> I dont know about your application, but I have had many problems with the
> BRGH set. My product must be used on any PC, and works fine on 95% on the PCs
> out there. A customer loaned me a laptop that would not work with my product.
> After doing all the engineering stuff, like analyzing data stream, code, and
> pulling my hair out, I decided to review the April 96 errata on the 74. One
>now-documented-feature was that the user "may see a high number of
>communication
> errors" if the BRGH is set.
>
> So, to make a long story short, I replaced the xtal with a 7.3728mhz,
> changed the
> firmware, and it fixed the problem. Thank you Mchip.
>
> Note that the problem only occured after sending about 7k of continuous data.
>


My application involved sending short packets, not 7K continuously,
maybe a small mismatch in baudrate over a long transmission makes a
difference. Also because the BRGH changes from x64 to x16 might make for
more sensitivity to slightly mismatched baud rates. Would changing sampling
frequency per bit increase susceptability to noise?

Yes I saw the Errata. But not until much later! (such is life)

We ran extensive error-rate tests to prove the system. The nett
result is that the 16C74 SCI works fine in my book. Only problem that
I can recall was getting the initialization sequence right.  And remembering
to toggle CREN!.....

I used 4.9152 Mhz xtal and spbrgh=01,brgh=1.




Ray Gardiner, 104 Macintosh Street, Shepparton, Victoria 3630,  Australia
ray@netspace.net.au