WO2010066207A1 - 在片自测试自修复方法 - Google Patents

在片自测试自修复方法 Download PDF

Info

Publication number
WO2010066207A1
WO2010066207A1 PCT/CN2010/000126 CN2010000126W WO2010066207A1 WO 2010066207 A1 WO2010066207 A1 WO 2010066207A1 CN 2010000126 W CN2010000126 W CN 2010000126W WO 2010066207 A1 WO2010066207 A1 WO 2010066207A1
Authority
WO
WIPO (PCT)
Prior art keywords
self
unit
test
repair
input
Prior art date
Application number
PCT/CN2010/000126
Other languages
English (en)
French (fr)
Other versions
WO2010066207A8 (zh
Inventor
林正浩
Original Assignee
上海芯豪微电子有限公司
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
Priority claimed from CN2008102046235A external-priority patent/CN101751317B/zh
Priority claimed from CN200810207673.9A external-priority patent/CN101763901B/zh
Application filed by 上海芯豪微电子有限公司 filed Critical 上海芯豪微电子有限公司
Publication of WO2010066207A1 publication Critical patent/WO2010066207A1/zh
Publication of WO2010066207A8 publication Critical patent/WO2010066207A8/zh

Links

Classifications

    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C7/00Arrangements for writing information into, or reading information out from, a digital store
    • G11C7/10Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
    • G11C7/1006Data managing, e.g. manipulating data before writing or reading out, data bus switches or control circuits therefor
    • 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
    • G01R31/317Testing of digital circuits
    • G01R31/3181Functional testing
    • G01R31/3185Reconfiguring for testing, e.g. LSSD, partitioning
    • G01R31/318533Reconfiguring for testing, e.g. LSSD, partitioning using scanning techniques, e.g. LSSD, Boundary Scan, JTAG
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C29/00Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
    • G11C29/04Detection or location of defective memory elements, e.g. cell constructio details, timing of test signals
    • G11C29/08Functional testing, e.g. testing during refresh, power-on self testing [POST] or distributed testing
    • G11C29/12Built-in arrangements for testing, e.g. built-in self testing [BIST] or interconnection details
    • G11C29/44Indication or identification of errors, e.g. for repair
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C29/00Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
    • G11C29/04Detection or location of defective memory elements, e.g. cell constructio details, timing of test signals
    • G11C29/08Functional testing, e.g. testing during refresh, power-on self testing [POST] or distributed testing
    • G11C29/12Built-in arrangements for testing, e.g. built-in self testing [BIST] or interconnection details
    • G11C29/44Indication or identification of errors, e.g. for repair
    • G11C29/4401Indication or identification of errors, e.g. for repair for self repair
    • 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
    • G01R31/2851Testing of integrated circuits [IC]
    • G01R31/2894Aspects of quality control [QC]

