WHY THIS COURSE?
Designing electronic circuits in the tens and hundreds of MHz range can
be a challenge because the presence of parasitics presents a lot of problems
in the physical circuits. This makes designing highfrequency circuits a rather
specialised subject, although much can still be resolved under the lumped
circuit assumption. But as the frequency moves up to GHz range, we have
serious trouble using lumped circuit models because voltage and current
change within the physical boundary of the circuit as a result of the
wavelength being comparable to the dimension of the physical circuits.
We therefore have to use a different approach to look at the problem.
In this course we will look mainly at circuit design in the tenshundreds
MHz range and will touch upon some basics for the GHz range design.
WHAT YOU WILL LEARN (Some philosophical notes about this course):
 Essential overview of analog electronics (3 weeks):
My experience of doing this course is that
students came from various backgrounds due to the many possible combinations of
study tracks that students took in previous years.
Sometimes, students claimed they hadn't
learnt this and that. So, I will assume that everyone knows just a little basics
about circuits, but not anything advanced. I will spend two to three weeks to
go over the essential concepts of analog circuits including devices, amplifier
configurations, feedback, etc. This will be like a compressed course of
all electronics fundamentals up to EC2 and ADIC. Then, you will have no excuse
of not knowing what feedback is, how driving impedance can be deduced, or
why Miller can reduce gain, etc. Moreover, in doing this revision, I will
put emphasis on highfrequency effects so that you will appreciate more
easily the problems to be studied in the later part of the course.
 Radio frequency circuit techniques (2 weeks):
This part of the course introduces you to the basic elements of desiging
circuits at highfrequency range. I will emphasize conceptual understanding
of the problems of highfrequency rolloff that limits the operation of
amplifiers at high frequencies. An important skill to acquire here is to
identify the vulnerable parts of a given circuit that can lead to rolloff.
With this knowledge, we can devise methods to combat it, and we will
systematically study a few popular amplifier configurations for highfrequency
applications.
 Highfrequency filter design (2 weeks):
This is a difficult part, because you have never done anything seriously
about filter synthesis and design. Obviously this would have been
due to a flaw in our
curriculum design. But as I have to take care of it anyway, I have to make
sure you will get the salient design concepts necessary for highfrequency
filter design. So, I will try to summarize the whole filter technology
background in a compact manner, up to a point where we can start looking
at highfrequency filter design. Essentially, our de facto starting
point is the inadequacy of operational amplifiers in handling highfrequency
signals. This leads to the introduction of a "new" element
(in the sense of usage rather than history of existence)
known as operational transconductance
amplifier or the OTA. Sometimes, we just call it transconductance
or even simply Gm. Using Gm as the basic amplifier block, we can design
filters in the hundreds of MHz range. We will study design examples
using signal flow graphs as tools and examine a few BJT and MOS OTAs.
 Impedance matching (2 weeks):
In highfrequency circuit design, an important procedure that must be
incorporated in every design is matching. The purpose of matching
is to prevent signals from being reflected as they move
along the various parts of a circuit or system. This in turn ensures
maximum power transfer from the signal source to the load. Matching also
has an important role to play in the stability of circuits. In this part
of the course, we focus ourselves on a few popular matching methods.
Essentially, we will derive, from very basic circuit theory,
some simple matching circuits that can achieve impedance matching for
a narrow bandwidth. In particular we will take a look at Lcircuits,
Tcircuits, picircuits, tapped capacitor circuits, etc.,
using a simple design approach employing the concept of Qfactor.
I will also briefly talk about the doubletuned circuit for wider band
matching.
 Transmission line matching (2 weeks):
At high frequencies (up to microwave range), transmission lines are
not just "nothing" as in lumped circuit analysis. They can affect power
transfer and even stability because they can reflect signals back and
create standing wave patterns. I will explain the concepts
of travelling waves in transmission lines and interpret the famous
Telegrapher's equations. What is important here is to understand
the various effects a transmission line has on the signal transmission.
I will use the Smith chart as a tool to make things appear
very intuitive so that you can easily visualize the many effects of a
transmission line. Then, the matching method can be easily
derived, of course with the help of the Smith chart. I
will try to cover some simple matching examples using the transmission line
itself as a matching element! This is called stub matching.
 Power amplifier design (2 weeks):
Designing power amplifiers at high frequencies has a rather different
set of criteria, which are developed to combat the easily unstable devices.
