Chapter 5

Test Generation for

Combination & Sequential
Circuits

Jin-Fu Li
Advanced Reliable Systems (ARES) Lab
Department of Electrical Engineering
National Central University
Jungli, Taiwan
 Outline
 Basics
 ATPG Algorithms for Combinational Circuits
 Boolean Difference
 Single-Path Sensitization
 D-Algorithm
 PODEM
 Redundancy Identification
 Problems of Sequential Circuit testing
 ATPG Approaches for Sequential Circuits
 Time-Frame Expansion
 Simulation-Based Approach
 Scan
 Summary

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 2

 Functional vs. Structural Test
 Consider a 64-bit adder as shown below

s/0, s/1
A B Cin0 Ai
Sumi
s/0, s/1 s/0, s/1
64 64 Bi
Cini s/0, s/1
s/0, s/1

s/0, s/1 s/1

Ai s/0
Bi
64 s/0, s/1 s/1
s/0
Cin i+1
s/0, s/1
Carry Sum Cin i s/0, s/1
s/1
s/0

s/1

Functional Test Structural Test

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 3

 Functional vs. Structural Test
 Functional test
 Generate complete set of tests for circuit input-
output combinations
 129 inputs & 65 outputs
 2129=680,564,733,841,876,926,926,749,214,
863,536,422,912 test patterns are required
 Using 1 GHz ATE, would take 2.15 x 1022 years
 Structural test
 64 bit slices and each slice has 27 faults (using
fault collapsing)
 At most 64x27=1728 faults, thus only 1728
test patterns are required
 Takes 0.000001728 seconds on 1 GHz ATE

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 4

 Circuit & Binary Decision Tree
 All ATPG programs need a data structure
describing the search space for test
patterns
 Binary decision tree

A’ A

A D
B’ B B’ B
B
C
C’ C C’ C C’ C C’ C
0 1 1 0 1 0 0 1

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 5

 Binary Decision Diagram
 Binary decision diagram (BDD)
 Follow path from source to sink node – product
of literals along path gives Boolean value at
sink
 Rightmost path: A B’ C’ = 1
 Problem: size varies greatly
A’ A
with variable order
B B
B’ B’

C C
C’ C’
0 1

Gate level
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 6
 Algorithm Completeness
 Algorithm is complete if it ultimately can
search entire binary decision tree, as
needed, to generate a test
 Untestable fault – no test for it even after
entire tree searched
 Combinational circuits only – untestable
faults are redundant, showing the presence
of unnecessary hardware

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 7

 Algorithm Types
 Exhaustive test generation
 Completely exercise the fault-free behavior
 Appropriate only when the number of PIs is
small
 Detects all the universal faults (i.e., all
combinational faults)
 Pseudoexhaustive test generation
 Test most of universal faults by applying
exhaustive test on subsets of PIs
 Pseudorandom test generation
 Generate test pattern deterministically
 Patterns have many characteristics of random
patterns but are repeatable
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 8
 Algorithm Types
 Algorithmic (deterministic) test generation
 Algebraic (symbolic) techniques
 SPOOFs
 Line condition equations
 Boolean difference
 Path-oriented techniques
 Single-path sensitization
 D-algorithm
 PODEM
 FAN
 Produces higher-efficiency test patterns, but its
cost is more expensive

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 9

 Boolean Difference
 Shannon’s Expansion Theorem
 An arbitrary Boolean function f(x1,x2,…,xn) can
be expanded about any variable
 For example, if we expand the function with
respect to x2, then the function can be
expressed as below
 f(x1,x2,…,xn)=x2.f(x1,1,…,xn)+x2’.f(x1,0,…xn)
 Boolean Difference
 Let f(x)=f(x1,x2,…,xn) be the normal (fault-free)
output function realized by network N, and
fa(x1,x2,…,xn) be the faulty output function
resulting from a fault a in N.
 The test set for a fault a is Ta=f(x)  fa(x)

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 10

 Boolean Difference
x 2 x3
 Example: x1 00 01 11 10
0 00 00 11 00
f ( X )  x1 x2  x2 x3  x1x3 1 10 1 0 1 1 00
f x1 / 0 ( X )  x2 x3
f ( X )  f x1 / 0 ( X )  ( x1 x2  x2 x3  x1x3 )  x2 x3  x1 x2
 Tx1 / 0  {100,101}
 Definition 1
 fi (0)  f ( x1, x2 ,, xi1,0, xi1,, xn ) : f(X) with xi  0(i.e., xi / 0)
 fi (1)  f ( x1, x2 ,, xi1,1, xi1,, xn ) : f(X) with xi  1(i.e., xi / 1)
 Definition 2
 The boolean difference of f(X) with respect to
xi is defined as
df ( X )
 f i (0)  f i (1)  f ( x1 ,, xi ,, xn )  f ( x1 ,, xi ,, xn )
dxi
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 11
 Boolean Difference
df ( X )
  0  fi (0)  fi (1)
dxi

