3 Replies Latest reply on Mar 31, 2016 11:54 AM by Amit Katz

    Power transistor thermal model

    Qingxuan Ma

      Hi there. I am new to flow simulation. I am now doing circuit cooling simulation and having trouble with my power transistor thermal model. In the circuit , I am using a GaN mosfet chip(5mm*5mm*2mm  1.5Watt) and I am adding "typical chip array" materials to the mosfet chip. The mosfet has a temperature of 50 degrees in simulation, while in my lab experiment the mosfet has a temperature of 70 degrees. Maybe the thermal model of my mosfet may not be proper.   Could anyone please tell me your transistor model used for flow simulation? Thanks!

        • Re: Power transistor thermal model
          Amit Katz

          Are you using just a generic material property from the engineering database? That could be a pretty big source of error. Also, are you accounting for thermal resistances in your chip-substrate joint? It's kind of hard to tell where the problem is without knowing your exact lab setup and what your model is.

          There are software packages specifically designed for this type of thermal problem, where you can model your transistors and circuits in detail to get an accurate thermal map, SolidWorks Flow Sim is fairly crude in comparison.

            • Re: Power transistor thermal model
              Qingxuan Ma

              model.jpg

              Thanks Amit! This is my model and the 6 black blocks represents 6 mosfet chips. I just add "Typical chip Array" from materials browser to them and set them each to be a volume heat source of 1.5Watts. Is there anythig wrong with the model?

                 I also find out that in my experiment I was using a 0.1m^3/min DC fan , while in my simulation I was using a 0.43 m^3/min DC fan. I guess this would be one cause for the difference result. But I still want to figure out whether I setup mosfet models correctly. Thanks!

                • Re: Power transistor thermal model
                  Amit Katz

                  Are you modeling the thermal junction resistance between your ICs and the board? I think that will be a significant factor in your cooling.

                   

                  As far as the fan goes, you should try to match your experiment as much as possible, but I think the real key here will be the fan curve, rather than the size of the geometry.