11 Replies Latest reply on Jun 13, 2013 4:17 AM by sebastien griveau

    Electronic components thermal properties

    sebastien griveau

      Hello everybody,

       

      I am working with the module electronic cooling on a heatsink for several electronic components, such as Cpu, DDR3, MOS, etc..

       

      However, I am having difficulties with the manufacturers for the thermal resistance "junction to board".

       

      On many website, I am able to find the "junction-to-case" and the "junction-to-ambiant" resistances but rarely the other one.

       

      So, my question: is there a website (like matweb for instance) where we can find these values ?

       

      Thank you all in advance.

       

      Best regards, Seb.

        • Re: Electronic components thermal properties
          Jared Conway

          are you trying to make them all 2R components?

            • Re: Electronic components thermal properties
              sebastien griveau

              Hello Jared,

               

              I intended to make them all 2R components but actually I have a FCBGA component that I will now treat differently.

              All my other components are still 2R though.

              For my FCBGA part, there is not casing, so I will put a 0.1 mm empty space from the PCB and use the Contact Resistance with Material and thickness option to simulate the solder balls. At least, I have these values.

              For the others I still cannot find the R(j-b) but I see that Mike gave me some formulas to approximate them.

              Do you think my parameters for FCBGA are correct ?

              Thanks for your answer.

                • Re: Electronic components thermal properties
                  Jared Conway

                  can you send an example link to what you've found so far? seems strange that if you have components that are generally idealized as 2R that th 2R information wouldn't be available. i could only think you're using something that is new or very uncommon.

                   

                  the alternative to 2R is modeling the component as closely like you're planning on doing with the FCBGA. depending on what you're looking to gain in the simulation, that should be significant.

                    • Re: Electronic components thermal properties
                      sebastien griveau

                      Jared,

                       

                      Below is a link to DDR3 memories for instance.

                       

                      https://dl.dropboxusercontent.com/u/54957119/AvenAo/DDR3%20V89C%20x16%20package%20thermal%20report.pdf

                       

                      On my board, I have two Cpus and two DDR3s. One of the Cpu does not have a top casing so I do not treat it as a 2R Components.

                      I am trying today to modelize all my components with casing, Silicium and solder balls to see the difference. I will let you know how it goes.

                      By the way, when I use contact resistance, do my components have to be coincident does it consider the contact until the next surface ?

                       

                      Thanks.

                        • Re: Electronic components thermal properties
                          Jared Conway

                          "By the way, when I use contact resistance, do my components have to be coincident does it consider the contact until the next surface?"

                           

                          on the first part, yes

                           

                          the second part i don't know if i understand

                           

                          one important point about contact resistance is only apply it to one face in the pair

                           

                          regarding using 2R, I think compact modeling using JEDEC is a completely separate standard vs 2R

                           

                          BUT, the benefit is, you have many examples in that document that you can use to come up with a conversion between the 2 standards and have something to compare it against.

                           

                          i think it really goes back to what you want from this analysis

                          if your goal is to udnerstand airflow patterns and heat dissipation, just model basic components and run the flow.

                  • Re: Electronic components thermal properties
                    Mike Pogue

                    Sebastien,

                     

                    Junction-to-ambient (Rj-a) is almost worthless, since it depends so heavily on the conditions of the test and it is so wildly conservative.

                     

                    For a first pass, you can substitute Junction-to-case (Rj-c) for Junction-to-board (Rj-b).

                     

                    If you want more accuracy, you can calculate Rpins and Rsolder directly from R = Integral [dL/(kA(L)]. A(L) means A as a function of L. For constant area, R = L/(kA). Then R(j-b) = R(j-c) + Rpins + Rsolder.

                     

                    This is conservative, because R(j-c) is to the plastic, while the pins have metal paths quite close to the junction. You'll start to see that Rpins can frequently be ignored. For components with large ground pads, R(j-c) is usually defined to the ground pad, so Rpins is obviously 0. Solder thickness depends on the type of solder. You can use a thickness of .002 inches, or you can contact the fabricator; he'll know what the thickness is. The solder resistance will be small compared to R(j-c), but sometimes not negligible.

                      • Re: Electronic components thermal properties
                        sebastien griveau

                        Hello Mike,

                         

                        Thank you for your answer.

                        I will directly start with your last formula. From the Flow database, I notice that the R(j-b) are always at least twice as much as the R(j-c), sometimes 10 times.

                        My DDR3 must not exceed 95°C in a 85°C environment, so I must make sure the proportion to which the heat is going is "realistic" in order to find out if a fan is mandatory or if a good heat sink would do it.

                        Thanks again.

                        Best regards, Seb.

                          • Re: Electronic components thermal properties
                            Mike Pogue

                            Not that I've been doing this forever, but I've never seen a junction to board resistance much more than twice the junction to case resistance. I tend to work on the same types of assemblies over and over though, so there's got to be a lot I haven't seen.

                             

                            I've also thought that there's a problem with the definitions of the different resistances. I use the definitions from a Texas Instruments white paper. But I know for certain that different engineers use different definitions--even in the same company.