| ??? 02/21/12 13:46 Read: times |
#186102 - But emi-filter, overvoltage protection etc may not be enough Responding to: ???'s previous message |
Kai Klaas said:
Per said:
You can't make a circuit immune to external events. Only adjust the amount of electromagnetic interference it takes to produce an error or kill the electronics. With todays filtering components you can filter out ESD to nearly any desired level. And you must do that, just because you cannot trust any actively running circuitry like watchdogs and µC. You MUST make the circuit immune to external events, at least to a level where the likelyhood of a failure is acceptable. But using a watchdog to cure an imporperly immunized circuit will straightaway run into disaster!! Kai Klaas But have you seen anyone talk about using a watchdog to cure bad electronics? An important issue here is that a number of regulations enforces a requirement for a clock-independant watchdog solution. In some situations multiple layers - one watchdog solution for the processor and another watchdog solution for output commuication making a link visibly dead if it isn't correctly serviced anymore. EMI filters, trisils etc are nice. But you can still not control everything. If lightning strikes an electrical central directly beside your device, it doesn't matter what filtering you have on inputs/outputs. Suddenly, every single trace on the board will look like an input if device isn't allowed to be 100% contained in a faradays cage. Or the contact wire for the train falls down and shorts, producing many, many, many kAmps in everything that was expected to be ground. ESD is just a very tiny part of the real-world problems. A short magnetic field pulse millions of times our normal uT levels will induce significant voltages where you don't want - and don't expect - them to be. |
