VLSI TESTING DESIGN FOR TESTABILITY FAULT DETECTION TECHNIQUES

DEFINITION OF TESTING § Testing in its broadest sense means to examine a product and to ensure that it functions and exhibits the properties and capabilities that it was designed to possess. § Main purpose of testing is to detect malfunctions in the product hardware and to locate their causes so that they may be eliminated. Testing terms: Otb. T : object to be tested DUT : device under test CUT : circuit under test Ls : Latches CN : Combinational Networks

CONVENTIONAL TEST METHODS These rely primarily on mechanical means and not on use of additional circuits in an otb. T for the purpose of facilitating its testing. Examples include use of extra I/O for additional test points, improvement of test features. Characteristics: § They are used for testing system parts only outside the system. § They rely on feeding signals directly through the test interface during listing. § They rely on the use of tester-driven timing.

DIFFICULTIES IN TESTING Shortage of I/O points. Signal distortions in interface connections. Noise disturbances. Uncertainties in input feeding. Uncertainties in output sensing. (Rejection of good parts reduces apparent yield. ) 6. Difficulty in synchronizing test objects timing with tester timing. 7. High costs for test equipment , test generation and execution. 8. Large volume of data to be processed. 1. 2. 3. 4. 5.

FAULTS If anything can go wrong, it will. Murphy’s Law

FAULT DEFINITION In any circuit composed of logic gates, there is the possibility of the occurrence of a fault. A fault is defined to have occurred when a circuit variable assumes a value(1, 0 or X) which differs from that expected that is violates the original circuit equation. Fault Types: § SAO : Stuck at ‘ 0’ (short with ground rail) § SA 1 : Stuck at ‘ 1’ (short with Vdd)

FAULT TYPES AND MODELS Examples of physical defects include § Defects in silicon substrate. § Photolithographic defects. § Mask contamination and scratches. § Process variation and abnormalities. § Oxide defects. Electrical faults caused: Shorts, opens, transistor stuck-on or stuck-off, Resistive shorts and opens, Excessive change in threshold voltage and excessive change in steady state currents.

KINDS OF FAULTS § Single faults § Multiple faults.

CIRCUIT FOR AO 1 EQUATIONS: X 7=X 6+X 5 X 5=X 1. X 2 X 6=X 3. X 4 X 7=X 1. X 2+X 3. X 4

KINDS OF FAULTS § Single faults § Multiple faults. No of single fault locations : 7 No of single faults : 2 * 7 = 14 No of double fault combs. : 2 * 7 C 2 = 84 Fault combinations are not unique. A test for SA 0 at x 1 also covers SA 0 at x 5 and x 7.

FAULT EQUIVALENCES § One or more inputs to an OR gate at SA 1 is equivalent to an OR gate whose output is at SA 1. § One or more inputs to an AND gate at SA 0 is equivalent to an AND gate whose output is at SA 0. § All inputs to an OR gate at SA 0 is equivalent to an OR gate whose output is at SA 0. § All inputs to an AND gate at SA 1 is equivalent to an AND gate whose output is at SA 1. Thus any gate output fault has an equivalent single stuck fault or multiple stuck fault.

MASKING OF FAULTS Definition: Let Tg be a test that detects a fault g. We can say that a fault f functionally masks the fault g iff the multiple faults (f, g) is not detected by any test in Tg.

SINGLE STUCK FAULT MODEL Single stuck-fault model (SSF) is the classical or standard fault model. Its usefulness results from the following attributes: § it presents many different physical faults; § it is independent of technology; § compared to other fault models, the number of SSFs in a circuit is small; § SSFs can be used to model other type of faults.

AND-NAND BLOCK X=AND(A, B, C, D) Y=NAND(A, B, C, D) A A B B C C D D sa 0 sa 1 A 1 0 1 1 1 B 1 1 1 0 1 1 C 1 1 1 0 1 1 D 1 1 1 1 0 X 1 0 1 0 Y 0 1 0 1

The test sequence can thus be obtained by finding out the combinations. Complementary circuits can be tested in the similar fashion. For a single stack model containing N nodes, where in each node can be in one of the 3 states (good, SA 0, SA 1) 3 N combinations are possible. For N=100 we get 5. 1047 combinations which is a very large data to process.

