Bill,

I've successfully made it work.  The devices can be counted on to
provide about +/- 2.5% accuracy over voltage, temperature and process.
If you get your bit timing to down the cycle, then it will be baud rate
independent if you've done it right.  I've played with rates up to MIDI
speeds (31.25 kHz, if I remember correctly), and never had a problem.
Double check your loop timing, or whatever mechanism you're using for
accurate bit times, based on an assumed 4 MHz and you'll be fine.

If you want to play with OSCCAL, just remember that it adjusts the
oscillator by approximately 2.5 Ns increments (1%).

Ciao,

Chuck Mauro
RAPID Engineering
KLA-TENCOR

> -----Original Message-----
> From: Michael Hagberg [SMTP:mhagberg@SIGNALHILL.NET]
> Sent: Friday, April 10, 1998 10:34 PM
> To:   PICLIST@MITVMA.MIT.EDU
> Subject:      Re: 12c5xx internal ocs
>
> depends on the baud rate, timing errors at 1200 might never be noticed
> but
> at 9600 could be critical.
>
> michael
>
>
> -----Original Message-----
> From: Bill Kennedy <hue@FRONTIERNET.NET>
> To: PICLIST@MITVMA.MIT.EDU <PICLIST@MITVMA.MIT.EDU>
> Date: Saturday, April 11, 1998 12:14 AM
> Subject: 12c5xx internal ocs
>
>
> >Fellow Pic'rs,
> >    Has anyone messed with the OSCCAL register in that, is it
> possible to
> >set the operating frequency to something known with some kind of
> precision?
> >The data book says 4MHz nominal, but with the calibration register,
> how
> >close to nominal can you get?  Can this be used for receiving rs-232,
> or
> >might a crystal be more proper?
> >
> >Thanks for any input.
> >