James R Albers wrote:
>

>
> I have included other software routines to notify the user of a power
> failure, and other problems, but I am looking for one more thing that I
> might add in order to give me or the user a clue that "Something" has
> happened.
>
> I'm using a 16F628 that has brown out detect.  But I am already using a
> MCP120 that resets the PIC at 4.5 Volts (I've got other TTL logic on the
> board, also).


Hi Jim, I have always been interested in temporary
failure modes in CPUs. The static memory, like
the registers (ram) and of course the program counter
which is one of those registers are of the flip-flop
dual semi type. Static stability relies on one
transistor (or cmos fet) being on and the other off.
So it is stable in each position. But a weird occurence
like power spike, a spike through an IO port, or
even the much-argued "cosmic rays" can cause an upset
making one or more of these register bits change state.

Most common results are obviously changes in ram
contents and program runaway (jumping to unknown
position in code). These can become quite frequent
when the device may be operated in an electrically
noisy environment.

Runaway code is easier to detect but detecting
changes in ram contents is harder. If the chip
has data eeprom internally this is much more
"mechanical" than the ram registers, so maybe you
could use some eeprom stored value to checksum
the registers. Interesting stuff, hope i'm not
raving on. :o)
-Roman

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu