MSc Electronic and Computer Engineering

Digital Design

1 Introduction to the module

This module looks at the techniques and challenges involved in the design of modern large scale
digital systems, for example the special purpose chips used in smartphones and tablets. These are
complex systems that may involve many millions of logic gates. These designs often involve the
integration of many different digital sub-systems, each of these which may be designed by a different
company with differing expertise and capability. The intellectual property from these various
suppliers is merged into a single system-on-chip.

To support this integration from diverse suppliers, standard design representations are needed, and
also standard ways to package intellectual property for re-use. Hardware description languages
provide the standard methodology for doing this. There are two widely used hardware description
languages (HDLs): VHDL and Verilog. HDLs also have the advantage that they allow designers to
write high level descriptions of digital systems which can then be automatically translated into
detailed logic design in an automatic process called logic synthesis. Synthesis provides an enormous
boost to designer productivity, and without it modern design involving hundreds of millions of logic
gates would not be feasible.

The type of system-on-chip used in smartphones is manufactured with fixed hardware function. Any
adaptation is purely through software, which can be slow. Also, fixed hardware function is only
economic if produced in large volumes. Programmable hardware technologies, such as FPGA, are
preferrable if the production volume is low. Very complex FPGAs are increasingly finding
applications in problem domains where the function of the hardware must frequently change to give
maximum performance on a dynamically changing problem, such as large data processing centres and
self-driving cars.

This module will introduce the skills necessary to be successful in modern digital system design. We
will start by looking at manual design techniques and seeing the fine detail of what happens in
systems constructed from logic gates and flip-flops. Then we will move on to look at the VHDL
language and how it synthesises to hardware. Finally, we will look at the properties of the components
of system on chip designs involving microprocessors and hardware accelerators, and the challenges
involved in verifying large scale systems

1
2 Basic Ideas of Combinational Logic
2.1 Types of Digital System
There are two main types of logic system:
 Combinational logic produces an output that is determined only by the state of the input at
that point in time. If we know what the present input is, then we know immediately what the
present output must be. An example is a standard up/down light switch.
 Sequential logic produces an output that is determined not only by the present input but also
by the device’s past history. An example would be a pull-cord light switch in a bathroom. If
we pull the cord (the input) when the light is off, then the light (the output) will turn on. If we
pull the cord when the light is on, then the light will turn off.
There are two types of activity we will need to know how to do:
 Analysis means we are given a circuit design and we must figure out what it would do and
how it would behave
 Synthesis means we are given a required function and we have to design the circuit that will
fulfil this function

Boolean algebra is a mathematical description of systems built from components that can take 2
possible signal values. Mastery of Boolean algebra enables us to take a requirement for the system
behaviour and translate it into a circuit design. It also gives us ways to ensure that our circuit designs
are efficient. Boolean algebra was originally developed to deal with problems in logic that considered
two-value propositions that are TRUE or FALSE, and only later became used for digital electronics.
Because of this basis in Boolean algebra, digital electronic systems are often referred to as logic
systems.

2.2 Our basic building blocks: logic gates

The basic building blocks for digital circuits are logic gates. In this session we will introduce the most
important logic gates and start to develop the algebra that will be used to represent the circuits that we
will design.

2.2.1 The AND gate

As an example, suppose we are designing an alarm system for our home. We have a switch that can
detect when the window is open. We also have a switch that enables the alarm system. When we go
out of the house we will switch the alarm on so that it will raise an alert if the window is open.
We represent this as two input signals:
A) Window is open? 1=yes 0=no
B) Alarm is enabled? 1=yes 0=no
We have one output signal:
C) Sound the alarm? 1=yes 0=no
The behaviour we want is
(Sound the alarm) = (Window is open) AND (Alarm is enabled)

One way we describe the behaviour of a logic system is a truth table. This is a list of all the possible
combinations of input condition and the corresponding condition that we require at the output. For our
simple alarm it looks like this:

2
 Truth table Symbol Boolean algebra
AND
A B C
0 0 0
A
0 1 0 C C = A.B
1 0 0 B
1 1 1

This function is called an AND function. The AND gate can have any number of inputs, but the
example shown above has 2 inputs. The AND gate outputs a 1 when all of its inputs are 1. This
operation is sometimes referred to as the logical product of A and B, because the 2-input AND gate is
very similar to a 1-bit multiplication

2.2.2 The OR gate

Now we want to extend our system to check for the upstairs window being open as well as the
downstairs. That gives us two input signals:
A) Upstairs window is open? 1=yes 0=no
B) Downstairs window is open? 1=yes 0=no
We have one output signal:
B) A window is open? 1=yes 0=no
The behaviour required is

(A window is open) = (Upstairs window open) OR (Downstairs window open)

For this case, the truth table, symbol and algebraic formula are:

Truth table Symbol Boolean algebra

OR
A B C
0 0 0 A
0 1 1 C C = A+B
1 0 1 B
1 1 1

This function is called an OR function. It outputs a 1 when any of its inputs is 1. A OR B is normally
written as A+B. This operation is sometimes referred to as the logical sum of A and B.

2.2.3 Inclusive OR, exclusive OR

The word “or” in the English language has two meanings
 Inclusive OR: one or the other or both
 Exclusive OR: either one or the other but not both
Normally in the English language we just figure out from the context whether the inclusive or
exclusive version is required. A mathematical representation requires greater precision and we regard
these as two completely different logical operations. The truth tables for these different functions are:

3
 OR C=A+B XOR C=A⊕B
A B C A B C
0 0 0 0 0 0
0 1 1 0 1 1
1 0 1 1 0 1
1 1 1 1 1 0
Inclusive OR Exclusive OR

Our OR gate implements the inclusive OR function. There is a logic gate that implements the XOR
function, but it is less fundamental than the OR gate and we’ll leave its consideration until later.