df ( X )
  1  f i (0)  f i (1)
dxi
 Let   xi / 0 and   xi / 1, then
T  f ( X )  f  ( X )  ( x i f i ( 0 )  x i f i (1))  f i ( 0 )
 x i f i (1) f i ( 0 )  x i f i (1) f i ( 0 )  x i ( f i (1)  f i ( 0 ))
df ( X )
 xi
dx i

T  f ( X )  f  ( X )  ( x i f i ( 0 )  x i f i (1))  f i (1)
 x i f i ( 0 ) f i (1)  x i f i ( 0 ) f i (1)  x i ( f i ( 0 )  f i (1))
df ( X )
 xi
dx i
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 12
 An Example of Boolean Difference
 The test set for   x1 /1 of the circuit shown
below can be derived as follows
x1 
x2 

x3

T  x1 df ( X )  x1 ( f ( 0 , x 2 , x 3 )  f (1, x 2 , x 3 ))
dx 1
 x1 [( x 2 x 3 )  ( x 2  x 3 )]
 x1 [( x 2  x 3 )( x 2  x 3 )  x 2 x 3 x 2 x 3 ]
 x1 x 2
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 13
 Single-Path Sensitization
 Definition
 We say a test t activates a fault  if it
generates an error (or a fault effect) by
creating different v (l ) and v (l ) values at the fault
site l . We say t propagates the error (fault
effect) to a PO z if it results in different v ( z )
and v (z) values
 Definition
 A line whose value in the test t changes in the
presence of the fault  is said to be sensitized
to  by t . A path composed of sensitized lines
is called a sensitized path.

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 14

 Procedure
 Fault activation or excitation
 Specify inputs so as to generate the
appropriate value at the fault site, i.e., 0 for
s/1 and 1 for s/0
 Fault propagation
 Select a path from the fault site to an output
and specify other signal values to propagate
the fault (error signal) along the path to the
output
 Line justification
 Specify input values so as to produce the signal
values specified in fault activation and fault
propagation, i.e., perform consistency check
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 15
 An Example
 Use single-path sensitization to derive a
test set for   a/1 in the following circuit
D b
G1
G5 f1
A a
B G2
C
G3
G6 f2

E G4
c

 Generate a appropriate value a=0. A=B=C=1

 Choose a path via G5b=1A=D=0.
Contradiction!
 Try another path via G6c=1C=1 and E=0.
OK! Therefore, T=ABCE’
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 16
 Definitions for Logic Manipulation
 A literal is defined as a single variable within a term that
may or may not be complemented
 A product term is an implicant of a function if the
function has the value 1 for all minterms of the product
term
 Any single 1 or group of 1s in the Karnaugh map of a
function F is an implicant of F
 If the removal of any literal from an implicant P results in
a product term that is not an implicant of the function,
then P is a prime implicant
 A product term is called a prime implicant of F if it cannot be
combined with another term to eliminate a variable (literal)
 If a minterm of a function is included in only one prime
implicant, that prime implicant is said to be essential
(essential prime implicant)
 A product term is an essential prime implicant of F if there is
a minterm that is only covered by that prime implicant
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 17
 Examples
 Karnough map of a function F
AB
CD 00 01 11 10
00 0 1 1 0
01 1 1 1 0
11 1 0 0 0
10 0 0 0 0

 Each of the coverings is a prime implicant

 BC’, A’C’D, A’B’D’

 F(A,B,C,D)=BC’+A’B’D’ (minimum # of PIs)

 Prime implicant A’C’D is a non-essential prime
implicant
 A PI is essential PI if it covers a minterm that
cannot be covered by any other PIs
 BC’ and A’B’D’ are essential PIs

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 18

 Definitions for ATPG Alg.
 Definitions
 The fault cone is the portion of a circuit whose
signals are reachable by a forward trace of the
circuit topology starting at the fault site
 A forward implication results when the inputs
to a logic gate are assigned to specific values
such that the output can be uniquely
determined
 Backward implication is the unique
determination of all inputs of a gate for given
output and possibly some of the inputs

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 19

 D-Algorithm
 Definition
 The singular cover of a logic gate is the
minimal set of input signal assignments needed
to represent essential prime implicants of the
logic gate
 D-Algorithm
 Based on a cubical algebra called D-calculus
 D: signal value that is 1 in normal circuit and 0
in faulty case (discrepancy), i.e., D=1/0
 D’: signal value that is 0 in normal circuit and 1
in faulty case, i.e., D’=0/1
 D and D’ may be defined the other way around

