Regarding the issue of motor torque not reaching the previous performance

[fanlikang]

Dear everyone：

    In our test on June 15th, using the configuration shown in the figure, the motor can achieve a torque of 520mN. m at 5500r/min.

       But when we tested again, using the same configuration, the torque could only reach 300mN. m. For the convenience of analysis, the following is the change process of speed and q-axis current during the test process.

      1）The change in q-axis current is normal before the torque reaches 300mN. m., as shown in the following figure

2）Once it reaches 300mN. m., the change in q-axis current immediately enters the set maximum limit, and the speed drops, making it impossible to further increase the torque.

3）We have checked the motor hardware and confirmed that it is normal. We suspect it is a software issue, but we are unable to solve this problem until now. Therefore, we seek your help

[RadekS · Accepted solution]

Hi fanlikang,
The only idea from my point of view is that the configuration isn't the same.
I guess that there might be some differences in limit values. The PI regulator parameters typically don't have such an effect. 
Please be aware that not all parameters might be changed in MCAT during the run state (e.g. motor parameters) - You should be able to see it by the color change in the category name. In another words - different parameters might be in MCU and different ones in MCAT.
If you change motor parameters, you have to generate the header file with static configuration, build project and upload sw into MCU.

Another idea: The output torque depend on input voltage (and power supply current limitation). Are you sure that you measure it on the same DCbus voltage?

Note: We would like to recomend to keep BEMF observer parameters the same as for current loop parameters - I mean F0 and attenuation. However, this should not have inffluence to described issue with limited torque.  

I hope it helps you.
Best regards

RadekS

[fanlikang]

Thank you very much for your patient answer. This issue has been resolved by us. In addition, during debugging, I found that the parameters of the Current loop, namely Ld and Lq, have a great impact on the motor performance. What is the principle? Also, the most important point is, what necessary measures should I take to reduce the power consumption of the motor controller

[RadekS]

Hello fanlikang,
the Ld and Lq parameters are used for calculating the current loop and also for the BEMF observer - therefore you saw the limited performance.

Regarding your second question about power consumption)
I am afraid that I am not sure what you mean by your question. The power consumption of MCU is typically negligible in comparison with the power consumed by the motor.
The highest power loss in an absolute ratio is on the motor - therefore the most effective way is an optimization of the motor construction.

The second place is typically the power stage (GDIC, transistors, DCbus capacitors,...). Here are switching loses and conductance loses. So, you may save some energy here by selecting diferent components, improving PCB layout,... Indirectly here play the role also the selected PWM mode (edge aligned vs double switching mode).

The MCU power consumption strongly depends on frequency, but this influence also the CPU performance. Therefore this isn't typical way. General recomendation is switch off (do not initialize) what you do not need and avoid having any floating digital inputs (initialized digital pins must be is some difined state - input with pull-up/down/output). Floating digital input with voltage arround Vdd/2 may cause additional power consumption - but this is rather critical for low power modes. You should also avoid the injection currents (voltage at pins higher than Vdd or lower than Vss)...

I hope it helps you.

BR
RadekS 

[fanlikang]

Thank you very much for your answer. Our problem has been resolved, and you can refer to my other replies for a detailed solution

[fanlikang]

Thank you very much for your patient answer. Our problem has been resolved. The reason for the problem is that one of the three phase wires on the control board is falsely connected, causing heat generation and thus diverting some of the power.

     However, we still have another issue, as shown in Figure 1, where the motor speed remains constant at 5500r/min. As the torque increases, the actual motor speed displayed on the upper computer and the actual current feedback of the q-axis can both follow the demand value well. However, as shown in Figure 2, when the torque continues to increase, the feedback speed of the motor fluctuates greatly, and the demand current of the q-axis fluctuates greatly. Is this normal? How to solve it?

A

fig.2

dditionally, although the upper computer displays significant fluctuations, the actual dynamometer display does not show significant fluctuations.

fig.1

[RadekS]

Dear fanlikang,

hard to say whether it is normal.
Some fluctuation is natural and the main principle of FOC is using current/speed feedback signals for calculating error corrections. Of course, the error corrections have limits defined by PI regulator's settings and current limits. It is possible that you are already on a limit of regulations. So, you may try to tune the PI regulators for slower response or increase the current limits (with respect to hardware).

The fluctuations in both pictures look quite regular. In the past, we also saw some regulation issues caused by vibrations in the mechanical setup of "high-power" applications. After adding the damping link between the motor and the dynamometer, the problem disappeared.

You have to try what will have the effect on it.

BTW: I can see from screenshots that you use some older version of software (Freescale logo).
It is possible that newer version may contain some improvements...
The latest sw for our motor control boards and kits should be available at https://www.nxp.com/AutoMCDevkits

Best regards

RadekS

[fanlikang]

Thank you very much for your patient answer. Our problem has been resolved. Power loss caused by wire virtual connection due to solder joint detachment

[RadekS]

Hello fanlikang,

I appreciate your notification.

Best regards

RadekS