PWM sine wave : A question

HI Elec. ss,

               A quick look in to the code, gave me an impression that, your TimerB [initially(1sec ) and later changes to (200us)] period is too short and you will loop in Main code and TimerB isr. try this, if u can c the wave form on the scope.

__interrupt void TB0_ISR(void)
{
  //TBCCR0 = 7-1;                           // 32 Khz clock, 999.8 ms delay, Clock period of TBCCR0

TBCCTL1 &= ~CCIFG;

//STOP timera AND timer b HERE

_DELAY_CYCLES(/*200 MICRO SECONDS DELAY HERE*/);

//  start timer a and b here
  TBCTL = TBSSEL_1 + MC_1;
  //TBCTL = 0;                                // Clear Timer_A control registers
  __bic_SR_register_on_exit(LPM0_bits);     // Exit LPMx, interrupts enabled
}

Regards,

Sri.

Hi Jens,

Thank you for the response.

I tried to fix the problem step by step as you have pointed out. I have made the following changes and everything worked fine. I am still testing it thoroughly though. I will let you know if there is more problems. Thank you for your suggestions

1. I configured Timer B interrupt using: 

TBCCTL0 = OUTMOD_3;                              // TBCCR1 set/reset
TBCCR0 = 32760-1;                                      // Clock period of TBCCR0 for 1 sec @ 32 kHz clock
TBCCTL0 |= CCIE;                                        // Enable timer B interrupt
TBCTL = TASSEL_1 + MC_1;                    // ACLK, up mode
__enable_interrupt();

2. I want to stop the sine wave once the 1 sec is reached. In the ISR for timer B, I made following changes

#pragma vector = TIMERB0_VECTOR
__interrupt void TB0_ISR(void)
{

TACTL = 0; //Clear Timer A
TBCTL = 0; //Clear Timer B

TACCTL1 &= ~CCIFG;                          //STOP timer A AND timer B 
TBCCTL0 &= ~CCIFG;

__delay_cycles(2000);
__delay_cycles(2000);

TBCTL = TBSSEL_1 + MC_1;      // start timer A and B here
TACTL = TBSSEL_2 + MC_1;
//TACCTL1 |= CCIE;

__bic_SR_register_on_exit(LPM0_bits); // Exit LPMx, interrupts enabled
}

Elec. ss said:

__delay_cycles(2000);

Using busy-waiting delays inside an ISR is a horrible thing. It contradicts the use of low-power-modes. So see this as a quick hack. A 'correct' way would be switchign between two TimerB configurations with a static flag. If the flag is clear, TiemrA is halted and TimerB is set up for a second delay and the flag is set. On next ISR entry, the flag is set and the timerB is configured for the long delay, TimerA is restarted and the flag is cleared.

I still don't see how the code could stop the DMA. Or was the trigger you use for the DMA a TimerA trigger? The comment was talking about DAC12IFG as trigger. (I didn't check the value). If so, keep in mind that you're halting (stretching) the sine wave somewhere in the middle, not at their zero-crossing point, giving the output a DC offset.

Jens,

Thank you for the comments,

I looked up my waveform and indeed, there is a DC when I stop the sine wave for the 200 us.

To fix this, 

I made the changes:

DMACTL0 = DMA0TSEL_7 + DMA1TSEL_7;        // Timer A triggers

Other thing I consider was: I made

TBCCR0 = 32768-1;    // an Integer multiple of 32, on a hunch

However I still see that, when the waveform is paused every 1 sec. it starts at a random point and not at the zero crossing. causing a DC voltage During the interval.

Also, for __delay_cycles() inside ISR, I plan to change it later. I have attached my code for your reference

Please suggest me on the these problems.

When you stop the PWM, you reset the DAC outpu tto 0 from whereever it currently is. This eliminates the DC during the pause. However, when the timer is enabled agian, teh DMA continues stuffing the DAC from where it was, making a jump back to the point where it was when you stopped the timer and manually set the DAC output to 0.

The correct way would be that instead of stopping the timer, you switch the DMA from repeated to signle cycle mode, so it finishes the current wave and then ends. When the signal is to be re-enabled, you have to restart the DMA in repeated block mode so the output begins anew from zero crossing point. This won't stop the wave output mid-wave when the 'off' time begins but lets it continue to the end (avoiding high-frequency transients). And starts the wave smoothly again when the 'off' time is over.