Abstract: An integrated circuit device includes an array of read/write memory cells, application logic circuitry, and address decoder circuitry coupled to receive input from the application logic circuitry and to provide output to the array of memory cells. The address decoder circuitry is reversible by having a bijective transfer function from the inputs to the outputs of the address decoder circuitry, and conservative by having the same number of 1's at the input and the output. During a test, the application logic circuitry provides a test value and test ancilla bits to the address decoder circuitry. During normal operation, the application logic circuitry provides an application memory address and constant ancilla bits to the address decoder circuitry.

Abstract: An integrated circuit device includes an array of read/write memory cells, application logic circuitry, and address decoder circuitry coupled to receive input from the application logic circuitry and to provide output to the array of memory cells. The address decoder circuitry is reversible by having a bijective transfer function from the inputs to the outputs of the address decoder circuitry, and conservative by having the same number of 1's at the input and the output. During a test, the application logic circuitry provides a test value and test ancilla bits to the address decoder circuitry. During normal operation, the application logic circuitry provides an application memory address and constant ancilla bits to the address decoder circuitry.

Abstract: A method testing an SRAM having a plurality of memory cells is disclosed. In a first step, a bit value is written into a cell under test (CUT). Subsequently, the first and second enabling transistors are disabled and the bit lines are discharged to a low potential. Next, the word line (WL) coupled to the memory cell under test is activated for a predetermined period. During a first part of this period, one of the bit lines (BLB) is kept at the low potential to force the associated pull up transistor in the CUT into a conductive state, after which this bit line (BLB) is charged to a high potential. Upon completion of this period, the bit value of the first cell is determined. The method facilitates the detection of weak or faulty SRAM cells without requiring the inclusion of dedicated hardware for this purpose.

Abstract: A new test pattern which consists of performing “very small jumps” and “very big jumps” within the matrix. The “very small jumps” are controlled by the row decoder, and have the effect of sensitizing the resistive open defects which lead to slow-to-fall behavior in the word line. A “very small jump” means that the memory position of two consecutive accesses remains in a unique sub-cluster until all rows in that sub-cluster have been tested, remains in the same cluster until all rows in that cluster have been tested, remains in the same U section until all rows in that U section have been tested, and finally, remains in the same Z block until all of the rows of that Z block have been tested. The “very big jumps” are intended to cover the class of resistive open defects which leads to slow-to-rise behavior, and is intended to mean that two consecutive memory accesses must never stay in the same subcluster, at the same cluster, or at the same U section.

Abstract: A method testing an SRAM having a plurality of memory cells is disclosed. In a first step, a bit value is written into a cell under test (CUT). Subsequently, the first and second enabling transistors are disabled and the bit lines are discharged to a low potential. Next, the word line (WL) coupled to the memory cell under test is activated for a predetermined period. During a first part of this period, one of the bit lines (BLB) is kept at the low potential to force the associated pull up transistor in the CUT into a conductive state, after which this bit line (BLB) is charged to a high potential. Upon completion of this period, the bit value of the first cell is determined. The method facilitates the detection of weak or faulty SRAM cells without requiring the inclusion of dedicated hardware for this purpose.

Abstract: The present invention relates to a method for detecting delay faults in a semiconductor memory. In an example embodiment, address bits and data bits are generated according to a test pattern suitable for testing the semiconductor memory. The address bits and the data bits are validated and then provided to input ports of the semiconductor memory. Memory operation is then started such that a time interval between the provision of the address bits and the data bits and the start of the memory operation is approximately equal to an operating clock cycle of the semiconductor memory. Such timing ensures that both the address decoder and the read/write circuitry are stressed in time appropriately, enabling detection of small delay faults.

