Simplifying Power Module Verification Using Compliance Checking

By Wilfried Wessel, Siemens EDA; Simon Liebetegger, University of Applied Sciences, Darmstadt; and Florian Bauer, Siemens EDA

Current simulation and verification methods for power modules are time-consuming. Each domain has specific solutions based on finite elements analysis, computational fluid dynamics and solvers for electric circuits like SPICE. This article investigates if it is possible to simplify the electrical verification using S-Parameter compliance checks. The usage of such checks speeds up the verification process. The performance enhancement can be achieved because time domain circuit simulation is no longer required. The first part of the article describes the problem to solve. Then the method of compliance checking in power modules is described. The article finally concludes with a future statement of work.

A power module has unique electrical characteristics such as high voltages, high current, and high power. For this reason, the simulation of the substrate parasitics is crucial [1]. This digital twin approach this article describes aims to reduce the number of physical prototypes. This directly relates to three of the 17 UN sustainability goals [2].

Fig. 1: Power Module PI2000/Siemens EDA

Power modules are the heart of each power electronic circuit. Over 70% of all electricity is processed by some form of power module. This shows the direct relation to Goal 7 (affordable and clean energy), Goal 9 (industry, innovation, and infrastructure), and Goal 13 (climate action) [2].

The compliance check tackles three out of four engineering challenges of a power module shown in Fig. 2. It can identify issues in the DC performance, such as current distribution or density issues. It also can identify issues in EMC caused by the coupling effect between the power and gate loop [1]. The compliance check is also suitable for predicting of AC performance criteria such as stray inductance.
The novelty of this article is that instead of running time-consuming electromagnetic analysis and SPICE time domain simulation independently, only the finite element methods or method of moments are performed.

For example, a simple change in the component placement affects all domains. The change could cause the current through all devices is no longer distributed equally (DC behavior), and the change could also cause the switching behavior for the different devices in the module will be affected (AC behavior), already these two changes in the electrical domain led to implicit changes in thermal and EMC behavior. The goal is to achieve equal static and switching losses, to ensure equal device temperatures for all components in the package [3]. Optimized electrical behavior also leads to low EMC radiation.

Fig. 2: Power modules, a four-dimensional design challenge.

Today, a high count of prototypes determines the power module design process. Power module compliance checking could drastically reduce this number. On top of this it can help to implement new technology trends for device such as silicon-carbide and gallium-nitride [4].

S-Parameter Compliance Check and Optimization

The work of the former paper “Enable S-Parameters for Power Modules” enabled the usage of an S-Parameter for time domain simulation. However, Tab. 1 shows that the simulation with S-Parameter still takes ten times longer than without any board or substrate parasitics.

Table 1: Simulation summary

For this reason, it is recommended to use an S-Parameter compliance check before going into the functional simulation. This kind of compliance check does not exist for power modules, but the method can be borrowed from high-speed SERDES analysis [5]. For the SERDES interface, the compliance check verifies if a board conforms to a certain standard, such as PCIe 4.0. Instead of full-time domain analysis, everything is modelled with S-Parameter and compared against worst-case conditions to meet the standard requirements. For power modules, these requirements are not available. Based on the performance figures of a power module, typical values can be extracted as lumped values, such as the stray inductance must not exceed 20 nH or the copper resistance from input to output must not exceed 200 µΩ. These values are for a single current path through one IGBT and are decreased by switching multiple IGBTs in parallel. To generate a reference S-Parameter, a typical reference circuit was needed.

Fig. 3: Simple circuit to create reference S-Parameter.

The circuit to generate the reference S-Parameter was developed based on Fig. 4 and with empiric experiments Fig. 16. In our example, R17 and L15 can set up the resistance and the inductance path. Tools such as HyperLynx LineSim can quickly generate the S-Parameter of this structure. This process can also be fully automated using Python.