2.2.4 The Inverter

Sometimes we need to reverse the value of a logic signal. Suppose we want to generate a signal that is
1 if the window is open. But we the only switches that we have available out a 1 when the window is
closed.

We can solve this with a device that has an input:

A) Window is closed? 1=yes 0=no
and an output:
B) Window is open? 1=yes 0=no

The behaviour required is

(Window is open) = NOT (Window is closed)
Its truth table, logic symbol and Boolean algebra representation are:

Truth table Symbol Boolean algebra

NOT
A B B=A
0 1
A B
1 0

The device is called an inverter (sometimes known as a NOT gate).

2.3 IEEE symbols

An alternative set of symbols for AND, OR and inverter are the IEEE symbols, shown below. Both
sets of symbols are widely used, but in this module we will use the traditional symbols.

The Traditional Symbol THE IEEE/IEC Symbol

The AND gate.

The OR gate.

The INVERTER

4
2.4 Putting it all together
Now we’ll design our alarm system.

Alarm enabled?
Upstairs window closed? Sound the alarm
Downstairs window closed?

The interface with the real world will be three input switches:
A) Alarm is enabled? 1=yes 0=no
B) Upstairs window is closed? 1=yes 0=no
C) Downstairs window closed? 1=yes 0=no
We have one output signal:
D) Sound the alarm? 1=yes 0=no

The required behaviour is that the alarm should be sounded if (the alarm is enabled and (the upstairs
window is not closed or the downstairs window is not closed))

It should be easy to see that a circuit that will accomplish this function is:

A
D
B
C

2.4.1 Inversion bubbles

The small circle at the output of an inverter is interpreted as a NOT operation and is called an
inversion bubble. We can use this in other locations in our diagrams to indicate the inversion of a
signal. For example, our alarm circuit can be represented like this:

A
D
B
C

The two different representations that we have seen are identical in their meaning.

5
2.4.2 Circuit analysis: figuring out what a circuit does
We have come up with a design for our burglar alarm circuit.

A
X D
B
C Z
Y

Now let’s analyse the circuit to make sure that it behaves in the way we expect. To help us with the
analysis we have given names to the additional circuit nodes X, Y, Z.

To see how the circuit responds to inputs, we could give it a set of input values, e.g. A=1, B=0, C=0
then work forward from the input to the output evaluating X, Y, Z, D. Evaluating the output of the
inverters tells us that X=1 and Y=1. Evaluating the OR gate’s response to these values of X and Y
tells us that Z=1. Finally we evaluate the AND gate’s response to the input Z=1, A=1 tells us that
D=1. So this set of values at the input will cause the alarm to sound.

2.4.3 Truth table for the circuit

A more thorough way to do this is to draw up the truth table. There are 3 inputs to the system, so there
are 23=8 possible input conditions.

A B C X Y Z D
0 0 0
0 0 1
0 1 0
0 1 1
1 0 0
1 0 1
1 1 0
1 1 1

We then move from the inputs toward the output, evaluating the columns. We start with X, which is
the inverse of B, and Y, which is the inverse of C. Now we can fill in the column for Z: it is 1 in any
row where we see X=1 or Y=1. Finally D is 1 in the rows where both A=1 and Z=1

A B C X Y Z D
0 0 0 1 1 1 0
0 0 1 1 0 1 0
0 1 0 0 1 1 0
0 1 1 0 0 0 0
1 0 0 1 1 1 1
1 0 1 1 0 1 1
1 1 0 0 1 1 1
1 1 1 0 0 0 0

This final truth table lists every possible input condition, and tells us what the output D would do in
each case.

6
2.4.4 Algebraic description of the circuit
To build up a Boolean logic representation of this circuit, we proceed as follows. The inverters give us
X=B
Y=C

The OR gate gives us

Z=X+Y=B+C

The AND gate gives us

D=A.(X+Y)=A.(B+C)

We use brackets, for example (X+Y) to show that we need first to perform the operation X OR Y and
then take the result of this operation and AND it with the value of A.

If we then want to evaluate the result for a particular set of inputs (say A=1, B=0, C=0) we substitute
these values into the equation.

D=A.(B+C)
D=1.(0+0)
D=1.(1+1)
D=1.1
D=1

2.5 Summary
Combinational logic circuits have an output value that is entirely determined by their inputs at that
instant of time. Boolean algebra gives us a mathematical foundation for representing and manipulating
digital circuits. The basic building block for logic circuits are logic gates. The AND gate, the OR gate
and the inverter are sufficient to construct any function that can be represented as a Boolean equation.

You should now know...

 The behaviour of the AND gate, OR gate and inverter
 The Boolean logic representation of the AND, OR and NOT function
 How to convert a logic circuit into a Boolean algebra equation
 How to convert a Boolean algebra equation into a logic cicruit
 How to analyse a circuit by constructing its truth table
 How to analyse a circuit using Boolean algebra

7
3 The Axioms of Boolean Algebra
So far, we have looked at simple examples. To tackle more complicated situations, we will need to
understand Boolean algebra. In this lecture we will look at some of the main rules of Booelan algebra.
We then put these rules to work in the next lecture to look at some examples of circuit simplification.

3.1 Relationship of Boolean Constants

A Boolean constant is a constant that can take a value of 0 or 1. We have already seen how the
fundamental operations of AND, OR and NOT operate on Boolean constants:

AND OR Inverter
0.0=0 0+0=0
0.1=0 0+1=1 0=1
1.0=0 1+0=1 1=0
1.1=1 1+1=1

3.2 AND and OR Functions of a Boolean variables

We continue our survey of Boolean algebra by looking at how our logic gates perform when we have
at least one input that is variable (free to change between 0 and 1) and other inputs are constant.

Suppose we have a digital signal (making transitions between 1 and 0 as time goes by) and we apply it
to the A input of an AND gate. The output C is 1 when both inputs are 1, and 0 at all other times.
Suppose the B input is fixed at 1 and will never change.

