A self-healing circuit system and method for use in MCU memory

By integrating self-testing and verification modules, fault recording and evaluation modules, and remapping and migration modules, the problems of single-bit errors and multi-bit faults in MCU memory during operation are solved, realizing self-diagnosis and repair of memory, and improving the reliability and lifespan of the system.

CN123072473APending Publication Date: 2026-11-10SHANGHAI CHIPON MICRO ELECTRONICS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611235360.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-14
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing technologies cannot effectively solve single-bit errors or multi-bit faults caused by process deviations, electromagnetic interference, and other factors during the operation of MCU memory, and lack self-healing capabilities, leading to system failure.

Method used

It integrates background periodic memory self-testing, automatic error correction and recording, and dynamic address remapping of storage blocks. It achieves self-diagnosis and repair through self-testing and verification modules, fault recording and evaluation modules, and remapping and migration modules.

Benefits of technology

This technology improves the reliability and lifespan of the memory during operation. It also enables real-time fault detection and dynamic management through an online self-healing circuit system, thereby enhancing the reliability and availability of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123072473A_ABST
    Figure CN123072473A_ABST
Patent Text Reader

Abstract

This invention discloses a self-healing circuit system and method for MCU memory, belonging to the field of integrated circuit and memory testing technology, and specifically relating to it. It includes: a self-test and verification module, a fault recording and evaluation module, and a remapping and migration module. The self-test and verification module includes an online self-test module and an error correction and verification module; the fault recording and evaluation module includes a fault map library and a health assessment module; and the remapping and migration module includes an address remapping module, a redundancy pool management module, and a migration controller. This invention extends fault handling from discrete post-manufacturing stages to a continuous closed loop throughout the entire chip operating cycle, achieving a functional leap; through a flexible address mapping strategy, it achieves dynamic and reversible management of faulty resources; and by shifting the core value objective from "yield" for producers to "reliability" and "availability" for end users, it better meets the needs of high-reliability application scenarios.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of integrated circuit and memory testing technology, and in particular to a self-healing circuit system and method for use in MCU memory. Background Technology

[0002] With the widespread application of MCUs in key fields such as automotive electronics, industrial control, and the Internet of Things, the reliability of their internal memory (such as SRAM and Flash) faces severe challenges. Memory may experience single-bit errors or multi-bit failures due to factors such as process deviations, electromagnetic interference, and aging, leading to system failure. The traditional solution is to perform read and write tests on the memory during the chip testing phase (such as writing all 0s, all Fs, a positive checkerboard, and a negative checkerboard before reading for verification). This common practice has the following limitations: Offline MBIST: Typically executed during chip testing or startup, it cannot cover soft errors or intermittent failures that occur during runtime.

[0003] Independent ECC (Error Correction Code): It can correct single-bit errors, but it cannot record the frequency and location of errors, nor can it preventive measures for memory cells that frequently malfunction.

[0004] Static bad block management: Commonly found in Flash memory, it usually requires pre-setting redundant blocks and has fixed remapping logic, which cannot adapt to dynamic error accumulation.

[0005] Lack of self-healing ability: Most existing solutions rely on external intervention or system reset, making it difficult to achieve self-diagnosis and repair during continuous operation; Summary of the Invention To address the shortcomings of existing technologies, the self-healing circuit system proposed in this invention integrates background periodic memory self-testing, automatic error correction and recording, and dynamic address remapping of storage blocks, thereby improving the reliability and lifespan of the memory during operation.

[0006] To achieve the above objectives, the present invention provides a self-healing circuit system applied to MCU memory, comprising: The self-test and verification module includes an online self-test module and an error correction and verification module. The online self-test module is used to periodically perform memory tests during MCU operation, detect faults and generate fault events. The error correction and verification module is used to correct single-bit or multi-bit errors in memory access in real time and record the number of error corrections. The fault recording and assessment module includes a fault map library and a health assessment module. The health assessment module is used to calculate the health value of the storage block based on the fault events and the number of error corrections. When the health value exceeds a preset degradation threshold, a migration trigger signal is generated for the remapping and migration module. The fault map library is used to store storage block degradation or migration marker information. The remapping and migration module includes an address remapping module, a redundancy pool management module, and a migration controller. The address remapping module is used to dynamically map access addresses from faulty storage blocks to redundant storage blocks according to migration trigger signals. The migration controller is used to control the migration of data from faulty storage blocks to redundant storage blocks and update the address mapping relationship. The redundancy pool management module is used to manage the allocation and status of redundant storage blocks.