Definitions

  • the invention relates to the field of boxing circuit design, and in particular to the field of integrated circuit testing.
  • the density of the memory is large, causing the failure rate of the memory to be higher than other logics of the same area, and as the ratio of the memory to the entire chip is increasing, the yield of the memory is getting lower and lower, so that the yield of the entire chip is also With sr, once the memory fails, the entire chip fails.
  • redundant rows or redundant columns are usually added to the memory so that the damaged rows or columns can be replaced with redundant rows or columns before, after, and during use. Yield and chip life.
  • the memory is tested by using an off-chip automatic test equipment to test whether the memory has failed. If a fault occurs and the memory is repairable, use laser or high voltage (usually 11 ⁇ 20 volts) according to the recorded fault location. The corresponding fuse or anti-fuse is processed in the same way to achieve repair.
  • BIST built-in self-test
  • a memory design that combines built-in self-test (BIST) circuitry and built-in self-healing (BISR) modules has been tested to test and repair corrupted memory.
  • BIST built-in self-test
  • BISR built-in self-healing
  • the built-in self-repair unit can automatically determine and record the fault location of the memory according to a certain algorithm, but in the repair, it still relies on the use of laser or high voltage to process the corresponding fuse or anti-fuse to achieve repair. This is a more time consuming and complex process and means an increase in cost.
  • the self-test, self-repair unit is more complicated, or the fuse and anti-fuse are used in the repair, which increases the cost. If there is a way to implement the self-test and self-repair of the chip in a common process, the failed chip can be repaired and the yield can be improved. If the self-test can be performed during the work, the replacement unit can be replaced by the backup unit. Increase chip and system reliability and extend service life.
  • the invention is directed to the deficiencies of the prior art, and proposes a self-testing and self-repairing method in a slice.
  • the simple test algorithm can be used to directly implement the memory repair using a simple circuit, which can be realized only by using an ordinary process. There is no need to use fuse or anti-fuse to achieve repair, no dependence on external equipment.
  • the basic idea of the invention is to test the inside of the chip by using a self-test unit, and then pass the test result to the self-repairing unit, and self-repair the failed unit by using the backup unit.
  • the self-test and self-repair are usually performed after the chip is powered on and before the work, that is, self-test and self-repair are performed immediately after each power-on, to ensure the normal operation of the chip afterwards, without using an external test system or a repair system.
  • the test time can be significantly shortened, and the test result is not required to be solidified by the physical mark, so that the self-test self-repair structure is simple and effective.
  • test method can not only significantly improve the test efficiency in the on-chip self-test, even if it is applied in the off-chip test system, it can significantly improve the test efficiency, therefore, on the off-chip
  • test methods described is also within the scope of the appended claims.
  • the invention is a self-test self-repairing method in a chip, which is characterized in that it can perform self-test of the chip without relying on an external device in the case of power-on operation, and can perform self-repair according to the test result; the chip is powered on and After entering the self-test self-repair mode, the method can be used to perform self-test self-repair without relying on external devices.
  • the way to enter the self-test self-repair mode can be triggered by an external pin or by using an existing signal.
  • the existing signal includes but is not limited to a reset signal (reset); the on-chip self-test self-repair method includes two stages:
  • the self-test unit tests the unit under test and passes the test result to the self-repair unit and stores it in the memory.
  • the self-repair unit operates according to the self-test result, and performs self-repair on the error part of the unit under test.
  • the test result controls the corresponding multiplexer or bypass switch to replace the failed unit under test with a backup unit or device; the corresponding switch includes but is not limited to a multiplexer, a three-state gate;
  • the self-test unit retests the unit under test. If no error is detected, the self-test self-repair is completed, and a test completion signal is generated for external detection. Otherwise, an unrepairable signal is generated for external detection or generated by the test result.
  • the test result table is for external detection, and a test completion signal is generated at the same time, and the system performs dynamic allocation of the task of the tested unit according to the signal or test result table information, combined with the dynamic allocation algorithm.
  • the self-test self-repair according to the present invention is characterized in that the number of the units to be tested may be non-specific.
  • the unit under test is a set of a basic element, an operation unit, or a part or all of a processor core having a specific relationship
  • the number of basic elements, arithmetic units, or processor cores in the set may be unfixed, and in the set
  • the connection relationship between the basic components, the arithmetic unit, or the processor cores may be serial, parallel, or serially parallel, or may be bypassed, looped, or looped.
  • the memory for storing test results of the present invention may be any volatile or non-volatile memory including, but not limited to, random access memory (RAM), read only memory (ROM), memristor (MEMSIST0R).
  • RAM random access memory
  • ROM read only memory
  • MEMSIST0R memristor
  • the corresponding switches of the present invention include, but are not limited to, a multiplexer, a three-state gate.
  • the self-test described in the present invention can be applied to all structures and parts within a chip, including but not limited to logic cells, analog cells, memories, internal wiring, input and output pins.
  • the self-repair described in the present invention can be applied to all structures and parts of a chip having a backup unit, the structure This includes but is not limited to logic cells, analog cells, memory, internal wiring, input and output pins.
  • the self-test for the logic unit and the analog unit according to the present invention can compare whether the operation result of the test vector is a specific relationship or the comparison logic unit, the operation result and the expected result of the comparison unit by the same logic unit and the simulation unit. Whether it is implemented for a particular relationship, including but not limited to equal, opposite, same or different.
  • the self-healing can be implemented by bypassing the failed logic unit, the analog unit, or replacing the failed logic unit, the analog unit with the backup logic unit, the analog unit.
  • the self-test self-repairing device for the logic unit and the analog unit according to the present invention is composed of a backup logic unit, a backup simulation unit, a self-test unit and a self-repair unit.
  • the backup logic unit and the backup analog unit are used to replace the failed logical unit and the analog unit.
  • the self-test unit includes, but is not limited to, test vector generation logic and a comparator.
  • the test vector generation logic generates a test vector
  • the comparator is used to determine whether the operation result of the logic unit and the simulation unit executing the test vector is correct.
  • Self-healing units include, but are not limited to, self-healing controllers, which are used to control backup logic units, backup analog units to replace failed logic units, analog units or bypass-failed logic units, and analog units.
  • the logic unit and the analog unit may be of any size, including but not limited to a basic element, an operation unit or a processor core; or may be any part of a logic circuit or an analog circuit, including but not limited to a data channel, a control unit, Analog circuit.
  • the structure may omit the self-repairing unit and the backup logic unit and the analog unit, and only perform self-test. If the failed logic unit and the analog unit are detected, an unrepairable signal is generated for external detection.
  • the method and apparatus of the present invention are characterized in that the logic unit and the analog unit may be basic elements, arithmetic units or processor cores having a specific relationship to be tested in a multi-operation unit/multi-core/many-core system, It may be a collection of partial or full components of a basic component, an arithmetic unit, or a processor core that have a particular relationship to be tested in a multi-operating unit/multi-core/many-core system.
  • the self-test unit may include a comparator for comparing whether the result obtained by the test unit from the test vector is a specific relationship, and determining the correctness of the unit under test according to the comparison result.
  • the particular relationship may be equal, opposite, reciprocal, complementary.
  • the self-test unit may further comprise part or all of the following means: vector generation logic for generating, storing or generating and storing a self-test vector, which may be generated by a vector
  • vector generation logic for generating, storing or generating and storing a self-test vector, which may be generated by a vector
  • the logic and vector memory are composed of two parts, or only vector generation logic or vector memory;
  • Test vector allocation logic including test vector allocation controller and physical connection, for configuring, establishing, or dynamically adjusting the input unit and vector generator of the unit under test in the tested multi-unit/multi-core/many-core system during the self-test Even Connection relationship
  • the operation result distribution logic includes an operation result distribution controller and a physical connection for configuring, establishing, or dynamically adjusting the output unit of the unit under test and the comparator in the tested multi-operation unit/multi-core/many-core system during the self-test process Connection relationship.
  • the vector generation logic of the present invention may be composed of two parts: vector generation logic and vector memory, or only vector generation logic or vector memory.
  • the vector memory in the vector generation logic of the present invention may use a storage medium which may be volatile or non-volatile, may be changed once in a write, or may be erased multiple times. Written.
  • the number of vector generation logics in the present invention may be one or more.
  • each part may be one or more.
  • each vector generation logic may rely on a synchronization mechanism to ensure that test vectors are synchronously sent to each unit under test; or test vectors may be asynchronously sent to each unit under test. However, relying on the synchronization mechanism to ensure the synchronization of the results of the comparison.
  • the vector generation logic of the present invention may be on-chip, on-chip, or off-chip.
  • the vector generation logic when it is composed of two parts: vector generation logic and vector memory, it may be part of the on-chip core, part of the on-chip core or part of the on-chip core, part of the off-chip, or part of the on-chip core, part of On the off-chip.
  • the test vector generated by the vector generation logic of the present invention covers the functions or instruction sets supported by the unit under test.
  • the vector generation logic generated by the present invention generates test vectors which may be generated according to certain rules or randomly generated. When the number of vector generation logic is plural, the test vectors generated or stored by different vector generation logic may have a specific relationship including, but not limited to, the same function and opposite functions.
  • the comparator of the present invention may be a comparator in the usual sense, or may be a comparator by a unit under test having a comparison function.
  • the comparator of the present invention may be a comparator with analysis decision logic whose analysis decision logic is characterized in that it can determine whether the unit under test participating in the comparison has failed.
  • the analysis and judgment logic can be implemented in hardware on-chip, or can be implemented in-chip or on-chip.
  • the operation result of each unit to be tested can be compared with the operation results of one, two or even more units to be tested.
  • the comparison can be performed simultaneously or at different times Row.
  • the comparator according to the present invention can compare all the calculation results when comparing the calculation results, or can compare only the partial feature calculation results or the final calculation results.
  • the test vector allocator of the present invention comprises a test vector distribution controller and a physical connection, and the test vector distribution controller configures the physical connection to form a connection relationship; wherein the test vector distribution controller can be implemented in hardware on the chip or in the slice.
  • the implementation of the software implementation can also be implemented off-chip.
