WO1995003589A1 - Robust delay fault built-in self-testing method and apparatus - Google Patents

Robust delay fault built-in self-testing method and apparatus Download PDF

Info

Publication number
WO1995003589A1
WO1995003589A1 PCT/US1994/008238 US9408238W WO9503589A1 WO 1995003589 A1 WO1995003589 A1 WO 1995003589A1 US 9408238 W US9408238 W US 9408238W WO 9503589 A1 WO9503589 A1 WO 9503589A1
Authority
WO
WIPO (PCT)
Prior art keywords
input
node
cut
integrated circuit
hazardous
Prior art date
Application number
PCT/US1994/008238
Other languages
French (fr)
Inventor
Michael L. Bushnell
Imtiaz Shaik
Original Assignee
Rutgers University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Rutgers University filed Critical Rutgers University
Priority to AU73699/94A priority Critical patent/AU7369994A/en
Priority to DE69423598T priority patent/DE69423598T2/en
Priority to EP94922668A priority patent/EP0663092B1/en
Priority to AT94922668T priority patent/ATE191091T1/en
Priority to KR1019950701125A priority patent/KR950703767A/en
Priority to JP7505311A priority patent/JPH08502365A/en
Priority to CA002145403A priority patent/CA2145403C/en
Publication of WO1995003589A1 publication Critical patent/WO1995003589A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/22Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing
    • G06F11/26Functional testing
    • G06F11/267Reconfiguring circuits for testing, e.g. LSSD, partitioning
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28Testing of electronic circuits, e.g. by signal tracer
    • G01R31/30Marginal testing, e.g. by varying supply voltage
    • G01R31/3016Delay or race condition test, e.g. race hazard test
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2201/00Indexing scheme relating to error detection, to error correction, and to monitoring
    • G06F2201/83Indexing scheme relating to error detection, to error correction, and to monitoring the solution involving signatures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2201/00Indexing scheme relating to error detection, to error correction, and to monitoring
    • G06F2201/88Monitoring involving counting

