Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
02/22/12 12:55
Read: times


 
#186134 - A watchdog should never be needed - but should be there
Responding to: ???'s previous message
I see a well working watchdog as good engineering. A safety feature that should not be needed but is there anyway just in case.

While the hope is that I will never need it, I expect it to work unless the board is toast. When people buy hobby equipment, they may be able to just cycle power if their equipment locks up. But if the equipment is located in a vehicle on another continent, or maybe far from the closest usable road, the cost of a hung unit can be extremely high. If a customer needs to spend hours and use a bandwaggon to visit the installation, I don't want any power cycle to solve their issue. I want them to find a melted antenna cable, or some other major and "interesting" major mishap outside my control.

I don't think I can remember the last time I saw a watchdog reset happen, except for test-bench/factory test setups where the code intentionally don't kick the watchdog, or I have just written some new - and broken - code that doesn't kick. Even better is that I can't remember any device that have hung and not being reset by the watchdog. The closest I have had, is devices that have entered power-save states at unplanned times because of unexpected input signals not being "compatible" with used sw-based filtering logic.

And good/bad practice aside they are sometimes required, or the equpment can't be certified.

List of 44 messages in thread
TopicAuthorDate
Pointless watchdog?            01/01/70 00:00      
   Sounds Like From What You Said...            01/01/70 00:00      
      Didn't want to point fingers.            01/01/70 00:00      
         A?            01/01/70 00:00      
            Don't forget AVR            01/01/70 00:00      
               One more argument ...            01/01/70 00:00      
                  Independently testable            01/01/70 00:00      
                     Slow but critical.            01/01/70 00:00      
                        Yes            01/01/70 00:00      
                           ESD            01/01/70 00:00      
                               You can't            01/01/70 00:00      
                                 You can...            01/01/70 00:00      
                                    But emi-filter, overvoltage protection etc may not be enough            01/01/70 00:00      
                                       ESD ...            01/01/70 00:00      
                                          It's not just ESD, it's "the works"            01/01/70 00:00      
                                             I do not trust the watchdog...            01/01/70 00:00      
                                                The simpler, the more trustworthy.            01/01/70 00:00      
                                                   Less complex does not mean less probable to fail!            01/01/70 00:00      
                                                   WDT and debug breadcrumbs            01/01/70 00:00      
                                                Lightning strikes            01/01/70 00:00      
                                                   A watchdog should never be needed - but should be there            01/01/70 00:00      
                                                   A watchdog isn't a fuse...            01/01/70 00:00      
                                                      correction            01/01/70 00:00      
                                                      I just can not agree with that reasoning            01/01/70 00:00      
                                                         Oh please, don't...            01/01/70 00:00      
                                                            so?            01/01/70 00:00      
                                                Why you need a watchdog            01/01/70 00:00      
                                                   And even if you had perfect software and hardware ...            01/01/70 00:00      
                                                      Reset versus power cycle            01/01/70 00:00      
                                                         Mars Pathfinder            01/01/70 00:00      
                                                         many modern chips            01/01/70 00:00      
                                                            The problem with bidirectional reset pins.            01/01/70 00:00      
                                                               The uC in question            01/01/70 00:00      
                                                            Latchup            01/01/70 00:00      
                                                               ONLY clocked circuits?            01/01/70 00:00      
                                                                  That kitchen exhaust fan ...            01/01/70 00:00      
                                                            link...            01/01/70 00:00      
                                                               Sales talk            01/01/70 00:00      
                                                   certified applications            01/01/70 00:00      
                  Using external watchdog            01/01/70 00:00      
                     It's a feature, actually.            01/01/70 00:00      
               Wet            01/01/70 00:00      
   Featuritis...            01/01/70 00:00      
   Looking at it from another perspective            01/01/70 00:00      

Back to Subject List