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 20

 D-Algorithm
 Composite logic values for D-Algorithm
v/va 0/0 1/1 0/1 1/0 x/x
Symbol 0 1 D’ D X

 For example, D’+0=0/1+0/0=(0+0)/(1+0)

= 0/1=D’
D’
D’
0

 The unspecified value (x) can be any value

in {0, 1, D, D’}, and it sometimes is
dented as u

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 21

 Intersection Rules
 Let v 1 , v 2 ,  , v n  { 0 ,1, x }, then
 1. x  v i  v i ,1  i  n vi if i  j
 2. if vi , v j  x then vi  v j 
D or D’ otherwise

 3. Composite logic values

 0 1 X
0 0 D’ 0
1 D 1 1
X 0 1 X

 For example:
 (1X1 01)  (X1X 01) = (111 01)
 (1X1 X1)  (01X X1) = (D11 11)

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 22

 Singular Cover & Singular Cubes
 Definition
 For a gate G realizing function f, the singular
cover of G is the set of the prime implicants of
f and f’. Each prime implicant of the singular
cover is called a singular cube or a primitive
cube
 A singular cube consists of the input subcube and the
output subcube: ( x1 ,, xn , y1 , ym )  (v1 , v2 ,, vn  m ),
where the xi’s are inputs, the yi’s are outputs, and vi
{0,1, x}
a b c a b c
0 0 1 0 0 1 Prime implicant of f
a 0 1 0 X 1 0
c PIs of f’
b 1 0 0 1 X 0
1 1 0

Truth table Singular cover

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 23
 Propagation D-Cubes (pdc)
 Definition
 The propagation D-cubes of a gate G are the
cubes that cause the output of G to depend
only on one or more of its specified inputs, i.e.,
cubes that propagate a fault on these inputs to
the output
a a b c
c 0 D D’
b
D 0 D’
D D D’
Propagation D-cubes

 Pdc’s can be derived by inspection or from the

singular cover (or truth table) using the intersection
rules

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 24

 Primitive D-Cubes for a Fault (pdcf)
 Definition
 The primitive D-cubes of a fault  associated
with a gate G are the cubes (with inputs
completely specified) that brings the influence
of  to the output of G, i.e., produce an error
signal (D or D’) at the output
a a b c a b c
c 1 X D’ 0 0 D
b
X 1 D’
for 
  c/ 0   c /1 for 
 Pdcf can be derived from the singular covers of
f and f

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 25

 Derive pdcf with Singular Covers
 Let p0 and p1 be the sets of singular cubes
of f with output coordinates of 0 and 1,
respectively, and q0 and q1 be the
corresponding sets in f
 p1  q0  D generated; p0  q1  D generated
 For example,
f f
a
b d a b c d a b c d
c 0 X X 1 0 X X 1
X 0 X 1 X X 0 1
  b/1 X X 0 1 1 X 1 0
1 1 1 0

p1  q0  {( D x1; D ), (101; D ), (1x D; D )}, p0  q1  {( D11; D ), (11D; D )}

Thus, pdcf of  is 101D
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 26
 D-Frontier
 Definition
 The set of gates whose output value is
currently x but have one or more D/D’ signals
on their inputs is called the D-frontier
 Fault propagation (D-drive) consists of selecting one
gate from the D-frontier and assigning values to the
unspecified gate inputs so that the gate outputs
become D or D’
 An empty D-frontier shows that backtracking should
occur
D-frontier

Multiple
sensitized paths

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 27

 D-Algorithm Procedure
 Procedure D-algorithm /*generating a test
for a single stuck fault  */
 1. Select a pdcf for  (fault activation/excitation)
 2. Sensitize all possible paths from the fault site
to a PO (fault propagation or D-drive)
 By intersections of the pdcf with pdc’s of successive
gates
 Continued (back to step 1 if necessary) until a primary
output has D or D’
 3. Develop a consistent set of primary input
values that will account for all lines set to 0 or 1
during the D-drive. If inconsistent, seek a
different path (or even a different pdcf). (line
justification, consistency check)
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 28
 Examples
 Use the D-algorithm to derive a test for   6/ 0 for the following
circuit
1 5
2 G1 7
G3
9
G5
3 G4
G2 8
4
 6

