4.

Simplification of Boolean Functions 4-1

Chapter 4. Simplification of Boolean

Functions

Boolean Cubes and Boolean Functions

(a) n = 1 0 1

00 01

(b) n = 2

10 11

010 011

000 001

(c) n = 3
110 111

100 101

0010 0011

0000 0001

0110

0111
(d) n = 4
0100
0101

1110
1111

1100

1110

1011

1000 1001

Figure 1: Boolean cubes [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-2

☞ A Boolean n-cube uniquely represents a Boolean function of n variables if

each vertex is assigned a 1 (marked) or 0 (unmarked).
☞ Each vertex of the n-cube represents a minterm (a row in the truth table).
Example 1
Fig. 2 shows the truth table and the corresponding cube representations of the carry
and sum functions. 2
ci xi yi c i+1 s i
0 0 0 0 0
0 0 1 0 1
0 1 0 0 1
0 1 1 1 0
1 0 0 0 1
1 0 1 1 0
1 1 0 1 0
1 1 1 1 1

(a) Truth table

010 011

000 001

110 111

100 101

(b) Carry function c i+1

010 011

000 001

110 111

100 101

(c) Sum function s i

Figure 2: Boolean-cube representations for carry and sum functions [Gajski].

☞ (a) Each m-subcube of the n-cube represents 2m minterms with the same
m
n m literals, where m < n; (b) each m-subcube with 2 1-minterms
represents a product term of n m literals.
☞ A prime implicant (PI) is a subcube (of 1-minterms) that is not contained in
any other subcube (of 1-minterms); an essential prime implicant (EPI) is a PI
that contains a 1-minterm that is not contained in any other PI.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-3

Map Representation

☞ Complexity of digital circuit (gate count) / complexity of algebraic expres-

sion (literal count).
☞ A function’s truth-table representation is unique; its algebraic expression is
not. Simplification by algebraic means is awkward (from algorithmic point
of view).
☞ A Karnaugh map (K-map) is an array of squares each representing one minterm.
Simplification by the map method is straightforward.

x
0 1
(a) n = 1

y
x 0 1
m0 m1
0 (b) n = 2

1
m2 m3

yz
x 00 01 11 10
m0 m1 m3 m2
0
(c) n = 3
1
m4 m5 m7 m6

zw 00 01 11 10
xy

00
m0 m1 m3 m2

01
m4 m5 m7 m6 (d) n = 4

11
m12 m13 m15 m14

10
m8 m9 m11 m10

Figure 3: Boolean cubes and corresponding Karnaugh maps [Gajski].

☞ Each K-map defines a unique Boolean function.

✏ A Boolean function can be represented by a truth table, a Boolean ex-
pression, an n-cube, or a map.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-4

☞ K-map is in fact a visual diagram of all possible ways a function may be

expressed—the simplest one can easily be identified.
✏ K-maps provide visual aid to identify PIs and EPIs.
✏ They are used for manual minimization of Boolean functions.
Exercise 1
How do you transform a K-map into a truth table? Is it unique? How do you
transform a K-map into an n-cube? Is it unique? 2

Two- and Three-Variable Maps

x y f y y y y
x 0 1 x 0 1 x 0 1 x 0 1
m0 m0 m 1
0 0
m1
0 xy xy
0 0 0
0 0 1 0 1
0 1 m2 m 3
1 0 m2 1 xy xy 0
1 1 1 1 1 1 1
m3 x+y
1 1 2-var map 2-var map f = xy (x  y) 0

Figure 4: Two-variable maps.

yz yz yz
x 00 01 11 10 x 00 01 11 10 x 00 01 11 10
m0 m1 m3 m2 0 x y z x y z x yz x yz
0 0 0 0 0 0 0 0
0 1 1 0 1 1
m4 m5 m7 m6 1 xy z xy z xyz xyz
0 0 0 0
1 1 1 1 1 1
3-var map 3-var map f=z x y + xy
0 0

Figure 5: Three-variable maps.

YZ Y
X 00 01 11 10

m0 m1 m3 m2 0 XY Z XY Z XY Z XY Z

m4 m5 m7 m6 X 1 XY Z XY Z XY Z XY Z

Z
(a) (b)
Fig. 2-10 Three-Variable Map

Figure 6: Three-variable map simplification [Mano & Kime].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-5

☞ Minterms are arranged in the Gray-code sequence. (Why?)