Engineers usually call it oscillation. Then, why is the amplifier so
easily become unstable or oscillatory? The problem is the internal
parasitics that form internal feedback paths. We know from EC1 that
whenever there is a roundtrip gain of just higher than 1, the circuit
oscillates. To study this problem, we begin with an appropriate
characteristionthe scattering parameters. I will
explain what scattering parameters are. Their physical meanings are
important to understanding the many aspects of highfrequency design.
With
scattering parameters, I will explain some concepts of power
gains in amplifiers and once again the matching problem in order to
maximize power gain. We will see later that what is important really
is the socalled transducer power gain because it measures how
much power that can be used relative to how much power that is available.
Our final task is to examine in detail the problem of
stability. I will explain, from the basic stability requirements,
how stability conditions can be visualized on the Smith chart and be
checked mathematically. To finish off, I will examine one way (out of
the many possibilities) of making an unstable amplifier stable by
neutralizing (cancelling) the internal feedback.
LECTURE AND TUTORIAL SCHEDULES:
This year, the weekly lecture and tutorial are scheduled on every
Friday evening. This is an unfortunate arrangment, I would say, because
three hours of continuous concentration is not practical. To make the
best use of the time, I will spend the first 2 hours every Friday
evening to present the lecture material, and will use the last hour
as tutorial where I will answer specific questions and discuss
homework problems. As indicated above, the teaching schedule is
roughly as follows:
Dates 
Topics  Notes
 9 September  30 September 
Overview of analog electronics

printed notes to be handed out 
7 October & 14 October  Radio frequency circuit techniques 
printed notes to be handed out 
21 October  Lecture and tutorial cancelled. 
28 October & 4 November  Highfrequency filter design 
(660KB)
printed notes to be handed out 
4 November & 11 November  Impedance matching 
(1MB)
printed notes to be handed out 
11 November & 18 November  Transmission line matching 
(773KB)
printed notes to be handed out 
25 November  Power amplifier design 
(1.6MB)
printed notes to be handed out 
2 December  Revision  
ASSIGNMENTS AND MINIPROJECTS:
From time to time, I will hand out problem sets. You must try to work
out solutions all on your own. During the tutorials, I will explain some
of the problems related to the assignments. Remember assignments do count
towards your continual assessments.
Assignment 1:
Please submit solution to Question 2 of Problem Set 1
by 7 October 2005.
Click this link to see the solution for this assignment.
Assignment 2:
Please submit solution to Question 6 of Problem Set 2
by 28 October 2005.
Click this link to see the solution for this assignment.
MidSemester Test:
The midterm test will be composed of only ONE problem: Question 3 of Problem
Set 3. This forms a "takehome test" to be handed in
on 11 November 2005.
That means each of you will have to reproduce the answer
all on your own. Your submission will be counted as midsemester test.
HINT: The following formula gives the drain current of MOSFET Q1 and Q2:
i_{D} = [ μ_{n} C_{ox} /2 ] [ W/L ] V_{eff}^{2}
where V_{eff} = v_{gs}  V_{TH}. This formula is true for MOSFET Q1 and Q2. But for Q3 and Q4, they are in triode
region. Then, the total current that flows through Q3 and Q4 is given by
2 i_{D3} = 2 μ_{n} C_{ox}
(W/L)_{3} [ V_{eff2}v_{ds3}  v_{ds3}^{2} / 2 ]
where we note that V_{eff3} = V_{eff2}.
Also, you should know that this total current through Q3 and Q4 plus the
current in Q2 must be equal to I_{1}. So, you should be able to
get a total of three equations containing the unknowns V_{eff1},
V_{eff2} and v_{ds3}.
In part (b), you may need to solve this set of equations, requiring numerical
solution procedures. To save your time, you may assume that you already know v_{ds3} = 0.0015648 V for the case v_{1} = 2.5 mV, and you may also assume that you know
v_{ds3} = 0.18297 V for the case v_{1} = 250 mV.
In part (c), you may take v_{ds3} = 0.0015648 V for the case v_{1} = 2.5 mV, and you may also take v_{ds3} = 0.15625 V for the case v_{1} = 250 mV.
Click this link to see the solution for this test.

Assignment 4:
Please submit solution to Question 5 of Problem Set 4
by 25 November 2005.
See lecture notes for matching procedure.