  • test vector distribution controller of the present invention can be omitted.
  • the connection relationship between the measured unit and the vector generator is in a fixed manner, and the fixed manner may be an existing connection relationship of the multi-operation unit/multi-core/nuclear system, or may be specifically designed for Test the connection of the design.
  • connection between the vector generator and the unit under test according to the present invention may be configured by the test vector allocator into any inter-connected relationship, including but not limited to a linear type, a network structure type.
  • the operation result distributor of the present invention includes a calculation result distribution controller and a physical connection, and the operation result distribution controller configures the physical connection to form a connection relationship.
  • the operation result distribution controller can be implemented in hardware on-chip, or can be implemented on-chip or on-chip.
  • the operation result distribution controller of the present invention can be omitted.
  • the connection relationship between the measured unit and the comparator is in a fixed manner, and the fixed manner may be a connection relationship existing in the multi-operation unit/multi-core/nuclear system, or may be specifically designed for Test the connection of the design.
  • connection relationship between the comparator and the unit under test according to the present invention can form an arbitrary interconnection relationship according to the operation result distributor, including but not limited to a linear type and a network structure type.
  • the test result table of the present invention is a data information table for storing information on whether or not the associated measured unit is invalid according to the result of the test comparison.
  • the storage medium of the test result table of the present invention may be volatile or non-volatile, and may be one-time write unchanged, or erasable multiple-write.
  • the self-testing method for the memory according to the present invention can be tested in accordance with a conventional memory testing method.
  • the self-test self-repair device for the memory according to the present invention is composed of a backup storage unit, a self-test unit, and a self-repair unit.
  • the backup storage unit is configured to replace the damaged unit during self-repair;
  • the self-test unit is configured to test whether the column in the storage unit sub-array is invalid, and pass the test result to the self-repairing unit.
  • the self-repairing unit includes state storage processing Unit, input splitter and output selector.
  • the state storage processing unit is used to store test results passed from the test unit.
  • the error state can be latched, and the address of the error storage unit can be latched at the same time.
  • the input distributor and the output selector can be controlled, and the damaged storage unit can be replaced with the backup storage unit to complete the repair of the memory.
  • the device When the device includes only a memory array and a self-test unit, the device performs only self-tests, and if a failed memory unit is detected, an unrepairable signal is generated for external detection.
  • the self-test for internal wiring described in the present invention can be implemented by giving an excitation at the input, comparing whether the output and the input conform to a particular relationship, including but not limited to equality, and vice versa. Self-healing can be implemented by replacing the failed connection with a backup connection.
  • the self-test self-repairing device for internal wiring is composed of a backup connection, a self-test unit and a self-repair unit.
  • the backup connection is an equivalent connection that can replace the tested connection.
  • the self-test unit is used to load the line excitation, obtain the output, and compare whether the output and the input conform to a certain relationship.
  • the self-healing unit is used to bypass the input of the failed connection to the backup connection and connect the output of the backup connection to the output of the failed connection.
  • the device When the device contains only internal wiring and self-test units, the device only performs self-tests. After the test of the connection is completed by the self-test unit, a signal indicating whether the connection is invalid is issued to the system, and self-repair is not implemented.
  • the self-test for the input and output pins of the present invention can be implemented by giving an excitation at the input, comparing whether the output and the input conform to a certain relationship, including but not limited to equal and opposite.
  • Self-healing can be implemented by replacing the failed input and output pins with the backup I/O pins.
  • the self-test self-repairing device for input and output pins comprises a backup input/output pin, a self-test unit, and a self-repairing unit.
  • the backup input/output pin is an input/output pin equivalent to the input/output pin to be tested, and may be a one-way input pin, a one-way output pin, or a bidirectional input/output tube. foot.
  • the self-test unit is used to load an excitation on the input and output pins, obtain an output, and compare whether the output and the input conform to a certain relationship.
  • the self-repairing unit is configured to bypass the input of the failed input and output pin to the backup input/output pin, and connect the output of the backup input and output pin to the output of the failed input/output pin.
  • the device When the device only contains input and output pins and a self-test unit, the device performs only self-testing. After the test circuit completes the test of the input and output pins, it signals the system whether the connection is invalid, and does not perform self-repair.
  • the self-test self-repair according to the present invention can be performed during wafer testing, or during integrated circuit testing after chip packaging or when the system including the chip is started; or self-testing conditions and cycles can be manually set at work. Regular period Conduct self-test and self-repair.
  • the test result table described in the present invention may be on-chip or off-chip.
  • the test result table of the present invention does not exist, the test result is sent out of the chip in the form of other information including, but not limited to, an information form indicating only whether there is a failed cell in the chip under test.
  • the unit state table of the present invention is an information table for storing state information of the associated measured unit generated on the basis of the test result table; the unit state table may be stored on a storage medium implemented by hardware in the chip, It can be transferred to off-chip storage after it is generated on-chip.
  • the dynamic allocation algorithm of the present invention is an algorithm for dynamically allocating a task of a measured unit according to test result table information.
  • the dynamic allocation algorithm may be implemented in hardware on-chip, or may be implemented in on-chip software, or Implemented off-chip.
  • the software includes, but is not limited to, an operating system.
  • the unit under test or the connection of the present invention has a bypass function, that is, the failed basic component or the arithmetic unit or the processor core input port is directly connected to the output port or is turned on by the line bypass.
  • the self-test according to the present invention can update the test result table in real time, and bypass the failed basic component or the arithmetic unit or the processor core according to the test result table information, for the basic component or the arithmetic unit or the processor core of the same structure, It can be a backup unit for each other, thus realizing the self-repair function of multiple arithmetic unit/multi-core/nuclear-core system.
  • the present invention uses a simple method and circuit, and can perform self-test and self-repair on the chip without relying on external equipment or any special process or device including laser, high voltage, and the like, fuse, anti-fuse, and the like.
  • the present invention tests the inside of the chip by using a self-test unit, and records the result in any volatile or non-volatile memory for controlling the corresponding semiconductor switch during self-repair. If the memory is volatile, the chip must be self-tested and self-repaired each time the power is turned on. If the memory is non-volatile, it can be self-tested and self-repaired. Test conditions and cycles, self-test and self-repair are performed periodically during work.
  • the invention can effectively shorten the test time. If an error occurs, the damaged unit is automatically replaced with the backup unit to implement self-repair in the slice.
  • the self-repair method of the structure is very simple, and can be realized only by a simple structure, and does not require complicated replacement algorithms and implementation methods; the structure can realize self-test and self-repair in the system startup, and does not require external force. Intervention.
  • the invention realizes on-chip self-test self-repair of the chip at a low cost, and improves the yield and reliability. It also effectively reduces the test cost of the chip during production.
  • FIG. 1 is a structural diagram of a chip system implemented in a self-test self-test.
  • Figure 2 is a general flow chart of the self-test self-test in the slice.
  • FIG. 3( a ) is a first embodiment of the self-test self-repair for the logic unit proposed by the present invention.
  • FIG. 3( b ) is a second embodiment of the self-test self-repair for the logic unit proposed by the present invention.
  • FIG. 4 is a flow chart of self-test self-repair for a logic unit according to the present invention.
  • FIG. 5 is an embodiment of the present invention for self-testing of memory self-test.
  • Figure 6 is a block diagram of a state storage processing unit (1 bit) of a multi-column in a repairable subarray.
  • Figure 7 (a) is the stored value of the state storage processing unit of Figure 6 before the self test.
  • Figure 7 (b) is the value stored in the state storage processing unit of Figure 6 after self-test.
  • Figure 8 (a) is a diagram of the structure of the input splitter in a repairable storage sub-array.
  • Figure 8 (b) is a block diagram of the output selector in a repairable memory subarray.
  • Figure 9 is an embodiment of the self-test self-repair for internal wiring of the present invention.
  • Figure 10 is an embodiment of the present invention in terms of input and output pins.
  • FIG 11 (a) shows an embodiment of the invention in terms of output pins.
  • Figure ⁇ (b) is an embodiment of the invention in terms of input pins.
  • Figure 12 is a diagram showing an example of the contents of the test result table in the present invention.
  • FIG. 13 is a diagram showing an embodiment of a processor core status table in the present invention. detailed description
  • the technical idea of the invention is to realize detection and self-repair of the unit under test by the self-test unit and the self-repair unit.
  • the self-test unit is used to test whether the unit under test fails, and the test result is transmitted to the self-repair unit.
  • the self-repair unit replaces the failed unit with the backup unit column to complete the self-repair. No external interventions or complex algorithms are required for repair.
  • FIG. 1 is a structural diagram of a chip system that implements self-testing in a slice self-test. The system consists of three parts: the unit under test (101), the self-test unit (102) and the self-repair unit (105).
  • the unit under test (101) includes but is not limited to a logic unit, a memory, an internal connection, and an input/output pin.
  • the self-test unit (102) gives the unit under test (101) an excitation (103), and the output from the unit under test returns to the self-test unit (102) for comparison with the expected result (104).
  • the result of the test is sent to the self-repairing unit (105). If there is an error in the unit under test, the self-repairing unit (105) replaces the error part of the unit under test with the backup unit according to the corresponding test result to achieve self-repair. If there is no backup unit in the wrong part of the unit under test, an unrepairable signal is generated for external detection.
  • FIG 2 is a general flow chart of the self-test self-test in the slice.
  • step one Perform self-test, and then go to step 2 (202) to determine whether there is a failed unit; if the test result is correct, the chip can work normally; if the test unit fails, go to step 3 (203) to determine whether the unit has Backup unit; If there is no backup unit, the chip cannot be repaired. If there is a backup unit, go to step 4 (204) Self-repair phase to repair the failed unit.
  • step 5 After the repair is completed, go to step 5 (205) to retest the repaired unit, and then go to step 6 (206) to determine whether there is a failed unit; if the test result is correct, the chip can work normally; if there is still a failed unit tested, It means that the chip cannot be repaired.
  • FIG. 3 and FIG. 4 are diagrams illustrating an embodiment of self-test self-repair for a logic unit according to the present invention.
  • the following embodiments are implemented in accordance with the technical solution of the present invention, but the scope of protection of the present invention is not limited to the embodiment.