[0007] Furthermore, the online self-test module includes: The scheduling controller is used to monitor the bus status in real time during normal MCU operation. When an idle state is detected, it starts the address generator and the stepper. Address generator and stepper are used to generate continuous or pseudo-random test addresses as logic blocks to be tested; Test mode generator, used to select the test mode; The write control unit is used to write test data to the logic block under test according to the selected test address based on the selected test mode. The read-write comparison unit is used to perform read operations on the logic block to be tested according to the test address, and compare the read data with the written data. If an error occurs, the error bit, address and error type information are recorded. The fault classification and reporting unit is used to combine error bit, address and error type information into a frame and send it to the fault recording and evaluation module.

[0008] Furthermore, the health assessment module generates a health value for each logic block by weighted accumulation of error correction counts and fault events. When the health value exceeds a preset degradation threshold, the logic block is marked as degraded, immediately written into the fault map library, and a migration trigger signal is generated.

[0009] Furthermore, the redundancy pool management module specifically reads the fault map library to obtain information on all logical blocks that have been marked as migrated or degraded, excludes the migrated or degraded logical blocks, and the remaining unused logical blocks form a redundant storage block resource pool. When a migration trigger signal is received, a redundant storage block is allocated from the redundant storage block resource pool for data migration.

[0010] Furthermore, the address remapping module includes: The logic block address extractor is used to decompose the MCU logic address into two parts: logic block ID and intra-block offset, in preparation for subsequent address matching. The address remapping parallel comparison array is used to compare the logical block ID with the allocated redundant storage block ID. If they match, the Match signal is set to 1, and the process enters the match arbitration and multiplexer. Match arbitration and multiplexers, using a priority encoder to ensure that only one address is selected when multiple addresses are hit simultaneously; The physical address synthesis and replacement logic unit combines and replaces addresses to form new physical addresses, which are the addresses of the redundant storage blocks to which they are mapped.

[0011] Furthermore, the update of the address mapping relationship in the migration controller specifically involves: directing the access of the faulty storage block to the mapped redundant storage block, while marking the redundant storage block with embedded data as being migrated in and writing it into the fault map library.

[0012] Furthermore, the self-healing circuit system also includes a configuration register for controlling the activation and deactivation of the self-healing circuit system.

[0013] The present invention also provides a self-healing method for MCU memory, based on the self-healing circuit system for MCU memory described above, characterized by comprising the following steps: (1) After the MCU is powered on, the self-healing control system can be enabled by configuring the register; (3) During MCU operation, the online self-test module periodically performs memory tests, detects faults and generates fault events; the error correction and verification module corrects single-bit or multi-bit errors in memory access in real time and records the number of error corrections; (4) The health assessment module calculates the health value of the storage block based on the cumulative number of fault events and error corrections. When the health value exceeds the preset degradation threshold, the storage block is marked as degraded, written into the fault map library, and a migration trigger signal is generated for the remapping and migration module. (5) Upon receiving the migration trigger signal, the redundancy pool management module reads the fault map library, obtains the redundant storage block pool based on all logical blocks that have been marked as migration or degradation, and selects an unused redundant storage block as the migration target; the address remapping module dynamically maps the access address from the faulty storage block to the redundant storage block; the migration controller controls the data to migrate from the faulty storage block to the redundant storage block, marks the redundant storage block as a migration, writes it into the fault map library, and updates the address mapping relationship.

[0014] The beneficial effects of this invention are: 1. From “offline detection” to “online self-healing”: This invention extends fault handling from the discrete stages after manufacturing to a continuous closed loop throughout the entire chip operating cycle, achieving a leap in functionality.

[0015] 2. From "physical replacement" to "logical management": It avoids the one-time nature and resource limitations of physical repair, and realizes dynamic and reversible management of faulty resources through flexible address mapping strategies.

[0016] 3. From “improving yield” to “ensuring reliability”: The core value objective has shifted from “yield” for producers to “reliability” and “availability” for end users, which is more in line with the needs of high-reliability application scenarios. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the self-healing circuit system applied to MCU memory according to an embodiment of the present invention.

[0018] Figure 2 This is a schematic diagram of the online self-test module structure in an embodiment of the present invention.

[0019] Figure 3 This is a schematic diagram of the address remapping module structure in an embodiment of the present invention.

[0020] Figure 4 This is a schematic diagram of the state machine of the migration controller according to an embodiment of the present invention. Detailed Implementation

[0021] The present invention will be further explained and described below with reference to the accompanying drawings and embodiments.

[0022] like Figure 1 As shown, this embodiment of the invention provides a self-healing circuit system for MCU memory, including: a self-test and verification module, a fault recording and evaluation module, and a remapping and migration module.

[0023] The self-test and verification module includes an online self-test module and an error correction and verification module.