Definitions

  • This invention relates to a method and apparatus for built-in self-testing of delay faults in hardware systems. This invention was made with Government support under Grant MIP 9058536, awarded by the National Science Foundation. The Government has certain rights in this invention. 2. Description of the Related Art
  • Integrated circuits can have permanent faults such as a grounded point or point connected to the positive supply. These permanent faults are often referred to as "stuck-at" faults. Many arrangements have been described for testing stuck-at faults. However, even though a integrated circuit does not have a stuck-at fault it may still not perform accurately when put in service.
  • a logic circuit may be expected to operate correctly during successive "clock” periods.
  • a "delay- fault” occurs when a circuit response requires more time than specified by design requirements. Delay-fault testing can be used to check if the circuit meets the required clock rates.
  • stuck-at fault and delay-fault testing can be performed immediately following circuit fabrication with automatic test equipment.
  • External testing of a circuit for delay-faults takes additional time for large circuits to check each path of the circuit.
  • Large circuits require faster test equipment in order to clock the outputs of the circuit under test at the right instant.
  • automatic test equipment can only test for delays up to 200 mHz.
  • automatic testing equipment for delay-faults at a frequency of up to 200 mHz is expensive, i.e., typically costing between $750,000 and $1 million.
  • U.S. Patent No. 5,056,094 describes an apparatus for testing the propagation delay between a driving device and a receiving device of an integrated circuit (IC) .
  • Test cells are disposed at boundaries of the IC to control data flow through application logic of the IC.
  • the test cell allows input data to be observed and output data to be controlled simultaneously.
  • Sampled input is compared to a known value to determine whether the signal propagated to the receiving device is within the time period between the first and second clock edges.
  • One conventional approach for testing without automatic test equipment is to add circuitry onto an IC chip for self-testing.
  • the Built-in Self-Testing (BIST) approach enables the circuit to test itself.
  • BIST Built-in Self-Testing
  • 5,138,619 relates to a built-in self-test circuit for on-chip testing of an integrated circuit memory.
  • An address pseudo-random pattern generator PRPG selectively furnishes test addresses to memory and a data PRPG selectively furnishes test data to memory.
  • a parallel signature analyzer PSA PRPG selectively furnishes data to the memory in normal mode and determines a signature in test mode.
  • a decoder compares the signature determined by the PSA PRPG with a known correct signature and sets a flag to indicate memory pass or fail. This testing method is used for determining stuck-at faults.
  • U.S. Patent No. 4,801,870 describes a method for testing complex integrated circuit devices. An advance simulation of the testing protocol determines a good signature for the device.
  • Pseudo-random pattern generators supply input test patterns to the device to be tested. Output responses of the device are combined to obtain a test signature. Thereafter, the test signature is compared to the good signature.
  • the method of this patent has the shortcoming of being used only for stuck fault testing.
  • U.S. Patent No. 4,672,307 describes a test system for testing combinational logic circuit faults using delay testing.
  • a test input circuit is connected to the combinational logic circuit.
  • the test input circuit has at least as many inputs as the combinational logic circuit.
  • the test input circuit applies a series of all possible single transitions of the combinational logic circuit. This is a set of binary numbers where there is only a one bit transition between successive numbers. For example, a two digit Gray code would be as follows: 00, 01, 11, 10. Validity of the outputs from the combinational logic circuit is checked for correctness during a predetermined time interval.
  • a Gray code generator has the disadvantage that the most significant bit changes only once for the entire sequence. In addition, a Gray code generator has long cycle lengths for circuits with large numbers of primary inputs.
  • Robust delay testing is used to test for an excessive path delay for a tested path independently of other path delays in the circuit, where the delays are variable or non-existent. It is desirable to provide a method and apparatus for robust delay-fault built-in self-testing of circuits.
  • the present invention comprises a method and apparatus for robust delay-fault testing of an integrated circuit (IC) by altering the circuit topology.
  • Hazardous nodes are determined for the IC circuit.
  • a hazardous node is defined as a gate whose output momentarily switches from the correct output value for a brief time period.
  • Cut points are inserted in the circuit for diverting input to the hazardous node to an observation point.
  • a succession of input patterns are applied to the IC during testing.
  • Output responses of the IC are processed in a multi-input signature register (MISR) to provide a first signature.
  • Output responses of the IC at the observation point are processed in a cut-point MISR to provide a second signature.
  • MISR multi-input signature register
  • the first and second signatures are compared to known correct signatures.
  • phase offset delays are added to the drivers of fanouts from hazardous nodes and the clock for the cut- point MISR is phase offset from the system clock.
  • an exclusive OR gate is used at the cut-point.
  • the exclusive OR gate eliminates inputs with opposite transitions for eliminating hazardous nodes during testing.
  • XOR equivalence gates in the IC are determined in both embodiments and are modified for eliminating inherent hazards. Cut-points are optimally inserted at common nodes of the IC for the eliminating the addition of excess hardware to the IC.
  • a modified Johnson counter is preferably used for generating hazard-free input patterns.
  • FIG. 1 is a flow diagram of the robust delay- fault testing method in accordance with the principals of the present invention.
  • Fig. 2 is a schematic diagram of a robust delay-fault testing system in accordance with the principals of the present invention.
  • Fig. 3 is a schematic diagram of a reconvergent fanout circuit.
  • Fig. 4 is a schematic diagram of a reconvergent fanout circuit with a common node.
  • Fig. 5 is a schematic diagram of a reconvergent fanout circuit with two common nodes.
  • Fig. 6 is a block diagram of a cut-point circuit.
  • Fig. 7 is a schematic diagram of a cut-point circuit with applied control signals.
  • Fig. 8A is a schematic diagram of a NMOS implementation of a cut-point circuit.
  • Fig. 8B is a schematic diagram of a CMOS implementation of a cut-point circuit.
  • Fig. 9 is a schematic diagram of a NAND realization of an XOR gate.
  • Fig. 10 is a schematic diagram of a hardware modified XOR gate as shown in Fig. 9 including CMOS switches.
  • Fig. 11 is a schematic diagram which is equivalent to Fig. 10 when a TEST signal is applied to the circuit under test.
  • Fig. 12 is a schematic diagram of an input generator for providing a modified Johnson counter.
  • Fig. 13 is a schematic diagram of the robust delay-fault testing system shown in Fig. 2 including clocks.
  • Fig. 14 is a schematic diagram of a circuit showing how application of a hazard-free input pattern to a circuit may still generate hazards in the response.
  • the figure illustrates the tracing of hazardous nodes using a simulation-guided approach. -jr.
  • Fig. 15 is a schematic diagram of a graph algorithm for tracing a reconvergent fanout circuit.
  • Fig. 16 is a schematic diagram of a graph algorithm for tracing a reconvergent fanout circuit.
  • Fig. 17 is a schematic diagram of the insertion of an exclusive OR gate in a circuit for eliminating hazards in the circuit during testing.
  • Fig. 1 is a flow diagram illustrating the basic concept of the robust delay-fault testing method in accordance with the principles of the present.
  • the testing concept is used to enable the circuit hardware to be hazard-free during testing in order to provide accurate robust testing of the circuit.
  • a hazard or hazardous node in a circuit can be defined as a non- Exclusive-OR, non-Equivalence gate having inputs with opposite transitions.
  • Exclusive-OR and Equivalence gates we define a hazardous gate or node as one having inputs with more than one transition. Depending on the speed of the transitions of the inputs, both inputs can have simultaneous transitions which can result in inaccurate testing of the circuit.
  • a hazard occurs on an OR gate when one input changes from 0 ⁇ 1 while a second input changes from 1 ⁇ 0.
  • both inputs to the OR gate are "0" and a momentary logic "0" is the OR gate's output.
  • XOR and Equivalence gates are inherently sensitive to multiple changes in inputs and are therefore hazard prone.
  • Signal parity is defined as the number of inversions a signal is subject to as it travels along a path. It is difficult to judge the parity of an XOR or Equivalence gate since a signal passing through one of the inputs of an XOR gate can either pass as it is or can be inverted depending upon the signal on the other input.
  • An XOR gate is hazard prone even if there are similar transitions on the inputs.
  • Paths in the circuit which are responsible for hazards are cut in order to achieve the objective of the robust testing concept.
  • a path is cut by inserting a cut-point on any of the lines of the path. For example, when a cut-point is inserted between lines connecting two nodes X and Y of the circuit, the cut-point prevents a signal at X from feeding Y in the test mode of the circuit. Instead, the signal of X is diverted to an observation point and Y is driven from a new hazard-free signal generated by an input pattern generator.
  • a circuit diagram 10 is generated which illustrates the circuit under test.
  • Circuit diagram 10 includes all internal functional elements and all interconnections between the input and output terminals of the circuit under test; i.e., AND gate, OR gate, NOR gate, NAND gate, or the like.
  • the circuit diagram is a logic model contained in a computer memory.
  • Module 11 determines if any XOR or Equivalence gates are present in circuit diagram 10.
  • XOR and Equivalence gates are modified in block 12 for providing definite signal parity between the input node and the output node, thereby eliminating hazards in the XOR and Equivalence gates.
  • Simulation or tracing of the circuit diagram 10 is performed in block 14 to determine if hazardous nodes are present in schematic diagram 10.
  • a cut-point is inserted in block 13 to eliminate one or more of the detected hazardous nodes.
  • a cut-point is inserted in an input path of a hazardous node for preventing the input signal from propagating through the hazardous node. The simulation or tracing is repeated until no more hazardous nodes are found in the circuit under test.
  • a TEST signal is applied to the circuit 16 for activating the cut-points and activating a hardware input generator for performing robust delay testing.
  • a hazard-free input pattern is applied to primary inputs of the circuit as well as to inputs of cut-points in order to avoid hazards in the response.
  • a hazard- free input pattern can be defined as a sequence of input vectors that differ successively by one bit. The hazard-free input pattern initiates only one transition at any one of the primary inputs while all other primary inputs are held at a constant value.
  • a hazard-free input pattern is a sequence that avoids multiple transitions on inputs to the circuit under test.
  • a hazard-free input pattern can be generated by a Gray code generator or a Johnson counter.
  • a Gray code counter has the disadvantage that for large numbers of inputs the cycle time is too large for practical testing.
  • a Johnson counter generates 2n basic patterns.
  • a Johnson counter has the disadvantage of not supplying sufficient patterns for all fault sites in the circuit.
  • a modified Johnson counter including controlled complementation of the Johnson counter outputs is used for providing sufficient numbers of hazard-free input patterns.
  • Fig. 2 illustrates an apparatus for a robust delay-fault built-in self-testing (BIST) system 20 in accordance with the principles of the present invention.
  • BIST delay-fault built-in self-testing
  • test vector generation and test result verification are accomplished by built-in circuitry.
  • Input pattern generators 21 are disposed at the boundary of integrated circuit (IC) 22 to generate an input pattern for testing IC 22.
  • Primary inputs 24 extend between input pattern generator 21 and IC circuit 22.
  • Primary input 24b is connected by path 26a to hazardous node 28.
  • a cut-point 23 is inserted in hazardous node 28 to direct input from path 26b to an observation point which is a cut-point multiple input signature register (MISR) 30.
  • Outputs 27 of IC 22 are received at output MISR 25.
  • Hazard-free input patterns are generated by input pattern generator 21 and are applied to primary inputs 24 and cut-point 23 during testing.
  • a new bit of hazard-free input pattern is applied to fanout circuit 31 through path. 26a in test mode.
  • responses from IC 22 are combined in output MISR 25.
  • Output MISR 25 receives a succession of input responses from primary outputs 27 of IC 22 and produces a derived function thereof or first signature.
  • Responses for cut-points 23 are combined in cut-point MISR 30 for providing a second signature.
  • Buffers 29 can be inserted in primary input path 24b in order to delay the hazard-free input pattern sufficiently to fairly test the primary input paths which are connected to a cut-point. It has been found that a transition in the hazard-free input pattern applied to primary input path 24b will arrive more quickly at output MISR 25 than a transition from an ordinary primary input 24a will arrive at output MISR 25. A signature computed at cut-point MISR 30 may not catch a delay-fault, if the fault is of a size large enough to delay the path from primary input 21 to output MISR 25. A delay-fault between input generator 21 and output MISR 25 can be larger than a delay-fault between cut-point 23 and output MISR 25.
  • a signature computed for the delay-fault between input signature generator 21 and output MISR 25 would not catch the shorter delay-fault of cut-point 23.
  • the number of buffers 29 used in system 20 depends on the computed nominal delay of each path driven from cut-point 23 to output MISR 25. Buffers 29 delay the transition sufficiently so that a signature can be effectively evaluated by output MISR 25.
  • Fig. 3 illustrates an example of a hazardous node 38 which is a reconvergent fanout circuit 31.
  • Reconvergent fanout circuit 31 splits into at least two paths before it reconverges at another point in the circuit.
  • Signal parity can be defined as the number of inversions a signal is subject to as it travels along a path.
  • a reconvergent fanout circuit can have a hazard if the paths are of opposite signal parity.
  • Output paths 34, 36 from NAND gate 32 reconverge at OR gate 38.
  • Path 34 is inverted with inverter 35 to produce odd signal parity on path 37.
