JAJSNU3 December 2023 AFE782H1 , AFE882H1
PRODUCTION DATA
The AFEx82H1 include a watchdog timer (WDT) that is used to make sure that communication between the system controller and the device is not lost. The WDT checks that the device received a communication from the system controller within a programmable period of time. To enable this feature, set WDT.WDT_EN to 1. The WDT monitors both SPI and UBM communications.
The WDT has two limit fields: WDT.WDT_UP and WDT.WDT_LO. The WDT_UP field sets the upper time limit for the WDT. The WDT_LO field sets the lower time limit. If the WDT_LO is set to a value other than 2’b00, then the WDT acts as a window comparator. If the write occurs too quickly (less than the WDT_LO time), or too slowly (greater than the WDT_UP time), then a WDT error is asserted. When acting as a window comparator, in the event of a WDT error, the WDT resets only when a write to the WDT register occurs. If the WDT_LO is set to 2'b00, then a write to any register resets the WDT time counter. In this mode, the WDT error is asserted when the timer expires.
If enabled, the chip must have any SPI or UBM write to the device within the programmed timeout window. Otherwise, the ALARM pin asserts low, and the ALARM_STATUS.WD_FLT bit is set to 1. The WD_FLT bit is sticky. After a WD_FLT has been asserted, WDT.WDT_EN must be set to 0 to clear the WDT condition. Then the WDT can be re-enabled. The WDT condition is also cleared by issuing a software or hardware reset. After the WDT condition is clear, WD_FLT is cleared by reading the ALARM_STATUS register.
When using multiple AFEx82H1 devices in a daisy-chain configuration, connect the open-drain ALARM pins of all devices together to form a wired-OR network. The watchdog timer can be enabled in any number of the devices in the chain; although, enabling the watchdog timer in one device in the chain is usually sufficient. The wired-OR ALARM pin can be pulled low in response to the simultaneous presence of different trigger conditions in the devices in the daisy-chain. The host processor must read the ALARM_STATUS register of each device to know all the fault conditions present in the chain.
The watchdog timeout period is based on a 1200-Hz clock (1.2288 MHz / 1024).