View topic - after timer compare interrupt doesnt return to the main

Search

after timer compare interrupt doesnt return to the main

AVR Freaks Forum Index » AVR (8-bit) Technical Forums » AVR forum

Author

Message

xnonamex

Posted: May 17, 2012 - 11:16 AM

Joined: May 17, 2012
Posts: 4

I am using timer2 compare match interrupt on Atmega32. The problem is that after executing the interrupt it does not return to the main(). But when the next compare match interrupt arises it goes to the interrupt routine again. Any suggestions? reti() does not help this issue.
my code:
initialization of timer2

Code:

SETBIT(TCCR2,CS22); //prescaling to 1024

SETBIT(TCCR2,CS21); //prescaling to 1024

SETBIT(TCCR2,CS20); //prescaling to 1024

OCR2=100;//gives ~15ms loop //100 gives ~10ms

SETBIT(TIMSK,OCIE2); //compare match interrupt

asm("SEI");

and ISR:

Code:

ISR(TIMER2_COMP_vect){

   turnGreenLedOn();

   timerCompflag=1;

   //reti();

}

JohanEkdahl

Posted: May 17, 2012 - 11:20 AM

Joined: Mar 27, 2002
Posts: 18530
Location: Lund, Sweden

Quote:

The problem is that after executing the interrupt it does not return to the main()

How are you determining this?

In order to get good answers here, or to actually solve it yourself, this is what you should do: Create a minimal but complete program that builds and runs and that displays the problem. Post that code here.

As for right now you have not even shown the main() that you claim execution does not return to. This leaves us in the usual guessing game, which not unlikely will be a waste of time.

Your call..

xnonamex

Posted: May 17, 2012 - 11:24 AM

Joined: May 17, 2012
Posts: 4

I determined it:
1) hardware experiment (with LEDs)
2) debuger

My code:

Code:

/*

 * SyncController.c

 *

 * Created: 2012.05.10. 9:49:33

 *  Author: xnonamex

 */ 

#include <avr/io.h>

#include "PWM.h"

#include "LCD.h"

#include "Tachometers.h"

#include "macros.h"

#include "controlBoard.h"

#define F_CPU 10000000UL

#include <util/delay.h>

#include <avr/interrupt.h>

#define READY_STRING "Ready"

#define FAIL "FL"

#define EMPTY " "

unsigned char timerCompflag=0;

ISR(INT0_vect)

{

   leftWheelPulses++;

}

ISR(INT1_vect)

{

   rightWheelPulses++;

}

ISR(TIMER2_COMP_vect){

   turnGreenLedOn();

   timerCompflag=1;

   //reti();

}

int main(void)

{   

   setUpPWM();

   setUpWheels();

   LCD_init();

   initTachometer();

   setUpControlBoard();

   //writeStringToDisplay(READY_STRING);   

   //_delay_ms(1000);

   //lcdClear();   

   SETBIT(TCCR2,CS22); //prescaling to 1024

   SETBIT(TCCR2,CS21); //prescaling to 1024

   SETBIT(TCCR2,CS20); //prescaling to 1024

   OCR2=100;//gives ~15ms loop //100 gives ~10ms

   SETBIT(TIMSK,OCIE2); //compare match interrupt

   TCNT2=0;

   asm("SEI");

    while(1)

    {

      if(timerCompflag==1)

      {   toggleRedLed();

         TCNT2=0;      

         timerCompflag=0;

         Control();         

      }      

    }

}

clawson

Posted: May 17, 2012 - 11:26 AM

Joined: Jul 18, 2005
Posts: 62228
Location: (using avr-gcc in) Finchingfield, Essex, England

You haven't said which AVR or which C compiler so the following are two wild guesses based on no evidence at all:

1) if it's a mega128 then the M103C fuse has not been cleared

20 if it is avr-gcc being used you have been caught by the "catch all" mechanism documented on this page of the manual:

http://www.nongnu.org/avr-libc/user-man ... rupts.html

If it turns out to be either of these I claim my £5 and also 3,000 points of added kudos for knowing how to operate a crystal ball!