  • Figure 3 is the structure diagram of the system under test in the self-test self-repair mode, and does not indicate the specific data input and output flow direction of the system in the working mode.
  • FIG. 3(a) is a first embodiment of the self-test self-repair for the logic unit proposed by the present invention.
  • a corresponding backup unit such as backup unit (302) as a backup of the unit under test (303), backup unit (306) as a backup of the unit under test (305), backup unit ( 310)
  • the backup unit (314) serves as a backup of the unit under test (313).
  • the vector generator (301) first generates a test vector and sends it to each of the tested logic cells.
  • Each of the tested logic cells executes a test vector, and the comparator compares the operation results of the tested logic cells, such as a comparator ( 304) for the measured logical unit (303) and the tested logical unit
  • the operation result of (305) is compared, and the comparator (307) compares the operation result of the measured logic unit (303) with the measured logic unit (311), and the comparator (309) pairs the logic unit to be tested (305).
  • the comparator (312) compares the operation results of the tested logic unit (311) and the measured logic unit (313), and the comparators compare the results.
  • the self-repair controller (308) judges the failed logic unit according to the comparison result of each comparator, and controls the backup unit to replace the failed logic unit, thereby realizing self-test of each measured logic unit. repair.
  • the test vector generated by the vector generator (301) covers the functions or instruction sets supported by the unit under test, and is as long as possible with multiple loops of test vectors with complicated running procedures and short program codes.
  • the operation results of each step may be compared in real time, or only the partial operation results selected in the operation process may be compared, or only the final test vector may be finalized. The results of the calculation are compared.
  • FIG. 3(b) is a second embodiment of the self-test self-repair for the logic unit proposed by the present invention.
  • the operation result of the test unit performing the test vector is compared with the expected result of the test vector, and the expected result may be the operation result of each step operation in the execution of the test vector, or may be an operation
  • the intermediate result selected in the process can also be the final result of the entire test vector. Therefore, during the comparison process, it is necessary to decide when to compare and how many comparisons to make based on the expected result.
  • the failure test unit is determined according to the comparison result, and the failed test unit is replaced by the backup unit for repair, thereby realizing self-test self-repair in the slice.
  • FIG. 4 is a flowchart of self-test self-repair for a logic unit according to the present invention.
  • the vector generator first generates a test vector (401), and the test vector is sent to each of the tested logic units, and each tested logic unit runs a test vector (402), and sends the operation result.
  • each comparator compares the operation results, and determines the failed logical unit (403) according to the comparison result, and the self-repair logic repairs the failed logical unit (404) by bypass or alternative method, after the repair is completed.
  • the logic unit will be retested (405). If the logic unit is tested correctly, the logic unit is available, otherwise an unrepairable signal is generated, thereby completing the self-test self-repair process of the on-chip logic unit.
  • FIG. 5, FIG. 6, FIG. 7, and FIG. 8 are diagrams of an embodiment of the present invention for self-testing of a memory self-test.
  • the following embodiments are implemented in accordance with the technical solution of the present invention, but the scope of protection of the present invention is not limited to the embodiment.
  • FIG. 5 is an embodiment of the present invention for self-testing of memory self-test.
  • This embodiment is a memory with column backup, which includes a storage array (501), a self-test unit (502) and a self-repairing unit (503); wherein the storage capacity is 1024 x 32 bits, and each bit contains 4 columns of storage units, ie, columns.
  • the address is two digits; it contains two sets of storage Sub-array (504), corresponding backup column (505) and decoder (506); self-test unit (502) mainly includes data generator (507), address generator (508) and comparator (509) three parts
  • the self-repairing unit (503) includes an input distributor (510), an output distributor (511), and a state storage processing unit (512); each storage sub-array (504) corresponds to a set of input distributors (510) and output selections; (511).
  • the self test unit (502) tests the storage array (501), generates a test result of the damaged column in the storage array (501), and passes the test result to the self-repair unit (503); the state storage processing unit (512) according to the test
  • the result is a certain logical process, and the result is controlled by the input control distributor (510) and the output selector (511).
  • the state storage processing unit (512) may have a different structure depending on the needs of the repair capability.
  • FIG. 6 is a structural diagram of a state storage processing unit (1 bit) of a plurality of columns in a repairable storage sub-array, which includes a storage unit (601) and a result processing unit (602).
  • the unit receives the results from the self test unit.
  • the test result and the value in the memory location (601) are initialized to 0; when the test result jumps from 0 to 1, that is, when the corresponding column fails, the column state of the failed state and the failed column can be stored in the storage unit.
  • (602) Medium.
  • the resulting processing unit (603) will generate a replacement signal.
  • each backup column in the memory using the structure can repair 4 columns whose column addresses are different from each other in the corresponding storage sub-array, that is, the memory of FIG. 5 using the structure can realize the repair of up to .8 columns of invalid columns. .
  • FIG. 7(a) is the stored value of the state storage processing unit in FIG. 6 before the self-test
  • FIG. 7(b) is the value stored in the state storage processing unit after the self-test in FIG.
  • the first row in Figure 7 (a) stores the flag for the corresponding bit in the memory array.
  • the second row is the address high of the failed column in the fail bit
  • the third row is the low column address of the failed column in the fail bit.
  • bit 4 bit 11 is the fail bit, where the memory cell with column address 10 in bit 4 and the memory cell with column address 01 in bit 11 are invalid and will be replaced by the backup column. The remaining columns are ready for use.
  • FIG. 8(a) is a structural diagram of an input distributor in a repairable storage sub-array; the structure is composed of a tri-state gate (801), and may also be composed of a multiplexer; wherein the data input is from outside the memory. Or self-test unit; determine which input is bypassed into the backup column based on the replacement information output by the state storage processing unit. This structure does not turn off the input of the failed column.
  • 8(b) is a structural diagram of an output selector in a repairable memory sub-array, which is composed of a tri-state gate (801) or a multiplexer; and the replacement information output by the processing unit is stored according to the state. Decide whether to use The output of the column replaces the output of a memory in the storage sub-array to form a memory.
  • FIG. 9 is an embodiment of the self-test self-repair of the internal connection of the present invention.
  • the internal connection in this embodiment is exemplified by the system bus, and is implemented according to the technical solution of the present invention, but the protection scope of the present invention is not limited to the system bus.
  • the self-test self-healing structure in Figure 9 includes a bus (901), a backup bus (902), a self-test unit (903), and a self-repair unit (904).
  • the self test unit (903) is mainly composed of a data generator (905) and a comparator (906).
  • the comparator (906) compares the result produced by the data generator (905) with the data read at the end of the bus, and generates a test result to be sent to the state processing unit (907) in the self-repairing unit (904), and the module utilizes the self.
  • the result of the test produces a repair signal. If a bus error is detected, the repair signal turns off the bus and automatically replaces it with the backup bus (902).
  • FIGS 10 and 11 illustrate an embodiment of the present invention for self-test of input and output pins.
  • the self-repair method for the input and output pins is similar to the bus self-repair, that is, replacing the failed input and output pins with the backup input and output pins.
  • the following embodiments are implemented in accordance with the technical solution of the present invention, but the scope of protection of the present invention is not limited to the embodiment.
  • FIG. 10 is an embodiment of the present invention in terms of input and output pins.
  • the input and output pins are mainly composed of an output gate (1001), an output control (1002), an input gate (1003), and a selector (1008), and are an input/output multiplexed pin.
  • the output gate (1001) is opened by the output control (1002), and the selector (1008) selects the output (1004), and is sent to the pin through the output gate (1001);
  • the output gate (1001) is turned off by the output control (1002), and the signal on the pin is sent to the input (1005) through the input gate (1003).
  • the output gate (1001) is opened by the output control (1002), the selector (1008) selects the excitation (1006), is sent to the pin through the output gate (1001), and is also sent to the input gate (1003).
  • the self-test unit obtains the signal through the input gate (1003) and compares it (1007), the self-test can be performed.
  • FIG. 11 is an embodiment of the present invention in terms of an output pin.
  • the output pin is mainly composed of an output gate (1101), an output control (1102), and a selector (1105), and is a one-way output pin.
  • the output gate (1101) is opened by the output control (1102), and the selector (1105) selects the output (1103), which is sent to the pin via the output gate (1101).
  • the output gate (1101) is opened by the output control (1102), the selector (1105) selects the excitation (1104), is sent to the pin through the output gate (1101), and the output gate is obtained from the test unit (1101). After the signal is compared (1106), the self-test can be performed.
  • Figure 11 (b) shows an embodiment of the invention in terms of input pins.
  • the input pin is mainly composed of an input gate (1108) and a selector (1105), and is a one-way input pin.
  • the signal on the pin is sent to the input (1107) via the input gate (1108).
  • the selector (1105) selects the excitation (1104), and sends it to the inside of the chip through the input gate (1108). After the unit obtains the signal through the input gate (1108) and compares it (1106), it can perform self-test.
  • FIG. 12 and FIG. 13 are embodiments of the contents of the test result table and the contents of the processor core state table in the present invention. Embodiments are implemented according to the technical solution of the present invention, but the protection content of the present invention is not limited to the two embodiments.
  • each label (1202) corresponds to a unit under test in the system, and the information at the position indicates the state of the unit under test, where "?" indicates that the corresponding unit under test is not tested," X" indicates that the corresponding unit under test has failed, and "0" indicates that the corresponding unit under test is normal.
  • the test result table can be on-chip or off-chip.
  • the storage medium may be volatile or non-volatile; it may be one-time writes that are no longer changed, or may be erasable and write-once. During system testing and operation, the failed unit under test is bypassed to ensure that the system can operate normally. Not only does it improve yield, but it also implements the system's self-healing capabilities.
  • FIG. 13 is an embodiment of a processor core status table in the present invention.
  • the processor core state is divided into untested, invalid, available, idle, and available and occupied, where "?" indicates that the corresponding unit under test is not measured, and "X" indicates that the corresponding unit under test is invalid. "0” indicates that the corresponding unit under test is available and idle, and "1” indicates that the corresponding unit under test is available and is being occupied.
  • the processor core state table (1301) is generated on the basis of the test result table, which may be implemented in hardware on-chip or in software by off-chip.
  • the system dynamically allocates each processor core task according to the state table through a dynamic allocation algorithm. For a failed processor core, the system bypasses it.
  • the system For a processor core that is available and occupied, the system records which process it is being processed. The basic information of the occupation and occupation process, and its tasks are not allocated during the occupation. For the available and idle processor cores, the system dynamically allocates tasks according to its own needs, thereby making full use of the system resources.