A=
C=
B=1

It is clear that the output C will just be a copy of the input A. We can say that C=A. On the other hand,
if we set the B input to 0 we get a 0 at the output. We can say that C=0.

A=
C=0
B=0
We can summarise these as
 A.1=A
 A.0=0

We can apply a similar thought process to all possible situations with AND gates and OR gates to get
this set of relationships:

AND relations OR relations

A.1=A A+1=1
A.0=0 A+0=A
A.A=A A+A=A
A.A=0 A+A=1

8
3.3 Relationships with inversion
It should be reasonably obvious that if we take a variable A and invert it, and then we invert it again
the output will be a copy of A.

A A

This sounds unremarkable, but actually will turn out to be very useful when we look combine this
with de Morgan’s laws.

3.3.1 De Morgan’s laws

De Morgan’s laws state that:
𝐴. 𝐵 = 𝐴̅ + 𝐵
𝐴 + 𝐵 = 𝐴̅. 𝐵

The laws tell us how to deal with AND, OR operations that are enclosed within inversions. The laws
are easy to prove with truth tables, but let’s think about an example that will illustrate what the laws
mean. Let’s go back to our burglar alarm example:

A) Upstairs window is closed? 1=yes 0=no

B) Downstairs window closed? 1=yes 0=no
S) The house is secure? 1=yes 0=no
Behaviour: S = A.B
Suppose we ask ourselves the question under what circumstances is the house not secure. In other
words, under what circumstances do we have 𝑆̅ = 1? A little thought should convince you that

NOT(House is secure) = NOT(Upstairs window closed AND Downstairs window closed)

= NOT(Upstairs window closed) OR NOT(Downstairs window closed)

𝑆̅ = 𝐴. 𝐵 = 𝐴̅ + 𝐵

So the AND condition becomes an OR condition when we take it outside of the NOT.

These laws also extend to gates with many inputs:

𝐴. 𝐵. 𝐶. 𝐷. … = 𝐴̅ + 𝐵 + 𝐶̅ + 𝐷 + ⋯
𝐴 + 𝐵 + 𝐶 + 𝐷 + … = 𝐴̅. 𝐵 . 𝐶̅ . 𝐷. …

3.4 Commutativity, associativity and distributivity

3.4.1 Commutativity
In our normal everyday arithmetic, some operations don’t care which way round we write the
operands. For example, 2×3 is the same as 3×2. We say that multiplication is commutative.

A quick inspection of the truth tables for AND and OR shows that both of these operations are
commutative:

A.B = B.A and A+B = B+A

9
3.4.2 Associativity
An associative operator means that if we have several occurrences of the same operator, we get the
same result regardless of how we group them. So in everyday arithmetic:
(2+3)+4 = 2+(3+4)
(2×3)×4 = 2×(3×4)

Addition and multiplication are associative. Because it doesn’t matter how we group, we normally
just say 2+3+4 or 2×3×4 without any brackets.

We are only free to change the groupings if all of the operations are the same. If the operators are
different, we are not allowed to change the grouping:
(2×3)+4 ≠ 2×(3+4)

It is easy to prove from the truth tables that AND and OR operations are associative:
(A.B).C = A.(B.C)
(A+B)+C = A+(B+C)

As a result, we would normally just say A.B.C, A+B+C without using brackets.

3.4.3 Operator precedence

In everyday arithmetic we may come across an expression like this:
2+3×4
Do we interpret this as (2+3)×4 or as 2+(3×4)? Mathematicians could have made the decision that
people should always use brackets, and disallow unbracketed expressions like 2+3×4. However,
that’s not the choice that was made. Instead the convention is that we allow unbracketed expressions
and interpret them by giving multiplication a higher precedence (i.e. we do it first). So
2+3×4 should be interpreted as 2+(3×4)
This isn’t a consequence of any fundamental property of multiplication and addition. It’s just a way to
save ourselves from having to always write brackets all of the time.

Similarly, in Boolean expressions AND has higher precedence than OR. So if we have
A+B.C
this should be interpreted as
A+B.C=A+(B.C)
This isn’t a consequence of any fundamental property of AND, OR. It’s just a way to save ourselves
from having to always write brackets all of the time.

3.4.4 Expanding out brackets (distributive property)

In everyday arithmetic, we are used to taking expressions like this:
2×(3+4)
and expanding out the bracket:
2×(3+4) = (2×3)+ (2×4)
We say that multiplication is distributive over addition.

10
In Boolean algebra, we have a similar result
A.(B+C) = A.B+A.C
We say that AND is distributive over OR

4 The complete list of axioms and theorems of Boolean algebra

We have now completed our list of the axioms of Boolean algebra:

AND relations Inverter relations OR relations

A.1=A A+1=1
A.0=0 𝐴̅=A A+0=A
A.A=A A+A=A
A.A=0 A+A=1

A.B=B.A A+B=B+A
(A.B).C=A.(B.C) (A+B)+C=A+(B+C)
A.(B+C)=A.B+A.C A+(B.C)=(A+B).(A+C)
A.B=A+B A+B=A.B

5 Some useful results

We will illustrate the use of Boolean algebra by proving a few useful results that will be used later in
the module.

5.1 The absorption method

A useful trick in in simplifying Boolean algebra equations (and their corresponding logic circuits) is
known as absorption. Suppose we have the following equation
𝐶 = 𝐴 + 𝐴. 𝐵

This can be simplified to just

𝐶=𝐴
We can understand this intuitively by reasoning that A.B can only be 1 when A is 1. But if A is 1, then
(because A appears as the first input to the OR function) C=1 regardless of the state of B. Conversely,
if A=0 then A.B must be 0, so C=0). We can prove this by truth table:

A B A.B A+A.B
0 0 0 0
0 1 0 0
1 0 0 1
1 1 1 1

The columns for A and A+A.B are identical for any possible input combination, so we have proved
the result. We also can prove it by Boolean algebra like this:

𝐶 = 𝐴 + 𝐴. 𝐵
𝐶 = 𝐴. 1 + 𝐴. 𝐵
𝐶 = 𝐴. (1 + 𝐵)
𝐶 = 𝐴. 1
𝐶 =𝐴
11
5.2 A+A.B=A+B
This is another useful result that will illustrate how we do Boolean manipulations. We start with

𝐶 = 𝐴 + 𝐴̅. 𝐵

The absorption methods says that we can replace A with A+A.B, so we have

𝐶 = 𝐴 + 𝐴. 𝐵 + 𝐴̅. 𝐵

Now we can simplify

𝐶 = 𝐴 + (𝐴 + 𝐴̅). 𝐵
𝐶 = 𝐴 + 1. 𝐵
𝐶 =𝐴+𝐵

So our result is proved

6 Summary
We have covered the axioms of Boolean algebra and illustrated their use with a few examples.
Mastery of Boolean algebra enables us to construct a circuit to accomplish a required functionality. It
also enables us to simplify a complex circuit to a circuit of equivalent function which requires fewer
resources to build. We will illustrate these two points in the following lectures.

You should now know...

 The rule of Boolean algebra
 How to perform algebraic manipulations with Boolean algebra

12
4 Circuit synthesis and Circuit Simplification
Circuit synthesis starts with a statement of required behaviour and figures out a circuit that will
provide this behaviour. Earlier we looked at examples where this could be done intuitively. Now we
will look at a systematic procedure that will always work, regardless of intuition.

4.1 An example
Suppose we have a system that takes 1-bit inputs (A, B, C) from three sensors. The required behaviour
is that the output T should form the majority decision of the sensors. So if the majority of sensors are
outputting a 1, then our system should output a 1. Otherwise it should output a 0. This can be
expressed as a truth table:

A B C T
0 0 0 0
0 0 1 0
0 1 0 0
0 1 1 1
1 0 0 0
1 0 1 1
1 1 0 1
1 1 1 1

To help us get to our final solution, we will start off by producing a few sub-circuits:

𝑍 = 𝐴̅. 𝐵. 𝐶 this outputs Z=1 whenever A=0, B=1, C=1; otherwise it outputs Z=0
𝑌 = 𝐴. 𝐵 . 𝐶 this outputs Y=1 whenever A=1, B=0, C=1; otherwise it outputs Y=0
𝑋 = 𝐴. 𝐵. 𝐶̅ this outputs X=1 whenever A=1, B=1, C=0; otherwise it outputs X=0
𝑊 = 𝐴. 𝐵. 𝐶 this outputs W=1 whenever A=1, B=1, C=1; otherwise it outputs W=0

T should take the value 1 whenever any of these sub-circuits outputs a 1. In other words

𝑇 = 𝑍 + 𝑌 + 𝑋 + 𝑊 = 𝐴̅. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅ + 𝐴. 𝐵. 𝐶

This immediately gives us the circuit to achieve the desired function T

A Z
B
C
Y

T
X

13
4.2 Canonical sum of product form
Our equation for the action of the circuit is

𝑇 = 𝐴̅. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅ + 𝐴. 𝐵. 𝐶 (1)

This way of expressing our behaviour is called a sum of products form. This is because we take our
inputs (maybe complemented, maybe not) then AND them together in various configurations to form
logical product terms 𝐴̅. 𝐵. 𝐶, 𝐴. 𝐵 . 𝐶, 𝐴. 𝐵. 𝐶, 𝐴. 𝐵. 𝐶. We then OR these terms together to form the
logical sum of these terms.

In the same way, we could also work out the equation for 𝑇, i.e. the list of all circumstances in which
our circuit will output a 0:

𝑇 = 𝐴̅. 𝐵 . 𝐶̅ + 𝐴̅. 𝐵 . 𝐶 + 𝐴̅. 𝐵. 𝐶̅ + 𝐴. 𝐵 . 𝐶̅ (2)

Equations (1) and (2) taken together are called the canonical sum of products. The characteristic of a
canonical equation is that every input appears (both inverted and uninverted) in every possible
combination with every other input in the product terms. The canonical sum of products is a Boolean
algebra way to describe the truth table.

4.3 Other representations for sum-of-products

Equations 1 and 2 can be represented in a concise form by noting that each of the product terms that
appears in (1) and (2) corresponds to a binary number. For example, the term 𝐴̅. 𝐵 . 𝐶 will be true when
ABC=001, the binary number 1. These product terms that appear in (1) and (2) and contain each of
the inputs, either in true or complemented form, are called minterms. Each of these minterms can be
assigned an identification number 0-7:

Minterm (Mi) Number Product

ABC 000 = 0 P0
A BC 001 = 1 P1
AB C 010 = 2 P2
AB C 011 = 3 P3
ABC 100 = 4 P4
A BC 101 = 5 P5
AB C 110 = 6 P6
ABC 111 = 7 P7

The sum-of-products expression may be expressed as a generic equation

2 N 1
f 
i 0
ai Pi

here ai is a Boolean constant and Pi represents the product term or minterm.

14
If we take our Boolean expression for T

T=A.B.C+A.B.C+A.B.C+A.B.C

We can express this as

T=0.P0+ 0.P1+ 0.P2+ 1.P3+ 0.P4+ 1.P5+ 1.P6+ 1.P7

This can be simplified to

T=P3+P5+P6+P7

Which we can write as

T= (3,5,6,7)

Similarly, our expression for T

T=A.B.C+A.B.C+A.B.C+A.B.C

Can be simplified to

T= (0,1,2,4)

Note that each of the minterms must appear once in either T or T

4.4 How do we measure the efficiency of a circuit design?

The canonical sum of products form will automatically generate a circuit for us to accomplish any
required logic function.

