??? 02/20/12 13:14 Read: times |
#186061 - One more argument ... Responding to: ???'s previous message |
... to support the use of an external watchdog then.
"The internal watchdog has two dozens of ways to configure it (timeout, type of reset, etc), and its critical failure mode is well documented in the datasheet (once you piece together snippets of information from various locations). It's nice, it's shiny, it's convenient, it works (and fails) exactly as specified in the datasheet." IMHO the "ideal" watchdog should require a clock only for resetting, not for reaching its timeout. If that means that I can't chose the timeout in 4096 steps between 1ms and 16s, then I'd happily trade this small inconvenience for a watchdog that isn't going to lock up with the rest of the chip. |