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Design And Manufacture Of Integrated Circuits (AREA)
  • Semiconductor Integrated Circuits (AREA)
  • Techniques For Improving Reliability Of Storages (AREA)
  • Tests Of Electronic Circuits (AREA)

Description

在片自测试自修复方法
技术领域
本发明涉及拳成电路设计领域, 尤其涉及集成电路测试领域。
背景技术
随着技术的进步,在集成电路设计领域,多核结构、在 S0C系统中集成更多的逻辑单元、 IP核、 存储器已经成为发展趋势, 随着对功能的需求和对性能的要求不断提升, 芯片中高 速高带宽系统总线的应用也越来越多, 32位、 64位的系统总线位宽已很常见, 更宽的 128 位片内总线也开始有较普遍的应用。系统集成度提高导致的另一个问题是芯片对外管脚数目 的增加。
在逻辑单元测试方面,由于被测单元较多,如果按照串行的顺序,依次测试各被测单元, 需要很长的测试时间, 增加了测试成本; 如果以并行的方式测试各被测单元, 则需要有大量 的测试引脚,增加了制造成本。此外,对于算术、逻辑等功能的测试与对存储器的测试不同, 与输出结果相对应的参考值很难通过简单逻辑得到,往往需要大量的高速存储器来保存参考 值, 也增加了芯片的成本。 同时, 逻辑单元数目的增加直接导致芯片良率的下降。如何在逻 辑单元自测试时用少量的存储空间保存参考值,并实现逻辑单元自修复,是一个需要解决的 问题。
随着逻辑单元、模拟单元数量的增多,一个或部分单元失效将引起整个芯片失效, 由此 导致生产出的芯片良率明显降低。在某些特殊的应用领域和场合, 如军事、航天等领域, 恶 劣的环境很容易导致在使用时, 芯片内一个或部分单元失效, 从而导致整个系统崩溃, 给用 户带来严重的损失或灾难性的后果。如果芯片中对良率影响较大的单元存在具有相同功能的 备份单元, 就可以用以修复失效的芯片。
存储器的密度较大,导致存储器的故障率要比同样面积的其他逻辑电 高,且随着存储 器占整个芯片的比例日益增大, 存储器的良率越来越低, 使得整个芯片的良率也随 . sr 一旦存储器发生故障, 整个芯片即失效。
为应对该问题, 通常都在存储器中加入冗余行或冗余列, 使其在封装前, 封装后, 以及 在使用时, 可以使用冗余行或列替换损坏的行或列, 提高存储器的良率和芯片的使用寿命。 通常存储器的检测都是使用片外的自动测试设备测试存储器是否发生故障,若发生故障 并检测到该存储器可修复, 则根据记录下来的故障位置使用激光或高电压 (一般为 11~20 伏) 等方式对相应的熔丝或反熔丝进行处理, 实现修复。
采用片外的自动测试设备对存储器进行测试需要花费较长的时间,为縮短测试时间,通 常在存储器中加入内建自测试(BIST)电路。自测试电路将自己生成的数据写入到存储器中, 再从存储器读出数据与期望值进行比较,并把测试结果传递给外部系统。外部系统通过传递 过来的数据, 确定存储器是否出现故障, 以及出现故障的存储单元的准确位置, 然后使用激 光或高电压等方式对相应的熔丝或反熔丝进行处理, 实现修复。
为了在提高存储器和芯片良率的基础上降低测试和修复成本, 己经有将内建自测试 (BIST)电路和内建自修复(BISR)模块结合起来的存储器设计,测试和修复损坏的存储器。 内建自修复单元可以根据一定的算法来自动判定并记录存储器的故障位置,但在修复时,仍 然依赖于使用激光或高电压等方式对相应的熔丝或反熔丝进行处理,实现修复。这是一个较 为耗时的复杂处理过程, 而且意味着成本的增加。
现有的自测试自修复方法或是采用复杂度较高的算法, 自测试、 自修复单元较为复杂, 或是在修复时采用熔丝、 反熔丝的方式, 增加了成本。如果能有一种方法普通工艺下, 简便 地实现芯片在片自测试自修复, 就可以修复失效的芯片, 提高良率; 如果还能在工作期间进 行自测试, 用备份单元替代失效单元, 就可以增加芯片和系统可靠性, 延长使用寿命。
发明内容
本发明针对现有技术的不足,提出了一种在片的自测试和自修复的方法,采用较为简单 的测试算法, 可以直接使用简单的电路实现存储器的修复, 仅使用普通工艺即可实现, 不需 要使用熔丝或反熔丝方式来实现修复, 完全不依赖外部设备。
本发明采用以下技术方案实现:
本发明的基本思路是采用自测试单元对芯片内部进行测试,然后把测试结果传递给自修 复单元,利用备份的单元对失效单元进行自修复。所述的自测试、自修复通常在芯片加电后、 工作前进行, 即每次加电后立即进行自测试和自修复, 保证之后芯片的正常工作, 不需使用 外接测试系统或修复系统, 在生产后测试 (production test ) 时, 能显著縮短测试时间, 同时不必用物理记号固化测试结果, 使该在片自测试自修复结构简单、有效。虽然本发明为 在片自测试自修复方法, 但所述的测试方法, 不仅在片内自测试中能显著提高测试效率, 即 使应用在片外的测试系统中, 也能明显提高测试效率, 因此, 在片外使用所述的测试方法也 应属于本发明所附权利要求的保护范围。
本发明为一种在片自测试自修复方法,其特征在于能够在加电工作的情况下不依赖于外 部设备进行芯片的自测试,并能根据测试结果进行自修复;芯片在加电工作并进入自测试自 修复模式后即可采用所述方法进行不依赖于外部设备的自测试自修复,进入自测试自修复模 式的途径可以是由外部引脚触发,也可以使用已有信号触发,所述已有信号包括但不限于复 位信号 (reset ); 所述在片自测试自修复方法包括两个阶段:
第一阶段,自测试单元测试被测单元并把测试结果传递给自修复单元,储存在存储器中; 自修复单元根据自测试结果进行操作,对被测单元的出错部分执行自修复, 由存储的测试结 果控制相应多路选择器或旁路开关, 以备份单元或器件取代失效的被测单元;所述相应开关 包括但不限于多路选择器、 三态门;
第二阶段, 自测试单元重测被测单元, 若未检测出错误, 则完成在片自测试自修复, 产 生测试完成信号,供外部检测,否则产生不可修复信号供外部检测或由测试结果生成测试结 果表供外部检测, 同时产生测试完成信号, 系统根据所述信号或测试结果表信息, 结合动态 分配算法进行被测单元任务的动态分配。
本发明所述的自测试自修复,其特征在于所述的被测单元的数目可以是非特定的。当被 测单元是具有特定关系的基本元件、运算单元或处理器核中的部分或全部组成的集合时,集 合中基本元件、运算单元或处理器核的数目可以是不固定的, 且集合中基本元件、运算单元 或处理器核间的连接关系可以是串行的、并行的或串行并行混合的, 也可以是带有旁路、循 环或回路的。
本发明所述用于存储测试结果的存储器可以是任意挥发性或不挥发性的存储器,包括但 不限于随机存储器 (RAM)、 只读存储器 (R0M)、 忆阻器 (MEMSIST0R)。 本发明所述相应开关包括但不限于多路选择器、 三态门。
本发明所述的自测试,可以应用于芯片内的所有结构和部分,所述结构包括但不限于逻 辑单元, 模拟单元, 存储器, 内部连线, 输入输出管脚。 本发明所述的自修复,可以应用于芯片内的具有备份单元的所有结构和部分,所述结构 包括但不限于逻辑单元, 模拟单元, 存储器, 内部连线, 输入输出管脚。
本发明所述的对于逻辑单元、模拟单元的自测试, 可通过相同的逻辑单元、模拟单元间 比较测试向量的运算结果是否为某种特定关系或比较逻辑单元、模拟单元的运算结果与预期 结果是否为某种特定关系来实施, 所述特定关系包括但不限于相等、 相反、 同或、异或。其 自修复可以通过旁路失效的逻辑单元、模拟单元或用备份的逻辑单元、模拟单元替换失效的 逻辑单元、 模拟单元来实施。 本发明所述的对于逻辑单元、模拟单元的自测试自修复装置由备份逻辑单元、备份模拟 单元、 自测试单元和自修复单元构成。其中, 备份逻辑单元、备份模拟单元用于替换失效的 逻辑单元、模拟单元。 自测试单元包括但不限于测试向量生成逻辑和比较器, 测试向量生成 逻辑生成测试向量,比较器用于判断逻辑单元、模拟单元执行测试向量的运算结果是否正确。 自修复单元包括但不限于自修复控制器, 自修复控制器用于控制备份逻辑单元、备份模拟单 元替换失效的逻辑单元、模拟单元或旁路失效的逻辑单元、模拟单元。所述逻辑单元、模拟 单元可以是任意大小的,包括但不限于基本元件、运算单元或处理器核;也可以是逻辑电路、 模拟电路中的任意部分, 包括但不限于数据通道、控制单元、模拟电路。所述结构可以省略 自修复单元和备份逻辑单元、 模拟单元, 只进行自测试, 若检测出失效的逻辑单元、模拟单 元, 则产生不可修复信号, 供外部检测。
本发明所述的方法和装置, 其特征在于所述逻辑单元、 模拟单元可以是多运算单元 /多 核 /众核系统中需要被测试的具有特定关系的基本元件、 运算单元或处理器核, 也可以是多 运算单元 /多核 /众核系统中需要被测试的具有特定关系的基本元件、运算单元或处理器核中 的部分或全部组成的集合。所述自测试单元可以包括比较器,用于比较被测单元执行自测试 向量得到的结果是否为特定关系, 并根据比较结果来确定被测单元的正确性。所述特定关系 可以是相等、 相反、 互逆、 互补。
本发明所述的方法和装置,其特征在于所述自测试单元还可以包含下列装置中的部分或 全部: 向量生成逻辑, 用于生成、存储或生成并存储自测试向量, 可以是由向量产生逻辑和向 量存储器两部分组成, 也可以是只有向量产生逻辑或向量存储器;
测试向量分配逻辑, 包括测试向量分配控制器和物理连接, 用于配置、建立或在自测试 过程中动态调整被测试多运算单元 /多核 /众核系统中被测单元输入端口与向量生成器的连 接关系;
运算结果分发逻辑, 包括运算结果分发控制器和物理连接, 用于配置、建立或在自测试 过程中动态调整被测试多运算单元 /多核 /众核系统中被测单元输出端口与比较器之间的连 接关系。
本发明所述的向量产生逻辑可以是由向量产生逻辑和向量存储器两部分组成,也可以是 只有向量产生逻辑或向量存储器。
本发明所述的向量产生逻辑中的向量存储器,其所用的存储媒介可以是挥发性的,也可 以是非挥发性的, 可以是一次写入不再更改的, 也可以是可擦除可多次写入的。
本发明所述的向量产生逻辑数量可以是一个或者多个,当向量生成逻辑是由向量产生逻 辑和向量存储器两部分组成时,,其每一部分均可以是一个或者多个。
本发明所述的向量产生逻辑数量为多个时,各个向量产生逻辑之间可以依靠同步机制来 保证将测试向量同步送到各个被测单元;也可以将测试向量非同步送到各个被测单元,但依 靠同步机制来保证同步进行运算结果的比较。
本发明所述的向量产生逻辑可以在片上核内、片上核外或者片外。当向量生成逻辑是由 向量产生逻辑和向量存储器两部分组成时,可以是部分在片上核内,部分在片上核外或者部 分在片上核外, 部分在片外, 或者部分在片上核内, 部分在片外。
