TPS65381-Q1: Various conditions for entering a safe state

Whether the failure analysis tree can be provided by mail

Hi Layne,

Thank you for your continued support.

Two more issues were encountered in testing the TPS65381, which do not match the datasheet description and require your assistance in reviewing:

1. Simulate entering safe state: 

Set SAFE_LOCK_THR[3:0] is 0xF, other registers remain in default configuration, injection failure (not configuring to exit diagnostics, but to allow the diagnostics to time out)

Test phenomenon: 65381 was found to be in a safe state and could not exit, with a value of 1 for DEV_ERR_CNT.

As described in datasheet 5.5.4.8, It will exit the safe state after a timeout, and should not be stuck in a safe state, please help checking if there are any incorrect configurations in the process?

The datasheet describes that  “Device remains locked in the SAFE state when the DEV_ERR_CNT[3:0] counter reaches SAFE_LOCK_THR[3:0] + 1 value” and “When the DEV_ERR_CNT[3:0] counter reaches SAFE_LOCK_THR[3:0] + 1 value, the device transitions to the RESETstate after 680 ms.”

2. Simulate watchdog failure:

Set SAFE_LOCK_THR[3:0] to 0xF, normally configure the watchdog and feed the watchdog, and wait until the watchdog count value is clear to stop feeding the dogs.

Test phenomenon: 

Through the above method to simulate the reset state, found that the 65381 did not enter the diagnostic state after reset state, but entered the safe state, and needs to be reset again to enter the diagnostic state.

The following two figures are the waveforms of the two abnormal phenomenon tests, please help analyze the cause of the phenomenon, and you can try EVM to reproduce the failure.

Please help analyze this problem, look forward to your reply, thanks~

Hi Layne,

Any updated for this issues? 

Hi Hangjie,

I am reviewing this question and will provide an update Tuesday.

Best regards,

Layne J