write_eeprom() hangs program when WDT is enabled

[Guest]

Hi All,

I am using PCML V4.095 with a PIC16F690.

Program I am working on is in final stages - passed all tests and debugging and I am in the process of enabling the watchdog for production - so the code before enabling the WDT works fine under all conditions.

I have a hard coded sleep() at the end of the main infinite loop as the chip spends most of its time asleep. I want the WDT to be on while the chip is awake and not when it is asleep - so from manuals and datasheets, easiest way seems to be to use software with setup_wdt().

All good so far - except the whole thing hangs when it hits a write_eeprom() function.

I have debugged this for hours and have managed to narrow it down to this function in a couple of spots - both in the main loop and inside an ISR. Put a setup_wdt(WDT_OFF) before the write_eeprom() and a setup_wdt(WDT_ON) after, and the program behaves.

So I guess, in a way I have fixed it - but my question to those with more experience with the CCS compiler is why??

Is this a bug, or am I configuring something wrong:

[PCM programmer]

1. How do you know that it hangs on that line ? What method are you
using to determine that ?

2. Are you using a hardware debugger (ICD-U40 or ICD2, etc.) ?
If so, are you using the special 16F690-ICD header board ?
Are you single-stepping through the code or using breakpoints ?

3. Ideally, you would post a stripped-down (but compilable) program
that shows the problem, but with the smallest possible amount of code.

[Guest]

Thanks for getting back PCM Programmer,

I am using printf's each step through the code. Not using a hardware debugger.

Sunday here in Oz and She Who Must Be Obeyed has banned me from working.

After sleeping on it, I am sure I must be doing something wrong - it doesn't make sense otherwise - will write some test routines tomorrow to see if I can reproduce the problem with a small chunk of code.

Will get back to you with the results.

Thanks,
Nigel

[big_nige]

Hi PCM Programmer,

Here is the smallest compilable chunk of code that produces the error. Realise that I could OR the setup_wdt commands, but that is not how they will be used in my program as the WDT is turned on and off in the main routine.

#include <16F690.h>
#fuses INTRC_IO,NOWDT,NOPROTECT
#use delay(clock=4000000, RESTART_WDT)
#use rs232(uart1, baud=57600, ERRORS, RESTART_WDT)
#use i2c(i2c1, master, FAST=400000)

void main() {
printf("in main\r\n");
setup_wdt(WDT_2304MS);
setup_wdt(WDT_ON);
delay_ms(1000);

while (TRUE) {
printf("start main loop\r\n");

output_high(PIN_C1);
delay_ms(1500);
output_low(PIN_C1);
delay_ms(1500);
printf("end of flash\r\n");
restart_wdt();
write_eeprom(0x00, 0xAF);
printf("%x written to eeprom\r\n", read_eeprom(0x00));
}
}


If you comment out the WDT setup "AF written to eeprom" gets printed out. If you enable them (as shown), it prints:

in main
start main loop
end of flash

and then starts to repeat, without getting to the other side of the write_eeprom.

Sorry about the 'hangs' in the subject text, it is not actually hanging the program, it is just resetting without executing the write_eeprom.

I am using the Microchip low pin count protoboard - hence the outputs to C1. They are mainly to test that the config for delay_ms is working.


#include <16F690.h>
#fuses INTRC_IO,NOWDT,NOPROTECT
#use delay(clock=4000000, RESTART_WDT)
#use rs232(uart1, baud=57600, ERRORS, RESTART_WDT)
#use i2c(i2c1, master, FAST=400000)

void main() {
printf("in main\r\n");
setup_wdt(WDT_2304MS);
setup_wdt(WDT_ON);
delay_ms(1000);

while (TRUE) {
printf("start main loop\r\n");

output_high(PIN_C1);
delay_ms(1500);
output_low(PIN_C1);
delay_ms(1500);
printf("end of flash\r\n");
restart_wdt();
write_eeprom(0x00, 0xAF);
printf("%x written to eeprom\r\n", read_eeprom(0x00));
}
}

[PCM programmer]

[big_nige]

aaahh - It all becomes clear

Thanks for the advice - much appreciated.

Cheers,
Nigel