1 2 5 3 4 6 3 5 7 2 6 8 7 8 9
G1 X 1 0
G2 X 0 D
G3 X
0 X 1
G4 D
0 D D’
G5 1 D D’
1 X 0 0 1 0 D’ D 1 D’
0 0 1 0 X D 1 1 0 D D D’ D D D’
sc pdcf sc pdc pdc

1 2 3 4 5 6 7 8 9
1. Select a pdcf- X X 1 0 X D X X X
2. Pdcf-  pdc-G4   X 0 1 0 X D X D’ X
3.   pdc-G5 (polarity inverted) X 0 1 0 X D 1 D’ D
4. Check line7=1 from sc-G3  line5=0 X 0 1 0 0 D 1 D’ D
5. Check line5=0 from sc-G1  line1=1 1 0 1 0 0 D 1 D’ D
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 29
 Examples
 Try it again for the fault   d /1 for the following circuit
a
G3 f
c G6
i
G2 e
h
b G4 g
G1 d G5
a

a b c d e f g h i
1 1 X D’ X X X X X
1 1 X D’ 1 X D X X Thus 110 is a test
1 1 0 D’ 1 X D X X
1 1 0 D’ 1 1 D D’ X
1 1 0 D’ 1 1 D D’ D

 D-algorithm is complete; it will find a test for a given fault

if such a fault exists (i.e., the fault is not redundant). Given
a fault list, the algorithm proceeds with 1 fault at a time.
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 30
 PODEM
 Path Oriented DEcision Making
 Similar in principle to the D-algorithm but
different in approach (more efficient)
 Complete, like D-algorithm
 Treats a value vl to be justified for line l as
an objective ( l , v l) to be achieved via PI
assignment – direct search
 Allows assigning values only to primary
inputs (PIs)  backtracing can occur only
at the PIs, i.e., examine all possible PIs
implicitly but exhaustively, and terminate
as soon as a test is found
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 31
 Backtracing
 Consider the circuit shown below and the
objective (f,1)
e

f
c
a
b d

1. Assume that backtracing follows the path (f,d,b). Simulating

the assignment b=1 does not achieve the objective (f,1).
2. Executing again backtracing follows the path (f,d,c,a). Now
simulating the assignment a=0 achieves f=1.

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 32

 PODEM
 Backtracing maps a desired objective into a PI
assignment that is likely to contribute to
achieving the objective
 No values assigned during backtracing; values
assigned only by simulating PI assignments,
i.e., only by forward implication of PI
assignments ( values always self-consistent
& automatically justified)
 Viewed as a branch and bound search
algorithm over an n-dimensional space (n
variables)
 Five to fifteen times faster than D-algorithm

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 33

 PODEM Procedure
 Procedure PODEM
 1. (Initialization)  PI, PI  x
 2. Assign a 0 or 1 to a PI (for a given objective)
 3. Determine whether the current combination
of values on the PIs, assigned or unassigned,
constitutes a test. Stop if a test is found
 4. (Backtracing) If it is possible to generate a
test with additional assigned PIs, go to 2
 5. (Backtracing) If  input pattern which has
not been examined as a possible test, go to 3,
else the fault is redundant (undetectable)

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 34

 Irredundant Hardware and Tests
 Combinational ATPG can find redundant
(unnecessary) hardware
 However, there are still circuits with
redundant hardware that are fully testable
 For example,
a/0, b/0,
a a/1 b b/1
A C

 Faults (a/0, a/1, b/0, b/1) can be tested by

tests (A=0, A=1)
 Therefore, these faults are not redundant

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 35

Redundant Hardware & Simplification

 Redundant hardware from testing definition

A D A
D/0
B B B E
E E

 The benefits of redundant hardware

removal
 Reduce area cost
 Improve performance
 Reduce power consumption
 Improve reliability
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 36
 Pernicious Fault Masking
 An example of a circuit with a redundant fault q/1
n
A
d q Z
l q/1
0 1
0
e 0 p conflict
0 f
B 0 r
0 h g 0 0 m
Y
C 0 0
0
k
j 0 0 s

n
A
1 d q Z
l D
D D
eD p
1 f D
B r
g m
C f/0 h Y
0 0
k
j s

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 37

 Pernicious Fault Masking
 For multiple stuck-faults, if one of the multiple
faults is redundant, the presence of the redundant
fault may mask the presence of other, testable
faults
 This is a serious compromise of circuit reliability

n
A
1 d q Z
l 1
D D
eD p q/1
1 f D
B r
g m
C f/0 h Y
0 0
k
j s

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 38

 Problems of Sequential Ckt Testing
A sequential circuit has memory in addition
to combinational logic
Test for a fault in a sequential circuit is a
sequence of vectors, which
 Initializes the circuit to a known state
 Activates the fault, and
 Propagates the fault effect to a primary output
Methods of sequential circuit ATPG
 Time-frame expansion methods
 Simulation-based methods