☞ Any 2 (horizontally or vertically) adjacent squares differ by exactly 1 vari-
able, which is complemented in one square and uncomplemented in the other.
☞ Any 2 minterms in adjacent squares that are ORed together will cause a re-
moval of the different variable, e.g., m5 + m7 = xy z + xyz = xz (y + y ) = 0 0

xz , because y + y = 1.
0

x iyi
ci 0 01 11 10
0 1 3 2

0 1
4 5 7 6

1 1 1 1

x iyi
ci 0 01 11 10
0 1 3 2

0 1 1
4 5 7 6

1 1 1

Figure 7: K-maps for the carry and sum functions [Gajski].

Example 2
P
F2 (X; Y ; Z ) =
P
Simplify the following Boolean functions: (a) F1 (X; Y ; Z )
(0; 2; 4; 5; 6).
= (3; 4; 6; 7); (b)

YZ Y YZ Y
X 00 01 11 10 X 00 01 11 10

0 1 0 1 1

X 1 1 1 1 X 1 1 1 1

Z Z
(a) F1 (X, Y, Z) = Σm (3, 4, 6, 7) (b) F2 (X, Y, Z) = Σm(0, 2, 4, 5, 6)
= YZ + XZ = Z + XY
Fig. 2-14 Maps for Example 2-4

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-6
Example 3
Simplify the Boolean function F (X; Y ; Z ) =
P (1; 3; 4; 5; 6).
YZ Y
X 00 01 11 10

0 1 1

X 1 1 1 1

Z
Fig. 2-15 F( X, Y , Z ) = Σm(1, 3, 4, 5, 6)

2
Example 4
0
Simplify the Boolean function F = X Z + X 0Y + X Y 0Z + Y Z .
YZ Y
X 00 01 11 10

0 1 1 1

X 1 1 1

Z
Fig. 2-16 F( X, Y , Z ) = Σ m (1 ,2 , 3, 5, 7)

Four-Variable Maps
yz yz yz
wx 00 01 11 10 wx 00 01 11 10 wx 00 01 11 10
00 0 1 3 2 00 1 1 1 00 1 1 1
01 4 5 7 6 01 1 1 1 01 1
11 12 13 15 14 11 1 1 1 11
10 8 9 11 10 10 1 1 10 1 1 1
wz 0 0
+ y 0
+ xz 0
xz
0 0
+ xy
0 0
+ w yz
0 0

Figure 8: Four-variable maps.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-7

☞ The map is considered to lie on a surface with the top and bottom edges, as
well as the right and left edges, touching each other to form adjacent squares.
✏ One square ) a minterm of 4 literals.
✏ Two adjacent squares ) a term of 3 literals.
✏ Four adjacent squares ) a term of 2 literals.
✏ Eight adjacent squares ) a term of 1 literal.
✏ Sixteen adjacent squares ) the constant ‘1’.
Example 5
For the maps shown above, f1 (w; x; y; z ) =
P (0; 1; 2; 4; 5; 6; 8; 9; 12; 13; 14) and
f2 = w x y + x yz + w xyz + wx y .
0 0 0 0 0 0 0 0 0
2
☞ A PI is a product term obtained by combining the maximum possible number
of adjacent squares in the map.
☞ Pick the EPIs that minimize the number of literals.

Minimize the number of groupings.

Maximize their sizes.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-8
Example 6
0 0
Show the region in the K-map that is represented by X Z .
YZ Y
WX 00 01 11 10 XZ

00 0 1 3 2 2
0 10
8
01 4 5 7 6
X
11 12 13 15 14
W 11
9 3
10 8 9 11 10 1

XZ Z
(a) (b)
Fig. 2-18 Four-Variable Map: Flat and on a Torus to Show Adjacencies

2
Example 7
Simplify the Boolean function F (W; X; Y ; Z ) =
P (0; 1; 2; 4; 5; 6; 8; 9; 12; 13; 14).
YZ Y
WX 00 01 11 10

00 1 1 1

01 1 1 1
X
11 1 1 1
W
10 1 1

Z
Fig. 2-19 Map for Example 2-5: F = Y + WZ + XZ

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-9
Example 8
Simplify the Boolean function F = 0
AB C
0 0
+ B 0 C D 0 + AB 0 C 0 + A0 BC D 0 .
CD C
AB 00 01 11 10

00 1 1 1

01 1
B
11
A
10 1 1 1

D
Fig. 2-20 Map for Example 2-6: F = BD + BC + ACD

*Five-Variable Maps
A=0 A=1 A=0 A=1
DE DE DE DE
BC 00 01 11 10 BC 00 01 11 10 BC 00 01 11 10 BC 00 01 11 10
00 0 1 3 2 00 16 17 19 18 00 1 1 00
01 4 5 7 6 01 20 21 23 22 01 1 1 01 1 1
11 12 13 15 14 11 28 29 31 30 11 1 11 1 1
10 8 9 11 10 10 24 25 27 26 10 1 10 1
f=AB E
0 0 0
+ BD E + ACE
0

Figure 9: Five-variable maps.

