I have successfully implemented the thermal coupling effect using the thermal package description. However, we are uncertain about how to incorporate frequency-dependent modeling into the system. Because the LPF frequency varies for each device within the target power module, we believe that each device should be assigned an individual thermal path, which then passes through its respective LPF before being connected to a common case node. Nevertheless, we are unsure how to implement this structure appropriately.
Hi J.S.Hwang,
It is not entirely clear what your question is. In the thermal package description, the off-diagonal parameters (e.g., Z₁₂) define the thermal impedance between two heat sources, which are typically modeled as lumped representations of the semiconductor junctions within the package. This path represents the heat flow between the junctions of two devices.
The diagonal parameters (e.g., Z₁₁) represent the junction-to-case thermal impedance. This corresponds to the primary heat flow path from the junction to the case as the device heats up. This would be the individual thermal path to a common case node.
Hope this helps.
Thanks you for the comments. I’m sorry that my explanation was unclear. So, I describe my problem for utilizing the frequency-domain thermal model more clearly.
The conventional thermal model of power device module based on Foster model is consist of the thermal impedances between the junction to case. In addition, as you mentioned, The thermal impedance that express thermal coupling effect can be applied to predict the junction temperature through thermal package description.
However, only considering the thermal coupling effect is not enough to evaluate the junction temperature due to heat flow of the case node. For this reason, frequency-domain thermal model has been proposed through second heat flow path between the junction to case with LPF and it is can be considered with thermal coupling effect.
My question is that how I could utilize for the frequency-domain thermal coupling model. I hope that the attached figure help you to understand my problem. Thanks your attention for this matter.
Hi,
Thanks for the detailed diagram. It makes your objective very clear.
For your application, I would recommend setting all thermal impedances in the device thermal description to zero and instead building the thermal network explicitly using PLECS thermal components such as the Thermal Chain, Thermal Capacitor, and Thermal Resistor. This approach gives you maximum flexibility in how the thermal behavior is modeled.
One important point to note is that in PLECS, switching losses are represented as Dirac impulse signals. Because of this, you will need to use a periodic impulse averaging block before injecting these losses into the thermal network. Otherwise, the results may be incorrect.
I have attached an example where I modified the “Buck Converter with Thermal Model” demo to compare the built-in PLECS thermal modeling approach with an implementation using explicit thermal components.
Hope this helps!
buck_converter_with_thermal_model.plecs (36.9 KB)
IGBT.xml (877.6 KB)
Thank you for your recommendation.
I wished to apply frequency modeling while using the thermal description function, but it doesn’t seem to have been implemented in PLECS yet. Fortunately, I have previously implemented the thermal bonding effect similarly, so I will implement the function I want as your advice.
I would appreciate your attention to this matter and consideration.
Hi,
It is not entirely clear what feature you are requesting. Could you please provide more details?