Hello, I have built a thermal simulation circuit for a single-leg H-bridge, using Infineon FF450R12ME7-B11 IGBT module. The simulation result shows that the IGBT junction temperature always stays around 30°C. I have tried modifying parameters and the thermal model, but there is no noticeable improvement. I cannot identify the root cause of this issue and would appreciate your guidance.
The attachments contain my plecs circuit and thermal simulation models.
Thank you!
FF450R12ME7_B11_switch.xml (8.6 KB)
H桥电路热仿真V4.plecs (145.0 KB)
Hi @anjunpeng
There are two separate issues here: one related to the simulation setup and another related to the thermal description.
What you are currently looking at is a transient simulation. Although the junction temperature starts to rise as soon as losses are generated, the thermal dynamics from junction to ambient are much slower than the electrical dynamics. Therefore, it is important to also consider the steady-state temperature. Performing a steady-state analysis, I obtain a maximum junction temperature of approximately 49 °C.
The second issue is the thermal impedance defined for the IGBT. The Foster coefficients are not fitted well and result in an extremely large thermal capacitance when converted to the equivalent Cauer network used for the simulation.
If you click on Fix Coefficients, you can see that PLECS removes one element. Converting the corrected coefficients to Cauer again results in much more reasonable thermal capacitance values.
This already has a significant influence on the transient temperature response:
For the steady-state result, however, the thermal capacitances are not relevant. Once steady state is reached, the thermal capacitors are fully charged, regardless of their unusually large values. Therefore, the maximum steady-state junction temperature remains approximately 49 °C:
There is one additional change that should be made in both thermal descriptions. At 0 A, the forward voltage for the conduction-loss characteristic should be 0 V. Non-zero values at the origin can lead to incorrect results when the data is interpolated or extrapolated. PLECS documentation similarly recommends defining zero loss values at the origin to avoid problematic extrapolation.
Please find the two modified thermal descriptions attached:
FF450R12ME7_B11_diode_RC.xml (7.0 KB)
FF450R12ME7_B11_switch_RC.xml (8.7 KB)
Thank you for your explanation. I think I roughly know where the problem is now.
