Thermal design propagation calculations for a DCDC converter

Question

So recently I've been doing some design for a DC/DC (this LMR33630) and now I'm doing the thermal design calculations using as a reference this Appnote, also from TI Accurate Thermal Calculations on the Back of a Napkin but there are a couple of things I don't get.

My calculations are as follow (with some rounding to make que maths faster and easier to read):

(1)

    Power_out = V_out*I_out = 12v*2A = 24W
    DCDC_efficiency (n) = 87%
    Power_in =  Power_out / n = 24w/0.87= 27.5w
    Power_dissipated = Power_in - Power_out = 3.5w

To calculate the junction temperature I assume that the whole heat goes to the board (not true, but very simplified)

(2) Tj= Tambient + Power_disipated*Thermal_impedance_to_board = 25C+3,5W*13.6C/W= 72.6C

but if I follow the App note: the DeltaT is outrageous different..(1oz copper as per App note has 125C/W themal impedance , 2sq.in of copper in my board)

(3) 125C/W*3.5w/2sq.in =  218.75 C !!!!

This is assuming that everything goes to the board, which don't because junction-to-case and case-to-ambient must be taken in to account.

To add, the board has been prototyped and with a DC load set to 12v@2A the temperature of the PCB is arround 55C

So my two questions:

• How should I interpret the data of 72.6C vs the 218.75C (I'm doing something wrong for sure).From the prototype I'm inclined to think that my first temp calculations in (2) is correct but I want to understand what I'm doing wrong with the Appnote.
• To make more precise my calculations, how I distribute the Power dissipated between the Temp to the board and Temp to the ambient?

Sorry for the long post! and If you have any other Appnote related to thermal design, it will be awesome to read it!

Answer 1

I believe the main point is that you consider all the power lost due to efficiency figure as dissipates by the IC.

This is, luckily, not true.

The inductor is for shure taking its, not negligible, part.

The example on the napkin application note you linked is somehow misleading, they do exactly what you have done, but they mostly refer to linear regulators where it's indeed true the power is wholly torched in the IC.

Only their last example deals with a switcher but it's a "3mm×3.8mm buck module with integrated inductor",