[0024] The online self-test module is used to periodically perform memory tests during MCU operation, detect faults, and generate fault events. For example... Figure 2 As shown, the online self-test module includes: The scheduling controller is used to monitor the bus status in real time during normal MCU operation. When an idle state is detected, it starts the address generator and the stepper.

[0025] Address generator and stepper are used to generate continuous or pseudo-random test addresses as logic blocks to be tested.

[0026] The test mode generator is used to select the test mode. The test modes are shown in the table below: Table 1 The write control unit is used to write the corresponding predefined test data (e.g., all 0s, all 1s, etc.) to the logic block under test according to the test address based on the selected test mode.

[0027] After the above actions are completed, the write control unit will take over the storage write port, and then insert a test mode write action. After the write action is completed, it will be automatically released.

[0028] The read-write comparison unit is used to perform read operations on the logic block to be tested according to the test address, and compare the read data with the written data. If an error occurs, the error bit, address and error type information are recorded.

[0029] The fault classification and reporting unit is used to combine error bit, address and error type information into a frame and send it to the fault recording and evaluation module.

[0030] The error correction and verification module is used to correct single-bit or multi-bit errors in memory access in real time and record the number of error corrections.

[0031] The fault recording and assessment module includes a fault map library and a health assessment module.

[0032] The health assessment module is used to calculate the health value of the storage block based on the cumulative number of fault events and error corrections. When the health value exceeds the preset degradation threshold, a migration trigger signal is generated for the remapping and migration module.

[0033] The health assessment module generates a health value for each logic block by weighting and accumulating the number of error corrections and fault events. When the health value exceeds a preset degradation threshold, the logic block is marked as degraded, written into the fault map library, and a migration trigger signal is generated.

[0034] The fault map library is used to store information on storage block degradation or migration markers.

[0035] The remapping and migration module includes an address remapping module, a redundancy pool management module, and a migration controller.

[0036] The redundancy pool management module specifically reads the fault map library to obtain information on all logical blocks that have been marked for migration or degradation. Based on this information, it obtains a redundant storage block resource pool. When a migration trigger signal is received, it allocates a redundant storage block from the redundant storage block resource pool for data migration.

[0037] The address remapping module is used to dynamically map access addresses from faulty storage blocks to redundant storage blocks based on migration trigger signals. For example... Figure 3 As shown, the address remapping module includes: The logic block address extractor is used to decompose the MCU logic address into two parts: the logic block ID and the block offset, in preparation for subsequent address matching.

[0038] The address remapping parallel comparison array is used to compare the logical block ID with the allocated redundant storage block ID. If they match, the Match signal is set to 1, and the process enters the match arbitration and multiplexer.

[0039] Matching arbitration and multiplexers, using a priority encoder, ensures that only one address is selected when multiple addresses are hit simultaneously.

[0040] The physical address synthesis and replacement logic unit combines and replaces addresses to form new physical addresses, which are the addresses of the redundant storage blocks to which they are mapped.

[0041] The migration controller is used to control the migration of data from faulty storage blocks to redundant storage blocks and update the address mapping relationship; the redundancy pool management module is used to manage the allocation and status of redundant storage blocks.

[0042] The specific steps for updating the address mapping relationship are as follows: the access of the faulty storage block is redirected to the mapped redundant storage block, and the redundant storage block with embedded data is marked as moved in and written to the fault map library.

[0043] The self-healing circuit system also includes a configuration register for controlling the activation and deactivation of the self-healing circuit system.

[0044] This invention provides a self-healing method for MCU memory, based on the aforementioned self-healing circuit system for MCU memory, comprising the following steps: After the S101 and MCU are powered on, the self-healing control system can be enabled through the configuration register.

[0045] After the MCU is powered on, the self-healing control system can be enabled via the configuration register, and the sensitivity of the health assessment module can also be configured.

[0046] S102. During MCU operation, the online self-test module periodically performs memory tests, detects faults and generates fault events; the error correction and verification module corrects single-bit or multi-bit errors in memory access in real time and records the number of error corrections.

[0047] The Online MBIST module selects a logic block under test in a low-priority polling manner and performs test mode write, read-back, and retention tests on the block without affecting the MCU's real-time access.

[0048] Each time the MCU accesses the memory, the error correction and verification module performs real-time error correction on the read data and records the number of error corrections.

[0049] When a data mismatch, hold failure, or write disturbance anomaly is detected, a fault event containing the logical block address, error type, and occurrence time is generated and submitted to the health assessment module.

