Discrepancy Between HIL-to-HIL and RT Box–STM32G474RE Results in Open-Loop Average Inverter Simulation

RT Box
1

Hello,
I am currently designing an open-loop control system using an average inverter model on an HIL RT Box, where the plant is modeled on the RT Box and the open-loop controller is implemented on an STM32G474RE. The inverter is designed to inject 100 A of current into the grid by regulating the reference voltage Uref. The offline simulation results shown in Fig 1 indicate that the inverter current injected into the grid matches the design target.
In addition, I performed an HIL-to-HIL simulation, where the plant is modeled on RT Box 2 Core 0 and the controller is implemented on RT Box 2 Core 1. The results shown in Fig 2 indicate that the injected inverter current is slightly different from the offline simulation results, but is acceptable. However, when I performed the simulation using the RT Box together with the STM32G474RE, the results became significantly different, as shown in Fig 3.

How could the discrepancy between the HIL-to-HIL simulation and the RT Box–STM32 simulation?

image
image1075×572 128 KB

Fig 1. Offline Simulation

image (1)
image (1)1075×560 80 KB

Fig 2. HIL-to-HIL Simulation

image (2)
image (2)1192×794 57.4 KB

Fig 3. RT Box-STM32 Simulation

Here model that i am using for RT Box-STM32 :

Open_Loop_Average_Model_Inverter.plecs (100.9 KB)

2

Hi Rizal

The sine wave running on the STM32G474 and the grid voltage phase angle (on the RT Box) drift, as the frequency is not exactly the same.

A more robust way would be to use a PLL that locks to the phase-angle of the measured three-phase voltages and recreates a sine wave based on the estimated grid phase-angle. See an example attached:

Open_Loop_Average_Model_Inverter_lc.plecs (112.1 KB)

Hope this helps.

Lino

3

Dear Mr. Lino

Thankyou for your explaining and solution.

Best regards
Rizal