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 39

 Sequential Circuit Representation
 A representation of a synchronous
sequential circuit
(primary inputs) (primary outputs)
X Z
Combinational Logic
Y (next state)
(present state) y

state
clk

 The concept of time frame

X(0) X(1) X(2)

y(0) Y(1) y(1) Y(2) y(2) Y(2)

Z(0) Z(1) Z(2)

Time frame
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 40
 Time-Frame Expansion
 Iterative Array Conversion
pseudo Flip Flop

X(0) X(1)

y(0) Y(1) y(1) Y(2) y(2)

C/L pFF C/L pFF

Z(0) Z(1)

Time (space) frame 0 Time (space) frame 1

 Sequence of inputs in time: x(1), x(2),…, x(n)

 Sequence of outputs in time: z(1), Z(2),…,z(n)
 Sequence of internal states in time: y(0),
y(1),…, y(n)

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 41

 Time-Frame Expansion
 For synchronous sequential circuit (FSMs)
 Transforming FSMs to iterative logic array (ILA)
by unrolling the Huffman model
 Applying D-alg (or PODEM, path sensitizer,
etc.) to the ILA
 Pseudo flip-flop (pFF) is a combinational circuit
mapping its excitation function onto its output
 E.g., a pseudo T-FF can be constructed by an XOR
gate as shown below

T Y y
0 0 0
T y
0 1 1
1 0 1 Y
1 1 0

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 42

 Time-Frame Expansion
 When a single stuck-at fault is present
in the real sequential circuit, it will
appear as a multiple fault, existing in
each unfolded iteration (time frame)

Fault

Unknown
Time State Time Time Next
or given
Frame 0 variables Frame n-2 Frame n-1 state
Init. state

Comb.
block

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 43

 An Example
 Derive a test sequence for the fault a/1 of the
following circuit. Assume that the initial state
of the sequential circuit is (y1,y2)=(0,0)
x
G1
y2 y1
G6 D1 y1’
x’
y1 G2

x’
G3
y2 y2
G7 D2 x (0)=1 x (1)=1 x (2)=1
x y2’
y2 G4
0 D’
y1(0)=0
D’ D’
y2(0)=0
x
G5 z
y1’
z (0)=1 z (1)=1 z (2)=D
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 44
 An Example
 Test pattern generation with D-alg.
 t0: initial state y1=y2=0; y2=0 and a/1 a=D’
 Set x(0)=1y2(1)=D’ and z(0)=1
 z != D or D’  Continue!
 t1: set x(1)=1y1(2)=y2(2)=D’ and z(1)=1
 z !=D or D’ Continue!
 t2: on G5, let x(2)=1
 y1(2)=D’z(2)=D
 Test sequence X=111
 Termination rules
 If z=D or D’ then a test sequence is found
 If k>4n, where n is the number of FFs of the
original circuit, then the circuit is redundant
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 45
 Simulation-Based Test Generation
 CONTEST
 A concurrent test generator for sequential circuits---
using concurrent fault simulator
 Pseudo code
Initialization:
{1. Start with an arbitrary vector and all FFs in unknown
state;
2. Generate new vectors to reduce cost by 1-bit changes in
the present vector
/*use only true-value simulation*/
/*cost=number of FFs in unknown state*/
3. Stop when cost drops below desired value; }
Tests for concurrent targets:
{1. Start with initialization vectors;

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 46

 Simulation-Based Test Generation
2. Fault simulate vectors and remove detected faults;
3. Compute cost function for the last vector;
/* cost(undetected f.)=min dist of its effect from an o/p
*/
/* cost(vector)=sum of costs of all undetected faults*/
4. New vectors=those 1-bit changes that reduce vector
cost; }
Test for remaining undetected single faults:
{1. Revise the cost function (a dynamic testability measure);
/*cost=K x activation cost + propagation cost */
/* activation cost = dynamic controllability of faulty line
*/
/*propagation cost = min dynamic observability of faulty
line */
/* k is a large weighting factor*/
2. Generate a test for a single fault;}

Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 47

 Summary
 Combinational ATPG algorithms
 D-algorithm
 PODEM
 single-path sensitization
 Boolean difference
 ATPG also can aid the designer to find the
redundant hardware of the circuit
 Sequential ATPG approaches
 Time-frame expansion
 Simulation-based ATPG
 Sequential ATPG are not widely used in the IC
industry
 Low fault coverage
 Long ATPG time
Advanced Reliable Systems (ARES) Lab. Jin-Fu Li, EE, NCU 48