☞ Imagine that the 2 maps are superimposed on one another.

☞ It is possible to construct a 6-variable map with four 4-variable maps by
following a similar procedure.
☞ Maps of 6 or more variables are hard to read, and thus are impractical.
✏ (1) Variable-entered maps (VEMs); (2) CAD programs.
☞ Any 2k adjacent squares, k = 0; 1; : : : ; n, in an n-variable map represent an
area that gives a product term of n k literals.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-10

v=0 v=1

{
{
zw
xy 00 01 11 10 00 01 11 10
0 1 3 2 16 17 19 18

{
00 m0 m1 m3 m2 m16 m17 m19 m18
4 5 7 6 20 21 23 22

01 m4 m5 m7 m6 m20 m21 m23 m22

u=0 12 13 15 14 28 29 31 30

11 m12 m13 m15 m14 m28 m29 m31 m30

8 9 11 10 24 25 27 26

10 m8 m9 m11 m10 m24 m25 m27 m26

32 33 35 34 48 49 51 50

{
00 m32 m33 m35 m34 m48 m49 m51 m50
36 37 39 38 52 53 55 54

01 m36 m37 m39 m38 m52 m53 m55 m54

u=1 44 45 47 46 60 61 63 62

11 m44 m45 m47 m46 m60 m61 m63 m62

40 41 43 42 56 57 59 58

10 m40 m41 m43 m42 m56 m57 m59 m58

v=0 v=1
{
{
zw
xy 00 01 11 10 00 01 11 10
0 1 3 2 16 17 19 18

{
00
4 5 7 6 20 21 23 22

01
u=0 12 13 15 14 28 29 31 30
x’v
11
8 9 11 10 24 25 27 26

10

32 33 35 34 48 49 51 50

{
00
36 37 39 38 52 53 55 54

01
u=1 44 45 47 46 60 61 63 62 xz
11
40 41 43 42 56 57 59 58

10

z ’w ’

Figure 10: Six-variable K-maps [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-11
Theorem 1
In a K-map, 2 minterms are adjacent iff they differ in exactly 1 variable.

✯ In general, for an n-variable function, where n = 2k , we can construct a k -D

K-map such that:
✩ the map is extended in each dimension on 2 of the variables;
✩ the extension sequence in each dimension is a Gray sequence.

Simplification Using K-Maps

☞ The complement of a function is represented in the map by the squares not

marked by 1s (they usually are marked by 0s).

CD CD
AB 00 01 11 10 AB 00 01 11 10
00 1 1 1 00 0
01 1 1 1 01 0
11 1 1 1 11 0
10 1 1 10 0 0
f = 0
A D
0
+ C + BD0 0
f
0
= CD + AB 0 C

0
Figure 11: The complement of f gives f in pos.

Example 9
We want to simplify the following function in sop & pos:

f (A; B; C; D ) =
X (0; 1; 2; 5; 8; 9; 10):

(a) Mark the map with 1s and 0s according to the function.

(b) Use the 1s to determine the EPIs of f , which immediately give the sop form:
0 0 0 0 0 0
f = B D +B C +AC D:

0
(c) Use the 0s to determine the EPIs of f , and complement it to give the pos form:
0 0
f = AB + C D + BD ;

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-12

B’  A’

D’ B’

C’
C’  f D’ f

A’
D D

Figure 12: Two-level implementation.

0 0 0 0 0
f = (A + B )(C + D )(B + D ):

The 2-level (sop/pos) implementations using the AND/OR gates are shown below.

