Posted on January 03, 2016 at 15:04The IWDG is a simple reset from somewhere in your program. The problem with this type of reset is what happens if the code that resets the IWDG is in the infinite loop that's crashed the code? The CPU never resets even though it isn't working.
The WWDG sets up a window with a minimum and maximum time between WWDG reset pulses. Clear the WWDG too soon (i.e. inside a loop as above) and it will force a reset. Same if the WWDG pulse is too late, resets the same as if the IWDG timed out.
Both watchdogs are useful. I use the IWDG to make sure the RTC interrupt is running, clearing IWDG at the one second interrupt. The WWDG reset is in a system monitor task that also handles power and temperature failures. The monitor task is the highest priority, so if SysTick (the FreeRTOS timebase) stops or is at the wrong frequency the WWDG won't reset within the window.
A second use of the WWDG is the early warning interrupt, shortly before the WWDG times out. The service routine can save relevant state information in non-volatile SRAM so some clue as to the cause can be used for troubleshooting.
The two watchdogs are cleared independently so I have some assurance one will fail if the other is trapped in a loop.
Jack Peacock
