> Is the internal clock of 16F88 and 16F628A sufficiently accurate for
> "RELIABLE" Asynchronous Serial Communication?

What is the clock's accuracy in the temperature range of the
application?  I think I've heard the number .5% bandied about-
total error in the async serial link should not exceed .5% (+/-)
for reliable communication.  Note that this figure includes both
stability and accuracy of the internal clock as well as any
error in the generated frequency of the serial data.

> I did some tests and seem it works fine for speeds up 19200 bps.

As Wouter said, speed doesn't matter- it's accuracy of the serial
clock you're worried about.  Of course, speed DOES matter, in a
way, because the baud rate generator will do some baud rates
better than others.  For example, the 18F2320 can do 1.2, 2.4, or
9.6 kbaud at .16% accuracy, but try and do 57.6 kbaud, and your
error value bumps up to 3.55%.

> Is it possible that I have problems with temperature or power supply
> variations?

Of course.  The datasheet will give you some idea of what those
variations will be, across your desired temperature range or the
range of acceptable power supply voltages.

Mike H.

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