A Z
B
C
Y

T
X

However, the resulting circuit is normally inefficient and requires more logic resource than is
necessary. To get a really exact measure of resource requirements is complex and requires a detailed
knowledge of the construction process, but we can get a good ballpark idea by measuring the number
of logic gate inputs

15
To get an estimate of hardware resource requirement it is tempting to just count the number of logic
gates. However, a four input AND gate normally requires twice as much hardware to build as a 2-
input AND gate. (The reason for this is the internal structure of a logic gate is made of switches called
transistors. Each separate input drives a separate transistor, so doubling the number of inputs doubles
the number of transistors used to build the gate.) So it makes more sense to count the total number of
inputs to logic gates and use this as our measure of hardware complexity.

Our crude measure of the complexity of our example circuit is

Inverter inputs 3
AND gates inputs 12
OR gate inputs 4
Total 19

4.5 Circuit Simplification

The canonical sum-of-products form can always give rise to a solution for a circuit, but the resulting
circuits are generally inefficient. We will look at how to use Boolean algebra identities to transform
complicated circuits into simpler ones. (There is an alternative approach to circuit simplification using
a graphical method called a Karnaugh maps. This method is usually a lot simpler than Boolean
simplification. We’ll look at this in the next lecture.)

4.5.1 Recognising terms that are unnecessary

The main trick that we will use is introduced by the following example:
A) It is raining? 1=yes 0=no
B) The weather is hot? 1=yes 0=no
C) Take an umbrella? 1=yes 0=no

Suppose we have the following equation

𝐶 = 𝐴. 𝐵 + 𝐴. 𝐵

This can be interpreted as meaning

(Take an umbrella) = (It is raining AND It is hot) OR (It is raining AND It is NOT hot)

It should be clear that when it is raining we will take an umbrella regardless of the weather.
Conversely, if it is not raining we will not take an umbrella regardless of the weather. So the weather
does not actually have any bearing on our decision whether to take an umbrella, i.e.

𝐴. 𝐵 + 𝐴. 𝐵 = 𝐴

This can also be demonstrated to be true by considering the truth table:

A B A.B A.B A.B + A.B

0 0 0 0 0
0 1 0 0 0
1 0 0 1 1
1 1 1 0 1

16
It’s easy to prove with Boolean algebra:
𝐶 = 𝐴. 𝐵 + 𝐴. 𝐵
𝐶 = 𝐴. (𝐵 + 𝐵 )
𝐶 = 𝐴. (1)
𝐶=𝐴
In analysing our circuits we need to be on the lookout for the opportunity to do this trick. We spot the
opportunity by looking for product terms that differ only in the complementation of one variable.

4.5.2 Using this method more than once

Consider this example:

𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅

The 𝐴. 𝐵. 𝐶 term differs from 𝐴. 𝐵 . 𝐶 in the complementation of only one variable. So here is an
opportunity to simplify. Likewise 𝐴. 𝐵. 𝐶 differs from 𝐴. 𝐵. 𝐶̅ by complementation of one variable. So
we can see two opportunities for simplification. However, if we start off by using 𝐴. 𝐵 . 𝐶 for
simplification, we get this:

𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅
𝐷 = 𝐴. 𝐶. (𝐵 + 𝐵 ) + 𝐴. 𝐵. 𝐶̅
𝐷 = 𝐴. 𝐶. 1 + 𝐴. 𝐵. 𝐶̅
𝐷 = 𝐴. 𝐶 + 𝐴. 𝐵. 𝐶̅

We have lost the 𝐴. 𝐵. 𝐶 term that we needed to do the simplification on the 𝐴. 𝐵. 𝐶̅ term, so we can’t
simplify further. Likewise, if we start by using 𝐴. 𝐵. 𝐶̅ for simplification, we get this:

𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅
𝐷 = 𝐴. 𝐵. (𝐶 + 𝐶̅ ) + 𝐴. 𝐵 . 𝐶
𝐷 = 𝐴. 𝐵. 1 + 𝐴. 𝐵 . 𝐶
𝐷 = 𝐴. 𝐵 + 𝐴. 𝐵 . 𝐶

Once again, we have lost the 𝐴. 𝐵. 𝐶 term that we needed to do the simplification on the 𝐴. 𝐵 . 𝐶, so we
can’t simplify further. How can we deal with this in w way that enables us to do both simplifications?
Remember that for any formula X we know that X+X=X. We can replace any instance of a formula by
two instances ORed together. So we take our original equation

𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅

And use A.B.C=A.B.C+A.B.C to get

𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅

Now we can simplify to simplest form:

𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅
𝐷 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵. 𝐶̅
𝐷 = 𝐴. 𝐶. (𝐵 + 𝐵 ) + 𝐴. 𝐵. (𝐶 + 𝐶̅ )
𝐷 = 𝐴. 𝐶. 1 + 𝐴. 𝐵. 1
𝐷 = 𝐴. 𝐶 + 𝐴. 𝐵

17
4.5.3 Simplifying our example circuit
Our canonical system equation for the circuit of section 4.1 is:

𝑇 = 𝐴̅. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅ + 𝐴. 𝐵. 𝐶

The final term A.B.C differs from each of the other three terms by the complementation of one
variable. So we would like to use it three times over. In Boolean algebra X+X=X, so we can say that
A.B.C = A.B.C + A.B.C + A.B.C. We insert this into our equation to get:

𝑇 = 𝐴̅. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶̅ + 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵. 𝐶

We can change the order and groupings to get:

𝑇 = (𝐴̅. 𝐵. 𝐶 + 𝐴. 𝐵. 𝐶) + (𝐴. 𝐵 . 𝐶 + 𝐴. 𝐵. 𝐶) + (𝐴. 𝐵. 𝐶̅ + 𝐴. 𝐵. 𝐶)

Now we can use our simplification method:

𝑇 = (𝐴̅ + 𝐴). 𝐵. 𝐶 + 𝐴. (𝐵 + 𝐵 ). 𝐶 + 𝐴. 𝐵. (𝐶̅ + 𝐶)

𝑇 = 1. 𝐵. 𝐶 + 𝐴. 1. 𝐶 + 𝐴. 𝐵. 1
𝑇 = 𝐵. 𝐶 + 𝐴. 𝐶 + 𝐴. 𝐵

This is a sum-of-products form, but it is much simpler than the original canonical sum-of-products.
Our simplified circuit is

A
B

Again, we can estimate the complexity of this circuit by counting the number of gate interconnections.