Miniproject topics and requirements
:
Miniprojects will be done in lieu of laboratory works. There are a few
reasons for this. First, many design approaches can be computerised and it
is an extremely good exercise for you to develop computer software to aid
design. Second, our laboratory is not (at the moment) well geared towards
highfrequency experiments. Third, because our course emphasizes conceptual
understanding, miniprojects are good tools to stretch your thinking.
The stipulated practical work for this subject is 9 hours of work related to
topics taught in the course. I believe it is reasonable to ask you to develop
a complete and fully working software to aid
the design of high frequency circuits. The 9 hour schedule is the main
constraint, which means that I can only specify relatively simple projects.
As more topics will be covered as we go along, I suggest that you do a quick
preview of the lecture materials to find out if you would like to pursue
your miniproject on a topic which will be taught at the later part of the
semester. The following topics are arranged in the sequence they are taught.
You may choose any ONE of these topics.
Topic 1: High frequency rolloff of transistor amplifiers
You are required to develop a complete computer software which can generate
all poles and zeros, and complete frequency response of the commonemitter
amplifier. Your software should input sufficient number of parameters and
produce complete list of results. Graphical results would be desirable in
the case of frequency responses. A clearly written and complete user manual
and a CDROM (or floppy diskette) containing the software will be required in the final submission.
Topic 2: Design of matching circuits
You are required to develop a complete software which can generate all
the circuit component values for the following types of matching circuits:
Lcircuit, picircuit, Tcircuit and tapped capacitor circuit. Your
software should input sufficient number of parameters (e.g., type of matching
circuits, Q factor, resistances to be matched, etc.) and produce complete
list of results. A clearly written and complete user manual
and a CDROM (or floppy diskette) containing the software will be required in the final submission.
Topic 3: Transmission line matching
You are required to develop a complete software which can generate the
information (e.g., position of inserting stub, length of stub, type of stub,
etc.) required to match a given load to a transmission of any characteristic
impedance. Incorporating the Smith chart
geometry is preferred over the use of firstprinciple equations. A clearly
written and complete user manual and a CDROM (or floppy diskette) containing the software
will be required in the final submission.
Note that this is NOT a group project. Each one of you is required to
submit a report outlining the objective and your approach (together with a
CDROM (or floppy diskette) of the software and a complete user manual).
The deadline of submission is
the day of my last lecture, 2 December 2005.
I will test your software and verify its functionality, and I will also
do a simple "independency check". If two or more of you are found to have
submitted essentially the same software, you will get zero mark for this
component.
ASSESSMENT:
There will be a midsemester test
for the purpose of assessment. But it
may be more appropriate to do a takehome test which consists of a difficult
design problem. This will save a week of lecture time.
The written examination will consist of a 2.5 hour paper. I will set
questions in a way that you will not be able to avoid knowing the basic
principles. I will put the basic things in one or two compulsory questions,
but will still allow choices so that students with
different interests and abilities will be able to pass the examination.
Altogether, the midsemester test, miniprojects and assignments will
account for 35% of the final marks, and the written examination will account for
65% of the final marks.
Let me stress that I never scale marks!
What you get is what you get! But you can be assured that my exam paper
is always
set to test what you know rather than to test what you don't know. So, you
should find the paper very answerable if you have followed my course;
in other words, I accept no excuse of failure.
STUDENT CONSULTATION HOURS:
I am usually available for consultation any time I don't have a class
or meeting. My timetable is posted outside my office. But since our class
is on every Friday, it seems to be convenient if
you can see me (if you have a question about your study) in the
afternoon before the lecture starts. This will allow me to replicate
interesting problems to the whole class, if it is appropriate to do so.
Also, Friday afternoons will be more convenient for parttimers as they
may just drop in a bit earlier in the same evening.
Preferred consultation time: Friday 3pm to 6 pm
SUPPLEMENTARY MATERIALS:
 Smith chart
 Semilog graph paper
FINAL ADVICE:
Learn with only your heart!
REFERENCES:
 Kenneth R. Laker and Willy M.C. Sansen, Design of Analog Integrated
Circuits and Systems, New York: McGraw Hill, 1994.
 Paul R. Gray, Paul J. Hurst, Stephen H. Lewis and Robert G. Meyer,
Analysis and Design of Analog Integrated Circuits, New York:
Wiley, 2001.
 W. Alan Davis and Krishna K. Agarwal, Radio Frequency Circuit
Design, New York: Wiley, 2001.
 Herbert L. Krauss, Charles W. Bostian and Frederick H. Raab,
Solid State Radio Engineering, New York: Wiley, 1980.
Michael Tse, 1 September 2005
