IGBT module selection can have a direct effect on inverter efficiency because the balance between conduction and switching losses changes with operating conditions. A module designed to reduce one type of loss may incur a compromise in the other, so a single device optimized for general or nominal operation may not be the best choice for every position in an inverter. In a three-level neutral-point-clamped (NPC) topology, matching IGBT variants to the roles they perform can provide a more tailored approach to system efficiency.
Why IGBT losses require an application-specific approach
IGBT module losses are broadly divided into two categories: conduction losses and switching losses. Conduction losses occur while a device is carrying current; switching losses arise as it transitions between conducting and blocking states. The balance between them depends on the operating conditions, including switching frequency.
For a given semiconductor process and voltage, reducing conduction losses generally involves a trade-off with switching losses, and reducing switching losses can involve the opposite trade-off. This means there is no single module that is optimized for every application and every point in an operating window. Selecting a device based only on nominal conditions can limit opportunities to optimize an inverter for its actual duty.
Instead, engineers can consider both the system-level application and the function of individual devices within a converter topology. Where the topology gives different IGBTs distinct loss profiles, using different variants in different positions may offer a better overall balance than fitting the same module throughout.
IGBT module selection in a three-level NPC inverter
In a three-level NPC phase leg, the IGBTs T1 through T4 account for most of the losses among the semiconductor devices described. The NPC diodes D5 and D6 contribute the next largest share, while losses in D1 through D4 are typically a relatively small part of the total.
The IGBTs do not all have the same loss profile. Switching losses dominate the overall losses of T1 and T4, while conduction losses dominate those of T2 and T3. That distinction creates an opportunity to assign modules according to the primary loss mechanism at each position, rather than using one IGBT variant for all four devices.
Matching module variants to switching frequency
The example compares two 3.3 kV IGBT variants: the DIM1500ESM33-MF, optimized for high-frequency operation, and the DIM1500ESM33-MS, optimized for conduction performance. In the described arrangement, the MF variant is used for T1 and T4, where switching losses dominate. The MS variant is used for T2 and T3, where conduction losses are the larger concern.
Comparisons of semiconductor losses per phase show how the results change with operating frequency. The conduction-optimized MS variant performs better than the compared competitor modules designed as average switching and conduction performers below approximately 550 Hz. Above approximately 550 Hz, the switching-optimized MF variant performs better than those competitors. These comparisons illustrate why the preferred individual module can depend on the frequency range in which the inverter operates.
Combining MS and MF variants in the phase leg further optimizes performance across the frequency spectrum in the reported comparison. The practical point is not that one configuration will suit every inverter, but that device position and operating frequency should both inform the selection. A topology-specific assessment can reveal efficiency opportunities that a single-module approach may miss.
Design considerations for power electronics teams
For engineers and system integrators evaluating an inverter design, the example suggests a structured selection process:
- Identify the dominant loss mechanism: Determine whether conduction or switching losses are more significant for each IGBT position.
- Account for operating frequency: Compare module performance over the system’s relevant frequency range rather than relying only on a nominal operating point.
- Assess the complete phase leg: Consider how individual device choices affect total semiconductor losses, including the contribution of NPC diodes.
- Evaluate mixed configurations: Where the topology and design allow, compare a position-specific combination of variants against using one variant throughout.
This approach makes module selection part of the broader inverter optimization process. It also highlights the importance of evaluating losses at the level of the actual circuit arrangement: the best choice for one device position may not be the best choice for another.
Conclusion: Optimize the system, not just the module
Conduction and switching losses respond differently to module design and operating frequency. In a three-level NPC inverter, the different loss profiles of T1–T4 provide a basis for using switching-optimized devices in some positions and conduction-optimized devices in others. The reported comparisons show that this type of tailored IGBT module selection can improve overall loss performance across a frequency range compared with relying on one variant for every slot.
For plant and system designers, the practical implication is to evaluate semiconductor choices in context: consider the converter topology, each device’s role, and the intended operating conditions. Optimizing at the individual-slot level, where appropriate, can provide a more informed path to inverter efficiency than selecting a module solely as a general-purpose solution.