本发明所述的向量产生逻辑产生的测试向量覆盖被测单元所支持的功能或者指令集。 本发明所述的向量产生逻辑产生测试向量可以是根据一定的规则产生,也可以是随机产 生。当所述向量产生逻辑的数目为多个时,不同向量产生逻辑产生或存储的测试向量可以是 有特定关系的, 所述特定关系包括但不限于功能相同、 功能相逆。
本发明所述的比较器可以是通常意义上的比较器,也可以是由具有比较功能的被测单元 充当比较器。
本发明所述的比较器可以是带有分析判断逻辑的比较器,其分析判断逻辑的特征在于其 可以裁定参与比较的被测单元是否失效。其分析判断逻辑可以在片内用硬件实现,也可以在 片内执行软件实现, 也可以在片外实现。
本发明所述的被测单元的运算结果进行比较时,每个被测单元的运算结果可以与其他一 个、两个甚至多个被测单元的运算结果进行比较。所述比较可以同时进行, 也可以不同时进 行。
本发明所述的比较器在进行运算结果比较时,可以对全部运算结果进行比较,也可以只 对部分特征运算结果或最终运算结果进行比较。
本发明所述的测试向量分配器包括测试向量分配控制器和物理连接,由测试向量分配控 制器配置物理连接构成连接关系;其中测试向量分配控制器可以在片内用硬件实现,也可以 在片内执行软件实现, 也可以在片外实现。
本发明所述的测试向量分配控制器可以省去。当没有测试向量分配控制器时,被测单元 与向量生成器的连接关系釆用固定方式,所述固定方式可以是多运算单元 /多核 /众核系统已 有的连接关系, 也可以是专为测试设计的连接关系。
本发明所述的向量生成器与被测单元的连接可以由测试向量分配器配置成任意相互连 接关系, 包括但不限于直线型、 网络结构型。
,本发明所述的运算结果分发器包括运算结果分发控制器和物理连接,由运算结果分发控 制器配置物理连接构成连接关系。其中运算结果分发控制器可以在片内用硬件实现,也可以 在片内执行软件实现, 也可以在片外实现。
本发明所述的运算结果分发控制器可以省去。当没有运算结果分发控制器时,被测单元 与比较器的连接关系采用固定的方式,所述固定的方式可以是多运算单元 /多核 /众核系统己 有的连接关系, 也可以是专为测试设计的连接关系。
本发明所述的比较器与被测单元间的连接关系可以根据运算结果分发器构成任意相互 连接关系, 包括但不限于直线型、 网络结构型。
本发明所述的测试结果表是用于存储根据测试比较结果产生的关联被测单元是否失效 的信息的数据信息表。
本发明所述的测试结果表的存储媒介可以是挥发性的,也可以是非挥发性的,可以是一 次写入不再更改的, 也可以是可擦除可多次写入的。
本发明所述的对存储器的自测试方法, 可以按照传统的存储器测试方法进行测试。 本发明所述的对存储器的自测试自修复装置由备份存储单元、 自测试单元、 自修复单元 构成。所述备份存储单元用于自修复时替换损坏单元;所述自测试单元用于测试存储单元子 阵列中的列是否失效,并把测试结果传递给自修复单元。所述自修复单元包括状态存储处理 单元,输入分配器和输出选择器。状态存储处理单元用于存储自测试单元传递过来的测试结 果。 当发现存储单元损坏时, 可以锁存错误状态, 还可以同时锁存错误存储单元的地址, 可 以控制输入分配器和输出选择器,使用备份存储单元替换损坏存储单元,完成存储器的修复。
当所述装置只包含存储阵列和自测试单元时,所述装置只进行自测试,若检测出失效存 储单元, 则产生不可修复信号, 供外部检测。 本发明所述的对于内部连线的自测试,可以通过在输入端给予激励, 比较输出端与输入 是否符合某种特定关系来实施, 所述特定关系包括但不限于相等、相反。 自修复可以通过用 备份的连线替换失效的连线来实施。
本发明所述的对于内部连线的自测试自修复装置由备份连线、自测试单元和自修复单元 构成。所述备份连线是可以替换被测试连线的等效连线。所述自测试单元用来对连线加载激 励、获取输出并比较输出与输入是否符合某种特定关系。所述自修复单元用来将失效连线的 输入旁路到备份连线上, 并将备份连线的输出连接到失效连线的输出端。
当所述装置只包含内部连线和自测试单元时,所述装置只进行自测试。由自测试单元完 成对连线的测试后, 向系统发出连线是否失效的信号, 不实施自修复。
本发明所述的对于输入输出管脚的自测试,可以通过在输入端给予激励, 比较输出端与 输入是否符合某种特定关系来实施, 所述特定关系包括但不限于相等、相反。 自修复可以通 过用备份的输入输出管脚替换失效的输入输出管脚来实施。
本发明所述的对于输入输出管脚的自测试自修复装置包含备份输入输出管脚、自测试单 元、 自修复单元。所述备份输入输出管脚是与被测输入输出管脚等效的输入输出管脚, 可以 是单向的输入管脚, 也可以是单向的输出管脚, 也可以是双向的输入输出管脚。所述自测试 单元用来对输入输出管脚加载激励、获取输出并比较输出与输入是否符合某种特定关系。所 述自修复单元用来将失效输入输出管脚的输入旁路到备份输入输出管脚上,并将备份输入输 出管脚的输出连接到失效输入输出管脚的输出端。
当所述装置只包含输入输出管脚和自测试单元时,所述装置只进行自测试。由测试电路 完成对输入输出管脚的测试后, 向系统发出连线是否失效的信号, 不实施自修复。
本发明所述的自测试自修复可以在晶圆测试时进行,也可以在芯片封装后集成电路测试 时或者包含该芯片的系统启动时进行;也可以人为设定自测试条件及周期,在工作期间定期 进行自测试和自修复。
当本发明所述的测试结果表存在时, 所述测试结果表可以在片内, 也可以在片外。 当本 发明所述的测试结果表不存在时,测试结果以其它信息形式送到片外,所述其他信息形式包 括但不限于仅表示被测芯片中是否有失效单元的信息形式。
本发明所述的单元状态表是用于存储在测试结果表的基础上产生的关联被测单元状态 信息的信息表;所述单元状态表可以在片内用硬件实现的存储媒介上储存,也可以在片内产 生后传输到片外储存。
本发明所述的动态分配算法为系统根据测试结果表信息进行被测单元任务动态分配的 算法, 所述的动态分配算法可以在片内用硬件实现, 也可以在片内执行软件实现, 也可以在 片外实现。 所述软件包括但不限于操作系统。
本发明所述的被测单元或连接具有旁路功能,即把失效的基本元件或运算单元或者处理 器核输入端口与输出端口直接导通或通过连线旁路导通。
本发明所述的自测试可实时更新测试结果表,并根据测试结果表信息将失效的基本元件 或运算单元或者处理器核旁路,对于相同结构的基本元件或运算单元或者处理器核,其可以 互为备份单元, 从而实现多运算单元 /多核 /众核系统的自修复功能。
有益效果:
本发明使用简单的方法和电路, 可以不依赖外部设备、不采用包括激光、高电压等任何 外力、熔丝、 反熔丝等任何特殊工艺或器件, 对芯片进行自测试与自修复。本发明通过采用 自测试单元对芯片内部进行测试,将结果记录在任何挥发性或非挥发性存储器中,用于自修 复时控制相应的半导体开关。若该存储器是挥发性的,则芯片每次加电工作时都要先进行自 测试自修复; 若该存储器是不挥发性的, 则可以只进行一次自测试自修复, 也可以人为设定 自测试条件及周期, 在工作期间定期进行自测试和自修复。
本发明可以有效的縮短测试时间。 若出现错误, 则自动使用备份单元替换掉损坏单元, 实现在片的自修复。该结构实现自修复的方法非常简单, 只需要简单的结构就可以实现, 不 需耍复杂的替换算法和实现方法;该结构可以在系统启动时实现在片自测试和自修复,不需 要外力的干预。 本发明以较低的代价实现芯片的在片自测试自修复, 提高了良率和可靠性, 也有效地降低了芯片的生产时测试成本。 附图说明
附图主要说明本发明的实施过程, 其中部件并非按照实际比例来制作。 同时, 实施例是 示意性的, 而不是限制性的。本发明所述的在片自测试自修复方法, 在下面具体实施例中进 行了具体的描述, 应当理解的是, 本发明并不受该实施例限制, 对于本领域普通技术人员来 说,可以根据本发明的技术方案和构思进行各种可能的替换、调整和改进,而所有这些替换、 调整和改进都应属于本发明所附权利要求的保护范围。
图 1为实现在片自测试自修复的芯片系统的结构图。
图 2为在片自测试自修复的一般流程图。
图 3 ( a) 为本发明所提出的针对逻辑单元自测试自修复的实施例一。
图 3 ( b ) 为本发明所提出的针对逻辑单元自测试自修复的实施例二。
图 4为本发明针对逻辑单元自测试自修复的流程图。
图 5为本发明针对存储器自测试自修复的一个实施例。
图 6为一种可修复存储子阵列中多列的状态存储处理单元 (1位) 结构图。
图 7 (a) 为图 6中状态存储处理单元在自测试前的存储的值。
图 7 (b ) 为图 6中状态存储处理单元自测试后存储的值。
图 8 ( a) 为一种可修复存储子阵列中输入分配器结构图。
图 8 ( b ) 为一种可修复存储子阵列中输出选择器结构图。
图 9为本发明针对内部连线的自测试自修复的实施例。
图 10为本发明在输入输出管脚方面的一个实施例。
图 11 (a) 为本发明在输出管脚方面的一个实施例。
图 ί ΐ ( b ) 为本发明在输入管脚方面的一个实施例。
图 12为本发明中测试结果表内容实施例。
图 13为本发明中处理器核状态表内容实施例。 具体实施方式
本发明的技术思路是通过自测试单元和自修复单元实现被测单元的检测与自修复。首先 通过自测试单元测试被测单元中是否失效,并把测试结果传递给自修复单元, 自修复单元利 用备份单元列替换失效单元, 完成自修复。 修复时不需要外来的干预和复杂算法。 请参阅图 1, 该图为实现在片自测试自修复的芯片系统的结构图。 该系统由被测单元 ( 101 )、 自测试单元 (102) 和自修复单元 (105 ) 三个部分组成。 其中, 被测单元 (101 ) 包括但不限于逻辑单元、 存储器、 内部连线、 输入输出管脚。 自测试单元 (102) 给予被测 单元( 101 )激励( 103 ),经被测单元输出返回自测试单元( 102)与预期结果进行比较( 104)。 测试的结果被送到自修复单元 (105), 若被测单元中有错误, 自修复单元 (105) 根据相应 测试结果用备份单元替换被测单元中错误部分, 以实现自修复。若被测单元中的错误部分没 有备份单元, 则产生不可修复信号, 供外部检测。
请参阅图 2, 该图为在片自测试自修复的一般流程图。 在晶圆测试时, 或封装后集成电 路测试时,或包含该芯片的系统启动时,或根据人为设定的自测试条件及周期在工作期间定 期进行自测试自修复时, 首先进入步骤一(201 )进行自测试, 再进入步骤二(202)判断是 否有失效单元; 若测试结果正确, 则说明该芯片可以正常工作; 若测试某单元失效, 则进入 步骤三(203)判断该单元是否有备份单元; 若没有备份单元,则说明该芯片无法修复, 若有 备份单元则进入到步骤四 (204) 自修复阶段, 对失效单元进行修复。 修复完成后, 进入步 骤五(205)重测已修复单元, 再进入步骤六(206)判断是否有失效单元; 若测试结果正确, 则说明该芯片可以正常工作; 若测试出仍有失效单元, 则说明该芯片无法修复。
图 3、 图 4为本发明针对逻辑单元自测试自修复的实施例。 下述实施例根据本发明技术 方案实施, 但本发明的保护范围不限于本实施例。其中, 图 3是被测系统在自测试自修复模 式下的结构图, 并未标出系统在工作模式下具体的数据输入、 输出结果流向等信息。
