TMS320F28P650DK: Motor control tuning on the UMCL project

Hi David,

1. Are you using sensored/sensorless method?

2. If you are using sensorless method, the motor parameters must be accurate for the speed estimator to work, and you can verify this in build level #2.

3. If the motor parameters are not available, you can try to run motor identification.

Thanks,

Jiaxin

Hi Jiaxin,

For the moment I am using sensorless method.

In Level2, my motor spins but not at the good speed (speedRef_Hz=10, speedForce_Hz=0.01 and physical speed=20rpm).

My motor is a 15 pole pairs PMSM.

During the identification procedure in Level 4, it seems that Ls_d_H and Ls_q_H cannot be calculated (I obtained 9.9999e-07, but the physical value is 0.00032).

However RoverL_rps is close to the physical value (Rs_Ohm=0.18):

I have tried to change the USER_M1_R_OVER_L_EXC_FREQ_Hz define value without any success.

Hi David,

David Blanc said:

In Level2, my motor spins but not at the good speed (speedRef_Hz=10, speedForce_Hz=0.01 and physical speed=20rpm).

Could you explain more on this? In the screenshot, it seems the actual speed doesn't match the reference speed. Meanwhile, if the motor pole pairs is 15, the physical speed should be 40rpm given reference frequency 10Hz.

David Blanc said:

During the identification procedure in Level 4, it seems that Ls_d_H and Ls_q_H cannot be calculated (I obtained 9.9999e-07, but the physical value is 0.00032).

How did you get the physical value? "Ls_d_H" and "Ls_q_H" should be the stator inductance on the d,q-axis.

 Thanks,

Jiaxin

My motor keeps spinning physically at 20rpm in level2 and level3, with a speedRef at 10Hz.

The motor manufacturer gives the line to line inductance in the motor datasheet:

Hi David,

David Blanc said:

My motor keeps spinning physically at 20rpm in level2 and level3, with a speedRef at 10Hz.

Yes. But with 15 pole pairs, the synchronous speed should be 40 rpm.

David Blanc said:

The motor manufacturer gives the line to line inductance in the motor datasheet:

The equivalent "Ls_d_H" and "Ls_q_H" are not directly accessible from the motor spec, you need to have the motor's dynamic model in the d,q frame to get them.

Thanks,

Jiaxin

I agree for the speed, that is why I need your help.

Regarding the Ld and Lq values, they are normally close to Ls or L0, it is not possible to have Ls=0.00032H and Ld=Lq=9.9e-07H. What paramter do I have to adjust if the Ld Lq is under estimated please?

Best regards,

David

David Blanc said:

Ls=0.00032H

Typically enter the same value for Ld user_motor1.h as you did for Lq. Often have to tweak (reduce) inductance a few hundred µH rotor to reach top speed without vibrations. The wait times of motor ID process (Ld/Lq) flux calculation may be due to the ADC PPB and data window combinations. We had to tweak the Ld/Lq flux wait times to stop 9.9e-07H, reset default parameter values.

user_motor1.c

pUserParams->LsWaitTime[EST_LS_STATE_ERROR] = 0;
pUserParams->LsWaitTime[EST_LS_STATE_IDLE] = 0;
pUserParams->LsWaitTime[EST_LS_STATE_RAMPUP] = (int_least32_t)(10.0 * USER_M1_ISR_FREQ_Hz); //10.0
pUserParams->LsWaitTime[EST_LS_STATE_COARSE] = (int_least32_t)(25.0 * USER_M1_ISR_FREQ_Hz); //30.0
pUserParams->LsWaitTime[EST_LS_STATE_FINE] = (int_least32_t)(25.0 * USER_M1_ISR_FREQ_Hz); //30.0
pUserParams->LsWaitTime[EST_LS_STATE_DONE] = 0;

defaults:

pUserParams->Ls_d_H = (float32_t)1.0e-6;
pUserParams->Ls_q_H = (float32_t)1.0e-6;
pUserParams->Ls_coarseDelta_H = (float32_t)0.0000001f;
pUserParams->Ls_fineDelta_H = (float32_t)0.00000001f;
pUserParams->Ls_min_H = (float32_t)0.000001f;
pUserParams->Ls_max_H = (float32_t)100.0f;

Hi David,

David Blanc said:

I agree for the speed, that is why I need your help.