After adjusting the parameters in the circuit and including all S-Parameters in the same plot, it can be verified if the general parameters were exceeded. In Fig. 4, no other S-Parameter crosses or undershoots the reference S-Parameter (pink). This means without any further simulation it can be guaranteed that the basic parameters are sufficient regarding the absolute values. The values are in the so-called high confidence region.

Fig. 4: Power module S-Parameter compliance check example S11 (red) U1-J1, S22 (green) U3-J1, S33 (blue) U5-U1, and the reference S-Parameter (pink)

The same technique can also be used to verify if the relative values are suitable. A lower and an upper S-Parameter limit is generated for this use case. The example in Fig. 5 shows that the deviation of the resistance value between U1, U3, and U5 is too big. This finding can now be used to optimize the placement in a very early verification phase. A similar approach can be applied to the inductance value.

Fig. 5: Power module S-Parameter compliance check example S11 (red) U1-J1, S22 (green) U3-J1, S33 (blue) U5-U1, and the reference S-Parameter (pink)

The desired workflow in the future aims for full automation. Currently, single parameters are analyzed to see if optimization is possible and to validate the order of changes.

Summary

For a fully automated and holistic optimization approach of power modules, relying on a single source of truth is essential. Lumped parasitics as SPICE subcircuits were used before to perform time domain functional simulation for power modules. The article illustrated that using S-Parameter compliance checks can reduce the design and verification time by 20 minutes per verification cycle. On top S-Parameters increased accuracy in a broad frequency range. This is important for new device technologies such as silicon carbide and gallium nitride. Other advantages are that the simulation result no longer depends on the engineer’s experience. Reference S-Parameters can be generated automatically based on simple lumped parameters. This allows easy comparison between the simulated and the reference S-Parameter.

It must be highlighted that there is no industry standard to specify power modules with S-Parameter currently. For this reason, the reference parameters are generated based on typical requirements for stray inductance and resistance. This makes it easy for the engineer or optimization algorithm to analyze if all simulated parameters are in the high confidence region. This verification method in the first step is borrowed from high-speed SERDES analysis and is called a compliance check.

The main advantage is that the same set of S-Parameter can be used to perform the time domain simulation. This guarantees data integrity. This enables to provide design methods and a solution to withstand the high demand for new efficient power modules in the future.

The article showed that the available software tools, such as Siemens EDA HyperLynx Advanced Solver, support an S-Parameter-based verification of power modules. It must be verified how this method can be applied in a productive environment and if the error detection rate is high enough to improve current design methods. Before entirely relying on an S-Parameter compliance check, a study must be performed to compare simulated and measured results for a given power module technology.

To bring this work back into the context of the 17 sustainability goals [2] the article showed that the verification process of power modules can be improved. The performance and usability improvements lead to higher acceptance in the engineering community. This drives Goal 9: industry, innovation, and infrastructure. The described method will lead to less physical prototyped, more efficient, and more robust power modules. This fulfills Goal 7 and Goal 13.

For even more on the topics introduced in this article, please download the new whitepaper from Siemens DISW, Why are S-Parameters superior for power module optimization?

References

1. Thomas D. Bayer. (2018). Inverter Power Module Parasitics Modeling with Cross-coupling Simplification for Fast Model Extraction and Switching Characteristics Simulation. IEEE.
2. United Nations – General Assembly Economic and Social Council. (2023, June 27). The 17 GOLAS Sustainable Development. https://sdgs.un.org/goals.
3. A. Griesinger, “Wärmemanagement in der Elektronik,” Springer Vieweg, Stuttgart, 2019.
4. S. Chowdhury, Z. Stum, Z. Li, K. Ueno, T.P. Chow, “Comparison of 600V Si, SiC and GaN power devices,” Trans Tech Publications Ltd, Switzerland, Vols. 778-780, pp 971-974, 2014.
5. Infineon, IGC193T120, https://www.infineon.com/dgdl/Infineon-IGC193T120T8RM-DS-v02_01-EN.pdf?fileId=db3a30433784a04001382e2b934d28b4, March 2023.