Note that NOT gates (inverters) are required if complemented inputs are not avail-
able. 2
☞ The 1s of the function in the K-map or the truth table represent the minterms,
and the 0s represent the maxterms.
☞ Entering a function in the map:
✏ Entering a function expressed in sop in the map is straightforward.
✏ To enter a function expressed in pos in the map, take the complement of
the function (to get sop) and from it find the squares to be marked by 0s.
The remaining squares are marked by 1s.
☞ Simplification procedure:
✏ Obtain truth table, canonical form, or standard form.
✏ Generate K-map.
✏ Determine PIs.
✏ Select EPIs.
✏ Find minimal cover (set) of PIs.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-13

CD C CD C
AB 00 01 11 10 AB 00 01 11 10

00 1 00 1

01 1 01 1
B B
11 1 1 1 11 1 1 1
A A
10 1 1 10 1 1

D D
(a) Plotting the minterms (b) Essential prime implicants

Fig. 2-22 Simplification with Prime Implicants in Example 2-8

Figure 13: Simplification with PIs for f =

P (0; 5; 10; 11; 12; 13; 15) [Mano & Kime].

CD C
AB 00 01 11 10 3
00 1 1 1

01 1 1
B
11 1 1
A
10
1 1

D
1 2
Fig. 2-23 Map for Example 2-9

Figure 14: Simplification with PIs for f =

P (0; 1; 2; 4; 5; 10; 11; 13; 15) [Mano & Kime].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-14
Exercise 2
Does the minimal cover of PIs contain only EPIs? 2
Example 10

F (A; B; C; D ) =
P
Simplify the following Boolean function in the pos form:
(0; 1; 2; 5; 8; 9; 10).
CD C
AB 00 01 11 10

00 1 1 0 1

01 0 1 0 0
B
11 0 0 0 0
A
10
1 1 0 1

D
Fig. 2-24 Map for Example 2-10: F = (A + B) (C + D) (B + D)

F
0
= AB + C D + BD 0 )F = (A0 + B 0 )(C 0 + D 0 )(B 0 + D ) 2
Exercise 3
Simplify F = (A0 + B 0 + C )(B + D ) in the pos form. 2

Don’t-Care Conditions

☞ In practice, there are applications where the function is not specified for cer-
tain combinations of the input variables. For example, in the 4-bit BCD code
for the decimal digits, the outputs are unspecified for the input combinations
1010-1111.
☞ Functions that have unspecified outputs for some input combinations are
called incompletely specified functions.
☞ The unspecified minterms of a function are called the don’t-care conditions,
or simply the don’t-cares, and are denoted as Xs.
☞ These don’t-care conditions can be used on a map to provide further simpli-
fication of the Boolean expression.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-15

☞ Each X can be assigned an arbitrary value, 0 or 1, to help the simplification

procedure.
Example 11
P
P
Simplify the Boolean function f (w; x; y; z ) =
care conditions d(w; x; y; z ) = (0; 2; 5).
(1; 3; 7; 11; 15) that has the don’t-

yz yz
wx 00 01 11 10 wx 00 01 11 10
00 X 1 1 X 00 X 1 1 X
01 0 X 1 0 01 0 X 1 0
11 0 0 1 0 11 0 0 1 0
10 0 0 1 0 10 0 0 1 0
f = yz + w 0 x0 f = yz + w0 z

Figure 15: Simplification using don’t-cares.

2
☞ Either one of the above expressions satisfies the conditions stated.
☞ Note that the above 2 expressions represent 2 functions that are algebraically
unequal: each covers different don’t-care terms.
☞ We may or may not include any of the Xs, while all the 1s must be included.
☞ It is also possible to obtain a simplified pos expression using the don’t-cares.

Tabulation (Quine-McCluskey) Method

☞ The map method is convenient when the number of variables does not exceed
5 or 6.
☞ The map method is essentially a trial-and-error method that does not offer
any guarantee of producing the best realization.
☞ The map method’s dependence on the somewhat intuitive human ability to
recognize patterns makes it unsuitable for design automation (programming
for a digital computer).

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-16

☞ The tabulation method is a specific step-by-step (algorithmic) procedure that

is guaranteed to produce a simplified standard-form expression for a function.
☞ It can be applied to problems with any number of variables.
☞ It is suitable for machine computation.
☞ It however is quite tedious for human use.
☞ The tabulation method was first formulated by Quine (1952) and later im-
proved by McCluskey (1956), thus known as the Quine-McCluskey method.
✏ Step 1: Exhaustive search of all PIs.
✍ Group minterms by number of 1s.
✍ Compare minterms and find pairs with distance 1.
✍ Generate subcubes.
✍ Repeat the above procedure on generated subcubes until no more
subcubes can be generated.
✏ Step 2: Choose among the PIs which give an expression with the least
amount of literals (minimal cover generation).
✍ Find EPIs through a selection table.
✍ Find minimal cover through the pos of the PIs.

Determination of PIs

☞ The goal of the Quine-McCluskey algorithm is a special case of the follow-

ing: Select the smallest possible subset, D, of a set of objects, A, so that some
criterion, f , is satisfied.
☞ The set D will consist of all products in the minimal sop realization of a given
Boolean function f , of which the set A contains all possible products.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
Let f =
P
4. Simplification of Boolean Functions

(0; 1; 2; 8; 10; 11; 14; 15).

4-17

(a) (b) (c)

wxyz wxyz wxyz

0 0000 ✓ 0,1 000– 0,2,8,10 –0–0

0,2 00–0 ✓ 0,8,2,10 –0–0
1 0001 ✓ 0,8 –000 ✓
2 0010 ✓ 10,11,14,15 1–1–
8 1000 ✓ 2,10 –010 ✓ 10,14,11,15 1–1–
8,10 10–0 ✓
10 1010 ✓
10,11 101– ✓
11 1011 ✓ 10,14 1–10 ✓
14 1110 ✓
11,15 1–11 ✓
15 1111 ✓ 14,15 111– ✓

① Group the minterms according to the number of 1s (see Column (a)).

② Combine any 2 minterms that differ from each other by exactly one variable
(i.e., dist-1), the unmatched variable removed (see Column (b)). Try this for
all possible pairs of minterms. A check (✓) is placed to the right of both
minterms if they have been used in a match.
③ Repeat the process. Combine any 2 product terms from Step ② that differ
from each other by exactly one variable, the unmatched variable removed
(see Column (c)).
④ The unchecked terms in the table form the PIs. Some of the product terms
may appear twice in the table. It of course is unnecessary to use the same
term twice.

☞ When comparing 2 terms to decide if they can be combined, the comparison

can be done directly on the decimal numbers. We combine 2 terms iff the
difference of their corresponding decimal numbers is a power of 2.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-18

☞ In most cases, the sum of PIs obtained from the procedure shown above is not
necessarily a standard form (which is minimized). The reason is that some of
the PIs may be redundant.
Exercise 4
Use the map method to simplify f =
P (0; 1; 2; 8; 10; 11; 14; 15), and compare the
result with the tabulation method. 2

Minimal Cover Generation

The selection of PIs that form the minimized function is made from a PI table:
each PI is represented in a row and each minterm in a column.
0 0 0 0 0 0
Suppose we have the following PIs: ,
x y z w xz , w xy , xyz , wyz , and wx .

PI Minterms 1 4 6 7 8 9 10 11 15
✓ 0
x y z
0
1,9 X X
✓ 0
w xz
0
4,6 X X
0
w xy 6,7 X X
xyz 7,15 X X
wyz 11,15 X X
✓ wx
0
8,9,10,11 X X X X
✓ ✓ ✓ ✓ ✓ ✓ ✓

① Check the PIs that cover minterms with a single X in their columns. These
PIs are the EPIs.
② Check each column whose minterm is covered by the selected EPIs.
③ If there are minterms left uncovered (7 & 15 in this example), some non-
essential PIs have to be selected to cover them. We of course will select the
PIs with a smallest total number of literals (xyz in this example).

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-19

*Multi-Output Minimization Using Maps

☞ Identify all possible PIs that cover each implicated minterms in each output
expression, and search for a minimal cover by using shared terms.
☞ The Quine-McCluskey method also can be extended for this purpose.

X
f 1 (A; B; C ) =
X (0; 2; 3; 5; 6)

f 2 (A; B; C ) =
X (1; 2; 3; 4; 7)

f 3 (A; B; C ) = (2; 3; 4; 5; 6)

BC BC BC
A 00 01 11 10 A 00 01 11 10 A 00 01 11 10
0 1 1 1 0 1 1 1 0 1 1
1 1 1 1 1 1 1 1 1 1
f1 2
f f3

Figure 16: Simplification of multiple output functions.

1
0 0 0 0 0
f = A C + BC + A B + AB C

f2
0 0 0 0
= A C + BC + AB C +AB
f3
0 0 0 0 0
= BC + A B + AB C + AB C

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-20

*XOR & XNOR Patterns on the Map

☞ Inspect the K-map to detect XOR/XNOR patterns: 1) kitty-corner adjacen-

cies; 2) offset adjacencies.

BC BC
A 00 01 11 10 A 00 01 11 10
0 1 0
1 1 1 1 1
kitty-corner offset
(A  B )C A(B  C)

CD CD CD
00 01 11 10 00 01 11 10 00 01 11 10
AB AB AB
00 1 1 00 1 00 1
01 1 1 01 1 1 1 01 1
11 11 1 11 1
10 10 1 1 1 10 1
f1 f2 f3

Figure 17: XOR/XNOR patterns on the map.

1
0 0 0 0 0 0 0 0 0 0
f = A B C D +AB CD + A BC D + A BC D
= A B
0 0
(C  D) + A B (C
0
 D)
0

= A
0
[(B )(C
0
 D) + (B )(C  D)
0
]
= A [B
0
(  C D )]

2
0 0 0 0 0 0
f = A C D + AC D + A BC + AB C

= C
0
(A  D) + C (A  B)

3
0 0 0 0 0 0 0 0
f = A B CD + A BC D + ABC D + AB C D

= (A  C )(B  D)

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-21

Two-Level Implementations

☞ Digital circuits are more frequently constructed with NAND/NOR gates than
with AND/OR/NOT gates due to ease of fabrication. For example, in gate
arrays, only NAND (or NOR) gates are used.
☞ The conversion process from an expression/schematic with AND, OR, and
NOT gates to one with only NAND or NOR gates is an example of technol-
ogy mapping.

NAND gate or f = x
0
+ y 0 + z 0 = (xyz )0

NOR gate or f = 0
x y z
0 0
= (x + y + z )0

Figure 18: NAND/NOR gates.

✯ NAND-NAND implementation:
✩ Simplify the function and express it in sop:

f = AB + CD + E:

✩ Transfer it to 2-level NAND-NAND expression (DeMorgan’s Thm):

0 0 0 0 0 0
f = ((AB + C D + E ) ) = ((AB ) (C D ) E ) :

✩ Draw the corresponding NAND gate implementation. Note that a 1-input

NAND gate can replace an inverter.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-22

☞ AND-OR  NAND-NAND
☞ The process can be done directly on the logic diagram.

✯ NOR-NOR implementation:
✩ Simplify the function and express it in pos:

f = (A + B )(C + D )E :

✩ Transfer it to 2-level NOR-NOR expression (DeMorgan’s Thm):

0 0 0 0 0 0
f = (((A + B )(C + D )E ) ) = ((A + B ) + (C + D ) + E ) :

✩ Draw the corresponding NOR gate implementation. Note that a 1-input

NOR gate can replace an inverter.

☞ It is the dual of the NAND-NAND implementation.

☞ OR-AND  NOR-NOR
☞ The process can be done directly on the logic diagram.
☞ The types of gates most often found in ICs are NAND & NOR, so NAND &
NOR logic implementations are the most important from a practical point of
view.

✯ NAND-AND & AND-NOR implementations:

✩ Both perform the AND-OR-INVERT (AOI) function.
✯ OR-NAND & NOR-OR implementations:
✩ Both perform the OR-AND-INVERT (OAI) function.

☞ Because of the INVERT part in each case, it is convenient to use the simpli-
0
fication of f (the complement of the function) instead of f .

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-23
yz
x 00 01 11 10
0 1 0 0 0 f = 0
x y z
0 0
+ xyz 0

1 0 0 0 1 f
0
= x y
0
+ xy 0 + z

Figure 19: NAND/NOR gates.

Example 12
Consider the function f given in Fig. 19.
AND-OR: f = x y z + xyz .
0 0 0 0

0 0 0 0 0 0 0 0 0
NAND-NAND: f = (f ) = [(x y z ) (xyz ) ] .
OR-NAND: f = [(x + y + z )(x + y + z )] . 0 0 0

NOR-OR: f = (x + y + z ) + (x + y + z ) . 0 0 0 0

0 0 0 0 0
AND-NOR: f = (f ) = (x y + xy + z ) .
NAND-AND: f = (x y ) (xy ) z . 0 0 0 0 0

0 0 0
OR-AND: f = (x + y )(x + y )z .
NOR-NOR: f = (f ) = [(x + y ) + (x + y ) + (z ) ] .
0 0 0 0 0 0 0 0 0
2
Example 13
Implement F (X; Y ; Z ) =
P (1; 2; 3; 4; 5; 7) with NAND gates.
YZ Y
X 00 01 11 10

0 1 1 1
F = XY + XY + Z
X 1 1 1 1

(a)

X X
Y
• Y
•
X • X
• • F • • F
Y • Y

Z • Z

(b) (c)

Fig. 2-30 Solution to Example 2-12

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-24
Example 14
Implement F = (AB 0 + A0 B )(E (C + D 0 )) with NOR gates.
A
B
A
F
B
E
C
D

(a) AND – OR gates

A
•
B •
A • • • F
B
•
E
C • X • •
D •
(b) NAND gates
Fig. 2-32 Implementing F = ( A B + AB ) E (C + D)

Exclusive-OR (XOR) Gates

☞ Some identities for XOR:

✩ X 0= ; 1= . X X X
0

✩ X  = 0;  = 1.
X X X
0

✩ X  = (  ); 
Y
0
X Y
0
X
0
Y = (X  Y )0 .

