| ??? 05/02/02 15:52 Read: times |
#22418 - RE: preventive actions fow sw crash |
As far as I am concerned, a badly designed 89Cxx series hardware have great tendency to be SW crashed when working under electricaly dirty environments.
With proper hardware design, all software crashes will be just that. For a well designed 89cxx equipment that must operate in a noisy environment the following must be done 1) Use a supervisor, not a cap for reset. 2) Install .1uF ceramic and 4.7uF tantalum across the processor WITH SHORT LEADS. Do the same for all other chips, there the tantalum may be omitted for chips not driving more than CMOS loads. 3) Put the circuit in a steel can connected to a clean ground (big enough to dissapate the heat) Have contactors etc. outside that can. 4) Install ferrite beads where wires enter or exit the can. 5) A must, use a 4 layer board with solid ground plane. 6) Use only differential inputs and outputs. 7) Use 5V devices, not 3.3 as far as software goes a) do not assume anything, e.g. if a routine does a table lookup and is called with a value outside the range of the lookup table, it must error in a predicatable way e.g. return a default value, return status to force corrective action or whatever. b) make sure per above that the software itself does not fail. c) This is not needed, but for Murphys sake, use a watchdog. Do it right, then Have fun, Erik |
| Topic | Author | Date |
| preventive actions fow sw crash | 01/01/70 00:00 | |
| RE: preventive actions fow sw crash | 01/01/70 00:00 | |
| RE: preventive actions fow sw crash | 01/01/70 00:00 | |
RE: preventive actions fow sw crash | 01/01/70 00:00 |



