> > Error percentage is constant, whatever the speed. The only way to do
> > serial with any confidence when the clock is liable to drift would be
> > to use bit-banging, timing based on a bit length of the external
> > transmitter. Not as convenient as the PIC's serial module but it can
> > be done fairly easily
>=20
> This can be done with auto baud rate detection in some of the PICs.  I be=
lieve
> PIC24 have it.  In this case, sender will send 1 byte to sync the clocks,=
 while
> receiver is in the auto baud detection mode.  I have a test setup that do=
es this
> with PIC24FJ32GA002 and an internal RC oscillator.  Works fine even with =
main clock
> frequency variation as high as 4x ( +/- PLL).  The auto baud sync can be =
done before
> each packet of information.  In this case, even crappy RC clock is stable=
 over
> short, ~10-100 ms, time.

But do note that to carry out the autobaud process you need to be using LIN=
 style communication - that is how the autobaud detector is designed to wor=
k. You end up starting each transfer with a break from the master device, t=
hen sending 0x55 character, then you transfer your actual message.

So to do the autobaud you need a polled system, with a master device, but c=
an have multiple slave devices. Instead of using RS232 as the physical laye=
r you need something like CAN or RS485 transceivers if you don't use 'prope=
r' LIN ones.
--=20
Scanned by iCritical.

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.