☞ The XOR function is both commutative and associative.

✩ A  B = B  A .
✩ (A  B)  C = A (  B C) = A   B C .
Exercise 5
(a) How do you implement the Even function with XOR (XNOR) gates?
(b) Design an n-bit even-parity generator. Calculate the hardware cost (in terms
of primitive gates) and performance (in terms of primitive-gate delays). 2

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-25

X •
•
• •
• • • F=X Y
•
Y •
Fig. 2-37 Exclusive-OR Constructed with NAND Gates

Figure 20: Implementing the XOR function with NAND gates [Mano & Kime].

YZ Y CD C
X 00 01 11 10 AB 00 01 11 10

0 1 1 00 1 1

X 1 1 1 01 1 1
B
Z 11 1 1
A
(a) X Y Z 10 1 1

D
(b) A B C D

Fig. 2-38 Maps for Multiple-Variable Odd Functions

Figure 21: Implementing the Odd function with XOR gates [Mano & Kime].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-26

*Technology Mapping

☞ A gate array is a 2-dimensional array of cells within which each cell contains
a single NAND (NOR) gate that has a fixed number (usually 3) of inputs.
☞ The conversion process from an expression/schematic with AND, OR, and
NOT gates to one with only NAND or NOR gates is an example of technol-
ogy mapping.

0 0
Rule 1 : xy = ((xy ) )
0 0 0 0 0
Rule 2 : x + y = ((x + y ) ) = (x y )
0 0 0 0 0
Rule 3 : xy = ((xy ) ) = (x + y )
0 0
Rule 4 : x + y = ((x + y ) )

Standard NAND NOR

form implementation implementation

sum−of−products

product−of−sums

Figure 22: Conversion standard forms to NAND and NOR implementations [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-27
Example 15
Consider the carry function again:

ci +1 = xiyi + xici + yi ci = (xi + yi )(xi + ci )(yi + ci ):

The NAND and NOR implementations of the carry function are shown in Fig 23.
2
2.4 1.4
xi ci+1 xi ci+1
yi 2.4 2.4 yi 1.4 1.4

ci ci
2.4 1.4

(c) NAND implementation

2.4 1.4
xi ci+1 xi ci+1
yi 2.4 2.4 yi 1.4 1.4

ci ci
2.4 1.4

(d) NOR implementation

Figure 23: NAND and NOR implementations of the carry function [Gajski].

☞ Replace AND and OR gates with NAND gates by using Rules 1 and 2, and
eliminate double inverters whenever possible.
☞ Replace AND and OR gates with NOR gates by using Rules 3 and 4, and
eliminate double inverters whenever possible.
☞ Term decomposition: each n-input gate is decomposed into a tree of m-input
gates, where n > m. For example, Fig 24 shows the decomposition of a
10-input AND gate into 3-input AND gates.
✏ The tree has dlogm ne levels and d(n 1)=(m 1) e m-input gates.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-28

2.4 2.4 2.4 2.4 2.4 2.4

2.4 2.4

2.4 2.4

(b) One possible decomposition (c) Alternative decomposition

Figure 24: Decomposition of a 10-input AND gate into 3-input AND gates [Gajski].

☞ Retiming: performance optimization (delay time minimization)—decomposition

of a large gate can produce a tree in which different paths may incur different
delays. For example, Fig 25 shows two NAND implementations of the 4-bit
carry-look-ahead function:

c 4 = g3 + p3 g2 + p3p2 g1 + p3 p2p1 g0 + p3p2p1 p0c0 :

✏ Carry generate function: gi  xi yi .
✏ Carry propagate function: pi  xi + yi .

✏ Carry look-ahead function: +1 = gi + pi ci .

ci

☞ Critical path: the delay on the longest input-output path.