Abstract: A method and a test arrangement for testing an SRAM having a first cell and a second cell coupled between a pair of bitlines is disclosed. In a first step, a data value is stored in the first cell being the cell under test (CUT), and its complement is stored in a second cell, being the reference cell. Next, the bitlines are precharged to a predefined voltage. Subsequently, the wordline of the reference cell is enabled for a predefined time period, for instance by providing the wordline with a number of voltage pulses. This causes a drop in voltage of the bitline coupled to the logic ‘0’ node of the reference cell. In a subsequent step, the wordline of the CUT is enabled, which exposes the CUT to the bitline with the reduced voltage. This is equivalent to weakly overwriting the CUT. Finally, the data value in the CUT is evaluated. If the data value has flipped, the CUT is a weak cell. Cells with varying levels of weakness can be detected by varying the reduced voltage on the aforementioned bitline.

Abstract: Hard-open defects between logic gates of, for example, an address decoder and the voltage supply which result in logical and sequential delay behavior render a memory conditionally inoperative. A method and apparatus for testing integrated circuits for these types of faults is proposed, in which two cells of two logically adjacent rows or columns are written with complementary logic data. If a read operation reveals the data in the two cells to be identical, the presence and location of a hard-open defect is demonstrated.

Abstract: A method and a test arrangement for testing an SRAM having a first cell and a second cell coupled between a pair of bitlines is disclosed. In a first step (410), a data value is stored in the first cell being the cell under test (CUT), and its complement is stored in a second cell, being the reference cell. Next, the bitlines are precharged to a predefined voltage (step 420). Subsequently, the wordline of the reference cell is enabled for a predefined time period (step 430), for instance by providing the wordline with a number of voltage pulses. This causes a drop in voltage of the bitline coupled to the logic ‘0’ node of the reference cell. In a subsequent step (440), the wordline of the CUT is enabled, which exposes the CUT to the bitline with the reduced voltage. This is equivalent to weakly overwriting the CUT. Finally, the data value in the CUT is evaluated. If the data value has flipped, the CUT is a weak cell.

Abstract: The present invention relates to a test system (100) interposed between a clock monitor self-timed memory. In an example embodiment, the test system (100) receives an internal clock signal (104) from the clock monitor (152), an external clock signal (CL) and a control signal (CS). A multiplexer (110) of the test system provides in dependence upon the control signal (CS) the internal clock signal (104) to the internal memory block (125) during a normal mode of operation of the self-timed memory and the external clock signal (CL) to the internal memory block (125) during a test mode (108) of the self-timed memory. The test system (100) enables control of the clock cycle of the internal memory block (125) by directly applying the external clock signal (CL) during test mode. Thus, the internal memory block is stressed properly enabling the detection of small delay faults.

Abstract: Hard-open defects between logic gates of, for example, an address decoder and the voltage supply which result in logical and sequential delay behavior render a memory conditionally inoperative. A method and apparatus for testing integrated circuits for these types of faults is proposed, in which two cells of two logically adjacent rows or columns are written with complementary logic data. If a read operation reveals the data in the two cells to be identical, the presence and location of a hard-open defect is demonstrated.

Abstract: A new test pattern which consists of performing “very small jumps” and “very big jumps” within the matrix. The “very small jumps” are controlled by the row decoder, and have the effect of sensitizing the resistive open defects which lead to slow-to-fall behavior in the word line. A “very small jump” means that the memory position of two consecutive accesses remains in a unique sub-cluster until all rows in that sub-cluster have been tested, remains in the same cluster until all rows in that cluster have been tested, remains in the same U section until all rows in that U section have been tested, and finally, remains in the same Z block until all of the rows of that Z block have been tested. The “very big jumps” are intended to cover the class of resistive open defects which leads to slow-to-rise behavior, and is intended to mean that two consecutive memory accesses must never stay in the same subcluster, at the same cluster, or at the same U section.

Abstract: The present invention relates to a method for detecting delay faults in a semiconductor memory. In an example embodiment, address bits and data bits are generated according to a test pattern suitable for testing the semiconductor memory. The address bits and the data bits are validated and then provided to input ports of the semiconductor memory. Memory operation is then started such that a time interval between the provision of the address bits and the data bits and the start of the memory operation is approximately equal to an operating clock cycle of the semiconductor memory. Such timing ensures that both the address decoder and the read/write circuitry are stressed in time appropriately, enabling detection of small delay faults.