Brushless Positioning Control using tm4c123 tiva

You've tossed "KISS" to the curb by starting w/the far more advanced, "Sine PWM."   For a decade & beyond - Trapezoid PWM - has proved quite adequate for most BLDC applications.   And it is far simpler, faster and easier to diagnose & troubleshoot.

May I suggest that you look instead at the "Basic BLDC" file groupings instead of the QS ones.   The QS ones "lock you into a GUI" and this competes w/your understanding & mastery of the (real) BLDC code running on your MCU - not buried w/in the GUI.

I must disagree w/vendor's statement that (only) two PWM Generators are required.  We've always used three - and (both) the QS and Basic BLDC code examples continue that "use of three" PWM Generators!

Think KISS - avoid use of the motor until you can view PWM output by your 3 PWM Generators.   And select "Trapezoid" as it's far faster and simpler.   Again - avoid use of the motor until all six of your PWM outputs (produced by (again) three PWM Generators - prove "alive & well."

[edit] Just noted the depth & complexity of your most recent post.   Three BLDC Motors - each under Sine PWM Control - AND IMU reading (itself quite an adventure) - proved by hundreds of "crash/burn" victims - right here!   Might your project be (just) a bit too extensive?  (I really, really think so - and I'm in this biz!)

We note your thread's titled "Brushless Positioning Control" - yet not one word has (yet) been devoted to "Positioning!"   KISS - torn/tattered - does not smile upon so many demands - and so few, "Known good" components, codings, and interconnects...

Hi Feliphe,

SPWM in qs-bldc is more square wave than sine appearing but does have sine artifacts. Normally Hall code changes are required to drive 120 phase shift.

For experiment Hall codes can be synthesized by a 60Hz timer and mock code generator inside the timer 1A timeout interrupt handler.  With out BLDC generated rotor position Hall codes the motor shaft can be spun by hand syncing it to the 60hz moving flux much like an induction motor.  I mainly wanted to see what 3 phase PWM sine looked like and how a BLDC motor behaves with sine wave forms. Well the BLDC vibrated at 60Hz but could be made to spin after the flux is synchronized by hand while the 3 phase induction motor rather enjoyed the sine wave.

Code below infers you have converted (qs-bldc: main.c, trapmod.c, pwm_ctrl.c) to run on the TM4C123.

    /* Configure TM1 for a 32 bit wide 16ms periodic timer
     * used in generating 60Hz sine wave commutation cycle. */  
    TimerConfigure(TIMER1_BASE, TIMER_CFG_PERIODIC);
    /* Set the Timer1A load value to 16ms or 60Hz. */
    TimerLoadSet(TIMER1_BASE, TIMER_A, 0xCB735);
    TimerIntEnable(TIMER1_BASE, TIMER_TIMA_TIMEOUT);
    TimerEnable(TIMER1_BASE, TIMER_A); 
    IntEnable(INT_TIMER1A);


void
Timer1AIntHandler(void)
{
     /* BLDC commutation sequence used to set
      * Hall to angle in MainWaveformTick*/ 
     unsigned char ucHallSequence[] =
                  {5, 2, 3, 4, 6, 1, 1, 6, 2, 5, 4, 3}; 

    /* Clear the Timer interrupt. */
    TimerIntClear(TIMER1_BASE, TIMER_TIMA_TIMEOUT);

       /* Timer 1A 60Hz interrupt monitor pin */
       GPIOPinWrite(GPIO_PORTC_BASE, GPIO_PIN_7, 0x00);
       GPIOPinWrite(GPIO_PORTC_BASE, GPIO_PIN_7, 0x80);
       GPIOPinWrite(GPIO_PORTC_BASE, GPIO_PIN_7, 0x00);

     if((g_sParameters.ucModulationType == MOD_TYPE_SINE) &&
    		                  (g_ulState & STATE_FLAG_RUN))
     {

        /* Punch the watchdog timer. */
        MainPunchWatchdog();

        /* Retrieve the last saved Hall value */
        g_ucSineHallIndex = g_ucSineHallPrevious;

        /* If Motor is running and NOT startup mode.
	* Get the next Hall code commutate the motor */
        TrapModulate(ucHallSequence[g_ucSineHallIndex]);

        /* Increment/Decrement the startup hall index for next time. */
        if((g_ulState & STATE_FLAG_RUN) == STATE_FLAG_BACKWARD)
        {
            g_ucSineHallIndex--;
            if(g_ucSineHallIndex > 11)
            {
                g_ucSineHallIndex = 11;
            }
        }
        else
        {
            g_ucSineHallIndex++;
            if(g_ucSineHallIndex > 11)
            {
                g_ucSineHallIndex = 0;
            }
        }

        /* Export the Hall code value for the
         * next MainWaveformTick update cycle */
        g_ulHallValue = ucHallSequence[g_ucSineHallIndex];

        /* Save the current and incremented Hall index
         * for the next timer 1A interrupt cycle. */
        g_ucSineHallPrevious = g_ucSineHallIndex;
      }
}

Case anyone was wondering - below shows a simulated PWM sine wave 2 axis (dq) in a 3 phase AC motor, wave appears very sinusoidal. The 2nd-3rd (qs-bldc) captured 20kHz and motor shaft speed roughly 10 RPM with 60Hz Hall code changes, 24vdc 200ma.  

25Khz PWM - rotor roughly 0-30 RPM

20kHz PWM:

Notice modulated flag sine wave center inverts, 12.5kHz PWM @60Hz Hall code changes.  

Hi Marko,
We have migrated software to TM4C1294 and it has similar issues remaining to (LM3S8971 NDNR) PWM switching modes. Recommend the time required to migrate LM3S code plus debug time it might benefit you to try Piccolo F28027F coupled to a DRV8305 and Instaspin motorware control suite. The F280x ePWM peripheral controller can produce many very complex wave forms, take a look.

If you must have the LM3S Gerber file perhaps can find available somewhere archived.

Marko you do know that MCU is NDNR no longer manufactured by TI. There was and still is an on going issue with PWM generator comparator B both MCU's. The BLDC software has recently been discovered was jamming comparator B in slow decay mode and basically Cloaks a half witted dead band pulse pwmB when it is actually disabled.

A recent WA patch (Not posted) for TM4C1294 seems to (finally) allow 1/2 bridge dead band to remain enabled during commutation. However the pulse width CMPB refuse to drop below the initial setting or follow pwmA pulse width during synchronous updates in FOC motor startup, without patch.  The good thing is the pulse width pwmB when set lower than 95% initially, produces low and high side PWM is (very efficient) than without a patch and forcing CMPB near 100% pulse width as the code originally did.

Testing this new patch setting 0.2us minimum pulse width seems to build a more (dynamic) slow decay mode for motors demanding large startup currents, mostly drives high side in very few PWM pulses.  FYI the FOC sensorless commutation transitions from an open to closed loop mode and forces slow decay during open loop. Hard to know about Halls mode and starting in fast decay though suspect there are similar issues in CMPB being jammed if ever starting a motor in slow decay, without patching. Suspect all that was due to the odd behavior of CMPB and the synchronous update patch now gives very even symmetry in the bridge PWM motor currents. That insane behavior of CMPB pulse width being static is still not proper and that is why I suggested Piccolo + DRV8305.

Highly advise not to get tangled in this can of smelly worms though much is far easier to debug using the TM4C1294 built in ICDI.

/cfs-file/__key/communityserver-discussions-components-files/908/Stellaris-LM3S8971-Gerber-RDK-TQFP_2D00_100.zip