[0050] S103. The health assessment module calculates the health value of the storage block based on the cumulative number of fault events and error corrections. When the health value exceeds the preset degradation threshold, the storage block is marked as degraded, written into the fault map library, and a migration trigger signal is generated for the remapping and migration module.

[0051] The health assessment module calculates a health score by weighting and summing the number of ECC error corrections, online self-test failure events, and aging counters.

[0052] When the health value exceeds the MCU's preset degradation threshold, the logic block is marked as a "potential degradation block". At this time, the health assessment module will send a degradation mark signal to the remapping and migration module.

[0053] S104. Upon receiving the migration trigger signal, the redundancy pool management module reads the fault map library, obtains the redundant storage block pool based on all logical blocks marked as migration or degradation, and selects an unused redundant storage block as the migration target. The address remapping module dynamically maps the access address from the faulty storage block to the redundant storage block. The migration controller controls the data to migrate from the faulty storage block to the redundant storage block, marks the redundant storage block as being migrated, writes it into the fault map library, and updates the address mapping relationship.

[0054] Upon receiving a degradation flag, the logic block is added to the migration queue, and write operations to that block are paused. The redundancy pool management module allocates an unused redundant block from the spare storage block pool as the migration target. The migration controller then enters a locked state, prohibiting the MCU from writing to that logic block. Data in the degradation block is copied byte-by-byte to the newly allocated redundant block using the internal DMA engine. The migration controller's state machine is as follows: Figure 4 As shown.

[0055] Example 1: An MCU storage system with SRAM self-healing capability.

[0056] I. Implementation Environment This embodiment uses a 32-bit industrial-grade MCU as an example, which integrates the following on-chip features: • 256KB SRAM main logic block to be tested; • 16KB redundant SRAM spare pool; • 64-bit data bus; • Single-cycle access timing.

[0057] II. System Structure This MCU integrates a memory self-healing control subsystem, including: • Online self-test module; • ECC error correction and verification module; • Fault map library; • Health assessment module; • Address remapping module; • Migration controller; • Redundancy pool management module; The above modules, together with the SRAM main array and redundant storage pool, form a self-healing closed loop.

[0058] III. Work Process (a) Power-on initialization After the MCU is powered on, the self-healing control unit loads the fault map library from the Shadow non-volatile memory area and writes the historical migration records into the CAM table of the address remapping module, so that all migrated logic blocks are automatically redirected to the redundancy pool.

[0059] (ii) Back-end inspection The online self-test module performs a background scan of the SRAM block at a period of 10ms, and uses the alternating write-read method of Walking-1 / 0 and 1010 / 0101 to detect weak cells and write disturbance errors.

[0060] (III) Error Accumulation and Health Assessment The error correction module corrects single-bit errors generated by MCU access and accumulates the number of error corrections. When the number of error corrections exceeds 32, the health assessment module determines that the SRAM block has entered a degraded state.

[0061] (iv) Autonomous migration When a logic block enters a degenerate state: 1. The redundancy pool management module allocates one free spare block; 2. The migration controller locks the write channel for this logic block; 3. Initiate internal DMA for hot data migration; 4. Verify data integrity using CRC checksum; 5. Update the CAM mapping; 6. Shield the original physical block.

[0062] (v) Operation recovery After the migration is complete, all MCU accesses to the logic block are automatically redirected to the redundant memory block, and the process is completely transparent to the software.

[0063] IV. Technical Effects SRAM lifetime increased by 8 times Single-event flip fault-tolerant fully automatic recovery Example 2: An MCU storage system with Flash self-healing capability.

[0064] I. Implementation Environment This embodiment is configured for an on-chip 2MB Flash program memory as follows: • 128KB redundant sector pool; • BCH multi-bit error correction module; • Sector-level address remapping.

[0065] II. Self-healing process 1. The online self-test module performs erase / write fatigue detection on the Flash sectors during system idle periods.

[0066] 2. When the number of erase / write failures exceeds the threshold, the health assessment module marks the sector as a degraded sector.

[0067] 3. Redundancy pool is allocated spare sectors.

[0068] 4. The migration controller performs sector-level data migration.

[0069] 5. Update the address remapping CAM table.

[0070] 6. Shield the original sector.

[0071] III. Technical Effects Flash lifespan ≥ 5 times.

[0072] Increase program storage stability and prevent crashes over long periods.

[0073] Example 3 A self-healing MCU with power-loss retention capability This embodiment adds a Shadow fault map library module to the existing embodiments one and two, so that the migration relationship can be automatically restored after power failure and restart, ensuring reliable operation of the MCU throughout its entire life cycle.