请参阅图 3 (a), 该图为本发明所提出的针对逻辑单元自测试自修复的实施例一。 对于 每一个被测逻辑单元, 均有一个相应的备份单元, 如备份单元 (302 ) 作为被测单元 (303) 的备份, 备份单元(306)作为被测单元(305)的备份,备份单元(310)作为被测单元(311 ) 的备份, 备份单元(314)作为被测单元(313) 的备份。 在自测试过程中, 首先向量生成器 ( 301 ) 产生测试向量并送到各个被测逻辑单元, 各被测逻辑单元执行测试向量, 比较器比 较各个被测逻辑单元的运算结果, 如比较器(304)对被测逻辑单元(303 )和被测逻辑单元 ( 305 )的运算结果进行比较, 比较器(307 )对对被测逻辑单元(303 )和被测逻辑单元(311 ) 的运算结果进行比较, 比较器 (309) 对对被测逻辑单元 (305 ) 和被测逻辑单元 (313 ) 的 运算结果进行比较, 比较器 (312 ) 对对被测逻辑单元 (311 ) 和被测逻辑单元 (313 ) 的运 算结果进行比较, 同时各比较器将比较结果送到自修复控制器(308), 自修复控制器(308) 根据各个比较器的比较结果判断失效逻辑单元,并控制备份单元替代失效的逻辑单元,从而 实现各被测逻辑单元的自测 自修复。 在该实施例中, 向量生成器 (301 ) 产生的测试向量 覆盖被测单元所支持的功能或者指令集,且尽量是带多重循环的运行过程复杂、程序代码精 短的测试向量。同时在进行运算结果相互比较时,可以是对每一步的运算结果实时进行比较, 也可以是只对在运算过程.中选定的部分运算结果进行比较,还可以是仅对整个测试向量的最 终运算结果进行比较。
请参阅图 3 (b), 该图为本发明所提出的针对逻辑单元自测试自修复的实施例二。 如图 所示, 在该实施例中, 被测单元执行测试向量的运算结果和测试向量的预期结果进行比较, 预期结果可以是测试向量执行过程中每一步操作的运算结果,也可以是在运算过程中选定的 部分中间结果, 还可以是整个测试向量的最终运算结果, 因此在比较过程中, 需要根据预期 结果来决定何时进行比较, 以及进行多少次比较。最后根据比较结果判定失效被测单元, 并 通过备份单元替换失效被测单元进行修复, 从而实现在片的自测试自修复。
请参阅图 4, 该图为本发明针对逻辑单元自测试自修复的流程图。 如图 4所示: 在自测 试开始后, 首先向量生成器产生测试向量(401 ), 测试向量送到各个被测逻辑单元, 各被测 逻辑单元运行测试向量(402 ), 并将运算结果送到相应的比较器, 各比较器对运算结果进行 比较, 并根据比较结果确定失效的逻辑单元(403 ) ,自修复逻辑通过旁路或者替代的方法修 复失效的逻辑单元(404), 修复完成后, 将对逻辑单元进行重测(405 ), 若逻辑单元测试正 确, 则说明该逻辑单元可用, 否则产生不可修复信号, 从而完成在片逻辑单元的自测试自修 复过程。
图 5、 图 6、 图 7、 图 8为本发明针对存储器自测试自修复的实施例。 下述实施例根据 本发明技术方案实施, 但本发明的保护范围不限于本实施例。
请参阅图 5, 该图为本发明针对存储器自测试自修复的一个实施例。 该实施例为一个拥 有列备份的存储器, 它包括存储阵列 (501 ), 自测试单元 (502 ) 和自修复单元 (503 ); 其 中的存储容量为 1024x32bit, 每位含有 4列存储单元, 即列地址为两位; 它含有两组存储 子阵列 (504), 相应备份列 (505)及译码器(506); 自测试单元(502)主要包括数据发生 器 (507)、 地址发生器 (508)及比较器 (509) 三个部分; 自修复单元 (503) 包含输入分 配器(510)、 输出分配器(511) 以及状态存储处理单元(512); 每个存储子阵列 (504)对 应一组输入分配器 (510) 和输出选择器 (511)。 自测试单元 (502) 测试存储阵列 (501), 产生存储阵列(501)中是否有损坏列的测试结果, 并把测试结果传递给自修复单元(503); 状态存储处理单元 (512)根据测试结果进行一定的逻辑处理, 并利用结果控制输入分配器 (510)和输出选择器(511)实现修复。根据修复能力的不同需求,状态存储处理单元(512) 可以拥有不同的结构。
请参阅图 6, 该图为一种可修复存储子阵列中多列的状态存储处理单元(1位)结构图, 它包括存储单元 (601)及结果处理单元 (602)。 该单元接收来自自测试单元的结果。 初始 化时, 测试结果和存储单元 (601) 中的值被初始化为 0; 当测试结果由 0跳转 1时, 即对 应的列失效时, 可以把失效状态和失效列的列地址存入存储单元 (602) 中。 当存储测试结 果的存储单元 (601) 中的值为 1时, 且当前列地址和存储的失效列的列地址相等, 结果处 理单元 (603)将产生替换信号。 把替换信号控制输入分配器和输出选择器把失效列的输入 和输出旁路到备份列。根据分析可以知道,使用该结构的存储器中每个备份列可以修复相应 的存储子阵列中列地址互不相同的 4列,即使用该结构的图 5存储器可以实现最多 .8列失效 列的修复。
请参阅图 7, 图 7 (a) 为图 6中状态存储处理单元在自测试前的存储的值, 图 7 (b) 为图 6中状态存储处理单元自测试后存储的值。 图 7 (a) 中的第一排存储的是存储阵列中 对应位是否失效的标志,第二排为失效位中失效列的地址高位,第三排为失效位中失效列的 列地址低位。 通过图 7 (b) 我们可以知道, 位 4, 位 11为失效位, 其中的位 4中列地址为 10的存储单元与位 11中列地址为 01的存储单元失效, 将被备份列替换, 其余列可正常使 用。
请参阅图 8, 图 8(a) —种可修复存储子阵列中输入分配器结构图; 该结构由三态门 (801) 构成, 也可以由多路选择器构成; 其中数据输入来自存储器外部或自测试单元; 根 据状态存储处理单元输出的替换信息来决定哪一路输入旁路到备份列中。本结构并不关断失 效列的输入。图 8(b)为一种可修复存储子阵列中输出选择器结构图,该结构由三态门(801) 构成,也可以由多路选择器构成;根据状态存储处理单元输出的替换信息来决定是否使用备 份列的输出替换存储子阵列中某一位形成存储器的输出。
图 9为本发明针对内部连线的自测试自修复的实施例,该实施例中的内部连线以系统总 线为例, 根据本发明技术方案实施, 但本发明的保护范围不限于系统总线。 图 9中自测试自 修复结构包括总线 (901), 备份总线 (902), 自测试单元 (903) 以及自修复单元 (904)。 自测试单元(903)主要由数据发生器(905)和比较器(906)组成。 比较器(906)把数据 发生器 (905) 产生的结果和总线末端读出的数据进行比较, 产生测试结果送到自修复单元 (904) 中的状态处理单元 (907), 则该模块利用自测试的结果产生修复信号。 若检测出某 根总线出错, 修复信号关断这根总线, 同时自动利用备份总线 (902) 进行替换。
图 10、 图 11为本发明针对输入输出管脚自测试的实施例。针对输入输出管脚自修复方 法与总线自修复类似, 即用备份输入输出管脚替换失效输入输出管脚。下述实施例根据本发 明技术方案实施, 但本发明的保护范围不限于本实施例。
请参阅图 10, 图 10为本发明在输入输出管脚方面的一个实施例。 该输入输出管脚主要 由输出门 (1001)、 输出控制(1002)、 输入门 (1003)和选择器(1008)组成, 是一个输入 输出复用的管脚。 在正常工作时, 如果工作在输出状态, 则由输出控制(1002)打开输出门 (1001), 选择器(1008)选择输出 (1004), 经过输出门 (1001)被送到管脚上; 如果工作 在输入状态,则由输出控制(1002)关闭输出门(1001),使管脚上的信号经过输入门(1003) 送到输入(1005)。在进行自测试时, 由输出控制(1002)打开输出门 (1001),选择器(1008) 选择激励(1006), 经过输出门 (1001)送往管脚, 同时也送往输入门 (1003), 自测试单元 获得经过输入门 (1003) 的信号后进行比较 (1007), 即可进行自测试。
请参阅图 11, 图 11 (a)为本发明在输出管脚方面的一个实施例。 该输出管脚主要由输 出门 (1101)、 输出控制 (1102) 和选择器 (1105) 组成, 是一个单向输出管脚。 在正常工 作时, 由输出控制 (1102)打开输出门 (1101), 选择器(1105)选择输出 (1103), 经过输 出门 (1101) 被送到管脚上。 在进行自测试时, 由输出控制 (1102) 打开输出门 (1101), 选择器 (1105) 选择激励 (1104), 经过输出门 (1101) 送往管脚, 自测试单元获得经过输 出门 (1101) 的信号后进行比较 (1106), 即可进行自测试。 图 11 (b) 为本发明在输入管 脚方面的一个实施例。 该输入管脚主要由输入门 (1108)和选择器(1105)组成, 是一个单 向输入管脚。 在正常工作时, 管脚上的信号经过输入门 (1108)送到输入 (1107)。 在进行 自测试时, 选择器 (1105) 选择激励 (1104), 经过输入门 (1108)送往芯片内部, 自测试 单元获得经过输入门 (1108) 的信号后进行比较 (1106), 即可进行自测试。
图 12、 图 13为本发明中测试结果表内容实施例和处理器核状态表内容实施例, 实施例 根据本发明的技术方案实施, 但本发明的保护内容不限于这两个实施例。
请参阅图 12, 该图为本发明中测试结果表内容实施例。 在该测试结果表 (1201 ) 中, 每一个标号(1202)对应系统中一个被测单元, 该位置上的信息表示被测单元的状态, 其中 "?"表示对应的被测单元未测, "X"表示对应的被测单元失效, "0"表示对应的被测单元 正常。该测试结果表可以在片内, 也可以在片外。其存储媒介可以是挥发性的, 也可以是非 挥发性的; 可以是一次写入不再更改的, 也可以是可擦除可多次写入的。在系统测试及运行 过程中, 失效的被测单元被旁路, 从而保证系统可以正常运行。 不但提高了良率, 而且实现 了系统的自修复功能。
请参阅图 13, 该图为本发明中处理器核状态表内容实施例。 在该图中, 处理器核状态 分为未测、失效、可用且空闲和可用且己被占用,其中"? "表示对应的被测单元未测, "X" 表示对应的被测单元失效, "0"表示对应的被测单元可用且空闲, " 1 "表示对应的被测单元 可用且正被占用。该处理器核状态表(1301 )是在测试结果表的基础上生成的, 其可以在片 内用硬件实现,也可以在片外由软件实现。系统根据状态表通过动态分配算法进行各处理器 核任务的动态分配,对于失效的处理器核,系统将其旁路,对于可用且已被占用的处理器核, 系统会记录其正被哪个进程占用以及占用进程的基本信息, 并在占用期间不会分配其任务, 对于可用且空闲的处理器核,系统根据自身需要实时进行任务的动态分配,从而充分利用系 统的资源。