Inverter inputs 0
AND Gate inputs 6
OR Gates inputs 3
Total 9

This represents 47% of the circuit complexity of the original design of section 4.1.

4.6 Summary
The canonical sum of products form is very important in the theory of Boolean algebra. It contains all
product terms that can be constructed from the inputs and allocates them to a sum that produces a one
at the output and a sum that produces a zero.

The canonical sum of products form can be used to directly construct a circuit, but the result will
almost always be inefficient. We have seen how to use Boolean algebra to the canonical sum of
products form to produce a simplified circuit. In the next lecture we will look at an easier approach to
simplification, the Karnaugh map.
18
 You should now know...
 How to construct a circuit from its truth table
 How to construct a canonical sum of products form
 How to represent a truth table using ∑(𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑡𝑒𝑟𝑚𝑠) notation
 How to estimate the resource cost of a digital circuit
 How to use Boolean algebra to derive a simplified circuit from a complex circuit

19
5 Minimisation by Karnaugh Map
In the last lecture we looked at a way to generate a circuit that will fulfil a user requirement. We are
given a required functionality (expressed as a truth table) and our task is to build a circuit that achieves
the desired function. The method was based on building a detector for each input condition that
contributes to a 1 at the output, then ORing the outputs of these detectors. We observed that this
technique will always work, but will normally produce a circuit that is inefficient. We then had to
simplify the resulting circuits using Boolean algebra. However, the use of Boolean algebra is tedious
and inefficient. An easier and quicker way is the Karnaugh map (also called K-map), which is a
graphical method that makes the required simplifications visually obvious.

Boolean algebra simplifies by looking for terms which differ only in the complementation of one
input, and eliminating that term. For example

𝑇 = 𝐴. 𝐵. 𝐶 + 𝐴. 𝐵 . 𝐶 = 𝐴. 𝐶

The Karnaugh map works by re-drawing the truth table onto a surface where any two neighbours
differ in the complementation of one input variable. The process of spotting opportunities for
simplification is then reduced to merging neighbours together to form larger groups.

In order to accomplish this, we will need to use Gray codes. Gray codes are sequences whose key
property is that each successor differs from its predecessor by the complementation of one bit. Here
are the 2-bit and 3-bit Gray code sequences:

2-bit 00, 01, 11, 10, 00 and repeats

3-bit 000, 001, 011, 010, 110, 111, 101, 100, 000, and repeats

5.1 The 2-input Karnaugh map

A Karnaugh map is way of re-drawing the truth table so that any pair of terms that differ by the
complementation of one input variable are adjacent to one another. The simplification can then be
done by eye without any need for Boolean algebra manipulations. Suppose we have the truth table
shown below:

Truth table: Karnaugh map:

A
A B T B 0 1
0 0 0 0 0
0
0 1 1
1 0 1
1 1 1
1 1 1

The Karnaugh map rearranges the truth table as a two-dimensional grid. The columns correspond to
the values of A, so the left column corresponds to A=0 and the right to A=1. The rows correspond to
the values of B; the top row is B=0; the bottom row is B=1. Inside the grid we write the values that the
output should have. So, for example, the top left cell in the K-map corresponds to A=0, B=0 resulting
in an output of T=0.

The purpose of this rearrangement is that the coordinates of any cell in the grid differ from the
coordinates of its horizontal or vertical neighbour by the complementation of a single input variable.
So the cell whose coordinates are A=0, B=0 is to the left of the cell with coordinates A=1, B=0.
20
Likewise, the cell A=0, B=0 is above the cell A=0, B=1. This is important in helping us to simplify our
circuit designs by spotting visual patterns.

Before we look at how to do simplification, let’s start off by seeing how the K-map turns into a circuit.
We start off our design process by putting a loop around every condition that causes an output of 1:

A 0 1
B
0 0 0

1
1 1

We now figure out what circuit would be needed to produce the 1. So, for example, the bottom left cell
outputs a 1 when A=0 and B=1. This can be done by a circuit that implements 𝐴̅. 𝐵. When we have
labelled all of the loops we have this:
A 0 1
B
0 0 1

1
1 1

If any of the conditions is true, then the overall output C must be 1. So we OR the outputs of the
detectors for these input conditions together to get the result:

𝐶 = 𝐴̅. 𝐵 + 𝐴. 𝐵

This is just the result that we got before from the truth table, so we haven’t gained any benefit yet.
However, when we come to simplifying the circuit the K-map will make the task easy.

To simplify the circuit, we look for any loops that are adjacent to one another horizontally or vertically
(no diagonals are allowed and we can’t go round corners). We then merge them into a single larger
loop. Here is an example:

A 0 1 A 0 1
B B
0 0 0 0 0 0

1 1
1 1 1 1

Throughout the loop, B has a constant value (B=1) and A can take any value (0, 1). This tells us that A
is irrelevant to the decision made by this loop, and its output is determined solely by the value of B.
So we can detect this loop simply by having a circuit that detects B=1.

21
When all of the loops are as big as they can be (remembering that we can’t go diagonally and can’t go
round corners) then we stop and our resulting sum-of-products circuit has been minimised (i.e. it is as
efficient as it can possible be).