  • Path 36 from AND gate ' 32 has even signal parity.
  • Hazardous node 38 is an OR gate.
  • a hazard can occur at hazardous node 38 since path 37 has opposite signal parity to path 36.
  • Path 36 or path 37 can be cut to avoid a hazard at hazardous node 38. If there are more than two paths reconverging, then we cut either all odd parity paths or all even parity paths. It has been found that inserting cut-points in every reconverging pair may result in inserting too many cut-points into IC 22. Preferably, cut-points are inserted at minimal locations in IC 22 for providing minimal additional hardware to the IC-.
  • Fig. 4 illustrates a reconvergent fanout 40 with the minimal number of cut-points inserted.
  • IC 22 can include a reconvergent fanout 40 that includes one or more "K" common nodes 41. Cycle 1 starts from X.
  • Source node 45 traverses paths 46 and 44 to Y 1 end node 54 and returns back to the X_, source node 45 along path 47 in the opposite direction.
  • Cycle 2 starts from X 2 source node 50, traverses path 43 to Y 2 end node 56 and returns along path 44 in the opposite direction.
  • Node 41. is common to cycles 1 and 2.
  • the number of cycles "n" that intersect at node K means that node K will have at least n fanout lines.
  • a cut-point is inserted at the fanout stem of node K in order to cut off "n" fanout lines from node K's output and thereby save n-1 cut points from the hardware overhead of the IC.
  • X 1 source node 45 is a NAND gate which includes output paths 46 and 47.
  • X 2 source node 50 is a OR gate which includes output paths 43 and 44.
  • Output path 46 is received at inverter 48 which provides input 44 to the Y 1 end node.
  • Y 1 end node 54 is an OR gate.
  • Output paths 47 and 43 provide input to node 41.
  • Output 51 from node 41 is received at Y 1 end node 54 and inverter 58.
  • Inverter 58 provides input to Y 2 end node 56.
  • cut- point 23 is inserted after node 41 between Y 1 end node 54 and inverter 58 for cutting input to both Y 1 end node 54 and Y 2 end node 56 with the minimal number of cut-points.
  • Fig. 5 illustrates a reconvergent fanout circuit 60 with K 1 node 74 and K 2 node 76 in common.
  • Path 61 is shared by cycles 1 and 2. Inserting cut-point
  • X 1 source node 62 is an exclusive OR which includes output paths 66 and 68.
  • Path 66 provides input to inverter 77 which provides input to Y_, end node 78.
  • Y, end node 78 is a NAND gate.
  • X 2 source node 64 is a NAND gate which includes output paths 70 and 72.
  • Output path 68 and output path 70 are received at node 74.
  • Node 74 is an AND gate and node 76 is an OR gate.
  • Path 72 provides input to inverter 75 which provides input to Y 2 end node 80.
  • Y 2 end node 80 is a NAND gate.
  • Output 61 from node 74 provides input to node 76.
  • Output 79 from node 76 provides input to Y 1 end node 78 and Y 2 end node 80.
  • cut-point 23 is inserted in path 61 between node 74 and node 76.
  • input paths to node Y 1 end node 78 and Y 2 end node 80 are cut by the insertion of cut-point 23.
  • a cut-point can be inserted anywhere on the reconvergent opposite parity paths. Accordingly, cut-points can be inserted away from critical paths of the circuit 2.
  • Fig. 6 is a block diagram of cut-point 23 inserted between X input node 90 and Y output node 92.
  • Output 91 from input node 90 is connected to cut-point 23 and output 93 of cut-point 23 is connected to output node 92.
  • Output 94 from input pattern generator 21 is connected to cut-point 23 and output 96 from cut-point 23 is connected to cut-point MISR 30.
  • Control inputs TEST and TEST are applied to cut-point 23 to respectfully provide or remove connections of cut-point 23 between input node 90 and output node 92.
  • Fig. 7 illustrates a circuit diagram of cut- point 23 including test switches 100a, 100b and 100c.
  • a control input of TEST is supplied to test circuit 100a for closing circuit 100a during normal operation of IC 22.
  • a control input of TEST is applied to circuits 100b and 100c for closing circuits 100b and 100c when IC 22 is in a test mode.
  • Fig. 8A illustrates implementation of cut-point 23 in a NMOS MOSFET logic system.
  • Fig. 8B illustrates implementation of cut-point 23 in a complementary MOSFET (CMOS) logic system.
  • Test is applied to transistor 110a during normal operation of IC 22.
  • TEST is applied to transistors 100b and 110c during TEST mode of IC 22. It will be appreciated that other logic systems such as TTL, ECL, BiCmos and it will be appreciated that other logic systems known in the art can be used with the teachings of the present invention.
  • Fig. 9 illustrates a NAND implementation 120 of an XOR Equivalence gate.
  • X, source node 122 and X 2 source node 124 have two reconverging paths to Y end node 142 with one inverted and one non-inverted path.
  • Path 123 from node 122 is non-inverted and path 121 from node 122 is inverted with inverter 126.
  • Path 125 from node 124 is non-inverted and path 127 is inverted with inverter 128.
  • Paths 129 and 125 are applied to NAND gate 132 and paths 123 and 130 are applied to NAND gate 134.
  • Outputs 136 and 138 are applied to NAND gate 140 which provides output 142 to Y end node 142.
  • FIG. 10 illustrates a hardware modification of the NAND implementation of the XOR gate 120 in an NMOS implementation.
  • Four NMOS switches are used to change the configuration of the XOR gate 120 such that source node X 1 has only non-inverting paths to end node Y and source node X, has only inverting paths to node Y.
  • NMOS switches 144a. 144b, 144c and 144d are inserted to change the configurations of NAND gates 132 and 134 so that input 122 has non-inverting paths to end node 142 and input 124 has inverting paths to end node 142.
  • NMOS switch 144a is inserted in the path 121 from node 122 to inverter 126 and NMOS switch 144b is inserted in path 133 from node 124 to inverter 126.
  • NMOS switch 144c is inserted in path 131 from node 122 to NAND gate 132 and NMOS switch I44d is inserted in path 135 from node 124 to NAND 132.
  • Fig. 11 illustrates an equivalent diagram of
  • Fig. 12 illustrates input pattern generator 160 for generating hazard-free input patterns which can be ⁇ used for input pattern generator 21, shown in Fig. 2.
  • a test pattern is a sequence of vectors in which each element, which represents the logic value to be given the primary inputs, takes one of the values, 0, 1, ⁇ a falling transition or ⁇ - a rising transition.
  • Input pattern generator 160 is a modified Johnson counter for providing more than 2n patterns. Single bits of the pattern are selectively inverted to result in a new pattern.
  • Outputs 164 of Johnson counters 161 are applied to a linear array of two-input XOR gates 162.
  • Input 166 of XOR gate 162 receives a control signal from input lines 168.
  • a corresponding bit of the Johnson counter 163 is complemented. For example, if the output of Johnson counter is 11111 and a complement is applied to the most significant bit the pattern applied would be 01111 and if no complement is applied the pattern would be 11111.
  • generated patterns cyclically alternate between complemented patterns and uncomplemented patterns to produce successive one-bit transition changes. An illustration of a modified Johnson scheme is shown in Table 1.
  • a pair-wise complementation scheme can be used to generate non-hazardous input patterns.
  • two bits can be complemented simultaneously.
  • the third and fourth bits can be selected for complementation.
  • An uncomplemented pattern of the third and fourth bits is applied as the first pattern.
  • the fourth bit is complemented as the second pattern.
  • the third bit complemented along with the fourth bit as the third pattern.
  • the fourth bit is uncomplemented and the third bit is complemented for the fourth pattern and both the third and fourth bits are uncomplemented for the fifth pattern.
  • a pair-wise complementation scheme has a cycle length of 5 x 2n patterns. It will be appreciated that other complementation schemes can be used for generating hazard-free input patterns.
  • Fig. 13 illustrates system 20 including clock 200 for cut-point MISR 30.
  • a signal from path 26b to cut-point MISR 30 can arrive before a signal . from primary inputs 24 to output MISR 25 on path 26a.
  • Cut-point MISR 30 should be clocked at the right instant for evaluating a fair signature in cut-point MISR 30 which reflects the exact delay of the paths which have been cut.
  • Clock 200 operates at an earlier instant than system clock 202 of IC 22.
  • clock 200 is offset by a phase that depends on the overall delay of paths being cut at cut- points 23.
  • a simulation of BIST system 20 is used to determine hazardous nodes and for optimally analyzing the insertion of cut-points 23 into BIST system 20.
  • a hazard modelling algebra can be used for simulating the hazardous nodes of the circuit under test.
  • Chakraborty's 13-valued hazard- modeling algebra can be used for modeling BIST system 20 as described in T.J. Chakraborty, Vishwani D. Agrawal and M.L. Bushnell, Path Delay Fault Testing For Logic Circuits, Proceedings of the 29th Design Automation Conference, p. 165-172, June 1992.
  • logic values of two consecutive clock periods are described by a transition state.
  • Three signal states (0, 1, and _ represent nine transition states and eighteen transitions with a hazard or no-hazard condition. These eighteen values eventually collapse into thirteen unique values and each is represented by a triplet. For example, 1
  • Table 2 shows the various transition states with a corresponding numerical encoding. The symbol ' * ' in the table stands for a condition where a hazard may or may not be present (unknown status) .
  • a truth table for each type of gate is implemented in the simulation for determining output values. For example, for a AND gate, if two opposite transitions without a hazard are input, the output must contain a hazard value.
  • a modified Johnson pattern sequence is applied to simulated inputs to determine hazardous nodes. If a hazardous node is found, paths leading to the hazardous node are traced to the primary input that had a transition for finding paths with opposite transitions. A signature is generated by the simulation and compared with a correct value.
  • Fig. 14 illustrates simulation guided path tracing by applying hazard-free input patterns for determining hazardous nodes. Hazardous nodes are determined by following transitions through IC 22.
  • a hazard-free input pattern "1100" is applied to primary inputs 302, 304, 306 and 308.
  • a falling transition is applied to primary input 310 which results in a hazard at NOR gate 324.
  • Path searching starts with primary input 310 through AND gates 318 and 320 to NOR gate 324. Paths leading to AND gate 314 and NOR gate 322 are avoided since there are no transitions on them. Results for robust delay-fault testing with
  • hazardous nodes are identified by enumerating the dominated gates in the circuit.
  • gate A dominates gate B if all paths from A to a circuit output flow through gate B.
  • Gate A can be defined as a dominator.
  • a graph algorithm can be used to enumerate paths in the circuit between dominators to trace IC 22. From the graph algorithm, the points in the circuit in which a cut-point should be inserted can be determined. Parity is assigned to each path of IC 22. If a node is encountered more than once it is assigned a parity of either "ALL EVEN" indicating all paths were even, "ALL ODD" indicating all paths were odd or "BOTH” indicating a mixture of even and odd parity paths.
  • D 1 dominator 352 has path 364 which is assigned ODD parity. Paths 354 and 356 to D 2 dominator 362 are assigned EVEN parity. D 2 dominator 362 has a parity defined as BOTH which is a hazardous node. The parity at D 2 dominator 362 is fixed at ODD or EVEN by a decoupling paths between D 1 dominator 352 and D 2 dominator 362. A cut-point 413 is inserted in path 358 for decoupling the paths between D 1 and D 2 .
  • node D is assigned a party of "SHADOW". Parity of SHADOW is defined as a node which is not considered in the current pass of graph analysis. Paths to node D will not be considered in the current pass until the parity of D 2 is decided to be either ODD or EVEN. ODD and EVEN parity paths are grouped starting and ending at the same nodes to avoid duplicating these paths for nodes driven by the same node.
  • Fig. 16 illustrates a graph algorithm for tracing circuit 400. D ⁇ dominator 402 has path 404 to D ⁇ dominator 406.
  • D 1 dominator 408 has paths 410, 412 and 414 to D 2 dominator 416. Paths 410 and 414 have EVEN parity. Inverter 418 provides ODD parity to path 412. D 2 dominator 416 is assigned parity BOTH. In the current pass of the graph algorithm, D ⁇ dominator 406 is assigned parity SHADOW. A cut-point 413 can be inserted in path 412 for decoupling. D 1 dominator 402 and D 2 dominator 406 to provide EVEN parity to D 2 dominator 402. After cut-point 413 is inserted, D ⁇ dominator 406 will have EVEN parity.
  • Nodes which decouple the maximum number of paths are chosen for the insertion of cut-points.
  • a group of EVEN parity paths with m ODD parity paths and n EVEN parity paths form a path group equivalent to mxn path pairs. If a cut-point is inserted in an EVEN (ODD) parity path, it is equivalent to decoupling m(n) path pairs. Weights are assigned to nodes of the path group to determine the optimum node to insert a cut-point. Each node in an EVEN (ODD) parity path is assigned weight m(n) . If the node is also in other path groups, the node weight is computed by adding all path group weights.
  • Fanout circuit 500 includes AND gate 501 connected by path 502 to NAND gate 504. PATH 506 is inverted with inverter 508. Exclusive OR gate 510 is inserted in path 508 to eliminate hazards in fanout circuit 500. Exclusive OR gate 510 has input 508, TEST input 512 and output 514. Output 516 is received at NAND gate 504.
  • a TEST signal of "0" is applied to input 512.
  • the signal on input 508 will be the same as the signal on output 514.
  • the NAND gate will have ODD parity from path 502 and EVEN parity from path 514 and is a hazardous node.
  • TEST signal is set to "1".
  • Exclusive OR gate 510 inverts path 508 to EVEN parity to eliminate a hazardous node in NAND gate 504.
  • the use of an Exclusive OR gate for eliminating hazards has the advantage of reducing the hardware added to the IC.
  • the use of the cut-point circuit for eliminating hazards has the advantage of providing additional information from the cut-point MISR for observing and controlling the IC.
  • the present invention has the advantage of providing robust built-in self-testing of delay faults in integrated circuits.
  • Hazardous nodes of the circuit are determined.
  • a cut- point is inserted to eliminate a hazard at the hazardous node.
  • the signal on the cut-line is diverted to an observation point cut-point MISR during testing of the circuit.
  • the signal that normally would fan out from the cut-point is instead driven by a hazard- free pattern generated by an extra primary input, which is added to the circuit.
  • a signature is computed at the output of the circuit and a second signature is computed at the observation point.
  • Hazard- free input patterns are applied to the circuit and the computed signatures are compared to known good signatures.
  • the system has the advantage of quickly determining delay faults for numerous integrated circuits.

