AC Current Unbalance at High Switching Frequencies in Simscape Electrical VSC Model – Snubber & Solver Choice

Hello,

I'm working with a 2-level Voltage Source Converter (VSC) in Simulink using the Simscape Electrical library. Here's my setup:

• DC Side: Modeled as a DC voltage source with a small resistance in parallel with two capacitors.
• AC Side: Connected to an three phase RL branch and a three-phase programmable voltage source (well balanced).
• Power Electronics: Universal Bridge block using ideal switches.
• Solver: Fixed-step
• Control technique of switches : PWM .

The issue:

• At lower switching frequencies (<2 kHz), the AC-side currents are balanced, and the simulation behaves correctly.
• At higher switching frequencies (>2 kHz), I observe a significant unbalance in the AC-side currents, even though the system setup remains symmetric.
• I’m confident this is a numerical artifact, not a physical modeling issue.
• Changing the snubber parameters (especially increasing the snubber resistance) noticeably improves current balance, but it requires unrealistically large values.

What I want to understand:

1. Why does this unbalance occur at higher switching frequencies when using ideal switches with the fixed-step solver?
2. How do the snubber parameters influence numerical stability and waveform accuracy at high frequencies?
3. Is there a recommended way to choose solver and snubber settings that ensures consistent and stable operation across different frequencies and operating conditions?
4. Is there a robust configuration or best practice that avoids these numerical issues without relying on excessive snubber resistance tuning?

Thank you

Answer 1

[Deepak]

Hi @Yasmine Ihcene,

At higher PWM switching frequencies (>2 kHz), the imbalance in AC-side currents you are observing is likely due to numerical artifacts caused by the fixed-step solver not accurately capturing rapid switching events when using ideal switches. This can lead to slight phase mismatches and artificial asymmetries in otherwise symmetric systems. Snubber resistors help by smoothing these transitions and reducing numerical stiffness, but requiring unrealistically large values often signals that the solver step size isn’t fine enough for the switching speed.

To improve stability and accuracy, try reducing the fixed-step size to at least 50–100 times smaller than the switching period (e.g. 1–2 µs for 5 kHz PWM), using a stiff solver like "ode14x", and enabling event detection or interpolation options if available. For even better robustness, consider replacing ideal switches with IGBT/MOSFET models or using switching-function/average converter models. These changes help maintain numerical stability across a wide frequency range without relying on excessive snubber tuning.

Please find attached the relevant documentation for reference:

www.mathworks.com/matlabcentral/answers/94961-how-does-using-multiple-simpowersystems-ideal-switch-blocks-in-series-with-snubber-impedances-compar

www.mathworks.com/help/simscape/ug/making-optimal-solver-choices-for-physical-simulation.html

www.mathworks.com/help/sps/powersys/ref/idealswitch.html

I hope this helps.