>I guess a lot depends on the accuracy of timekeeping needed.  If the 5-minute
> interval is somewhat flexible, you could simply count watchdog resets, and
> approximate your 5 minutes without adding an extra chip (the RTC module).
It is possible to improve accuracy by calibrating the WDT against the
crystal/resonator - every so often (mins/hours). Set the prescaler to
minimum, set a flag & start a counter, waiting for the chip to be
reset by the wdt. When it resets, check the flag to see if it was a
calibrate operation, if so, look at the counter value to determine the
WDT period. 
Obviously 'flag' should be more than 1 bit, to avoid accidental
triggering by the PIC's random power-up state. If you want a 5 min
interval, you could do a cal cycle every 5 mins, which determines how
many wdt timeouts to wait for the next period to get close to the 5
mins you require. This should be very accurate as long as the temp &
supply is fairly stable over the 5 min period.
Another method would be a 32K crystal on a 4060 cmos counter, which
would give an accurate (I think 0.5 sec) pulse, which could be divided
further by the PIC or another 4020/4040/4060 chip - nice & cheap & low
power. (you could use this to reset the PIC, so you could turn off the
WDT, saving the 15-20uA that it takes - not what the CMOS counter
takes, but it won't be much).
     ____                                                           ____
   _/ L_/  Mike Harrison / White Wing Logic / wwl@netcomuk.co.uk  _/ L_/
 _/ W_/  Hardware & Software design / PCB Design / Consultancy  _/ W_/
/_W_/  Industrial / Computer Peripherals / Hazardous Area      /_W_/