Abstract

This invention relates to a method and apparatus for robust delay fault testing of integrated circuits (IC) (22) with built-in self-testing. Hazardous nodes (28) of the IC (22) are determined. The topology of the IC (22) can be modified to include cut-point (23) at the hazardous nodes (28) of the circuit (22). Input (24) to the cut-point (23) is diverted to an observation point (30). A first output MISR (25) provides a signature for the outputs (27) of the IC (22). A cut-point multi-input signature register (MISR) (30) at the observation point generates a first signature. During testing a hazard-free input pattern is applied to the IC (22) and the generated first and second signatures are compared to known correct signatures.

Description

TITLE: ROBUST DELAY FAULT BUILT-IN SELF-TESTING
METHOD AND APPARATUS
BACKGROUND OF INVENTION 1. Field of the Invention
This invention relates to a method and apparatus for built-in self-testing of delay faults in hardware systems. This invention was made with Government support under Grant MIP 9058536, awarded by the National Science Foundation. The Government has certain rights in this invention. 2. Description of the Related Art
Integrated circuits can have permanent faults such as a grounded point or point connected to the positive supply. These permanent faults are often referred to as "stuck-at" faults. Many arrangements have been described for testing stuck-at faults. However, even though a integrated circuit does not have a stuck-at fault it may still not perform accurately when put in service.
A logic circuit may be expected to operate correctly during successive "clock" periods. A "delay- fault" occurs when a circuit response requires more time than specified by design requirements. Delay-fault testing can be used to check if the circuit meets the required clock rates.
Conventionally, stuck-at fault and delay-fault testing can be performed immediately following circuit fabrication with automatic test equipment. External testing of a circuit for delay-faults takes additional time for large circuits to check each path of the circuit. Large circuits require faster test equipment in order to clock the outputs of the circuit under test at the right instant. Typically, automatic test equipment can only test for delays up to 200 mHz. In addition, automatic testing equipment for delay-faults at a frequency of up to 200 mHz is expensive, i.e., typically costing between $750,000 and $1 million.
A method for delay-fault testing is described in U.S. Patent No. 5,056,094. This patent describes an apparatus for testing the propagation delay between a driving device and a receiving device of an integrated circuit (IC) . Test cells are disposed at boundaries of the IC to control data flow through application logic of the IC. The test cell allows input data to be observed and output data to be controlled simultaneously. Sampled input is compared to a known value to determine whether the signal propagated to the receiving device is within the time period between the first and second clock edges. One conventional approach for testing without automatic test equipment is to add circuitry onto an IC chip for self-testing. The Built-in Self-Testing (BIST) approach enables the circuit to test itself. U.S. Patent No. 5,138,619 relates to a built-in self-test circuit for on-chip testing of an integrated circuit memory. An address pseudo-random pattern generator (PRPG) selectively furnishes test addresses to memory and a data PRPG selectively furnishes test data to memory. A parallel signature analyzer (PSA PRPG) selectively furnishes data to the memory in normal mode and determines a signature in test mode. A decoder compares the signature determined by the PSA PRPG with a known correct signature and sets a flag to indicate memory pass or fail. This testing method is used for determining stuck-at faults. U.S. Patent No. 4,801,870 describes a method for testing complex integrated circuit devices. An advance simulation of the testing protocol determines a good signature for the device. Pseudo-random pattern generators supply input test patterns to the device to be tested. Output responses of the device are combined to obtain a test signature. Thereafter, the test signature is compared to the good signature. The method of this patent has the shortcoming of being used only for stuck fault testing.
U.S. Patent No. 4,672,307 describes a test system for testing combinational logic circuit faults using delay testing. A test input circuit is connected to the combinational logic circuit. The test input circuit has at least as many inputs as the combinational logic circuit. The test input circuit applies a series of all possible single transitions of the combinational logic circuit. This is a set of binary numbers where there is only a one bit transition between successive numbers. For example, a two digit Gray code would be as follows: 00, 01, 11, 10. Validity of the outputs from the combinational logic circuit is checked for correctness during a predetermined time interval. A Gray code generator has the disadvantage that the most significant bit changes only once for the entire sequence. In addition, a Gray code generator has long cycle lengths for circuits with large numbers of primary inputs.
Of possible general relevance are U.S. Patent Nos. 5,095,483 and 5,051,996 directed to signature comparisons and 4,635,261 and 4,893,072 directed to testing devices for integrated circuits. Excessive delays in circuits under test are often due to device parameter variations caused by random fluctuations during fabrication of the circuits. Variations in delay characteristics can be found in several devices or paths in the circuit. Accordingly, path delay-faults can be caused by devices not in the tested path. Non-robust delay testing is a test which detects faults under the assumption that only the paths passing through a given delay-fault site can cause excessive path delay and all other paths are delay-fault free. Non-robust delay-fault testing is invalidated if a delay occurs on a path other than the one passing through a given delay-fault site. The above-described patents relate to non-robust testing.
Robust delay testing is used to test for an excessive path delay for a tested path independently of other path delays in the circuit, where the delays are variable or non-existent. It is desirable to provide a method and apparatus for robust delay-fault built-in self-testing of circuits.
SUMMARY OF THE INVENTION Briefly described, the present invention comprises a method and apparatus for robust delay-fault testing of an integrated circuit (IC) by altering the circuit topology. Hazardous nodes are determined for the IC circuit. A hazardous node is defined as a gate whose output momentarily switches from the correct output value for a brief time period. Cut points are inserted in the circuit for diverting input to the hazardous node to an observation point. In a first embodiment, a succession of input patterns are applied to the IC during testing. Output responses of the IC are processed in a multi-input signature register (MISR) to provide a first signature. Output responses of the IC at the observation point are processed in a cut-point MISR to provide a second signature. The first and second signatures are compared to known correct signatures. In order to ensure fair testing, phase offset delays are added to the drivers of fanouts from hazardous nodes and the clock for the cut- point MISR is phase offset from the system clock.
In a second embodiment, an exclusive OR gate is used at the cut-point. The exclusive OR gate eliminates inputs with opposite transitions for eliminating hazardous nodes during testing. In addition, XOR equivalence gates in the IC are determined in both embodiments and are modified for eliminating inherent hazards. Cut-points are optimally inserted at common nodes of the IC for the eliminating the addition of excess hardware to the IC. A modified Johnson counter is preferably used for generating hazard-free input patterns. The invention will be further understood by reference to the following drawings.
BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a flow diagram of the robust delay- fault testing method in accordance with the principals of the present invention.
Fig. 2 is a schematic diagram of a robust delay-fault testing system in accordance with the principals of the present invention.
Fig. 3 is a schematic diagram of a reconvergent fanout circuit. Fig. 4 is a schematic diagram of a reconvergent fanout circuit with a common node.
Fig. 5 is a schematic diagram of a reconvergent fanout circuit with two common nodes.
Fig. 6 is a block diagram of a cut-point circuit.
Fig. 7 is a schematic diagram of a cut-point circuit with applied control signals.
Fig. 8A is a schematic diagram of a NMOS implementation of a cut-point circuit. Fig. 8B is a schematic diagram of a CMOS implementation of a cut-point circuit.
Fig. 9 is a schematic diagram of a NAND realization of an XOR gate.
Fig. 10 is a schematic diagram of a hardware modified XOR gate as shown in Fig. 9 including CMOS switches.
Fig. 11 is a schematic diagram which is equivalent to Fig. 10 when a TEST signal is applied to the circuit under test. Fig. 12 is a schematic diagram of an input generator for providing a modified Johnson counter. Fig. 13 is a schematic diagram of the robust delay-fault testing system shown in Fig. 2 including clocks.
Fig. 14 is a schematic diagram of a circuit showing how application of a hazard-free input pattern to a circuit may still generate hazards in the response.
The figure illustrates the tracing of hazardous nodes using a simulation-guided approach. -jr.
Fig. 15 is a schematic diagram of a graph algorithm for tracing a reconvergent fanout circuit.
Fig. 16 is a schematic diagram of a graph algorithm for tracing a reconvergent fanout circuit.
Fig. 17 is a schematic diagram of the insertion of an exclusive OR gate in a circuit for eliminating hazards in the circuit during testing.
DETAILED DESCRIPTION OF THE INVENTION
During the course of this description like numbers will be used to identify like elements according to the different figures which illustrate the invention. Fig. 1 is a flow diagram illustrating the basic concept of the robust delay-fault testing method in accordance with the principles of the present. The testing concept is used to enable the circuit hardware to be hazard-free during testing in order to provide accurate robust testing of the circuit. A hazard or hazardous node in a circuit can be defined as a non- Exclusive-OR, non-Equivalence gate having inputs with opposite transitions. For Exclusive-OR and Equivalence gates, we define a hazardous gate or node as one having inputs with more than one transition. Depending on the speed of the transitions of the inputs, both inputs can have simultaneous transitions which can result in inaccurate testing of the circuit. For example, a hazard occurs on an OR gate when one input changes from 0 → 1 while a second input changes from 1 → 0. In this case, there can be a momentary instant when the first input has not yet changed to "1" and the second input has changed to "0", therefore, both inputs to the OR gate are "0" and a momentary logic "0" is the OR gate's output.
During testing of the circuit, a transition can be applied to the circuit which propagates through the OR gate and ultimately produces a logic "0" as the OR gate output. However, if the circuit is tested during the momentary logic condition a logic "0" would be produced which would be a false determination of the output condition since the transition had not yet propagated through the system.
XOR and Equivalence gates are inherently sensitive to multiple changes in inputs and are therefore hazard prone. Signal parity is defined as the number of inversions a signal is subject to as it travels along a path. It is difficult to judge the parity of an XOR or Equivalence gate since a signal passing through one of the inputs of an XOR gate can either pass as it is or can be inverted depending upon the signal on the other input. An XOR gate is hazard prone even if there are similar transitions on the inputs.
Paths in the circuit which are responsible for hazards are cut in order to achieve the objective of the robust testing concept. A path is cut by inserting a cut-point on any of the lines of the path. For example, when a cut-point is inserted between lines connecting two nodes X and Y of the circuit, the cut-point prevents a signal at X from feeding Y in the test mode of the circuit. Instead, the signal of X is diverted to an observation point and Y is driven from a new hazard-free signal generated by an input pattern generator.
