Why does DQ Current control of 3 phase inverter work with Simscape SPS components but not with Simscape Electrical components?

Hi Community,

I am building a current controller for a three phase inverter in the dq frame. I have two models. Both are attached here.

• Model 1 is built with Simscape Specialised Pwer systems library components &
• Model 2 has components from Simscape electrical.

Expected Results - Measured Id and Iq should follow the setpoint/references provided for each respectively. As there is PID control involved there will be some overshoot, rise time etc as expected, but eventually steady state values should converge to the reference.

With all the parameters and controller design same (to the best of my knowledge) I see desired results with Model 1 but not with Model 2. Any ideas why this could be happening?

Model 1

Model 1 results

Model 2

Model 2 results

This result not does not follow the setpoint but also becomes unstable. The PID parameters are the same but could there be a need for different tuning? If so why, since the input output ranges should be similar as all the other parameters are the same.

Due credit for Model 1

"Syed Abdul Rahman Kashif (2025). Grid Tied Inverter with Current Controller (https://www.mathworks.com/matlabcentral/fileexchange/130854-grid-tied-inverter-with-current-controller), MATLAB Central File Exchange. Retrieved January 31, 2025."

Answer 1

[Sabin]

The two models are not the same. In SPS you are using the Universal Bridge while in Simscape you use the Modular Multilevel Converter (Three-Phase). To do a fair comparison you will need to use the Converter (Three-Phase) instead of the MMC.

The two models you try to compare are not similar, The SPS runs at a fundamental sample-time of 1e-6 seconds while the Simscape model runs at 1.5e-4 seconds which is wrong because the switching frequency for the MMC is 10kHz (1e-4 secconds). There may be other issues with the parameterization of the MMC. I have replaced the MMC with a two-level converter on my end, I've configured the threshold voltage in the converter to 0.5 and run it at 1e-6 seconds sample time. It works as expected.

[Ankita]

Hi @Sabin

Thank you for your response. I tried using it with Converter (Three-Phase) as you suggested. Unfortunately, this did not help. My current outputs have now become drastically reduced in magnitude (in the mA range).

Also could I ask you to kindly elaborate a bit on your suggestion? I would like to better understand why a certain control strategy would work with one inverter configuration and not the other. Granted, the internal configurations would be different, but the DC and AC voltage are the same between MMC and Converter in this case, why does that affect the controller so drastically?

[Sabin]

You are correct that both the two-level converter and the MMC are doing similar things, but the modulation will be different. Since I assume your plan is to control the two-level converter it would be more suitable to provide the model that or to rephrase the question as I don’t see the point of comparing two models that are modeling different systems.

[Sabin]

@Ankita

The two models you try to compare are not similar, The SPS runs at a fundamental sample-time of 1e-6 seconds while the Simscape model runs at 1.5e-4 seconds which is wrong because the switching frequency for the MMC is 10kHz (1e-4 secconds). There may be other issues with the parameterization of the MMC. I have replaced the MMC with a two-level converter on my end, I've configured the threshold voltage in the converter to 0.5 and run it at 1e-6 seconds sample time. It works as expected.

[Ankita]

I did adjust the PWM generation strategy differently for the universal bridge as well as the MMC in the originally shared models (image below). Does this cover what you meant by "the modulation will be different" or are there more elements to this that I am missing?

I am working towards a final implementation using Simscape electrical (the blue blocks) instead of SPS library components. The MMC block is what I finally need to be able to control. The idea behind sharing the two comparison models is that with underlying control strategy more or less similar (with different PWM generation) I am unable to get desired results with MMC.

I adjusted the sample time of simscape to 1e-6, but I do still see unstable outputs with MMC.