SPRUHM8K December 2013 – May 2024 TMS320F28374D , TMS320F28375D , TMS320F28376D , TMS320F28377D , TMS320F28377D-EP , TMS320F28377D-Q1 , TMS320F28378D , TMS320F28379D , TMS320F28379D-Q1
When ECC is programmed and enabled, Flash single-bit errors are corrected automatically by ECC logic before giving data to the CPU, but the errors are not corrected in Flash memory. Flash memory still contains wrong data until another erase/program operation happens to correct the Flash contents. Irrespective of whether the error interrupt is enabled or disabled, single-bit errors are always corrected before giving data to the CPU. When the interrupt is disabled, users can check the single-bit error counter register for any single-bit error occurrences. The error counter stops incrementing once the value is equal to the threshold+1. It is always suggested to set the threshold register to a non-zero value so that the error counter can increment. It is up to the user to decide the threshold value at which to reprogram the Flash with the correct data. A typical threshold setting to avoid triggering on transient errors that are corrected, but identify persistent faults is 10. When ECC is programmed and enabled, Flash uncorrectable errors end up triggering an NMI to the respective CPU. Please refer to Section 3.11 for more details on Flash error correction and error catching mechanisms.