'(Fwd) Re: Read Modify Write commands'
|John Payson is apparently involved in a heated battle with his own
mailreader; he's asked me to send the following to the list at
It's a little hard to follow... It may help to know that any text
that's either UNMARKED with ">" characters or with an EVEN number of
">" characters is from John; text preceded by an ODD number of ">"
characters is mine.
------- Forwarded Message Follows -------
Subject: Re: Read Modify Write commands
Date: Mon, 20 May 1996 22:56:39 -0500 (CDT)
From: "John Payson" <mcs.com> supercat
[note: my previous message was accidentally just sent to Andy Warren;
I'm incuding it, his reply, and my reply to that]
[mumble mumble--why can't the PicList have consistent Reply field
handling mumble mumble? Preferred format, IMHO, is to have all
messages "From:" and "Reply-To:" the author, "CC:"'ed to the PicList.
Since most mailers allow including or not including CC's in replies,
this would allow both list replies and private replies easily.
> It really IS a good question... A bunch of people have had problems
> with the built-in serial-comm hardware on the 16C74 -- problems
> unrelated to the known "BRGH=1" bug -- and until now I've had no
> clue how to answer them. Hmm...
I don't have any 16C74's handy, but I'd suggest you try writing a
simple program which, e.g., takes every byte received from the host
port while PA0 is set and sends it out the serial port, and every byte
received via the serial port PA1 is set and sends it out the host
port. While PA0 or PA1 are not set no not poll the respective ports.
This software should then let you see whether a byte that arrives at
the serial port between the last poll of the port and the transmission
of a byte via that port will get lost; ditto for the host port. If
so, then I think the following workarounds might help:
 Serial port: whenever writing to the serial port, clear the serial
port timer and poll the input first. If you clear the serial
port timer, then this will guarantee that no character will
finish reception for a "few" [depending upon baud rate] cycles.
Note that this will dwibble the serial port timing by 1/16 bit
but this should not be an issue in most cases since it can only
happen once per byte time anyway.
 Host port: Simply ensure that your protocol is such that the PIC
knows whether the host will be writing or reading on any given
cycle. Real kludge, but if the movwf instruction works the way I
think it does this is probably the best way to get reliable port
------- End of Forwarded Message ------------------
Andrew Warren - ix.netcom.comfastfwd
Fast Forward Engineering, Vista, California
More... (looser matching)
- Last day of these posts
- In 1996
, 1997 only
- New search...