Good day, colleagues,
My customer is using a TPS3850 window watchdog in a FuSi application (PL-c, ideally PL-d in the future). You can immediately find many explanations of how a window watchdog works and how it differs from a normal watchdog, but in principle there is no information on how to trigger it from the software.
Our current solution tries to trigger the WWDG every now and then. However, we have built a kind of "watchdog" into the software, which then looks at the clock and throws most of the pulses away again because they come too early. Only when the time window at the lower edge has opened does the guard let a toggle pulse through. In principle, however, this bypasses the lower limit that characterizes the WWDG.
Our watchdog has a lower / upper time window of 22 / 55 ms. Our SW is a closed-loop control, one cycle of the main loop takes less than 1 ms in normal operation. A sign of life is now sent for each run and the watchdog module in the software then checks whether the lower time window has already been exceeded. If not, the life sign is discarded, otherwise the WD pin is toggled. However, there are also irregular special functions (e.g. writing parameters to the flash, calculating checksums, etc.), which take considerably longer than a normal cycle and, depending on the amount of data, can generally vary greatly in length.
If half of the functions in the main loop are not executed and a cycle is completed in 0.4ms instead of 0.8ms, the WD-SW module has to throw away twice as many life signs, but the external window watchdog does not notice this. You could count or generally calculate in advance how long a run takes and then toggle the watchdog every 28 runs, for example. But then there are these special functions where it is difficult to estimate when and how long they will run, which then confuse everything.
It would be great to get a few more useful tips on how to trigger a WWDG from the SW in the most sensible way.
Thank you,
Daria
