We want to use TPS3813 for UART communication monitoring. We cannot form our own triggering impulses as impulses belong to an application - i.e. UART communication, which MASTER-SLAVE perform between each other.
The method of monitoring is well known and is as following:
If MASTER doesn't see responses from SLAVE it wants to perform reset of the SLAVE. In order to do that, RX of the SLAVE will be connected to WDI and once MASTER doesn't receive responses from SLAVE's TX it will stop transmitting from it's own TX (this stop is a part of coding in the app), which will cause WDI seeing impulses of UART communication. This will cause timeout on WDT and SLAVE will be reset.
The problem is that regular WDTs are available for maximum 1.6s typ (were 1.12s is min). In our case MASTER is quiet for a bit more than 1.12s (but less 2s), so we need higher than 1.12s timeout. With TPS3813 we seems can configure for that. However it's quite complicated to understand boundaries, especially lower window frame, and how exactly it works as this is not as other "normal" Watchdogs.
We configured WDT=WDR=Vdd (window typ 2.5s)
Boundary violations, lower window frame and other things from datasheet absolutely mind blowing. So basically, is that enough just to trigger WDI anyhow without much care about boundaries?
What we can guarantee is the following: some impulses will always exist during WINDOW FRAME. The question is as simple as this: "If there are multiple impulses exist within WINDOW FRAME would they keep RESET high all the time regardless of either rise or fall edge was the first in the beginning of the window"?