AVR-Chat

On Fri, Jun 03, 2005 at 08:14:06AM -0000, ttse7 wrote:
> Hi all,
> 
> First of all, I really double check that crystal 32.768K is plugged
> between PC7 and PC6.  

What tools do you have to verify that crystal is operating?
Oscilliscopes are very handy in these situations.

> Actually, I put a variable inside the overflow2 function, every time
> overflow occurs, the counter increases by 1.  Then the main function
> keeps sending out the content of that counter.  But I find the counter
> did not increase at all.

I don't hear you confiming the counter variable was declared "volatile".
What you describe is a classic symptom of what volatile is used for.

> If the cross out the statement ASSR|=BV(AS2), every thing is OK.

The BV() macro is another whose use is frowned upon. Try (1<<AS2) in its
place. Same meaning only other C programmers would understand it
quicker.

> "Is anyone has any sucessful experience in using Atmega16 together
> with real time clock. Please share with me."

On 64L and 128, but not the 16. Notice I didn't always have a 32 kHz
crystal to play with, so I faked it the best I could.

//
//	Do the dirty work to initialize hardware for use of Timer0
//	as a Real Time Clock. Want a 32.768 kHz crystal on TOSC1/2.
//
//	Timer 0 is 8 bits. Config for SIG_OVERFLOW0 exactly 1 Hz.
//
void
t0_init( void )
{
	TIMSK &= ~((1<<TOIE0)|(1<<OCIE0));	//  clear OCIE0 and TOIE0
#ifdef USE_32khz
	ASSR = (1<<AS0);			//  crystal on TOSC1/2
	TCNT0 = 0;
	TCCR0 = (1<<CS02)|(1<<CS00);		//  clock/128 
#else
	ASSR = 0;				//  Use I/O clock
	TCNT0 = 0;				//  seems like a good idea
	TCCR0 = (1<<CS02)|(1<<CS01)|(1<<CS00);	//  clock/1024 
#endif
	TIFR  = (1<<TOV0);	//  oddly, writing 1 here clears

	//  Apnote "doc1259.pdf" says we should loop here until not busy:
	//  If this loop doesn't resolve quickly the watchdog will bite.
	while( ASSR & ((1<<TCN0UB)|(1<<OCR0UB)|(1<<TCR0UB)) )
		;

	TIMSK |= (1<<TOIE0);	//  set Timer OVerflow 0 enable
}

SIGNAL( SIG_OVERFLOW0 )
{
#ifndef USE_32khz
        //  If not the 32 kHZ clock then must use main CPU XTAL.
        //  Can only divide by 1024 in hardware which still
        //  results in ~61 overflows per second for 16 MHz.
        //  Approximate the seconds clock here.
        static uint8_t subsec;

        if( subsec ) {
                subsec--;
                return;
        }
        subsec = XTAL/1024L/256L;
#endif
        seconds.u32++;
}


Had a board's 32 kHz to fail. Software got stuck on this and the watchdog
circuit kept the board in infinite reset loop:

 while( ASSR & ((1<<TCN0UB)|(1<<OCR0UB)|(1<<TCR0UB)) )
                ;

-- 
David Kelly N4HHE, dkelly@HiWAAY.net
========================================================================
Whom computers would destroy, they must first drive mad.