EDIT: our posts crossed so at least we now know it is avr-gcc so it could be (2) but then we don't know what's really happening here:

Code:

   setUpPWM();

   setUpWheels();

   LCD_init();

   initTachometer();

   setUpControlBoard();

_________________

xnonamex

Posted: May 17, 2012 - 11:35 AM

Joined: May 17, 2012
Posts: 4

As I understood catch-all means that you don't have an ISR for an activated interrupt and the result is that the program restarts all the time, which is not the case this time.

setUpPWM:

Code:

void setUpPWM(void){

   /*Setup Timer Counter 1 in Fast PWM Mode*/

   SETBIT(TCCR1A,WGM11);

   SETBIT(TCCR1B,WGM12);

   SETBIT(TCCR1B,WGM13);

   /*Set the PWM Mode to Non inverted mode for the output OC1A and OC1B*/

   SETBIT(TCCR1A,COM1A1);

   SETBIT(TCCR1A,COM1B1);

   /*pre-scaling*/

   SETBIT(TCCR1B,CS10);

   /*Set PD5 and PD4 as an output*/

   SETBIT(DDRD,PD4);

   SETBIT(DDRD,PD5);   

   ICR1=255; //f=10kk/250=40k

}

setupwheels:

Code:

void setUpWheels(void){

   //output on pins for directions

   SETBIT(DDRD,PD0);

   SETBIT(DDRD,PD1);

}

LCD_init():

Code:

void LCD_init(){

   DDRB|=0xFF; //bit 7 down to 0 set to output

   _delay_ms(70);

   LCD_nibble_transfer(0x0C);

   _delay_ms(24);

   LCD_nibble_transfer(0x0C);

   _delay_ms(24);

   LCD_nibble_transfer(0x0C);

   _delay_ms(24);

   LCD_nibble_transfer(0x04);

   _delay_ms(24);

   LCD_cmd_write(0x28); // 4 bit data and 2 lines display

   _delay_ms(2);

   LCD_cmd_write(0x08); //set display

   _delay_ms(2);

   LCD_cmd_write(0x01); //clear display

   _delay_ms(2);

   LCD_cmd_write(0x06); //entry mode move cursor to the right

   _delay_ms(2);

   LCD_cmd_write(0x0e); //display on,and add cursor

   _delay_ms(2);

   LCD_cmd_write(0x02); //set cursor to first digit; (addr 0x80)

   _delay_ms(2);

}

initTachometer:

Code:

void initTachometer(){

   SETBIT(MCUCR,ISC11);//Int on rising edge

   SETBIT(MCUCR,ISC10);//Int on rising edge

   SETBIT(MCUCR,ISC01);//Int on rising edge

   SETBIT(MCUCR,ISC00);//Int on rising edge

}

setUpControlBoard:

Code:

void setUpControlBoard(void){

   SETBIT(DDRC,DDC0);//set as output

   SETBIT(DDRC,DDC1);//set as output

   SETBIT(DDRC,DDC2);//set as output

   //rest is input

}

clawson

Posted: May 17, 2012 - 11:40 AM

Joined: Jul 18, 2005
Posts: 62228
Location: (using avr-gcc in) Finchingfield, Essex, England

This is a bit like pulling teeth but can we see

Code:

turnGreenLedOn();

next please?

To be honest I'd just follow Johan's advice. Cut the code down to the minimal test program that still shows the fault and post the entire thing. 90% of the time while doing such a cut down you will encounter a "slaps forehead" moment where you finally see what you did wrong anyway.

EDIT: oh wait a minute I'm an idiot this is FAQ#1. Time to use volatile!

(guess why it's FAQ#1 in my signature!). It's also FAQ#1 in the manual for the compiler:

http://www.nongnu.org/avr-libc/user-manual/FAQ.html

and I also wrote some words on this to explain even further:

Optimization and the importance of volatile in gcc

I think we were mislead by the thread title - it DOES return to main(), it's just main() does not "see" the change (to the flag) which was made in the interrupt.

_________________