Digital

Di it l IIntegrated
t t d Ci
Circuits
it &
Microcontrollers

Chapter 2. Binary representation

Numeration Systems

 Decimal
 the set of symbols is:
{0, 1, 2, 3, 4, 5, 6, 7, 8, 9};
 The integer number 493 in base 10:
49310 = 4•102 + 9•101 + 3•100 =
= 400 + 90 + 3
 The rational number 35,54 in base 10:
35.6410 = 3•101 + 5•100 + 6•10-1 + 4•10-2 =
= 30 + 5 + 0 0.6
6 + 0
0.04
04

2
Numeration Systems

 Binary
 the set of symbols is:
{0, 1};
 110012 = 1 1•2
24 + 1
1•2
23 + 0
0•2
22 + 0
0•2
21 + 1
1•2
20 =
= 16 + 8 + 1 = 25
 110.012 = 1•22 + 1•21 + 0•20 + 0•2-1 + 1•2-2=
= 4 + 2 + 0.25 = 6.25

Conversion between numeration systems

 Conversion from binary into hexadecimal

 Group into 4 bits (nibbles)
 Each nibble corresponds to a hexadecimal symbol:

1101101,10011012

0110 1101,1001 10102

6 D , 9 A16

4
Conversion between numeration systems
 Conversion from decimal into base b
 For the integer part we can write:
i  d b  d
k k -1  b  ...  d3  b  d2  b  d1  b  d0
 d0 is the remainder of division of i to b
 The quotient (another integer) is divided to b
 Repeat until reach 0.
 The remainder obtained after each division is the symbol dk of
representing into the base b.

i  d k b k  dk -11 bk -1  ...  d 3b3  d2 b 2  d1b1  d0 b0

Conversion between numeration systems

 C
Conversion
i from
f decimal
d i l into
i t base
b b
 For the fractional part:
f  d 1 b1  d 2 b 2  d 3 b 3  ...  d  k b  k  ...
 Multiply f with b
b  f  d 1  d 2 b 1  d 3 b 2  ...  d  k b  k 1  ...
 Keep the integer part from the right part, d–1, which is
subtracted from the left part.
 Continue by multiplying the remaining fractionary part to b until
reach 0.

b  b  f  d 1  =d 2  d 3 b 1  ...  d  k b  k  2  ...

6
Conversion between numeration systems
 EXAMPLE
EXAMPLE:
 Represent the number 23.65 into base 2:
 Th integer
The i t part:
t
23 : 2 = 11 1 LSB
11 : 2 = 5 1
5:2= 2 1
2:2= 1 0
1:2= 0 1 MSB
2310 = 101112

Conversion between numeration systems

 Th fractional
The f i l part:
0.65  2 = 1.3 1 MSB
03
0.3  2 = 0.6
06 0
0.6  2 = 1.2 1
02
0.2  2 = 0.4
04 0
0.6510=0.10(1001)2
0.4  2 = 0.8 0
0.8  2 = 1.6 1
0.6  2 = 1.2 1
0.2  2 = 0.4 0 ...
 Results that number 0.65 cannot be exactly represented on a
finite number of bits.
bits

8
Negative numbers representation

 MSB – sign bit.

0 for positive numbers (+);
 1 for negative numbers (–).

 The rest of N–1 bits are for value representation.

2N–2 ... 20
s m
1 bit  N–1 bits 

Negative numbers representation

 The representation: sign bit, magnitude

 EXAMPLE:
9 = 0 01001
–9
9 = 1 01001
 The range of representation:
 2NN–11 positive values between 0 and 2NN–11–1.
1
 2N–1 negative values between –(2N–1–1) and 0.

2N–2 ... 20
s m
1 bit  N–1 bits 

10
Negative numbers representation

 Two’s complement representation

 The negative numbers representation is obtained by
addition of 2N.
 EXAMPLE:
EXAMPLE
 For N=6 bits (2N = 64).
13 = 0011012
–13 corresponds to 64 + (–13) = 51 = 1100112

11

Two’s complement representation

 For obtaining negative numbers:

 Each bit is complemented;
 Add 1.
 EXAMPLE:
 for 13 = 0011012
complement each bit: 1100102
add 1: 1100112 = –13
 for –13 = 1100112
complement
l t each
h bit
bit: 0011002
add 1: 0011012 = +13

12
Two’s complement representation

 The range of two’s complement representation

 2N–1 positive values between 0 and 2N–1–1.
 2N–1 negative values between –2N–1 and -1.