EXISTENCE FUNCTION Developing a test sequence: x 6=x 3. x 4 x 5=x 1. x 2 x 7=x 5+x 6 Rules for labeling the nodes: § Primary inputs are labeled with the lowest indexed variables. § Fan outs are labeled separately.

CIRCUIT FOR AO 1 EQUATIONS: X 7=X 6+X 5 X 5=X 1. X 2 X 6=X 3. X 4 X 7=X 1. X 2+X 3. X 4

EQUATIONS Fi(x 0, x 1…. xp)=Gi(xo, x 1…. xp) F=G F. G + G. F = 0 where F is the set of inputs and G is the set of outputs. F G x 1. x 2 x 5 x 3. x 4 x 6 x 5+x 6 x 7 F. G : x 5’x 1 x 2 x 6’x 3 x 4 x 7’x 5 x 7’x 6 ‾G. F : x 5 x 1’ x 6 x 3’ x 5 x 2’ x 6 x 4’ x 7 x 6’x 5’

EXISTENCE FUNCTION GENERATOR Mark all the points which are covered by at least one of the terms. Instead of 7 variable K map use a Marquand chart. After cancellation, take the remaining points. These are the ones in the existence function circuit. ( No of ones = 16). Move from one point to a place where there is a change in output.

Longest chain will produce the desired test sequence. The complete test sequence is 5 -7 -6 -14 -10 -11 -9 -13 -5. Each of the input variables is tested independently for a change in value from 0 to 1 and again from 1 to 0. Each of the intermediate variable is also tested in the process. Each output variable is thereby tested for its ability to change value from a 1 to 0 and from a 0 to 1.

ADVANTAGES OF TEST SEQUENCES § Test sequence can be produced by a hardware unit § § instead of the usual software unit. Continuous resetting between tests is not necessary. Since at least one of the outputs change on the application of an input, detection of a failure is logically straightforward. The test sequence covers all detectable single faults. The test sequence is closed i. e it returns to the initial state. This helps in reducing resetting.

DRAWBACKS OF SINGLE STACK FAULT MODEL § § Does not take into account other kinds of faults such as AC-faults. Bridging circuits. Faults in CMOS circuits. Multiple faults simultaneously presented in the system.

DESIGN FOR TESTABILITY Testable means capable of being ascertained as being fault free or not. The aim of testability is to make the parts testable not only on test fixtures separately from the system but also within the system when the parts are connected. It should also include diagnosability i. e the capability of locating faults at least down to the smallest repair-replaceable unit

THREE KEY FUNCTIONS 1. Control Setting the conditions for the tests so that stimuli can be supplied to the object to be tested. 1. Observation Obtaining the response to the stimuli so that the behavior can be evaluated. 1. Isolation Making the control and observation possible and more reliable.

TECHNIQUES § Ad-hoc Testable Design Techniques § Initialize sequential circuit § Avoid redundancy logic § Avoid asynchronous logic § Avoid redundant circuits. § Built in Self Testing. (BIST)

AD-HOC DESIGN The three main features are § Partition and Multiplexer techniques. § Use of switches Ex: For a 32 bit counter checking is very difficult. But if we have sub-circuits , testing will be easier. Switches will be placed throughout.

PARTITION TECHNIQUE

BUILT IN SELF TECHNIQUES In built-in-self-techniques (BIST) parts of the circuits are used to test the circuit itself. On line BIST is used to perform test under normal operation where as off line BIST for testing offline. Components: § Pseudo Random Pattern Generator (PRPG) § Output Random Analyzer (ORA)

BUILT IN SELF TEST

OUTPUT RANDOM ANALYZER Cyclic Redundancy Check G(x) = Q(x) P(x) + R(x) where P(x) is the characteristic polynomial (output of the CUT). R(x) is the remainder and Q(x) is the quotient. P(x) = x^5 + x^4 + x^2 + 1 G(x) with the sequence {1 1 0 1 0 1} G(x) = x^7 + x^6 + x^5 + x^4 + x^2 + 1 and R(x) = x^4 + x^2 which corresponds to the register(0 0 1)

The on chip storage of a fault dictionary containing all the test inputs with the corresponding outputs is prohibitively expensive in terms of the chip area. Alternative is to compare the outputs of 2 identical circuits for the same inputs assuming that the probability that the two devices will have the same kind of faults is less.

END