This gives us the same answer as the Boolean algebra simplification, but is much easier. The K-map
method also makes it much easier to be sure that we have found all possible simplifications of our
circuit and that the final result is minimised. With Boolean algebra it is harder to be sure that we have
truly found the most efficient solution.

5.2 Why does this work?

The K-map puts terms that differ only in the complementation of one variable into adjacent cells. So
for example 𝐴̅. 𝐵 is right next to 𝐴. 𝐵. By merging these two neighbouring cells, we have visually
made the simplification 𝐴̅. 𝐵 + 𝐴. 𝐵 = 𝐵 without needing to use Boolean algebra.

5.3 Karnaugh maps with 3 inputs

The K-map with three inputs A, B, C looks like this:
AB
C 00 01 11 10
0

The rows correspond to the values of input C. The columns correspond to the values of the inputs A
and B: respectively A=0,B=0 then A=0,B=1, then A=1,B=1 then A=1,B=0. The whole point of a
Karnaugh map is to put terms that differ only in the complementation of one variable next to each
other. A binary sequence 00, 01, 10, 11 would not achieve this. The Gray code sequence 00, 01, 11, 10
does achieve this requirement. It is essential to the success of the K-map method that the variables
must be indexed using a Gray code sequence.

The basic procedure is the same as for two inputs, but there are two additional pieces of information
needed:
 Loops must enclose rectangles or squares whose sides are an exact power of 2 (i.e. 1, 2 or 4). This
was also true for the 2-input case, but we didn’t mention it because it was automatically true as any
loop that we could find would automatically be 1 or 2 long.
 The right-hand edge of the K-map is connected to the left hand edge. So the column AB=10 is a
neighbor of the column AB=00. You can think this shape as a cylinder that has been flattened out
for ease of representation on the 2-dimensional page, but in actual operation the right edge would
connect up to the left.

5.3.1 A 3-input example

In section 4.1 we looked at an example system that takes 1-bit inputs (A, B, C) from three sensors. The
required behaviour is that the output T should form the majority decision of the sensors. So if the
majority of sensors are outputting a 1, then our system should output a 1. Otherwise it should output a
0. This has the truth table and K-map as follows:

22
 Truth table: Karnaugh map:
A B C T
0 0 0 0 AB
C 00 01 11 10
0 0 1 0
0 1 0 0 0 0 0 1 0
0 1 1 1
1 0 0 0
1 0 1 1
1 0 1 1 1
1 1 0 1
1 1 1 1

To solve this, we go through the same loop-connecting procedure as we used before (remembering
that we can’t go diagonally, can’t go round corners, and that we are only allowed rectangles or squares
that are 1, 2 or 4 long). The final result contains three loops:

AB
AB C 00 01 11 10
C 00 01 11
0 0 0 1 0
0 0 0 1 0
1 0 1 1 1
1 0 1 1 1

 𝐴. 𝐵: within this loop, A=1, B=1 and we don’t care about C (C can take values of 0 or 1 without
any consequence on the decision that this loop takes). An AND gate with inputs A, B will detect
the input pattern A=1, B=1.
 𝐵. 𝐶: within this loop, B=1, C=1 and we don’t care about A. An AND gate with inputs B, C will
detect this condition.
 𝐴. 𝐶: within this loop, A=1, C=1 and we don’t care about B. An AND gate with inputs A, C will
detect this condition.

This process immediately gives us our minimised circuit without the need for Boolean algebra:

A
B

T=A.B+B.C+A.C

This gives us the same result that we derived using Boolean algebra in section 4.5.3, but the Karnaugh
map gives us the solution more quickly and easily. Moreover, with the Karnaugh map it is easy to
convince ourselves that we truly have found the most efficient possible solution (because all loops are
as large as possible), whereas with Boolean algebra it isn’t always obvious.

23
5.4 Karnaugh maps with 4 inputs
The K-map with four inputs A, B, C, D looks like this:
AB
CD 00 01 11 10
00

01

11

10

The rows and columns both proceed in a Gray code sequence (00, 01, 11, 10) in order to ensure that
the coordinates of adjacent cells differ only in the complementation of one input variable. The rows
correspond to C=0,D=0 then C=0,D=1 then C=1,D=1 then C=1,D=0. The columns correspond to
A=0,B=0 then A=0,B=1, then A=1,B=1 then A=1,B=0.

The right hand edge of the K-map is linked to the left hand edge. Also the top edge of the K-map is
connected to the bottom edge. You can think of it as a doughnut shape1 that has been flattened out for
ease of representation on the 2-dimensional page, but in actual operation the right edge would connect
up to the left and the top to the bottom.

4.5 An example with 4 inputs

Let’s consider an example with the following truth table. We transfer it to our Karnaugh map:

Truth table: Karnaugh map:

A B C D T
0 0 0 0 0 AB
0 0 0 1 0 CD 00 01 11 10
0 0 1 0 0 00 0 0 1 0
0 0 1 1 1
0 1 0 0 0
01 0 0 1 0
0 1 0 1 0
0 1 1 0 0
0 1 1 1 1 11 1 1 1 1
1 0 0 0 0
1 0 0 1 0 10 0 0 1 0
1 0 1 0 0
1 0 1 1 1
1 1 0 0 1
1 1 0 1 1
1 1 1 0 1
1 1 1 1 1

1
The formal mathematical name of a doughnut shape is a torus
24
Using our looping procedure we get the following result

AB
CD 00 01 11 10
00 0 0 1 0

01 0 0 1 0

11 1 1 1 1

10 0 0 1 0

We have two loops:

 One encloses the values where A=1, B=1 and C and D do not contribute to the decision. This is
detected by an AND gate with inputs A, B.
 The other encloses the values where C=1, D=1 and A and B do not contribute to the decision. This
is detected by an AND gate with inputs C, D. Our final minimised circuit is:

