Hi guys,

I have code here which basically does some operands, checks the validity of results and lights up a led sequence and moves onto the next operand in between a delay.

each operand is run x amount of times for integrity.

I am wondering is it feasible to have a sleep routine which puts the atmega328 to sleep after each operand is done, and to wake up and do the next operand etc.. ie: replace the delays with a sleep routine.?

I can use a wdt/interrupt etc to wake it up i assume?

I am interested in doing a power consumption comparison between case a/ and case b/.


thanks for listening.

int main (void)

{
int repeat = 0;

 //SetSystemClockPrescaler(2);

 while(1) // repeat all functions 3 x times
 
 { 
         for(repeat = 0; repeat < 4; repeat++){
         array(); // run array function
        _delay_ms(3000);
        reset_d();
        }

           for(repeat = 0; repeat < 3; repeat++){
        pwm_led();  // run PWM Led function
        _delay_ms(3000);
        reset_d();
        }
        
        for(repeat = 0; repeat < 3; repeat++){
        maths(); //  run Math function
       _delay_ms(3000);
        reset_d();
        }

        for(repeat = 0; repeat < 3; repeat++){
        simpsons(); // run simpson algorithm
        _delay_ms(3000);
        reset_d();
        }

        for(repeat = 0; repeat < 3; repeat++){

        newton_sqrt(20); // run newton rhapson for square root of 20
        _delay_ms(3000);
        reset_d();
        }
        
        

//   sleep_mode();



     

 }
return (1);


}

 

I would check with the datasheet to see the amount of power it takes to wake up.

AND if the sleep/wake period is longer than your desired delay.

If you actually do save power by sleeping through the 3 second delays, that cant be bad.

I do not know off hand, but I bet you could find sleep functions prewritten.

There may even be a low power mode for the 328, but I cannot recall.

The newer ATMEL and PIC micropower uCs have the ability to hibernate for a set time.

For instance, If you want a data logger to take the temperature at 8am every morning for a year, the uC need only be awake for a few mS every day.

Thing is, you still have to be counting the time passing, so I dont really see how you could save power. The going to sleep and waking up may eat any gains.

 

That is true, seems the gains to be had would be minimal, if anything.

I could add up all the delays (15 sceonds) and replace if with 1 sleep/wake up instead for part two.?

makes more sense.

scenario 1 / 5 routines with 15 seconds delay

scenario 2 / 5 routines with a sleep/wake up over 15 seconds.


what do you think?