☞ The rule of thumb is to minimize delay on critical paths and minimize cost
on non-critical paths. A technology mapping example is shown in Fig 26.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-29

g3
p
g2 3

p
p2 3
g1
p
3 p c4
p1 2
g0
p
3 p
p1 2
p
c0 0
(a) AND−OR implementation
g3

p
3
g2
p
p2 3
g
p 1
3 c4
p2
p1
g0
p
3
p
p1 2
p0
c0

(b) Decomposition of (a)

g3

p
3 1.8
g2
p
p2 3
1.8
g
p 1
3
p2 1.8 1.8 c4
p1 1.8 1.8
g0
p
3
p 1.8
p1 2 Max delay = 9.2 ns
p0 1.8
c0

(c) NAND implementation of (b)

g3

p
3
g2
p
p2 3
g
p 1
3
p2 c4
p1
g0
p
3
p
p1 2
p0
c0

(d) Performance optimized decomposition

g3

p
3 1.8 1.8
g2
p
p2 3
1.8
g
p 1
3
p2 1.8 c4
p1 1.8 1.8
g0
p
3
p 1.8
p1 2 Max delay = 6.4 ns
p0 1.8
c0

(e) Performance optimized NAND implementation

Figure 25: Two NAND implementations of the 4-bit carry-look-ahead function [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-30

y 2.4
2.4

w 1
2.4 F
2.4
z 1

(a) AND−OR implementation (Delay = 7.2ns, Cost = 28)

y 1
1.4
1.4

1.4 F
w 1
1.4
z 1

(b) NAND implementation (Delay = 5.2ns, Cost = 22)

B
y 1 A
1.4
1.4

1.4 F
w 1
1.4
z 1

(c) Two possible conversions

y 1 1.4

2.0 1 F
w 1
z 1

(d) Alternative A (Delay = 5.4ns, Cost = 20)

y
2.0

1.4 F
w 1
1.4
z 1

(e) Alternative B (Delay = 3.8ns, Cost = 20)

y
2.0

1.4 F
w
1.4 1
z

(f) Cost optimized alternative B (Delay = 3.8ns, Cost = 18)

Figure 26: Technology mapping example [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-31

Hazard-Free Design
Definition 1
Hazards are unwanted switching transients (glitches) that appear at the output of
a circuit because of different propagation delays on different but converging paths
through the circuit.

.. .. static 0-hazard static 1 hazard dynamic hazards

Figure 27: Hazards.

☞ Hazards can cause malfunction.

☞ Static-1 hazard: there are two 1-minterms that differ in only one variable but
are not covered by a common product term in a sop implementation (Fig. 28).
✏ It can be eliminated by including an additional PI covering both adjacent
1-minters (called the consensus term) (Fig. 29).
☞ Static-0 hazard: there are two 0-minterms that differ in only one variable but
are not covered by a common sum term in a pos implementation.
☞ Static hazards are caused by two complementary signals which become equal
for short periods of time due to different delays on different paths through the
circuit.
☞ Dynamic hazard: two signals that always have the same value (even during
the transition) become different for a short period of time.
✏ It is a static hazard that occurs during the output transition.
✏ It occurs when the same variable value propagates through the circuit on
two different paths with different delays.
✏ See Fig. 30 for an example.
✍ It can be eliminated by introducing a redundant (consensus) PI.
✍ It also can be eliminated by inserting a delay element.

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-32

yz
x 0 01 11 10
0 1 3 2

0 1
4 5 7 6

1 1 1 1

F = xy’ + yz
(a) Map representation

x a
2.4
1
y 2.4 F
2.4
z b
(b) Logic schematic

a 3.4

b 2.4

F 4.8 1.0

t0 t1 t 2
(c) Timing diagram

Figure 28: A design with static-1 hazard [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-33

yz
x 0 01 11 10
0 1 3 2

0 1
4 5 7 6

1 1 1 1

F = xy’ + yz + xz

(a) Map representation

x a
2.4
1
y
b
z
2.4 2.4 F

2.4
c

(b) Logic schematic

a 3.4

b 2.4

(c) Timing diagram

Figure 29: Hazard-free design [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005
4. Simplification of Boolean Functions 4-34

Static 1−hazard
x a
2.4
1 e
y 2.4
2.4
z b 2.4 F

w 2.4 d

(a) Logic schematic

4.8
e 1.0

3.4
d

2.4
F 1.4 1.0

t0 t1 t2

(b) Timing diagram

Figure 30: A design with dynamic hazard [Gajski].

c Cheng-Wen Wu, Lab for Reliable Computing (LaRC), EE, NTHU 2005