Sean Breheny wrote:
> A
> standard 8 bit serial transmission with 1 start and 1 stop bit can
> tolerate about 10% total error.

Not even close.  Do the math.

The 0 time reference is the leading edge of the start bit.  The center of
the last data bit is therefore 8.5 bit times later.  That should be within
1/4 bit for good reliability.  0.25 / 8.5 = 2.9%.  The guaranteed to fail
drift of the last bit is 1/2 bit time, which is 5.9% clock error between
receiver and transmitter.  Then add the fact that most receivers sample the
incoming line at 16x the baud rate.  That means the receiver could be off by
1/16 of a bit time in measuring the 0 reference time for the character.  So
the guaranteed to fail clock error becomes (1/2 - 1/16) / 8.5 = 5.15%.
Again, you want to not exceed about half that in a real system, so a total
of 2.5% error between transmitter and receiver is a good rule of thumb.

Note that some of the newer PICs have internal oscillators that are good to
this level, except maybe at extremes of voltage and temperature.  Keep in
mind however that this eats up the entire error budget at one end, so such
systems rely on the other end being derived from a baud rate crystal.  This
is not something you'd generally want to do in a commercial product, but
it's OK for one off hobby use when you know the PIC will only be
communicating with your PC.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist