Yet another RS232 problem

sorin · Joined: 14 Mar 2005 Posts: 2

Hi there.

I have the following code for serial interrupt testing. It works fine, but when I copy/paste alot of data the PIC hangs and then is restarted by WDT. I assume it's some kind of RS232 buffer overflow, but I thought ERRORS should take care of this... any thoughts ?

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

I didn't look closely at your code, but my very first thought is to
add the line shown in bold, below:

#include <16F876.h>
#use delay(clock=12000000, restart_wdt)
#fuses WDT, HS, NOPROTECT, NOLVP, NOBROWNOUT

Ttelmah · Guest

If a serial interrupt has occured, there is always a character 'waiting'. So move the line to read the buffer outside the test part of the routine. Effectively you are stopping the routine from handling characters if the 'process data' flag is set. The 'errors' flag, only operates when a character is read. It tells the code to clear the UART error flags if they are set at this point, but you are preventing this from happening...
You can also get rid of the 'restart_wdt', in the RS232 setup. This is only needed if you are sitting 'waiting' in getc, and with the interrupt driven comms this should never happen.
Beware the order you do things. You should reset the buffer counter before resetting the 'process data' flag, otherwise if a character arrives between these two statements, it will be appnded to the already processed buffer.

Best Wishes

sorin · Joined: 14 Mar 2005 Posts: 2

you are both absolutely right, thanks, it's ok now.

the reason I am stopping the buffer from being added to in the serial interrupt while processing is that I am afraid the buffer might get overwritten by new data, and it will not be able to process correctly the old data.