 The result of adding a number with its two’s

complement is 0:

13+ 0011012
–13 1100112
=0 10000002

13

Integer numbers representation

 Binary representation is considered right

aligned (decimal point is right to LSB).
 MSB represents the sign bit.
–2
2N–1 2N–2 ... 20
s i .
 N biţi 

 Integer two’s complement range:

2 N 1 ,..., 1,0,...,2 N 1  1

14
Integer numbers representation

 EXAMPLE: (for N=4 bits)

Decimal Binary Decimal Binary
0 0000 -8 1000
1 0001 -7
7 1001
2 0010 -6 1010
3 0011 -5
5 1011
4 0100 -4 1100
5 0101 -3
3 1101
6 0110 -2 1110
7 0111 -1 1111

15

Integer numbers representation

 Sign integers addition

1111 011 11
+3 0011 –5 1011 –3 1101
–2 1110 +3 0011 –4 1100
1 0001 –2 1110 –7 1001
 An overflow occurs if the result is outside of the N bits
representation range:
+3 0011 –3
3 1101
+6 0110 –6 1010
9 1001=–7 –9 0111 =7

16
Integer numbers representation

 Sign bit extension

 Needed when increasing the number of bits for the
integer part.
 Sign
Si bit iis copied
i d tto th
the lleft
ft ttoward
d MSB
MSB.

N=4 biţi N’=8 biţi

+3 0011 +3 0000 0011
–3 1101 –3 1111 1101

17

Integer numbers representation

 Multiplying by a power of 2
 Multiplying by 2k is equivalent with shifting left k
bits and filling with 0 toward LSB.

N=8 biţi N=8 biţi

3 0000 0011 –3 1111 1101
3.22 0000 1100 –3.22 1111 0100

18
Integer numbers representation

 Dividing by a power of 2
 Dividing by 2k is equivalent with shifting right k bits
and sign bit extension.

N=8 biţi N=8 biţi

24 0001 1000 –24
24 1110 1000
24/23 0000 0011 –24/23 1111 1101

19

Integer numbers representation

 Integers multiplication
 unsigned integers
 The result in double
6 0110
precision representation 5 0101
 Binary multiplication with 0110
0 and 1 0000
0110
__ 0000 +
30 0011110

20
Integer numbers representation

 Multiplication is equivalent with consecutive shift

and
d add
dd operations.
ti
 For example 5 can be expressed:
5  20  2 2
 Multiplication can be computed:
6  5  6   20  2 2   6  1  6  2 2

61 0110
622 0110 +
0011110

21

Integer numbers representation

 Integers division
 Successive subtractions of the divisor from the
dividend.

15| 3 1111 11
5 –11 101
0011
–11
00

22
Fractional numbers representation

 A fractional part f is any number who’s modulus

satisfies
ti fi ththe iinequality:
lit
0.0  | f |  1.0
 Left aligned: binary point is at the right of MSB
–20 2–1 2–2 ... 2–B
s. f
1bit  B=N–1 bits 

 Fixed point fractional representation range:

1, ..., 2 B ,0, 2 B...,1  2 B
23

Fractional numbers representation

 EXAMPLE:
 for 4 bits
N=4
N
Decimal Binary Decimal Binary
0 0000 -1 1000
000
0.125 0001 -0.875 1001
0.250 0010 -0.750 1010
0.375 0011 -0.625 1011
0.500 0100 -0.500 1100
0.625 0101 -0.375 1101
0.750 0110 -0.250 1110
0.875 0111 -0.125 1111
24
Fractional numbers representation

 Qm.n format
n bits for fractional part;
p
 (optional) specify the number of bits m for the integer
part, excluding the sign bit (MSB);
 The complete binary representation has 1+m+n bits.
 EXAMPLE (for N=16 bits):
 Q15 means 15 bits for the fractional part (16 bits with the sign
bit)
 Q1.14
Q1 14 has 1 bit for the integer part
part, 14 bits for the fractional part
and the sign bit.

25

Fractional numbers representation

 Quick conversion of fractional numbers into

binary
f represented on N=B+1 bits is an integer multiple of
2–BB 0 –2
–1 –22 2 ...
–B 2
s. f

 Let i the corresponding integer multiple:

i  f  2B
 Equivalent with a left shifting of f with B bits.

26
Fractional numbers representation

–20 2–1 ... 2–B

s . f

–2B 2B–1 ... 20

s i .
 N bits 

 Converting f from binary to decimal

 consider the binary representation for the corresponding
signed integer i and then, divide by 2B

27

Fractional numbers representation

 EXAMPLE:
 Binary to decimal (N=8 bits, B=7),
 For: 0.010 0110
d i l iinteger:
decimal t 0010 0110 = 3810
divide by 27=128: 38/128 = 0.296875

 For: 1.110 1100

two’s
two s complement: –0001
0001 0100 = –20
2010
divide by 27=128: –20/128 = –0.15625

28
Fractional numbers representation
–20 2–1 ... 2–B
s . f

–2B 2B–1 ... 20

s i .
 N biţi 

 Converting f from decimal to binary

 multiply f by 2B (shift left B bits)
 Represent the integer part i as a signed integer on N bits.