A circuit diagram 10 is generated which illustrates the circuit under test. Circuit diagram 10 includes all internal functional elements and all interconnections between the input and output terminals of the circuit under test; i.e., AND gate, OR gate, NOR gate, NAND gate, or the like. Preferably, the circuit diagram is a logic model contained in a computer memory.
Module 11 determines if any XOR or Equivalence gates are present in circuit diagram 10. XOR and Equivalence gates are modified in block 12 for providing definite signal parity between the input node and the output node, thereby eliminating hazards in the XOR and Equivalence gates.
Simulation or tracing of the circuit diagram 10 is performed in block 14 to determine if hazardous nodes are present in schematic diagram 10. A cut-point is inserted in block 13 to eliminate one or more of the detected hazardous nodes. A cut-point is inserted in an input path of a hazardous node for preventing the input signal from propagating through the hazardous node. The simulation or tracing is repeated until no more hazardous nodes are found in the circuit under test. •
During testing of the circuit a TEST signal is applied to the circuit 16 for activating the cut-points and activating a hardware input generator for performing robust delay testing. In a tester signature compare module 17, a hazard-free input pattern is applied to primary inputs of the circuit as well as to inputs of cut-points in order to avoid hazards in the response. A hazard- free input pattern can be defined as a sequence of input vectors that differ successively by one bit. The hazard-free input pattern initiates only one transition at any one of the primary inputs while all other primary inputs are held at a constant value. A hazard-free input pattern is a sequence that avoids multiple transitions on inputs to the circuit under test. A hazard-free input pattern can be generated by a Gray code generator or a Johnson counter. A Gray code counter has the disadvantage that for large numbers of inputs the cycle time is too large for practical testing. A Johnson counter generates 2n basic patterns. A Johnson counter has the disadvantage of not supplying sufficient patterns for all fault sites in the circuit. Preferably, a modified Johnson counter including controlled complementation of the Johnson counter outputs is used for providing sufficient numbers of hazard-free input patterns.
After the hazard-free input pattern is appli^ to the circuit, primary outputs are checked with a multiple input signature register for comparing the outputs to a correct value during any clock period. Outputs from the cut-points at the observation point are checked with a second multiple input signature register. If a wire or gate is too slow, then a signal being sampled by tester signature compare module 17 will be incorrect. If the signature has an incorrect value an error signal will be generated in block 18 and if the signature has a correct value the circuit testing is complete and the circuit is tested in module 19.
Fig. 2 illustrates an apparatus for a robust delay-fault built-in self-testing (BIST) system 20 in accordance with the principles of the present invention. In BIST system 20, test vector generation and test result verification are accomplished by built-in circuitry. Input pattern generators 21 are disposed at the boundary of integrated circuit (IC) 22 to generate an input pattern for testing IC 22. Primary inputs 24 extend between input pattern generator 21 and IC circuit 22.
Primary input 24b is connected by path 26a to hazardous node 28. A cut-point 23 is inserted in hazardous node 28 to direct input from path 26b to an observation point which is a cut-point multiple input signature register (MISR) 30. Outputs 27 of IC 22 are received at output MISR 25. Hazard-free input patterns are generated by input pattern generator 21 and are applied to primary inputs 24 and cut-point 23 during testing. A new bit of hazard-free input pattern is applied to fanout circuit 31 through path. 26a in test mode. After application of the hazard-free input pattern, responses from IC 22 are combined in output MISR 25. Output MISR 25 receives a succession of input responses from primary outputs 27 of IC 22 and produces a derived function thereof or first signature. Responses for cut-points 23 are combined in cut-point MISR 30 for providing a second signature.
Buffers 29 can be inserted in primary input path 24b in order to delay the hazard-free input pattern sufficiently to fairly test the primary input paths which are connected to a cut-point. It has been found that a transition in the hazard-free input pattern applied to primary input path 24b will arrive more quickly at output MISR 25 than a transition from an ordinary primary input 24a will arrive at output MISR 25. A signature computed at cut-point MISR 30 may not catch a delay-fault, if the fault is of a size large enough to delay the path from primary input 21 to output MISR 25. A delay-fault between input generator 21 and output MISR 25 can be larger than a delay-fault between cut-point 23 and output MISR 25. A signature computed for the delay-fault between input signature generator 21 and output MISR 25 would not catch the shorter delay-fault of cut-point 23. The number of buffers 29 used in system 20 depends on the computed nominal delay of each path driven from cut-point 23 to output MISR 25. Buffers 29 delay the transition sufficiently so that a signature can be effectively evaluated by output MISR 25.
Fig. 3 illustrates an example of a hazardous node 38 which is a reconvergent fanout circuit 31. Reconvergent fanout circuit 31 splits into at least two paths before it reconverges at another point in the circuit. Signal parity can be defined as the number of inversions a signal is subject to as it travels along a path. A reconvergent fanout circuit can have a hazard if the paths are of opposite signal parity. Output paths 34, 36 from NAND gate 32 reconverge at OR gate 38. Path 34 is inverted with inverter 35 to produce odd signal parity on path 37. Path 36 from AND gate' 32 has even signal parity. Hazardous node 38 is an OR gate. A hazard can occur at hazardous node 38 since path 37 has opposite signal parity to path 36. Path 36 or path 37 can be cut to avoid a hazard at hazardous node 38. If there are more than two paths reconverging, then we cut either all odd parity paths or all even parity paths. It has been found that inserting cut-points in every reconverging pair may result in inserting too many cut-points into IC 22. Preferably, cut-points are inserted at minimal locations in IC 22 for providing minimal additional hardware to the IC-. Fig. 4 illustrates a reconvergent fanout 40 with the minimal number of cut-points inserted. IC 22 can include a reconvergent fanout 40 that includes one or more "K" common nodes 41. Cycle 1 starts from X. source node 45, traverses paths 46 and 44 to Y1 end node 54 and returns back to the X_, source node 45 along path 47 in the opposite direction. Cycle 2 starts from X2 source node 50, traverses path 43 to Y2 end node 56 and returns along path 44 in the opposite direction. Node 41. is common to cycles 1 and 2. The number of cycles "n" that intersect at node K means that node K will have at least n fanout lines. A cut-point is inserted at the fanout stem of node K in order to cut off "n" fanout lines from node K's output and thereby save n-1 cut points from the hardware overhead of the IC.
X1 source node 45 is a NAND gate which includes output paths 46 and 47. X2 source node 50 is a OR gate which includes output paths 43 and 44. Output path 46 is received at inverter 48 which provides input 44 to the Y1 end node. Y1 end node 54 is an OR gate. Output paths 47 and 43 provide input to node 41. Output 51 from node 41 is received at Y1 end node 54 and inverter 58. Inverter 58 provides input to Y2 end node 56. Preferably, cut- point 23 is inserted after node 41 between Y1 end node 54 and inverter 58 for cutting input to both Y1 end node 54 and Y2 end node 56 with the minimal number of cut-points.
Fig. 5 illustrates a reconvergent fanout circuit 60 with K1 node 74 and K2 node 76 in common.
Path 61 is shared by cycles 1 and 2. Inserting cut-point
23 between nodes 74 and 76 is optimal for reducing hardware overhead.
X1 source node 62 is an exclusive OR which includes output paths 66 and 68. Path 66 provides input to inverter 77 which provides input to Y_, end node 78. Y, end node 78 is a NAND gate. X2 source node 64 is a NAND gate which includes output paths 70 and 72. Output path 68 and output path 70 are received at node 74. Node 74 is an AND gate and node 76 is an OR gate. Path 72 provides input to inverter 75 which provides input to Y2 end node 80. Y2 end node 80 is a NAND gate. Output 61 from node 74 provides input to node 76. Output 79 from node 76 provides input to Y1 end node 78 and Y2 end node 80. Preferably, cut-point 23 is inserted in path 61 between node 74 and node 76. In this arrangement input paths to node Y1 end node 78 and Y2 end node 80 are cut by the insertion of cut-point 23. A cut-point can be inserted anywhere on the reconvergent opposite parity paths. Accordingly, cut-points can be inserted away from critical paths of the circuit 2.
Fig. 6 is a block diagram of cut-point 23 inserted between X input node 90 and Y output node 92. Output 91 from input node 90 is connected to cut-point 23 and output 93 of cut-point 23 is connected to output node 92. Output 94 from input pattern generator 21 is connected to cut-point 23 and output 96 from cut-point 23 is connected to cut-point MISR 30. Control inputs TEST and TEST are applied to cut-point 23 to respectfully provide or remove connections of cut-point 23 between input node 90 and output node 92.
Fig. 7 illustrates a circuit diagram of cut- point 23 including test switches 100a, 100b and 100c. A control input of TEST is supplied to test circuit 100a for closing circuit 100a during normal operation of IC 22. A control input of TEST is applied to circuits 100b and 100c for closing circuits 100b and 100c when IC 22 is in a test mode.
Fig. 8A illustrates implementation of cut-point 23 in a NMOS MOSFET logic system. Fig. 8B illustrates implementation of cut-point 23 in a complementary MOSFET (CMOS) logic system. Test is applied to transistor 110a during normal operation of IC 22. TEST is applied to transistors 100b and 110c during TEST mode of IC 22. It will be appreciated that other logic systems such as TTL, ECL, BiCmos and it will be appreciated that other logic systems known in the art can be used with the teachings of the present invention.
Fig. 9 illustrates a NAND implementation 120 of an XOR Equivalence gate. X, source node 122 and X2 source node 124 have two reconverging paths to Y end node 142 with one inverted and one non-inverted path. Path 123 from node 122 is non-inverted and path 121 from node 122 is inverted with inverter 126. Path 125 from node 124 is non-inverted and path 127 is inverted with inverter 128. Paths 129 and 125 are applied to NAND gate 132 and paths 123 and 130 are applied to NAND gate 134. Outputs 136 and 138 are applied to NAND gate 140 which provides output 142 to Y end node 142. Fig. 10 illustrates a hardware modification of the NAND implementation of the XOR gate 120 in an NMOS implementation. Four NMOS switches are used to change the configuration of the XOR gate 120 such that source node X1 has only non-inverting paths to end node Y and source node X, has only inverting paths to node Y. NMOS switches 144a. 144b, 144c and 144d are inserted to change the configurations of NAND gates 132 and 134 so that input 122 has non-inverting paths to end node 142 and input 124 has inverting paths to end node 142. NMOS switch 144a is inserted in the path 121 from node 122 to inverter 126 and NMOS switch 144b is inserted in path 133 from node 124 to inverter 126. NMOS switch 144c is inserted in path 131 from node 122 to NAND gate 132 and NMOS switch I44d is inserted in path 135 from node 124 to NAND 132. Fig. 11 illustrates an equivalent diagram of
Fig. 10 with a control signal TEST asserted. If node X1 is input and node X2 is input the logic function for the circuit is Y = X.,X2. Paths 150 and 152 from node X. provide non-inverted inputs to NAND gates 132 and 134. and paths 154 and 156 are inverted with respective inverters 126, 128 to provide inverted inputs to gates 132 and 134. Paths from X1 and X2 to Y have definite signal parity.
