Raffaele Rialdi wrote:

> All you have to do from your PC software is to let for example the DSR
> high.  Then you can use a simple resistor or a 7805 to power your PIC
> and the MAX232.

  Now there's a trap for young players!  A 7805 has a drop-out voltage
of 2 volts, so it only works down to 7V, while its quiescent current is
rated at up to 5 mA, which is probably more than your circuit would
draw!

  What you *actually* do is to take a diode from *each* sourcing line
(RTS, TXD, DTR) to feed Vdd, which you buffer with a capacitor (100-200
µF) and regulate with a 5.1V Zener or shunt regulator across this.  Now,
you have also limited your positive excursion on TXD which you feed to
the PIC as data through a resistor (10 k ohm or so).

  You do not need a series resistor as this will only introduce an
unpredictable (certainly, not predictable from machine to machine)
voltage drop and the RS-232 standard defines current limiting.  Rate
your Zener however on the possibility of as much as 60mA - this is 300mW
at 5V, and of course the smallest usual Zeners are 450mW, so should be
no problem.

  You may similarly derive a -5V supply.  In fact, many mice use a -5V
supply to power the PIC or Z8, and a positive supply to power the LEDs
in the opto-interrupters (in series, of course).  This allows an NPN
transistor to control the data going back on the PC's RXD line.  In
general, a PNP transistor on Vdd can switch the RXD against either a
resistor or current driver on the negative line.

  The other status lines should be grounded (*not* connected to either
of these derived internal supplies which only wastes current) to
prevent cable cross-talk generating spurious signals and possibly
causing Status-change interrupts in the PC's UART.  I realise most mice
do not connect these lines at all which is OK as long as they are not
wired in the cable and you use no extension!
-- 
  Cheers,
        Paul B.