29

Fractional numbers representation

 EXAMPLE:
 Decimal to binaryy (N=8
( biţi,
ţ B=7))
 For: 0.875
multiply by 27: 0.875.128=11210
represented binary: 11210 = 011100002

 For: –0.625
multiply by 27: –0.625.128= –8010
represented in 8010 = 010100002
two’s complement: –8010 = 101100002

30
Fractional numbers representation

 For 0.65 (N=8 bits, B=7)

multiply by 27: 0.65.128= 83.210
t k the
take th integer
i t part:
t 8310 = 010100112

 In
I the
th last
l t example
l the
th quantization
ti ti error (by(b
truncation) appears since the fractional 0.65 cannot
be exactly represented on 8 bits
bits.
 The result 0.10100112 is equal to 0.6484375.
 The q
quantization error is: 0.65
- 0.6484375
= 0.0015625

31

Floating point representation

 The following are equivalent
representations of 1,234

123,400.0 x 10-2
12 340 0
12,340.0 x 10-11
The representations differ
1,234.0 x 100 in that the decimal place –
123
123.4 x 101 th “point”
the “ i t” – “floats”
“fl t ” to
t the
th
left or right (with the
12.34 x 102
appropriate
pp p adjustment
j in
1.234 x 103 the exponent).
0.1234 x 104

32
Parts of a Floating Point Number

Exponent
-0.9876
0 9876 x 10-33
Sign of
Sign of Location of exponent
mantissa decimal point Mantissa

Base

33

IEEE 754 Standard

 Single precision: 32 bits, consisting of...

 Signbit (1 bit)
 Exponent (8 bits)
 Mantissa (23 bits)

 Double precision: 64 bits, consisting of…

 Signbit (1 bit)
 Exponent
p ((11 bits))
 Mantissa (52 bits)

34
Single Precision Format

32 bits

Mantissa (23 biţi)

Exponent (8 biţi)

Sign of mantissa (1 bit)

35

Normalization

 The mantissa is normalized

 Has an implied decimal place on left
 Has an implied “1” on left of the decimal place

 E.g.,
 Mantissa 10100000000000000000000
 Represents… 1.1012 = 1.62510

36
Excess Notation

 To include +ve and –ve exponents, “excess”

notation
t ti isi used
d
 Singleprecision: excess 127
 Double precision: excess 1023

 The value of the exponent stored is larger than

the actual exponent
 E.g.,
g , excess 127,,
 Exponent  10000111
 Represents… 135 – 127 = 8
37

Example
 Si l precision
Single i i
0 10000010 11000000000000000000000

1 112
1.11

130 – 127 = 3

0 = positive mantissa

+1.112 x 23 = 1110.02 = 14.010

38
Hexadecimal

 It is convenient and common to represent the

original
i i l flfloating
ti point
i t number
b iin h
hexadecimal
d i l
 The preceding example…

0 10000010 11000000000000000000000

4 1 6 0 0 0 0 0

39

Converting from Floating Point

 What decimal value is represented by the

following
o o g332-bit
b t floating
oat g po
pointt number?
u be

C17B000016

40
Converting from Floating Point

 Step 1
 Express in binary and find S, E, and M

C17B000016 =

1 10000010 111101100000000000000002
S E M

1 = negative
0 = positive

41

Converting from Floating Point

 Step 2
 Find“real” exponent, n
 n = E – 127
= 100000102 – 127
= 130 – 127
=3

42
Converting from Floating Point

 Step 3
 PutS, M, and n together to form binary result
 (Don’t forget the implied “1.” on the left of the
mantissa.)

-1.11110112 x 2n =
-1.11110112 x 23 =
-1111.10112

43

Converting from Floating Point

 Step 4
 Express result in decimal

-1111.10112
-15
15 2-1 = 0.5
2-3 = 0.125
2-44 = 0.0625
0.6875

Answer: -15.6875
15.6875
44
Converting to Floating Point

 Express 36.562510 as a 32-bit floating point

number (in hexadecimal)

45

Converting to Floating Point

 Step 1
 Express original value in binary

36.562510 =

100100.10012

46
 Converting to Floating Point

 Step 2
 Normalize

100100 10012 =
100100.1001

1.0010010012 x 25

47

Converting to Floating Point

 Step 3
 Determine S, E, and M

+1.0010010012 x 25
S M
n E = n + 127
= 5 + 127
= 132
= 100001002

S = 0 (because the value is positive)

48
Converting to Floating Point

 Step 4
 Put S, E, and M together to form 32-bit
32 bit binary result

0 10000100 001001001000000000000002
S E M

49

Converting to Floating Point

 Step 5
 Express in hexadecimal

0 10000100 001001001000000000000002 =
0100 0010 0001 0010 0100 0000 0000 00002 =

4 2 1 2 4 0 0 016

Answer: 4212400016
50