[0074] Finally, it should be noted that the above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A self-healing circuit system applied to MCU memory, characterized in that, include: The self-test and verification module includes an online self-test module and an error correction and verification module. The online self-test module is used to periodically perform memory tests during MCU operation, detect faults and generate fault events. The error correction and verification module is used to correct single-bit or multi-bit errors in memory access in real time and record the number of error corrections. The fault recording and assessment module includes a fault map library and a health assessment module. The health assessment module is used to calculate the health value of the storage block based on the fault events and the number of error corrections. When the health value exceeds a preset degradation threshold, a migration trigger signal is generated for the remapping and migration module. The fault map library is used to store storage block degradation or migration marker information; The remapping and migration module includes an address remapping module, a redundancy pool management module, and a migration controller. The address remapping module is used to dynamically map access addresses from faulty storage blocks to redundant storage blocks according to migration trigger signals. The migration controller is used to control the migration of data from faulty storage blocks to redundant storage blocks and update the address mapping relationship. The redundancy pool management module is used to manage the allocation and status of redundant storage blocks.

2. The self-healing circuit system applied to MCU memory according to claim 1, characterized in that, The online self-test module includes: The scheduling controller is used to monitor the bus status in real time during normal MCU operation. When an idle state is detected, it starts the address generator and the stepper. Address generator and stepper are used to generate continuous or pseudo-random test addresses as logic blocks to be tested; Test mode generator, used to select the test mode; The write control unit is used to write test data to the logic block under test according to the selected test address based on the selected test mode. The read-write comparison unit is used to perform read operations on the logic block to be tested according to the test address, compare the read data with the written data, and record the error bit, address and error type information if an error occurs. The fault classification and reporting unit is used to combine error bit, address and error type information into a frame and send it to the fault recording and evaluation module.

3. The self-healing circuit system applied to MCU memory according to claim 1, characterized in that: The health assessment module generates a health value for each logic block by weighted accumulation of error correction counts and fault events. When the health value exceeds a preset degradation threshold, the logic block is marked as degraded, immediately written into the fault map library, and a migration trigger signal is generated.

4. The self-healing circuit system applied to MCU memory according to claim 1, characterized in that, The redundancy pool management module specifically reads the fault map library to obtain information on all logic blocks that have been marked as migrated or degraded, excludes the migrated or degraded logic blocks, and the remaining unused logic blocks form a redundant storage block resource pool. When a migration trigger signal is received, a redundant storage block is allocated from the redundant storage block resource pool for data migration.

5. The self-healing circuit system applied to MCU memory according to claim 4, characterized in that, The address remapping module includes: The logic block address extractor is used to decompose the MCU logic address into two parts: logic block ID and intra-block offset, in preparation for subsequent address matching. The address remapping parallel comparison array is used to compare the logical block ID with the allocated redundant storage block ID. If they match, the Match signal is set to 1, and the process enters the match arbitration and multiplexer. Match arbitration and multiplexers, using a priority encoder to ensure that only one address is selected when multiple addresses are hit simultaneously; The physical address synthesis and replacement logic unit combines and replaces addresses to form new physical addresses, which are the addresses of the redundant storage blocks to which they are mapped.

6. The self-healing circuit system applied to MCU memory according to claim 1, characterized in that, The specific steps for updating the address mapping relationship in the migration controller are as follows: the access of the faulty storage block is directed to the mapped redundant storage block, and the redundant storage block with embedded data is marked as migrated in and written into the fault map library.

7. The self-healing circuit system applied to MCU memory according to claim 1, characterized in that, The self-healing circuit system also includes a configuration register for controlling the activation and deactivation of the self-healing circuit system.

8. A self-healing method for MCU memory, based on the self-healing circuit system for MCU memory as described in any one of claims 1-7, characterized in that, Includes the following steps: (1) After the MCU is powered on, the self-healing control system can be enabled by configuring the register; (3) During MCU operation, the online self-test module periodically performs memory tests to detect faults and generate fault events; The error correction and verification module corrects single-bit or multi-bit errors in memory access in real time and records the number of error corrections. (4) The health assessment module calculates the health value of the storage block based on the cumulative number of fault events and error corrections. When the health value exceeds the preset degradation threshold, the storage block is marked as degraded, written into the fault map library, and a migration trigger signal is generated for the remapping and migration module. (5) Upon receiving the migration trigger signal, the redundancy pool management module reads the fault map library, obtains the redundant storage block pool based on all logical blocks that have been marked as migration or degradation, and selects an unused redundant storage block as the migration target; the address remapping module dynamically maps the access address from the faulty storage block to the redundant storage block; the migration controller controls the data to migrate from the faulty storage block to the redundant storage block, marks the redundant storage block as a migration, writes it into the fault map library, and updates the address mapping relationship.