Fig. 12 illustrates input pattern generator 160 for generating hazard-free input patterns which can be^ used for input pattern generator 21, shown in Fig. 2. A test pattern is a sequence of vectors in which each element, which represents the logic value to be given the primary inputs, takes one of the values, 0, 1, η a falling transition or \- a rising transition. Input pattern generator 160 is a modified Johnson counter for providing more than 2n patterns. Single bits of the pattern are selectively inverted to result in a new pattern. Outputs 164 of Johnson counters 161 are applied to a linear array of two-input XOR gates 162. Input 166 of XOR gate 162 receives a control signal from input lines 168. If a high control signal is received at XOR gate 162 a corresponding bit of the Johnson counter 163 is complemented. For example, if the output of Johnson counter is 11111 and a complement is applied to the most significant bit the pattern applied would be 01111 and if no complement is applied the pattern would be 11111. In the modified Johnson ' counter, generated patterns cyclically alternate between complemented patterns and uncomplemented patterns to produce successive one-bit transition changes. An illustration of a modified Johnson scheme is shown in Table 1.
TABLE 1
CONTROLLED SWITCHING FOR SINGLE-SIT COMPLEMENTATION SCHEME
Output of Operational Pattern Johnson Counter Mode A-pliei
11111 C (Bit 4) 01111
NC 11111 01111 NC 01111
C (Bit 4) 11111 00111 C (Bit 4) 10111 NC 00111
00011 NC 00011
C (Bit 4) 10011
00001 C (Bit 4) 10001 NC 00001
00000 NC 00000
C (Bit 4) 10000
10000 C (Bit 4) 00000 NC 10000
11000 NC 11000
C (Bit 4) 01000
11100 C (Bit 4) 01100 NC 11100
11110 NC 11110
C (Bit 4) OHIO
11111 C (Bit 4) 01111 NC 11111
C (Bit . 4) - Complement bit -4
NC - No complement
Each bit complemented produces 2x (# of Johnson counter patterns) or 2 x (2n) = 4n patterns A pair-wise complementation scheme can be used to generate non-hazardous input patterns. In a pair-wise scheme two bits can be complemented simultaneously. For example, the third and fourth bits can be selected for complementation. An uncomplemented pattern of the third and fourth bits is applied as the first pattern. The fourth bit is complemented as the second pattern. The third bit complemented along with the fourth bit as the third pattern. The fourth bit is uncomplemented and the third bit is complemented for the fourth pattern and both the third and fourth bits are uncomplemented for the fifth pattern. A pair-wise complementation scheme has a cycle length of 5 x 2n patterns. It will be appreciated that other complementation schemes can be used for generating hazard-free input patterns.
Fig. 13 illustrates system 20 including clock 200 for cut-point MISR 30. A signal from path 26b to cut-point MISR 30 can arrive before a signal.from primary inputs 24 to output MISR 25 on path 26a. Cut-point MISR 30 should be clocked at the right instant for evaluating a fair signature in cut-point MISR 30 which reflects the exact delay of the paths which have been cut. Clock 200 operates at an earlier instant than system clock 202 of IC 22. Preferably, clock 200 is offset by a phase that depends on the overall delay of paths being cut at cut- points 23.
In a first embodiment, a simulation of BIST system 20 is used to determine hazardous nodes and for optimally analyzing the insertion of cut-points 23 into BIST system 20. A hazard modelling algebra can be used for simulating the hazardous nodes of the circuit under test. Preferably, Chakraborty's 13-valued hazard- modeling algebra can be used for modeling BIST system 20 as described in T.J. Chakraborty, Vishwani D. Agrawal and M.L. Bushnell, Path Delay Fault Testing For Logic Circuits, Proceedings of the 29th Design Automation Conference, p. 165-172, June 1992. In this algebra, logic values of two consecutive clock periods are described by a transition state. Three signal states (0, 1, and _ represent nine transition states and eighteen transitions with a hazard or no-hazard condition. These eighteen values eventually collapse into thirteen unique values and each is represented by a triplet. For example, 1 | h | 0 represents a logic 1 to 0 transition with a hazard in between, while 1 | nh | 0 represents the same transition with no hazard. Table 2 shows the various transition states with a corresponding numerical encoding. The symbol ' * ' in the table stands for a condition where a hazard may or may not be present (unknown status) . Other hazard capturing algebras that can be used in the present invention are described as: Smith Algebra, "Model for Delay Faults Based Upon Paths", Gordon L. Smith, IBM, Proceedings of the 1985 International Test Conference, pp. 342-349, 1985; Reddy Algebra, "On Delay Fault Testing in Logic Circuits", C.J. Lin and S.M. Reddy, IEEE Transactions on Computer-Aided Design, 6(5) : 694-703, Sept., 1987; and Schulz et al. Algebra, "Advanced Automatic Test Pattern Generation Techniques for Path Delay Faults", M.H. Schulz, F. Fuchs, and F. Fink, Proceedings of the 19th Symposium on Fault- Tolerant Computing, pp. 44-51, June 1989.
TABLE 2 THIRTEEN-VALUED ALGEBRA Values Codes
0 I h I 0 1 0 h 0 2
0 - X 3
1 h 0 4
1 h 1 5
1 - X 6
X - 0 7
X - 1 8
X - 1 9
0 nh 1 o 10
00 I nnhh 1 I i1 11
0 nh 1 i 12
0 nh 1 o 13
A truth table for each type of gate is implemented in the simulation for determining output values. For example, for a AND gate, if two opposite transitions without a hazard are input, the output must contain a hazard value. During circuit simulation, a modified Johnson pattern sequence is applied to simulated inputs to determine hazardous nodes. If a hazardous node is found, paths leading to the hazardous node are traced to the primary input that had a transition for finding paths with opposite transitions. A signature is generated by the simulation and compared with a correct value. Fig. 14 illustrates simulation guided path tracing by applying hazard-free input patterns for determining hazardous nodes. Hazardous nodes are determined by following transitions through IC 22. A hazard-free input pattern "1100" is applied to primary inputs 302, 304, 306 and 308. A falling transition is applied to primary input 310 which results in a hazard at NOR gate 324. Path searching starts with primary input 310 through AND gates 318 and 320 to NOR gate 324. Paths leading to AND gate 314 and NOR gate 322 are avoided since there are no transitions on them. Results for robust delay-fault testing with
BIST system 20 are shown in Table 3.
TABLE 3
DELAY FAULT BIST COVERAGES FOR TEST CIRCUITS
Initial Sizes Robust Total Final Sizes Circuit # # # Delay Cut- # # # Gates Pi's PO's Fault Points Gates Pi's PO's
Coverage Added
Roth 8 4 3 100% 2 10 6 5
Smith 6 6 2 100% 0 6. 6 2 cl7 6 5 2 100% 1 7 6 3
ALU181 58 14 8 97.69% 8 74 22 16
C432 160 36 7 99.89% 12 190 48 19
C880 383 60 26 99.46% 42 425 102 68
Results indicate that greater than 97% coverage was performed for the tested benchmark circuits.
In a second embodiment, hazardous nodes are identified by enumerating the dominated gates in the circuit. For example, gate A dominates gate B if all paths from A to a circuit output flow through gate B. Gate A can be defined as a dominator. After do inators in IC 22 have been identified, a graph algorithm can be used to enumerate paths in the circuit between dominators to trace IC 22. From the graph algorithm, the points in the circuit in which a cut-point should be inserted can be determined. Parity is assigned to each path of IC 22. If a node is encountered more than once it is assigned a parity of either "ALL EVEN" indicating all paths were even, "ALL ODD" indicating all paths were odd or "BOTH" indicating a mixture of even and odd parity paths.
;; Fig. 15 illustrates a use of a graph algorithm system 350 for tracing fanout circuit 31 shown in Fig. 3. D1 dominator 352 has path 364 which is assigned ODD parity. Paths 354 and 356 to D2 dominator 362 are assigned EVEN parity. D2 dominator 362 has a parity defined as BOTH which is a hazardous node. The parity at D2 dominator 362 is fixed at ODD or EVEN by a decoupling paths between D1 dominator 352 and D2 dominator 362. A cut-point 413 is inserted in path 358 for decoupling the paths between D1 and D2.
If a node D in a fanout region of another node includes node D2 dominator 362 which has a parity of BOTH, node D is assigned a party of "SHADOW". Parity of SHADOW is defined as a node which is not considered in the current pass of graph analysis. Paths to node D will not be considered in the current pass until the parity of D2 is decided to be either ODD or EVEN. ODD and EVEN parity paths are grouped starting and ending at the same nodes to avoid duplicating these paths for nodes driven by the same node. Fig. 16 illustrates a graph algorithm for tracing circuit 400. Dχ dominator 402 has path 404 to Dγ dominator 406. D1 dominator 408 has paths 410, 412 and 414 to D2 dominator 416. Paths 410 and 414 have EVEN parity. Inverter 418 provides ODD parity to path 412. D2 dominator 416 is assigned parity BOTH. In the current pass of the graph algorithm, Dγ dominator 406 is assigned parity SHADOW. A cut-point 413 can be inserted in path 412 for decoupling. D1 dominator 402 and D2 dominator 406 to provide EVEN parity to D2 dominator 402. After cut-point 413 is inserted, Dγ dominator 406 will have EVEN parity.
Nodes which decouple the maximum number of paths are chosen for the insertion of cut-points. A group of EVEN parity paths with m ODD parity paths and n EVEN parity paths form a path group equivalent to mxn path pairs. If a cut-point is inserted in an EVEN (ODD) parity path, it is equivalent to decoupling m(n) path pairs. Weights are assigned to nodes of the path group to determine the optimum node to insert a cut-point. Each node in an EVEN (ODD) parity path is assigned weight m(n) . If the node is also in other path groups, the node weight is computed by adding all path group weights.
After a cut-point is inserted at a node, all paths of the node are determined to be hazard-free. If additional paths in the circuit are present, weights for insertion of additional cut-points are calculated and additional cut-points are inserted. The graph algorithm is continued until the entire circuit has been traced.
Fig. 17 illustrates an alternative to using cut-points to eliminate hazardous nodes in IC 22 in block 13 of Fig. 1. Fanout circuit 500 includes AND gate 501 connected by path 502 to NAND gate 504. PATH 506 is inverted with inverter 508. Exclusive OR gate 510 is inserted in path 508 to eliminate hazards in fanout circuit 500. Exclusive OR gate 510 has input 508, TEST input 512 and output 514. Output 516 is received at NAND gate 504.
In normal operation of circuit 500, a TEST signal of "0" is applied to input 512. The signal on input 508 will be the same as the signal on output 514. The NAND gate will have ODD parity from path 502 and EVEN parity from path 514 and is a hazardous node. During testing, TEST signal is set to "1". Exclusive OR gate 510 inverts path 508 to EVEN parity to eliminate a hazardous node in NAND gate 504. The use of an Exclusive OR gate for eliminating hazards has the advantage of reducing the hardware added to the IC. The use of the cut-point circuit for eliminating hazards has the advantage of providing additional information from the cut-point MISR for observing and controlling the IC. The present invention has the advantage of providing robust built-in self-testing of delay faults in integrated circuits. Hazardous nodes of the circuit are determined. A cut- point is inserted to eliminate a hazard at the hazardous node. The signal on the cut-line is diverted to an observation point cut-point MISR during testing of the circuit. The signal that normally would fan out from the cut-point is instead driven by a hazard- free pattern generated by an extra primary input, which is added to the circuit. A signature is computed at the output of the circuit and a second signature is computed at the observation point. Hazard- free input patterns are applied to the circuit and the computed signatures are compared to known good signatures. The system has the advantage of quickly determining delay faults for numerous integrated circuits.
While the invention has been described with r -.erence to the preferred embodiment thereof, it will be appreciated by those of ordinary skill in the art that modifications can be made to the structure and form of the invention without departing from the spirit and scope thereof.

