Hi Dennis - thanks for the reply.  I've clarified a few things below as I
wasn't too clear in my first posting -

>
>Hi Lewis
>
 the batteries are in series, and are currently being replaced
>(Old one in while a new one in the other location is being inserted), then
>the old cell should have a high enough impedance to not provide enough
>current to run the circuit. When the older battery is removed, the power
>will go and the PIC will enter a reset state when the new battery is
inserted.

The batteries are in series in the holder so that you get a full 6V (6.3
open circuit) on the terminals of the holder.  The sequence that I have
some concerns about is as follows;
1) The battery holder is empty - no batteries are installed.
2) User slips in the first coin cell - the holder now has only one battery,
the voltage on the terminals of the holder (and thus the micro) is 3.0V.
3) The user installs the second battery - the holder now has two batteries
in series and has 6 volts on the terminals of the holder - and thus the micro.

It's the period of time between 2) and 3) that I am wondering about - i.e.
am I going to cause "funny business" with the micro and then when the
second cell is installed, the micro has gone "south" on me or something
like that.

The code is being written so that it sets up everything and then goes to
sleep when powered up.  When it awakes (via the change on port B) it does
it's main task as long as the button is held down by the operator - then
when the button is released - it goes to sleep again.



>Things to look out for:-
>
>1/      CMOS latch up, ensure that the PIC is not attempting to drive high
>capacitive or inductive loads during the power up phase

No problem here - in fact, all other "circuits" are being powered by the
port pins on the PIC and therefore won't be fired up until the micro does
it in the main task code.



>2/      Check the operating range of the PIC, some will operate down to 2.5V

I want to use the '620 series but the F84 is also on the plate for usage as
well.  From the spec sheets it's 4-6V for the Vdd on the F84- I'm running
them at 4MHz.  The '620 looks better in that it specs a Vdd down to 3V -
this would quell any concerns about powering up the micro with one lithium
cell for a brief time before the second one was loaded into the holder...

>3/      On power up (Check the status bits) DON'T trust the reset circuits
>to fully enable all the on board stuff. i.e. Do your own resetting!

Not sure what you mean by this.  I'm new to PICs - is there a bit in the
status register to check for a power-up condition?  *grabbing for the data
book as I type this :)


>4/      Don't trust the on board RAM, i.e. The RAMS contents may be corrupted
>

No problem here - there's only a handful of bytes and they are all loaded
with counter values etc. each time the routines use them...


>The last two are very important, if the PIC loses power, then the internal
>reset (Not like some other micros) only resets the internal stuff and
>attempts to set the oscillator is running. The RAM may have had data written
>any where, so be ready for this.

Thanks for the tips - will keep this in mind.

>On the oscillator, if you are having problems choosing the correct C and R
>(For Xtal) then look at using an RC, these are cheep, start faster and if
>designed correctly use less power for the same frequency (Up to 500KHz),
>note that this excludes the 32KHz stuff

I want to bit-bang a port bit for a serial data stream coming out of the
micro - no receiver - just transmit.  Was thinking of the ceramic
oscillator approach here - cheaper than xtals but more accurate than RC -
will be fixed at 4800 baud...



>
>For a last piece of advice (Standard thing that I give to all people using
>battery powered stuff) Always ensure that the internal stuff and the TRIS
>registers are constantly re enabled.

Duly noted - thanks for assistance!