>
>I do understand now, i.e. RS232 is usefull to extend the u(s)art interface
>over a long distance, i.e. 1 or 2 or 3 meters, but when our signals travel
>only a few centimeters on a pcb, we do not need rs232 and in fact rs232
>will simply be a hinderance.

Exactly.

I would also argue that async serial is a waste of resources at that level.
You can talk clocked serial over a couple of general I/O pins, 
letting the slower device be the master.
This has many advantages. In async serial, you need to maintain 
precise timing, and you need to process the char you've received 
before the next one arrives.  You can add handshaking pins, but it's 
pretty much guaranteed that you will receive at least one more char 
after you assert the handshake line to get the other guy to shut up.

In clocked serial, the recipient toggles a clock line, and the sender 
puts the data bits out when it sees the clock change.
So, the recipient can go off and do something that takes even several 
seconds, and come back and resume the conversation right where it 
was, even in the middle of a char.  SPI is a hardware based version 
of clocked serial, but some chips are not blessed with those internal 
peripherals, and they have to "bit-bang" it, doing the protocol in 
software as opposed to hardware.

You can bit-bang async serial as well, but you have to keep those 
timings accurate, and that can be pretty difficult without a timer, 
and you may not want to use your timer for that.

Engineering is the art of compromise. :)

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