C2000WARE-MOTORCONTROL-SDK: How should I control the minimum duty cycle of the PWM output

Hi,

I have found this snippet in HAL_writePWMData

        float32_t V_pu = -pPWMData->Vabc_pu.value[pwmCnt];
        float32_t V_sat_pu = MATH_sat(V_pu, 0.5, -0.5);
        float32_t V_sat_dc_pu = V_sat_pu + 0.5;
        int16_t pwmValue  = (int16_t)(V_sat_dc_pu * period);
        ...
        EPWM_setCounterCompareValue(obj->pwmHandle[pwmCnt],
                                    EPWM_COUNTER_COMPARE_A,
                                    pwmValue);

I can see that the duty cycle has the ability to go all the way from 0% to 100% (V_sat_dc_pu.) Should I close the saturation window here? Although this would reduce the maximum voltage vector and maybe casue some distorsion, this would prevent the duty cycle to go very low. Am I missing something? What other option are there available? It appears that indeed the PI controller is going to saturation.

Hi,

I have found the set point definition for the PI controls. As I am using an induction motor, should I specify a set point different than zero for the Id reference? and how should I calculate this set point to be able to run  is04_signal_chain_test_eabi?

MATH_Vec2 IdqSet_A = {0.0, 0.75};

You may need to set the correct "USER_ADC_FULL_SCALE_CURRENT_A" value in user.h and the sign of "current_sf" in HAL_readADCDataWithOffsets() in hal.h according your own hardware board for running lab04.

static inline void
HAL_readADCDataWithOffsets(HAL_Handle handle,HAL_ADCData_t *pADCData)
{
HAL_Obj *obj = (HAL_Obj *)handle;

float32_t value;

float32_t current_sf = HAL_getCurrentScaleFactor(handle);
float32_t voltage_sf = HAL_getVoltageScaleFactor(handle);

Hello Yamming,

Thank you for your response.

I changed to a PM motor. Now, when I run is04_signal_chain_test_eabi, the motor is activated at Zero Speed with Full Torque. The gate driver no longer detects a desaturation of the IGBT.

I have reviewed what you have suggested. Although, I am using a floating shunt resistor (x3) - Could this be an issue? This is my set-up

The shunt resistor is in-line with the phase with the positive side of the amplifier connected to the motor side. So I assumed that the sign of "current_sf" should be negative so I changed HAL_readADCDataWithOffsets() in hal.h as shown

static inline void
HAL_readADCDataWithOffsets(HAL_Handle handle,HAL_ADCData_t *pADCData)
{
HAL_Obj *obj = (HAL_Obj *)handle;

float32_t value;

float32_t current_sf = - HAL_getCurrentScaleFactor(handle);   // I Placed the negative sign here
float32_t voltage_sf = HAL_getVoltageScaleFactor(handle);

I also double checked USER_ADC_FULL_SCALE_CURRENT_A and calculated it with the following equation

Ifs [A] = 3.3 [V] / ( 0.005 [V/A] * Isolator Gain * Differential Gain) = 100.88 [A]

Where:

   Isolator Gain = 8.2

   Differential Gain =  0.79787

Hi Yanming, thanks for your support.

I finally found out what I was doing wrong. As I had not idenfied the motor at the time I was first trying to run  is04_signal_chain_test_eabi, I was not providing values for R and L

#define USER_MOTOR_Rr_Ohm                 (NULL)
#define USER_MOTOR_Rs_Ohm                 (NULL)
#define USER_MOTOR_Ls_d_H                 (NULL)
#define USER_MOTOR_Ls_q_H                 (NULL) 

This was causing all Kp_Id, Ki_Id, Kp_Iq and  Ki_Iq to resolve to zero in setupControllers() in labs.h (although it is weird beacuse there's a division by zero, I assume it is due to floating point nuisances.) For this reason, the output of the PI controller was zero at all times.

    float32_t RdoverLd_rps = Rs_d_Ohm / Ls_d_H;    // RdoverLd_rps is resolving to zero
    float32_t RqoverLq_rps = Rs_q_Ohm / Ls_q_H;    // RqoverLq_rps is resolving to zero
    float32_t BWc_rps = userParams.BWc_rps;
    float32_t currentCtrlPeriod_sec =
                (float32_t)userParams.numCtrlTicksPerCurrentTick /
                userParams.ctrlFreq_Hz;
    float32_t outMax_V = userParams.Vd_sf * userParams.maxVsMag_V;

    float32_t Kp_Id = Ls_d_H * BWc_rps;
    float32_t Ki_Id = RdoverLd_rps * currentCtrlPeriod_sec;

    float32_t Kp_Iq = Ls_q_H * BWc_rps;
    float32_t Ki_Iq = RqoverLq_rps * currentCtrlPeriod_sec;

So I went ahead and executed is05_motor_id_eabi. This program calculated R and L and now the motor is running smoothly.

Hi Yanming, thanks for your support.

I finally found out what I was doing wrong. As I had not idenfied the motor at the time I was first trying to run  is04_signal_chain_test_eabi, I was not providing values for R and L

#define USER_MOTOR_Rr_Ohm                 (NULL)
#define USER_MOTOR_Rs_Ohm                 (NULL)
#define USER_MOTOR_Ls_d_H                 (NULL)
#define USER_MOTOR_Ls_q_H                 (NULL) 

This was causing all Kp_Id, Ki_Id, Kp_Iq and  Ki_Iq to resolve to zero in setupControllers() in labs.h (although it is weird beacuse there's a division by zero, I assume it is due to floating point nuisances.) For this reason, the output of the PI controller was be zero at all times.

    float32_t RdoverLd_rps = Rs_d_Ohm / Ls_d_H;    // RdoverLd_rps is resolving to zero
    float32_t RqoverLq_rps = Rs_q_Ohm / Ls_q_H;    // RqoverLq_rps is resolving to zero
    float32_t BWc_rps = userParams.BWc_rps;
    float32_t currentCtrlPeriod_sec =
                (float32_t)userParams.numCtrlTicksPerCurrentTick /
                userParams.ctrlFreq_Hz;
    float32_t outMax_V = userParams.Vd_sf * userParams.maxVsMag_V;

    float32_t Kp_Id = Ls_d_H * BWc_rps;
    float32_t Ki_Id = RdoverLd_rps * currentCtrlPeriod_sec;

    float32_t Kp_Iq = Ls_q_H * BWc_rps;
    float32_t Ki_Iq = RqoverLq_rps * currentCtrlPeriod_sec;

So I went ahead and executed is05_motor_id_eabi. This program calculated R and L and now the motor is running smoothly.