CCS/TMS320F28335: How to configure the ECAP register to realize the function: detect the rising edge and the falling edge when capturing a 50HZ pwm with a 50% duty?

sorry, the code

ECap1Regs.ECCTL1.bit.CAP1POL     = EC_RISING; // rising edge
ECap1Regs.ECCTL1.bit.CAP1POL     = EC_FALLING; // falling edge

is 

ECap1Regs.ECCTL1.bit.CAP1POL     = EC_RISING; // rising edge
ECap1Regs.ECCTL1.bit.CAP2POL     = EC_FALLING; // falling edge

Serenity_pjb,

The eCAP was not intended to be used in this manner, but I think you could get this to work with a reasonable software delay. It looks like you have configured one-shot mode, for this purpose I would suggest continuous mode. One-shot mode is really so that you can capture a fast signal and process the result several cycles later. It sounds like you would like to trip reliably every cycle.

If you would like to continue using the eCAP interrupt I would suggest added a GPIO toggle to the eCAP interrupt and observe this output on an oscilloscope. This should point out if its correctly tripping when you expect. It should not be delayed unless another interrupt has priority.

Are you already using the Trip Zone submodule (TZ) for something else? It sounds like the TZ may work better for this purpose. Please know that you would still need to synchronize the PWM's period, because the trips would be cycle-by-cycle or one-shot. The TZ will be able to force high or force low, not both. You could use the action qualifier to force the signal high at the beginning at the PWM period and then use the TZ to force it low when you receive the zero crossing signal.

Unfortunately the PWM is not setup to be a signal follower as you have described above. 

Regards,
Cody 

thanks for your reply, Cody,

Cody Watkins said:

The eCAP was not intended to be used in this manner

but I have found someone else to use ECap to lock phase with grid. and you said

Cody Watkins said:

One-shot mode is really so that you can capture a fast signal and process the result several cycles later.

you mean when working at  one-shot mode, the ECap captures an edge signal then the dsp needs to several cycles to process something. Is this cycle the system's cycle? If so, even 10 cycles costs only 10/150000000s. but the edge signal comes every 0.01s.

I didn't use the TZ module, for the action qualifier has the control bit to force high or low.

And yesterday, I tried another idea(I suppose one-shot mode can be used in this manner.).

At first, I configured CAP1POL = RISING_EDGE. when interrupt is triggered, configured the CAP1POL to FALLING_EDGE. The next time interrupt is triggered, change back to RISING_EDGE, and this will be cycled.

void SetCap1Mode(void)
{
    ECap1Regs.ECEINT.all = 0x0000;                 // disable all interrupts
    ECap1Regs.ECCLR.all = 0xFFFF;                  // clear all flag

    ECap1Regs.ECCTL1.bit.CAP1POL = EC_RISING;        // rising edge
    ECap1Regs.ECCTL1.bit.CAPLDEN  = EC_DISABLE;   // disable CAP1-CAP4 register loads
    ECap1Regs.ECCTL1.bit.PRESCALE = EC_DIV1;          // input / 1

    ECap1Regs.ECCTL2.bit.CAP_APWM      = EC_CAP_MODE;      // work at cap mode
    ECap1Regs.ECCTL2.bit.CONT_ONESHT = EC_ONESHOT;     // work at one-shot mode
    ECap1Regs.ECCTL2.bit.SYNCO_SEL   = EC_SYNCO_DIS;     // Disable sync out signal
    ECap1Regs.ECCTL2.bit.SYNCI_EN    = EC_DISABLE;              // Disable sync in
    ECap1Regs.ECCTL2.bit.TSCTRSTOP   = EC_FREEZE;            // stop the counter when not used
    ECap1Regs.ECCTL2.bit.STOP_WRAP   = EC_EVENT1;           // stop at event 1
    ECap1Regs.ECCTL2.bit.REARM       = EC_ARM;                         // arm one-shot

    ECap1Regs.ECEINT.bit.CEVT1 = 1;


}

interrupt void ISRCap1(void)
{
    if( ( ECap1Regs.ECFLG.bit.CEVT1 == 1 )&&( ECap1Regs.ECCTL1.bit.CAP1POL == EC_RISING ) )
    {
        capcount1++;
        LED1 = ~LED1;
        EPwm1Regs.AQCSFRC.bit.CSFB = 2;

        ECap1Regs.ECCLR.bit.CEVT1  = 1;            // clear the flag for event 1
        ECap1Regs.ECCTL1.bit.CAP1POL = EC_FALLING;
    }

    if( ( ECap1Regs.ECFLG.bit.CEVT1 == 1 )&&( ECap1Regs.ECCTL1.bit.CAP1POL == EC_FALLING ) )
    {
        capcount2++;
        LED2 = ~LED2;

        EPwm1Regs.AQCSFRC.bit.CSFB = 1;

        ECap1Regs.ECCLR.bit.CEVT1  = 1;            // clear the flag for event 1
        ECap1Regs.ECCTL1.bit.CAP1POL = EC_RISING;
    }

    ECap1Regs.ECCLR.bit.INT    = 1;            // clear the global interrupt flag;
    ECap1Regs.ECCTL2.bit.REARM = EC_ARM;    // acknowledge next edge detect
    PieCtrlRegs.PIEACK.all     = PIEACK_GROUP4;

}

I tested it in the debug mode with a breakpoint at the first line of interrupt function. Then I run step by step, watching the corresponding register bit at the same time. The logic is right. But when the DSP run with no breakpoints, EPwm1B keeps outputing high level.

Regards,

Serenity_pjb

Serenity,

I do think that this will work, but we did not plan on someone using the device in this way so you will not find many hardware implementations which make this easier. You will need to use software intervention as you have done.

Serenity_pjb said:

Cody Watkins

The eCAP was not intended to be used in this manner

but I have found someone else to use ECap to lock phase with grid.

I don't believe one-shot mode is required for this system. But it is fine to use, I don't see any potential issues with this.

Are any other PWM sub-modules that are setup? It is possible that they are influencing the output.

Regards,
Cody 

Serenity_pjb,

the GPIO mux selects different functions, if a PWM is configured then the GPIO's output will be blocked. If the GPIO is configured then the PWM's output is blocked.

Glad you got it solved. Additionally, being that this requires so much software intervention the best way to do this is probably a GPIO, as you have done.

Regards,
Cody