2894-2905, December 2011.
[19] Zelda B. Zabinsky, Random Search Algorithms, in Wiley Encyclopedia of Operations Research and Management Science, John Wiley and Sons, 2011.
[20] Koichi Ogawa, Tsukasa Takahashi, Yoshio Koyanagi and Koichi Ito, Automatic
Impedance Matching of an Active Helical Antenna Near a Human Operator, in
Proc. European Microwave Conference, Vol. 3, pp. 1271-1274, October 2003.
[21] R. J. M. Vaessens, E. H. L. Aarts and J. K. Lenstra, A local search template,
Computers and Operations Research, Vol. 25, No. 11, pp. 969979, November
1998.
[22] Sean Luke, Essentials of Metaheuristics, A Set of Undergraduate Lecture Notes,
Department of Computer Science, George Mason University, February 2012.
[23] Daigo Muramatsu, Online Signature Verification Algorithm Using HillClimbing Method, in Proc. IEEE International Conference on Embedded and
Ubiquitous Computing, Vol. 2, pp. 133-138, December 2008.
[24] Francisco Javier Ovalle Martnez, Julio Solano Gonzalez and Ivan Stojmenovic, A parallel hill climbing algorithm for pushing dependent data in clientsproviders-servers systems, in Proc. International Symposium on Computers and
Communications, pp. 611-616, November 2002.
64
[25] Renjeng Su and Noureddine Kermiche, A learning scheme for open-loop and
closed-loop control, in Proc. IEEE International Symposium on Intelligent Control, pp. 523-528, August 1988.
[26] Shen Jianhe, Chen Shuhui and Cai Jianping, An Open-plus-closed-loop Control
for chaotic Mathieu-Duffing Oscillator, in Proc. Chinese Control Conference,
pp. 492-496, July 2007.
[27] Robert M
ullner, Carsten F. Ball, Kolio Ivanov, Johann Lienhart and Peter
Hric, Contrasting Open-Loop and Closed-Loop Power Control Performance in
UTRAN LTE Uplink by UE Trace Analysis, in Proc. IEEE International Conference on Communications, pp. 1-6, June 2009.
[28] Jin Lu and Lyndon J. Brown, A combination of open and closed-loop control for
disturbance rejection, in Proc. Canadian Conference on Electrical and Computer
Engineering, pp. 563-568, May 2011.
[29] Howard T. Russell, A Loop-Breaking Method for the Analysis and Simulation
of Feedback Amplifiers, IEEE Transactions on Circuits and Systems I, Fundamental Theory and Applications, Vol.49, No. 8, pp. 1045-1061, August 2002.
[30] Rob Kopmeiners, Time Accurate Control of RF over a DIGRF Interface, United
States Patent Application 20100074311, http://www.freepatentsonline.
com/y2010/0074311.html.
[31] Johan Bergman, M
arten Ericson, Dirk Gerstenberger, Bo Goransson, Janne
Peisa and Stefan Wager, HSPA Evolution Boosting the performance of mobile
broadband access, Ericsson Review, No 1, 2008.
[32] Ranjit Gharpurey, Managing Linearity in Radio Front-Ends, in Proc. IEEE
Custom Integrated Circuits Conference, pp. 1-8, September 2011.
[33] Ichirou Ida, Jun-ichi Takada, Takeshi Toda and Yasuyuki Oish, An Adaptive
Impedance Matching System and its application to Mobile Antennas, in Proc.
IEEE Region 10 Conference (TENCON), Vol. 3, pp. 543-546, November 2004.
[34] Jesus de Mingo, Antonio Valdovinos, Alfredo Crespo, Denis Navarro and
Paloma Garcia, An RF Electronically Controlled Impedance Tuning Network
Design and Its Application to an Antenna Input Impedance Automatic Matching System, IEEE Transactions on Microwave Theory and Techniques, Vol. 52,
No. 2, pp. 489-497, February 2004.
[35] Luke Sankey and Zoya Popovic, Adaptive Tuning for Handheld Transmitters,
International Microwave Symposium Digest, pp. 225-228, June 2009.