A
B
T=A.B+C.D
C
D

5.5 Karnaugh map terminology

In our K-map method, we have been looking for loops whose contents are all 1, and whose sides are
1, 2 or 4 long. The technical name for these loops is an implicant. If the implicant is true, then our
function is true. Consider the following example:

AB
CD 00 01 11 10
00 0 0 0 0

01 0 1 1 0

11 0 1 1 0

10 0 0 0 0

25
If we look for loops that contain only ones, we can find 9 implicants in all:
 𝐴̅. 𝐵. 𝐶̅ . 𝐷, 𝐴. 𝐵. 𝐶̅ . 𝐷, 𝐴̅. 𝐵. 𝐶. 𝐷, 𝐴. 𝐵. 𝐶. 𝐷
 𝐵. 𝐶̅ . 𝐷, 𝐵. 𝐶. 𝐷, 𝐴̅. 𝐵. 𝐷, 𝐴. 𝐵. 𝐷,
 𝐵. 𝐷

To construct our Boolean representation of the function, we construct a logical sum of implicants. We
don’t need to use all of the implicants, we just need enough to make sure that every 1 in the K-map is
enclosed within at least one implicant. In order to guide us in which implicants to use, there are
another two definitions that will be helpful:

A prime implicant is an implicant that is not enclosed within any other implicant. For example,
𝐴̅. 𝐵. 𝐶̅ . 𝐷 is an implicant, but it is not a prime implicant because it is entirely enclosed within B.D.
B.D is a prime implicant because it is not entirely enclosed within another implicant.

An essential prime implicant is a prime implicant which has some 1s which are not enclosed within
any other implicant. Because we need enough implicants to cover all of the 1s in our K-map, that
means that any essential prime implicant must be one of the terms in our final representation.

A minimal sum is the smallest number of prime implicants require to describe the function. It must
contain all essential prime implicants, plus enough prime implicants to cover the remaining 1s.

5.6 Karnaugh maps with 5 inputs

For a function with 5 inputs, we take two, 4-variable K-maps and mirror the indices. The vertical
dashed line indicates a mirroring, or fold, line such that the two K-maps overlay one-another with a
one-variable change between maps.

We then follow the standard process of labelling the terms that are constant throughout each loop:
Blue loop: 𝐴. 𝐷
Red loop: 𝐴̅. 𝐵. 𝐷
Green loop: 𝐴. 𝐶̅ . 𝐷

Our overall representation of the function is a logical sum of these terms:

𝐴. 𝐷 + 𝐴̅. 𝐵. 𝐷 + 𝐴. 𝐶̅ . 𝐷

26
5.6 Karnaugh maps with 6 inputs
For a 6-input function, we take four 4-variable K-maps and mirror the indices in both horizontal and
vertical directions. Here is an example:

5.7 Summary of the Karnaugh map procedure

Karnaugh maps are a way of drawing the truth table so that circuit can quickly and easily be
minimised by eye. This avoids the complexity of Boolean algebra manipulations. The key rules to
remember are:
 We find the largest loop we can that encloses output values of 1
 The loops must be rectangular or square and have sides that 1, 2 or 4 long (we can’t go diagonally,
can’t go round corners)
 The rows and columns are labelled using the Gray code sequence (0,1 for 1 variable; 00, 01, 11,
10 for 2 variables)

You should now know...

 What a Gray code is
 How to construct Karnaugh maps for circuits with 2,3,4,5 or 6 inputs
 How to extract a minimal sum of products representation from a Karnaugh map
 The meaning of the following terms:
o Implicant
o Prime implicant
o Essential prime implicant
o Minimal sum of products

27
 Index

1 Overview of the module

2 Basic Ideas of Combinational Logic 2

2.1 Types of Digital System 2
2.2 Our basic building blocks: logic gates 2
2.2.1 The AND gate 2
2.2.2 The OR gate 3
2.2.3 Inclusive OR, exclusive OR 3
2.2.4 The Inverter 4
2.3 IEEE symbols 4
2.4 Putting it all together 5
2.4.1 Inversion bubbles 5
2.4.2 Circuit analysis: figuring out what a circuit does 6
2.4.3 Truth table for the circuit 6
2.4.4 Algebraic description of the circuit 7
2.5 Summary 7

3 The Axioms of Boolean Algebra 8

3.1 Relationship of Boolean Constants 8
3.2 AND and OR Functions of a Boolean variables 8
3.3 Relationships with inversion 9
3.3.1 De Morgan’s laws 9
3.4 Commutativity, associativity and distributivity 9
3.4.1 Commutativity 9
3.4.2 Associativity 10
3.4.3 Operator precedence 10
3.4.4 Expanding out brackets (distributive property) 10
3.4 The complete list of axioms and theorems of Boolean algebra 11
3.5 Some useful results 11
3.5.1 The absorption method 11
3.5.2 A+A.B=A+B 12
3.6 Summary 12

4 Circuit Synthesis and Circuit Simplification 13

4.1 An example 13
4.2 Canonical sum of product form 14
4.3 Other representations for sum-of-products 14
4.4 How do we measure the efficiency of a circuit design? 15
4.5 Circuit Simplification 16
4.5.1 Recognising terms that are unnecessary 16
4.5.2 Using this method more than once 17
4.5.3 Simplifying our example circuit 18
4.6 Summary 18

5 Minimisation by Karnaugh Map 20

28
5.1 The 2-input Karnaugh map 20
5.2 Why does this work? 21
5.3 Karnaugh maps with 3 inputs 21
5.3.1 A 3-input example 22
5.4 Karnaugh maps with 4 inputs 24
5.4.1 An example with 4 inputs 24
5.5 Karnaugh map terminology 25
5.6 Karnaugh maps with 5 inputs 26
5.6 Karnaugh maps with 6 inputs 27
5.7 Summary of the Karnaugh map procedure 27

29