I've just been working with break detection using a 16C63A, and ran into
some surprises.  I am not certain whether the USART on the 17Cxx's are the
same as the 16Cxx's (Microchip usually reuses designs as much as possible,
though sometimes a later-designed chip will incorporate some improvements).

On the framing error bit, the docs say that it is reset after reception of a
valid character, and reading RCREG.  It seems to mean exactly that.  It
stays set until there is a valid data byte in RCREG.  You don't get any more
interrupts until there is a valid character in RCREG.  Reading RCREG when
the framing error bit is set does not yield the all-zeroes byte you would
expect.

So, you can probably assume you got a break character for at least one
character time whenever you see FERR, but there doesn't seem to be any easy
way to double-check it other than reading the serial input line yourself and
looping to make sure it stays low for a while.

Does anybody know any better break detection method?  Are all the USART's
alike in this respect?  How about the new USARTS with the 9th-bit stuff for
RS485?

> -----Original Message-----
> From: Jerry [SMTP:jam@AMESLAB.GOV]
> Sent: Tuesday, October 26, 1999 12:22 PM
> To:   PICLIST@MITVMA.MIT.EDU
> Subject:      DMX512 Break
>
> I'm attempting a design on a DMX512 receiver.  DMX is a protocol used in
> stage lighting control systems.  It is asynchronous serial, 1 start, 8
> data, and 2 stop, no parity, 250kbaud.  A packet of up to 512 8bit dimmer
> values is preceded by a break character, specified as between 88uS and 1
> second.
>
> This design will be using the 17C43 or 44.  I have a couple thoughts on
> detecting the break.  I have seen a few posts here on the subject, and I
> have some questions.  If I use the framing error bit to detect the break,
> how will I know if I have a real framing error?  I thought about
> programming a 12C5xx to just look for a 88uS low, and then issue an
> 'interrupt' to the 17C43....but this sounds like overkill to me.
>
> I did one of these on an 8051, and I looked for a null character with the
> 9th bit (1st stop bit) a zero.  I can't seem to get that concept to work
> on
> a PIC.
>
> BTW, clock is 32MHZ
>
> Thanks for any info out there...
>
> Jerry Musselman                         jam@ameslab.gov
> Electronics Technician                  office:(515)294-7906
> 258 MD Bldg                                     fax   : (515)294-0568
> Ames Laboratory, ISU                    ESG / Electronics