Claims

WE CLAIM : 30
1. A method for robust delay-fault testing of an integrated circuit comprising the steps of: determining hazardous nodes in said integrated circuit; inserting cut-points at said hazardous nodes for diverting input to said hazardous nodes to an observation point; applying a succession of first input patterns to said integrated circuit on paths of said IC having said cut-points; applying a succession of second input patterns to said integrated circuit on paths of said IC without cut-points; processing the output responses at said observation point of said integrated circuit to said succession of first input patterns to produce a first signature; processing the output responses of said integrated circuit to said succession of second input patterns to produce a second signature; and, comparing said first and second signatures with a known correct first and second signature for said integrated circuit.
2. The method of claim 1 further comprising the steps of: identifying XOR and equivalence gates in said integrated circuit, said XOR and equivalence gates including a first path from a first input node to an output node and a second path from a second input node to said output node; and modifying said XOR and equivalence gates so that said first path is inverted and said second path is non-inverted for eliminating hazards in said-gates before said step of applying a succession of input patterns to said integrated circuit.
3. The method of claim 2 wherein said input patterns are hazard-free input patterns.
4. The method of claim 3 wherein said input patterns are generated by a modified Johnson counter.
5. The method of claim 4 wherein said step of inserting cut-points at said hazardous nodes includes determining common hazard nodes and inserting said cut- point before said common hazardous node.
6. The method of claim 5 wherein a first and second of said common nodes is determined and wherein said method further comprises the step of: inserting said cut-point between said first and second common nodes.
7. The method of claim 6 wherein said hazardous nodes are determined by applying a hazard simulation algebra to said integrated circuit.
8. The method of claim 7 wherein said hazard simulation Algebra is chosen from the group comprising Chakraborty's 13 valued algebra, Smith Algebra, Reddy Algebra and Schultz et al. algebra.
9. The method of claim 6 wherein said hazardous nodes are determined by a graph algorithm for tracing said integrated circuit.
10. A system for robust delay fault testing of an integrated circuit comprising: inputs coupled to said integrated circuit; means for generating a succession of first and second input patterns coupled to said inputs; means for determining hazardous nodes in said integrated circuit; a first multiple input signature register for receiving outputs to said succession of first input patterns from said integrated circuit; means for inserting cut-points at said hazardous nodes for diverting said succession of second input patterns to said hazardous node to a second multiple input signature register; and means for applying a hazard-free input pattern o said inputs.
11. The system of claim 10 further comprising: means for determining XOR and equivalence gates in said integrated circuit, said XOR and equivalence gate including a first path from a first input node to an output node and a second path from second input node to said output node; and, means for modifying said XOR and equivalence gates in said integrated circuit so that said first path is inverted and said second path is non-inverted for eliminating hazards on said gates.
12. The system of claim 11 further comprising: at least one buffer located between said inputs and said cut-points.
13. The system of claim 12 further comprising: a modified Johnson counter for generating said hazard- free input pattern.
14. The system of claim 13 wherein said modified Johnson counter includes one bit complementation.
15. The system of claim 14 wherein said modified Johnson counter includes a two bit complementation.
16. The system of claim 15 wherein said hazardous node is a reconverged fanout circuit, said fanout circuit having inputs with opposite polarity.
17. The system of claim 16 further comprising: a phase-offset clock on said second MISR register for ensuing fair testing of path propagation delays.
18. The system of claim 17 wherein said cut- points and said first and second multiple input signature registers and said means for generating a succession of input patterns are implemented in a NMOS MOSFET or CMOS MOSFET logic system.
19. A method for robust delay fault testing of an integrated circuit comprising the steps of: determining hazardous nodes in said integrated circuit; and eliminating said hazardous nodes during testing of said circuit.
20. The method of claim 19 wherein said step for eliminating hazardous nodes comprises inserting an exclusive OR gate at said hazardous node for eliminating opposite parity at said hazardous node during said testing.
PCT/US1994/008238 1993-07-23 1994-07-21 Robust delay fault built-in self-testing method and apparatus WO1995003589A1 (en)

Priority Applications (7)

Application Number Priority Date Filing Date Title
AU73699/94A AU7369994A (en) 1993-07-23 1994-07-21 Robust delay fault built-in self-testing method and apparatus
DE69423598T DE69423598T2 (en) 1993-07-23 1994-07-21 METHOD AND DEVICE FOR THE ROBUST AND AUTOMATIC TESTING OF DELAY ERRORS
EP94922668A EP0663092B1 (en) 1993-07-23 1994-07-21 Robust delay fault built-in self-testing method and apparatus
AT94922668T ATE191091T1 (en) 1993-07-23 1994-07-21 METHOD AND DEVICE FOR ROBUST AND AUTOMATIC CHECKING OF DELAY ERRORS
KR1019950701125A KR950703767A (en) 1993-07-23 1994-07-21 Independent self-test method and device for delayed failure [ROBUST DELAY FAULT BUILT-IN SELF-TESTING METHOD AND APPARATUS]
JP7505311A JPH08502365A (en) 1993-07-23 1994-07-21 Robust delay fault built-in self-test method and apparatus
CA002145403A CA2145403C (en) 1993-07-23 1994-07-21 Robust delay fault built-in self-testing method and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US08/096,731 US5422891A (en) 1993-07-23 1993-07-23 Robust delay fault built-in self-testing method and apparatus
US08/096,731 1993-07-23

Publications (1)

Publication Number Publication Date
WO1995003589A1 true WO1995003589A1 (en) 1995-02-02

Family

ID=22258821

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US1994/008238 WO1995003589A1 (en) 1993-07-23 1994-07-21 Robust delay fault built-in self-testing method and apparatus

Country Status (9)

Country Link
US (1) US5422891A (en)
EP (1) EP0663092B1 (en)
JP (1) JPH08502365A (en)
KR (1) KR950703767A (en)
AT (1) ATE191091T1 (en)
AU (1) AU7369994A (en)
CA (1) CA2145403C (en)
DE (1) DE69423598T2 (en)
WO (1) WO1995003589A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109388839A (en) * 2017-08-14 2019-02-26 龙芯中科技术有限公司 Clock system method for analyzing performance and device