Claims

1、 一种在片自测试自修复方法, 其特征在于能够在加电工作的情况下不依赖于外部设 备进行芯片的自测试,并能根据测试结果进行自修复;芯片在加电工作并进入自测试自修复 模式后即可采用所述方法进行不依赖于外部设备的自测试自修复,进入自测试自修复模式的 途径可以是由外部引脚触发,也可以使用已有信号触发;所述在片自测试自修复方法包括两 个阶
第一阶段,自测试单元测试被测单元并把测试结果传递给自修复单元,储存在存储器中; 自修复单元根据自测试结果进行操作,对被测单元的出错部分执行自修复, 由存储的测试结 果控制相应多路选择器或旁路开关, 以备份单元或器件取代失效的被测单元;
第二阶段, 自测试单元重测被测单元, 若未检测出错误, 则完成在片自测试自修复, 产 生测试完成信号,供外部检测,否则产生不可修复信号供外部检测或由测试结果生成测试结 果表供外部检测, 同时产生测试完成信号, 系统根据所述信号或测试结果表信息, 结合动态 分配算法进行被测单元任务的动态分配。
2、 根据权利要求 1所述的方法, 其特征在于所述自测试可以应用于芯片内的所有结构 和部分, 所述结构可以是逻辑单元, 模拟单元, 存储器, 内部连线, 输入输出管脚。
3、 根据权利要求 1所述的方法, 其特征在于所述自修复可以应用于芯片内的具有备份 单元的所有结构和部分, 所述结构可以是逻辑单元, 模拟单元, 存储器, 内部连线, 输入输 出管脚。
4、 根据权利要求 1所述的方法, 其特征在于对于逻辑单元、 模拟单元的自测试, 可通 过相同的逻辑单元、模拟单元间比较测试向量的运算结果是否为某种特定关系或比较逻辑单 元、模拟单元的运算结果与预期结果是否为某种特定关系来实施,所述特定关系可以是相等、 相反、同或、异或;其自修复可以通过旁路失效的逻辑单元、模拟单元或用备份的逻辑单元、 模拟单元替换失效的逻辑单元、 模拟单元来实施。
5、 根据权利要求 4所述的方法, 其特征在于可以构成对于逻辑单元、 模拟单元的自测 试自修复装置; 所述装置由备份逻辑单元、 备份模拟单元、 自测试单元和自修复单元构成; 所述备份逻辑单元、备份模拟单元用于替换失效的逻辑单元、模拟单元; 所述自测试单元用 于生成测试向量并判断逻辑单元、模拟单元执行测试向量的运算结果是否正确;所述自修复 单元用于控制备份逻辑单元、备份模拟单元替换失效的逻辑单元、模拟单元或旁路失效的逻 辑单元、模拟单元; 所述逻辑单元、模拟单元可以是任意大小的基本元件、运算单元或处理 器核; 也可以是逻辑电路、 模拟电路中的任意部分, 包括数据通道、 控制单元、 模拟电路。
6、 根据权利要求 5所述的方法和装置, 其特征在于所述逻辑单元、 模拟单元可以是多 运算单元 /多核 /众核系统中需要被测试的具有特定关系的基本元件、 运算单元或处理器核, 也可以是多运算单元 /多核 /众核系统中需要被测试的具有特定关系的基本元件、运算单元或 处理器核中的部分或全部组成的集合;所述自测试单元可以包括比较器,用于比较被测单元 执行自测试向量得到的结果是否为特定关系,并根据比较结果来确定被测单元的正确性;所 述特定关系可以是相等、 相反、 互逆、 互补。
7、 根据权利要求 6所述的方法和装置, 其特征在于所述自测试单元还可以包含下列装 置中的部分或全部:
向量生成逻辑, 用于生成、存储或生成并存储自测试向量, 可以是由向量产生逻辑和向 量存储器两部分组成, 也可以是只有向量产生逻辑或向量存储器;
测试向量分配逻辑, 包括测试向量分配控制器和物理连接, 用于配置、建立或在自测试 过程中动态调整被测试多运算单元 /多核 /众核系统中被测单元输入端口与向量生成器的连 接关系; ·
运算结果分发逻辑, 包括运算结果分发控制器和物理连接, 用于配置、建立或在自测试 过程中动态调整被测试多运算单元 /多核 /众核系统中被测单元输出端口与比较器之间的连 接关系。
8、 根据权利要求 1所述的在片自测试自修复方法, 其特征在于可以构成对存储器的自 测试自修复装置; 所述装置由备份存储单元、 自测试单元、 自修复单元构成; 所述备份存储 单元用于自修复时替换损坏单元; 所述自测试单元用于测试存储单元子阵列中的列是否失 效,并把测试结果传递给自修复单元;所述自修复单元用于存储自测试单元传递过来的测试 结果; 当发现存储单元损坏时, 可以锁存错误状态、错误存储单元地址或错误状态及错误存 储单元地址, 还可以使用备份存储单元替换损坏存储单元, 实现存储器的修复。
9、根据权利要求 1所述的在片自测试自修复方法,其特征在于对于内部连线的自测试, 可以通过在输入端给予激励, 比较输出端与输入是否符合某种特定关系来实施,所述特定关 系可以是相等、 相反; 自修复可以通过用备份的连线替换失效的连线来实施。
10、根据权利要求 9所述的在片自测试自修复方法,其特征在于可以构成对内部连线的 自测试自修复装置; 所述装置由备份连线、 自测试单元和自修复单元构成; 所述备份连线是 可以替换被测试连线的等效连线;所述自测试单元用来对连线加载激励、获取输出并比较输 出与输入是否符合某种特定关系; 所述自修复单元用来将失效连线的输入旁路到备份连线 上, 并将备份连线的输出连接到失效连线的输出端。
11、根据权利要求 1所述的在片自测试自修复方法,其特征在于对于输入输出管脚的自 测试, 可以通过在输入端给予激励, 比较输出端与输入是否符合某种特定关系来实施, 所述 特定关系可以是相等、相反; 自修复可以通过用备份的输入输出管脚替换失效的输入输出管 脚来实施。
12、 根据权利要求 11所述的在片自测试自修复方法, 其特征在于可以构成对输入输出 管脚的自测试自修复装置; 所述装置包含备份输入输出管脚、 自测试单元、 自修复单元; 所 述备份输入输出管脚是与被测输入输出管脚等效的输入输出管脚;所述自测试单元用来对输 入输出管脚加载激励、获取输出并比较输出与输入是否符合某种特定关系;所述自修复单元 用来将失效输入输出管脚的输入旁路到备份输入输出管脚上,并将备份输入输出管脚的输出 连接到失效输入输出管脚的输出端。
13、根据权利要求 1所述的在片自测试自修复方法,其特征在于所述自测试自修复可以 在晶圆测试时进行, 也可以在芯片封装后集成电路测试时或者包含该芯片的系统启动时进 行; 也可以人为设定自测试条件及周期, 在工作期间定期进行自测试和自修复。
14、根据权利要求 1所述的在片自测试自修复方法,其特征在于所述的动态分配算法为 系统根据测试结果表信息进行被测单元任务动态分配的算法,所述的动态分配算法可以在片 内用硬件实现,也可以在片内执行软件实现,也可以在片外实现;所述软件可以是操作系统。
PCT/CN2010/000126 2008-12-12 2010-01-28 在片自测试自修复方法 WO2010066207A1 (zh)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN200810204623.5 2008-12-12
CN2008102046235A CN101751317B (zh) 2008-12-12 2008-12-12 多运算单元/多核/众核系统的自测试自修复机制与结构
CN200810207673.9 2008-12-23
CN200810207673.9A CN101763901B (zh) 2008-12-23 2008-12-23 在片自测试自修复方法

Publications (2)

Publication Number Publication Date
WO2010066207A1 true WO2010066207A1 (zh) 2010-06-17
WO2010066207A8 WO2010066207A8 (zh) 2010-08-05

Family

ID=42242363

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2010/000126 WO2010066207A1 (zh) 2008-12-12 2010-01-28 在片自测试自修复方法

Country Status (1)

Country Link
WO (1) WO2010066207A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108804261A (zh) * 2017-05-05 2018-11-13 中兴通讯股份有限公司 连接器的测试方法及装置
CN111833959A (zh) * 2020-07-20 2020-10-27 北京百度网讯科技有限公司 存储器的测试的方法、装置、电子设备和计算机可读存储介质
CN113360323A (zh) * 2021-07-02 2021-09-07 西安紫光国芯半导体有限公司 一种众核计算电路、堆叠芯片和容错控制方法
CN113806144A (zh) * 2021-09-08 2021-12-17 瑞芯微电子股份有限公司 芯片修复方法、系统及控制终端
CN115480960A (zh) * 2021-05-31 2022-12-16 西安紫光国芯半导体有限公司 具自测试功能的众核计算电路、及其测试方法、装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1332459A (zh) * 2000-06-30 2002-01-23 三菱电机株式会社 半导体集成电路、半导体集成电路的存储器修复方法
US6343366B1 (en) * 1998-07-15 2002-01-29 Mitsubishi Denki Kabushiki Kaisha BIST circuit for LSI memory
US6667918B2 (en) * 2002-05-01 2003-12-23 Mellanox Technologies Ltd. Self-repair of embedded memory arrays
CN1478282A (zh) * 2000-12-01 2004-02-25 �Ƚ�΢װ�ù�˾ 存储器存储阵列的内置自修复的方法与装置
US6728910B1 (en) * 2000-09-20 2004-04-27 Lsi Logic Corporation Memory testing for built-in self-repair system
US7216277B1 (en) * 2003-11-18 2007-05-08 Xilinx, Inc. Self-repairing redundancy for memory blocks in programmable logic devices

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6343366B1 (en) * 1998-07-15 2002-01-29 Mitsubishi Denki Kabushiki Kaisha BIST circuit for LSI memory
CN1332459A (zh) * 2000-06-30 2002-01-23 三菱电机株式会社 半导体集成电路、半导体集成电路的存储器修复方法
US6728910B1 (en) * 2000-09-20 2004-04-27 Lsi Logic Corporation Memory testing for built-in self-repair system
CN1478282A (zh) * 2000-12-01 2004-02-25 �Ƚ�΢װ�ù�˾ 存储器存储阵列的内置自修复的方法与装置
US6667918B2 (en) * 2002-05-01 2003-12-23 Mellanox Technologies Ltd. Self-repair of embedded memory arrays
US7216277B1 (en) * 2003-11-18 2007-05-08 Xilinx, Inc. Self-repairing redundancy for memory blocks in programmable logic devices

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108804261A (zh) * 2017-05-05 2018-11-13 中兴通讯股份有限公司 连接器的测试方法及装置
CN108804261B (zh) * 2017-05-05 2023-05-19 中兴通讯股份有限公司 连接器的测试方法及装置
CN111833959A (zh) * 2020-07-20 2020-10-27 北京百度网讯科技有限公司 存储器的测试的方法、装置、电子设备和计算机可读存储介质
US11393552B2 (en) 2020-07-20 2022-07-19 Beijing Baidu Netcom Science And Technology Co., Ltd. Method, apparatus and electronic device for texting memory, and computer-readable storage medium
CN111833959B (zh) * 2020-07-20 2022-08-02 北京百度网讯科技有限公司 存储器的测试的方法、装置、电子设备和计算机可读存储介质
CN115480960A (zh) * 2021-05-31 2022-12-16 西安紫光国芯半导体有限公司 具自测试功能的众核计算电路、及其测试方法、装置
CN113360323A (zh) * 2021-07-02 2021-09-07 西安紫光国芯半导体有限公司 一种众核计算电路、堆叠芯片和容错控制方法
CN113806144A (zh) * 2021-09-08 2021-12-17 瑞芯微电子股份有限公司 芯片修复方法、系统及控制终端

Also Published As

Publication number Publication date
WO2010066207A8 (zh) 2010-08-05

Similar Documents

Publication Publication Date Title
US7284166B2 (en) Programmable multi-mode built-in self-test and self-repair structure for embedded memory arrays
US11119857B2 (en) Substitute redundant memory
US6601218B2 (en) Semiconductor integrated circuit device
KR100388550B1 (ko) 멀티쓰레드 컴퓨터 프로세서의 쓰레드 능력 변경
US8988956B2 (en) Programmable memory built in self repair circuit
KR100749187B1 (ko) 보편적으로 액세스 가능하고 완전히 프로그램 가능한메모리 내장 자체 테스트 시스템 및 방법
US6769081B1 (en) Reconfigurable built-in self-test engine for testing a reconfigurable memory
JP2003505814A (ja) エンベデットメモリのための予備行列置換ビルトイン解析システム
JPH1069799A (ja) 集積回路メモリ・デバイス用組込み自己テスト装置
KR20160148347A (ko) 셀프 리페어 장치 및 방법
KR20150030430A (ko) 셀프 리페어 장치
EP1389315A2 (en) Hierarchical built-in self-test for system-on-chip design
US6353563B1 (en) Built-in self-test arrangement for integrated circuit memory devices
US20120230136A1 (en) Selectable repair pass masking
US7213186B2 (en) Memory built-in self test circuit with full error mapping capability
JP2006512698A (ja) 直接アクセスモードによって埋め込みdram回路を試験するための回路および方法
US9640279B1 (en) Apparatus and method for built-in test and repair of 3D-IC memory
US9318222B2 (en) Hierarchical, distributed built-in self-repair solution
JP2004505395A (ja) マイクロコード・ベースのメモリ・ビルトイン・セルフテストのシステム初期化
WO2010066207A1 (zh) 在片自测试自修复方法
CN112562770B (zh) 具有测试电路的半导体装置
US20080013389A1 (en) Random access memory including test circuit
KR100383479B1 (ko) 디지털 반도체 회로를 테스트하기 위한 회로 및 방법
KR20130104732A (ko) 테스트 회로, 메모리 시스템 및 메모리 시스템의 테스트 방법
JPS61292299A (ja) オンチツプメモリテスト容易化回路

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 10721263

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 10721263

Country of ref document: EP

Kind code of ref document: A1