| ??? 12/20/09 13:33 Read: times |
#171794 - Oh yes it can! Responding to: ???'s previous message |
Richard Erlacher said:
Investigating watchdog-based timeouts requires that the watchdog has the means to signal that the last reset occurred due to watchdog timeout, else the information is lost. Some MCU's have that facility to support their built-in watchdog. Indeed. I think most chips that have an internal watchdog have a way to detect when a reset was due to the watchdog. Some 8051s also have the ability to distinguish a "warm" start (ie, reset without losing power) from a "cold" start (ie, power-up or power-cycle). I think there's a Philips app note on the subject... I know of no diagnostic tool that one can buy, to plug into one's application and divine out what's causing failures. One way is to record information in an area of RAM that is not cleared by a reset - that way, information can be retained over a reset that can help in determining the cause of the reset. If non-volatile RAM is available, that information can even be retained over a power-cycle. This does, of course, rely on the cause of the problem not also corrupting the log! |
| Topic | Author | Date |
| Micro D.I.I. Reset question | 01/01/70 00:00 | |
| if the wind blows from the west | 01/01/70 00:00 | |
| Monostable multivibrator or watchdog | 01/01/70 00:00 | |
| the problem with watchdogs | 01/01/70 00:00 | |
| Watchdogs are useful even if they don't help with debugging | 01/01/70 00:00 | |
| it's not just for debugging ... | 01/01/70 00:00 | |
| Oh yes it can! | 01/01/70 00:00 | |
| It's not as widespread as I'd hoped | 01/01/70 00:00 | |
| the best I have seen | 01/01/70 00:00 | |
| watchdog interrupt | 01/01/70 00:00 | |
That would help, but not in production | 01/01/70 00:00 |