The MFR_SPECIFIC_08 command is used to identify catastrophic faults which occur first, and store this information to NVM. See the product datasheet for more information. The MFR_SPECIFIC_08 command must be accessed through Write Byte/Read Byte transactions. MFR_SPECIFIC_08 is a shared register. Transactions to this register do not require specific PAGE settings. However, note that channels A and B have independent bit fields within the command.

Whenever a catastrophic fault occurs, the first event detected will trigger the MFR_SPECIFIC_08 command to update according to the tables below. This recording happens independently for channel A and channel B. If the PMBus host issues a STORE_DEFAULT_ALL, this information will be committed to NVM, and may be retrieved at a later time. In order to clear the record for either channel, the PMBus host must write the corresponding bits (CF_CHA for channel A, CF_CHB for channel B) to 000b, and issue STORE_DEFAULT_ALL.

Attempts to write any non-zero value to this command will be treated as invalid data - data will be ignored, the appropriate flags in STATUS_CML, and STATUS_WORD, will be set, and the PMB_ALERT pin will be asserted to notify the host of the invalid transaction.