> I thought to use a serie diode, but the voltage drop bring the PIC power volta
ge
> near the power supply limit of the MCU.

I've  just been through this exercise for a similar circuit to
protect against the batteries being put in backwards.

Initially I used a Zetex ZVP4424A low Vgs, P-channel, enhancement
mode FET. Just tie the Gate to your negative terminal. When the
circuit is connected backwards, the FET is biased off and the
Drain/Source diode is reverse biased. When it is the right way
around, the diode is forward biased and gets shunted by the FET
turning on when the voltage starts to pass the Vgs threshold.

Eventually I changed this to just a schottky diode because all of the
loads could cope with reverse connection (eg. LEDs, dc motor switched
by transistor, etc) and I could put those on the battery side of the
diode. That just left the PIC drawing about 25uA which puts the
voltage drop well down on the curve so the diode is only costing
around < 0.15V. You have to look carefully at your circuit and
calculate what will happen at power up (reset time vs. circuit
capacitance).
The diode is cheaper but it comes with trade-offs. The FET is simple
and works well. It just costs a bit more (which wasn't acceptable for
my case).

One thing that caught me was that the 16C54 and 16C54A (LP) are
spec'd down to 2.5V but the later revisions (B & C) are only spec'd
(and operate)  down to 3V.This is the opposite to what I thought was
the effect of a die shrink (ie. specs stay the same or get better)
but the Microchip man at ESC had a plausable explanation so I'm
better educated now. Watch out for die shrunk parts.

Steve.

======================================================
Steve Baldwin                Electronic Product Design
TLA Microsystems Ltd         Microcontroller Specialists
PO Box 15-680, New Lynn      http://www.tla.co.nz
Auckland, New Zealand        ph  +64 9 820-2221
email: steveb@tla.co.nz      fax +64 9 820-1929
======================================================