If the speed was the same for build level 2 and build level 3, it seems the open loop control in build level 2 doesn't work as expected as the actual speed is not following the given reference. Did you verify that in build level 2, the voltage and current feedback signals are correct before continuing?

In build level 2, did you modify the motor's v/f profile to match the motor you have?

Thanks,

Jiaxin

Hi Jiaxin,

Thanks a lot, now I have more realistic value for Ld Lq after following your suggestions.

So now my motor spins in closed loop of speed (level4), and speed_Hz follows speedRef_Hz, but the physical speed is always half the speedRef. Do you have any idea to solve that problem?

How can I use SFRA tool to tune my controllers please? my project is the Universal Motor Control Lab and it seems to integrate that tool but I don't how to use it.

Hi David,

In build level 2, did you check if the actual angular speed matches the estimated angular speed?

The SFRA functions are not directly included through changing the build levels in the universal motor control lab, but you could refer to other projects which has build level with SFRA and add the functions into your software.

Thanks,

Jiaxin

Thank you Jiaxin for your reply!

I have used datalog.iptr[0]=motorVars_M1.angleGen_rad and datalog.iptr[0]=motorVars_M1.angleEST_rad.

I didn't reach to use the Graph tool to see datalogBuff1 and datalogBuff2 contents, but I extract them to an Excel file and I confirm that angleEST_rad and angleGen_rad are quite the same @ speedRef_Hz=30.0

But the physical speed of the rotor is 60rpm there (that motor has a 15 pole pairs).

I also see a weird thing with the same type of scale factor of 0.5.

I configure USER_M1_PWM_FREQ_kHz to 20.0, but I see on the oscilloscope a communation frequency of 10kHz:

The picture look good.

Hi David,

Let me think about it.

Thanks,

Jiaxin

Hi Jiaxin,

I have an idea. I wonder if my dual problem would be due to a bad CPU clock frequency settings.I use a LAUNCHXL-F28P65X (with a 25MHz oscillator).

Please find below my settings:

Hi David,

You can confirm the PLL and SYSCLK speed via GPIO port after configuring XCLKOUT divisor.

    /* Select XCLKOUT source is SYSCLK */
    SysCtl_selectClockOutSource(SYSCTL_CLOCKOUT_PLLSYS); //SYSCTL_CLOCKOUT_SYSCLK
    /* Set XCLKOUT divider for 25MHz Signal
     * using GPIO_16_XCLKOUT J2-15 */
    SysCtl_setXClk(SYSCTL_XCLKOUT_DIV_4);
    
    // x49C MCU
    // GPIO16->SPIA-SDI for J1/J2 connection
    // XCLOCK_OUT test INTOSC1 J2-pin 15
    GPIO_setMasterCore(16, GPIO_CORE_CPU1);
    GPIO_setPinConfig(GPIO_16_XCLKOUT); //GPIO_16_SPISIMOA
    GPIO_setDirectionMode(16, GPIO_DIR_MODE_OUT);
    GPIO_setPadConfig(16, GPIO_PIN_TYPE_STD);
    GPIO_setQualificationMode(16, GPIO_QUAL_ASYNC);    
    

David,

How did you measure the actual speed in RPM of the motor?

Thanks,

Jiaxin

Hi Jiaxin,

I measure the actual speed in RPM counting the number of rounds the rotor makes in one minute.

Hi David,

I didn't aware of any problem in the motor's speed since the estimated angle and the given angle matches each other. You can double verify this through the frequency of current waveform. Probably there is something wrong with the actual speed RPM measurement.

Thanks,

Jiaxin