Looks at this web forum and see if it helps any. http://www.physicsforums.com/showthread.php?t=330314
My pcb boards and electronic assemblies typically use smaller surface mount chip components and IC's. As you commented about the detail of the components affecting the analysis time, I try to keep my model parts as simple as possible. For chip resistors, capacitors, and inductors I use extruded blocks and have defined them as ceramic material (Alumina AL2O3). For IC's, transistors, etc.. I use a leadframe/footprint made out of copper and a plastic body made out of Epoxy Die encapsulate. I have found that www.matweb.com is a great source for material thermal properties.
I do agree that sometimes it is very difficult to get this information from vendors. I think they are mostly trying to protect their product from competitors. I have had some success by contacting customer service reps and explaining why I need thermal information to evaluate new design that incorporate their products.
It would be hard to determine where you apply the heat sources without knowing your assembly and goals. In my cases I typically will have 1-2 watts in power transistors but very low power thru the bulk of the resistors and other chip components, therefore I usually only apply the heat sources to the transistors.
I am always interested in how others are applying Flow Simulation to electronic models. Please reply or contact me if you have further questions. Hope this helps.
Thanks for the reply.
Regarding the conversion of thermal resistance to thermal conductivity: I know that I need the length parallel to the charge flow and the cross sectional area perpendicular to the charge flow, but I think this depends on where the junction point is located in the IC. For example: If I have an IC with ΘJC = 0.12 K/W and package dimensions of w=35mm l=35mm h=3mm. Is the junction always in the center of the IC? Or basically is this enough information to caluculate the thermal conductivity? Also, should I use ΘJC or ΘJB?
What do you consider to be an expected error rate when you compare to actual test results? If your maximum temperature of an assembly is +/-3 degrees C compared to measured results, is this a sucess?
I would assume that the junction is centered. This is one of the problems that I have run into. You really don't know what exactly is inside of encapsulated device such as an IC.
I have use both ΘJC and ΘJB before but now I usually just use ΘJC. In my models I didn't notice a significant difference with both in the model.
As for the acceptable error rate, I think it is really going to be determined by temperature ranges you are working with along with the goals that you are trying to achieve. I am not trying to be elusive here as i really don't know exactly what you are trying get out of your study. I am assuming that you are putting power in to certain devices and looking to see how much temperature rise and heat distribution/heat sinking properties you have. In my case I am trying to use specific components as heat sources to bring the assembly up to a specific temperature, and then determine how much power was consumed, how the heat sources temperature affected the temperature of other surrounding components, and how these results are affected by different ambient temperatures.
I have only been using Flow for a little over a year now and still have a lot to learn. As I learn more on what details are most important to model as well as interpreting and absorbing all of the data that flow creates, I believe that this program is fairly accurate. There are still a lot of outside sources and variables that have been affecting both my models as well as actual bench products.
Hi Laura and Ron,
I have been working about heatsink and enclosure optimization for better cooling performance at pcb assembly . I am having same problem with you also. I have different related question as well. did you get higher temperature rate than expected according to definition because ı am having higher results around 100 C degere. It can be related definition of components materials? Could you share your experiences about that?
what's your goal for the analysis?
quick and dirty type analyses, put heat on the block, pick a material that is the major composition and go. that should give you good data about what is happening at the system level.
if you want to go deeper, usually people start at least going to a 2R model. this is something you can do with the electronics cooling module.
going deeper than that, you start talking about breaking the components into multiple bodies, applying materials to them, contact resistances....etc.
my experience is that if you pay attention to the general system and use the first approach, you can get really decent results. make sure you have things like radiation enabled if you have high temp differences and also make sure to have a decent mesh.