Families Citing this family (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
BE1004959A3 (en) * 1991-06-28 1993-03-02 Bell Telephone Mfg Method and device for testing atm connections.
JP3124417B2 (en) * 1993-07-13 2001-01-15 三菱電機株式会社 Logic simulation system and logic simulation method
JPH07114580A (en) * 1993-10-18 1995-05-02 Fujitsu Ltd Delay time analysis system for logical device
US5583787A (en) * 1994-03-08 1996-12-10 Motorola Inc. Method and data processing system for determining electrical circuit path delays
GB9417602D0 (en) * 1994-09-01 1994-10-19 Inmos Ltd A controller for implementing scan testing
US5974579A (en) * 1996-09-03 1999-10-26 Credence Systems Corporation Efficient built-in self test for embedded memories with differing address spaces
WO1998018077A1 (en) * 1996-10-23 1998-04-30 Rutgers University Method and system for identifying tested path-delay faults
US6018813A (en) * 1997-04-21 2000-01-25 Nec Usa, Inc. Identification and test generation for primitive faults
KR100292821B1 (en) * 1997-09-08 2001-06-15 윤종용 Parallel signature compression circuit
US5903577A (en) * 1997-09-30 1999-05-11 Lsi Logic Corporation Method and apparatus for analyzing digital circuits
US6148425A (en) * 1998-02-12 2000-11-14 Lucent Technologies Inc. Bist architecture for detecting path-delay faults in a sequential circuit
WO1999045667A1 (en) * 1998-03-03 1999-09-10 Rutgers University Method and apparatus for combined stuck-at fault and partial-scanned delay-fault built-in self test
US6308300B1 (en) * 1999-06-04 2001-10-23 Rutgers University Test generation for analog circuits using partitioning and inverted system simulation
JP2001042012A (en) * 1999-07-29 2001-02-16 Mitsubishi Electric Corp Test pattern-generating apparatus, method for cutting loop, method for cutting propagation path, method for detecting delay failure and computer-readable recording medium with program recorded for making computer execute the method
US6760873B1 (en) * 2000-09-28 2004-07-06 Lsi Logic Corporation Built-in self test for speed and timing margin for a source synchronous IO interface
US6728914B2 (en) * 2000-12-22 2004-04-27 Cadence Design Systems, Inc Random path delay testing methodology
US20020194558A1 (en) * 2001-04-10 2002-12-19 Laung-Terng Wang Method and system to optimize test cost and disable defects for scan and BIST memories
JP2003233639A (en) * 2002-02-06 2003-08-22 Mitsubishi Electric Corp Failure verification device, failure verification method and failure analysis method
US7082558B2 (en) * 2002-11-25 2006-07-25 Texas Instruments Incorporated Increasing possible test patterns which can be used with sequential scanning techniques to perform speed analysis
JP2005308471A (en) * 2004-04-20 2005-11-04 Matsushita Electric Ind Co Ltd Path delay test method
US20060107055A1 (en) * 2004-11-17 2006-05-18 Nesvis, Networks Method and system to detect a data pattern of a packet in a communications network
US8051352B2 (en) * 2006-04-27 2011-11-01 Mentor Graphics Corporation Timing-aware test generation and fault simulation
JP5223735B2 (en) * 2009-03-10 2013-06-26 富士通株式会社 Memory test circuit and processor
JP5353679B2 (en) * 2009-12-17 2013-11-27 富士通株式会社 Failure diagnosis support program and failure diagnosis support device
US8990760B2 (en) * 2011-08-26 2015-03-24 Mentor Graphics Corporation Cell-aware fault model generation for delay faults

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4787062A (en) * 1986-06-26 1988-11-22 Ikos Systems, Inc. Glitch detection by forcing the output of a simulated logic device to an undefined state
US4965474A (en) * 1988-09-16 1990-10-23 Texas Instruments Incorporated Glitch suppression circuit

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS58106464A (en) * 1981-12-21 1983-06-24 Advantest Corp Glitch detecting and measuring instrument
NL8303536A (en) * 1983-10-14 1985-05-01 Philips Nv LARGE-INTEGRATED CIRCULATION WHICH IS DIVIDED IN ISOCHRONIC AREAS, METHOD FOR DESIGNING SUCH AN INTEGRATED CIRCUIT, AND METHOD FOR TESTING SUCH AS INTEGRATED CIRCUIT.
US4688223A (en) * 1985-06-24 1987-08-18 International Business Machines Corporation Weighted random pattern testing apparatus and method
US4745355A (en) * 1985-06-24 1988-05-17 International Business Machines Corporation Weighted random pattern testing apparatus and method
US4687988A (en) * 1985-06-24 1987-08-18 International Business Machines Corporation Weighted random pattern testing apparatus and method
US4801870A (en) * 1985-06-24 1989-01-31 International Business Machines Corporation Weighted random pattern testing apparatus and method
US4635261A (en) * 1985-06-26 1987-01-06 Motorola, Inc. On chip test system for configurable gate arrays
US4672307A (en) * 1985-12-20 1987-06-09 University Of Southern California Simplified delay testing for LSI circuit faults
JP2582250B2 (en) * 1986-10-03 1997-02-19 日本電信電話株式会社 Timing signal delay circuit device
JPS63256877A (en) * 1987-04-14 1988-10-24 Mitsubishi Electric Corp Test circuit
JPS647400A (en) * 1987-06-29 1989-01-11 Hitachi Ltd Ic tester
US5184067A (en) * 1988-07-12 1993-02-02 Kabushiki Kaisha Toshiba Signature compression circuit
US5065090A (en) * 1988-07-13 1991-11-12 Cross-Check Technology, Inc. Method for testing integrated circuits having a grid-based, "cross-check" te
US5051996A (en) * 1989-03-27 1991-09-24 The United States Of America As Represented By The United States Department Of Energy Built-in-test by signature inspection (bitsi)
US5095483A (en) * 1989-04-28 1992-03-10 International Business Machines Corporation Signature analysis in physical modeling
US5023875A (en) * 1989-05-26 1991-06-11 Hughes Aircraft Company Interlaced scan fault detection system
US5056094A (en) * 1989-06-09 1991-10-08 Texas Instruments Incorporated Delay fault testing method and apparatus
US5091872A (en) * 1989-06-23 1992-02-25 At&T Bell Laboratories Apparatus and method for performing spike analysis in a logic simulator
US5138619A (en) * 1990-02-15 1992-08-11 National Semiconductor Corporation Built-in self test for integrated circuit memory
JP2561164B2 (en) * 1990-02-26 1996-12-04 三菱電機株式会社 Semiconductor integrated circuit
US5504432A (en) * 1993-08-31 1996-04-02 Hewlett-Packard Company System and method for detecting short, opens and connected pins on a printed circuit board using automatic test equipment

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4787062A (en) * 1986-06-26 1988-11-22 Ikos Systems, Inc. Glitch detection by forcing the output of a simulated logic device to an undefined state
US4965474A (en) * 1988-09-16 1990-10-23 Texas Instruments Incorporated Glitch suppression circuit

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109388839A (en) * 2017-08-14 2019-02-26 龙芯中科技术有限公司 Clock system method for analyzing performance and device
CN109388839B (en) * 2017-08-14 2023-05-30 龙芯中科技术股份有限公司 Clock system performance analysis method and device

Also Published As

Publication number Publication date
EP0663092A1 (en) 1995-07-19
CA2145403A1 (en) 1995-02-02
CA2145403C (en) 2002-01-29
DE69423598T2 (en) 2000-09-14
EP0663092A4 (en) 1996-02-14
US5422891A (en) 1995-06-06
EP0663092B1 (en) 2000-03-22
JPH08502365A (en) 1996-03-12
DE69423598D1 (en) 2000-04-27
KR950703767A (en) 1995-09-20
AU7369994A (en) 1995-02-20
ATE191091T1 (en) 2000-04-15

Similar Documents

Publication Publication Date Title
US5422891A (en) Robust delay fault built-in self-testing method and apparatus
CA1273062A (en) Programmable logic array
EP0006328B2 (en) System using integrated circuit chips with provision for error detection
Greenstein et al. E-PROOFS: a CMOS bridging fault simulator
EP0737337B1 (en) Apparatus and method for testing integrated circuits
Li et al. Diagnosis of sequence-dependent chips
Muehldorf et al. LSI logic testing—An overview
JPH0544690B2 (en)
Park et al. A statistical model for delay-fault testing
US6173426B1 (en) Method of locating faults in LSI
Favalli et al. Sensing circuit for on-line detection of delay faults
Pomeranz et al. Unspecified transition faults: a transition fault model for at-speed fault simulation and test generation
Chakraborty et al. On variable clock methods for path delay testing of sequential circuits
US6904554B2 (en) Logic built-in self test (BIST)
Gong et al. Locating bridging faults using dynamically computed stuck-at fault dictionaries
US7500165B2 (en) Systems and methods for controlling clock signals during scan testing integrated circuits
Chakraborty et al. Path delay fault simulation algorithms for sequential circuits
Chakraborty et al. Path delay fault simulation of sequential circuits
JP3104739B2 (en) LSI tester used for LSI failure analysis
Chakraborty et al. Robust testing for stuck-at faults
Sziray Test calculation for logic and delay faults in digital circuits
Chakraborty et al. Simulation of at-speed tests for stuck-at faults
JPH09159727A (en) Cmos semiconductor device
Chandramouli et al. A new BIST scheme for CMOS PLAs
Gong et al. A Diagnosis Algorithm for Bridging Faults in Combinational Circuits

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AT AU BB BG BR BY CA CH CZ DE DK ES FI GB HU JP KP KR KZ LK LU MG MN MW NL NO NZ PL PT RO RU SD SE SK UA VN

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): AT BE CH DE DK ES FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN ML MR NE SN TD TG

WWE Wipo information: entry into national phase

Ref document number: 2145403

Country of ref document: CA

WWE Wipo information: entry into national phase

Ref document number: 1994922668

Country of ref document: EP

121 Ep: the epo has been informed by wipo that ep was designated in this application
WWP Wipo information: published in national office

Ref document number: 1994922668

Country of ref document: EP

REG Reference to national code

Ref country code: DE

Ref legal event code: 8642

WWG Wipo information: grant in national office

Ref document number: 1994922668

Country of ref document: EP